diff --git a/3rdparty/crc32c_adler/crc32c_adler.c b/3rdparty/crc32c_adler/crc32c_adler.c
index 72542222b5430ee0feafa0544f7c9dd72eb0966b..ce23887e4fd1ad28466fd57f84ca559d8a4b6523 100755
--- a/3rdparty/crc32c_adler/crc32c_adler.c
+++ b/3rdparty/crc32c_adler/crc32c_adler.c
@@ -499,7 +499,7 @@ uint32_t crc32c(uint32_t crc, const void *buf, size_t len)
{
#if defined(__x86_64__) /* @RRL: to compile for ARM */
return crc32c_hw_support() ? crc32c_hw(crc, buf, len) : crc32c_sw(crc, buf, len);
-#elif
+#else
return crc32c_sw(crc, buf, len);
#endif
}
diff --git a/3rdparty/cuttdb/CMakeLists.txt b/3rdparty/cuttdb/CMakeLists.txt
deleted file mode 100755
index 2fb2bf49fcd21fd80fb7fa7f79a133ffcb2775fc..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/CMakeLists.txt
+++ /dev/null
@@ -1,27 +0,0 @@
-cmake_minimum_required(VERSION 3.10)
-
-project(dap_cuttdb C)
-
-add_definitions ("-D_GNU_SOURCE")
-set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -fPIC")
-
-file(GLOB cuttdb_src RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} src/*.c)
-file(GLOB cuttdb_h RELATIVE ${CMAKE_CURRENT_SOURCE_DIR} src/*.h)
-list(APPEND cuttdb_src ${cuttdb_h})
-
-# the server part ain't ported, and thus not built, so are tests.
-list(FILTER cuttdb_src EXCLUDE REGEX "ae_")
-list(FILTER cuttdb_src EXCLUDE REGEX "server.")
-list(FILTER cuttdb_src EXCLUDE REGEX "dump.")
-list(FILTER cuttdb_src EXCLUDE REGEX "builddb.")
-list(FILTER cuttdb_src EXCLUDE REGEX "test_mt.")
-
-if(UNIX)
- list(FILTER cuttdb_src EXCLUDE REGEX "mman.")
-endif()
-
-add_library(${PROJECT_NAME} STATIC ${cuttdb_src})
-set_target_properties(${PROJECT_NAME} PROPERTIES LINKER_LANGUAGE C)
-set_target_properties(${PROJECT_NAME} PROPERTIES COMPILER_LANGUAGE C)
-target_include_directories(${PROJECT_NAME} INTERFACE src)
-target_compile_options(${PROJECT_NAME} PRIVATE -Wno-sign-compare)
diff --git a/3rdparty/cuttdb/src/ae_epoll.c b/3rdparty/cuttdb/src/ae_epoll.c
deleted file mode 100644
index ff8591d86c3d7962c1b01e6ad51b09144ba74f04..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/ae_epoll.c
+++ /dev/null
@@ -1,109 +0,0 @@
-/* Linux epoll(2) based ae.c module
- * Copyright (C) 2009-2010 Salvatore Sanfilippo - antirez@gmail.com
- * Released under the BSD license. See the COPYING file for more info. */
-
-#include <sys/epoll.h>
-#include <errno.h>
-
-typedef struct aeApiState {
- int epfd;
- struct epoll_event events[AE_SETSIZE];
-} aeApiState;
-
-static int aeApiCreate(EventLoop *eventLoop) {
- aeApiState *state = malloc(sizeof(aeApiState));
-
- if (!state) return -1;
- state->epfd = epoll_create(1024); /* 1024 is just an hint for the kernel */
- if (state->epfd == -1) return -1;
- eventLoop->apidata = state;
- return 0;
-}
-
-/*
- be not referenced anywhere
-static void aeApiFree(EventLoop *eventLoop) {
- aeApiState *state = eventLoop->apidata;
-
- close(state->epfd);
- free(state);
-}
-*/
-
-static int aeApiAddEvent(EventLoop *eventLoop, int fd, int mask) {
- aeApiState *state = eventLoop->apidata;
- struct epoll_event ee;
- ee.events = EPOLLONESHOT;
- if (mask & AE_READABLE) ee.events |= EPOLLIN;
- if (mask & AE_WRITABLE) ee.events |= EPOLLOUT;
- ee.data.u64 = 0; /* avoid valgrind warning */
- ee.data.fd = fd;
- if (epoll_ctl(state->epfd, EPOLL_CTL_ADD,fd,&ee) == -1 && errno != EEXIST) {
- fprintf(stderr, "epoll_ctl(%d,%d) failed: %d\n", EPOLL_CTL_ADD,fd,errno);
- return -1;
- }
- return 0;
-}
-
-static int aeApiUpdateEvent(EventLoop *eventLoop, int fd, int mask) {
- aeApiState *state = eventLoop->apidata;
- struct epoll_event ee;
- ee.events = EPOLLONESHOT;
- if (mask & AE_READABLE) ee.events |= EPOLLIN;
- if (mask & AE_WRITABLE) ee.events |= EPOLLOUT;
- ee.data.u64 = 0; /* avoid valgrind warning */
- ee.data.fd = fd;
- if (epoll_ctl(state->epfd, EPOLL_CTL_MOD,fd,&ee) == -1) {
- fprintf(stderr, "epoll_ctl(%d,%d) failed: %d\n", EPOLL_CTL_ADD,fd,errno);
- return -1;
- }
- return 0;
-}
-
-static int aeApiDelEvent(EventLoop *eventLoop, int fd) {
- aeApiState *state = eventLoop->apidata;
- struct epoll_event ee;
-
- ee.events = 0;
- ee.data.u64 = 0; /* avoid valgrind warning */
- ee.data.fd = fd;
- /* Note, Kernel < 2.6.9 requires a non null event pointer even for
- * EPOLL_CTL_DEL. */
- if ( epoll_ctl(state->epfd,EPOLL_CTL_DEL,fd,&ee) == -1
- && errno != ENOENT && errno != EBADF) {
- fprintf(stderr, "epoll_ctl(%d,%d) failed: %d\n", EPOLL_CTL_DEL,fd,errno);
- return -1;
- }
- return 0;
-}
-
-int aeApiPoll(EventLoop *eventLoop, struct timeval *tvp) {
- aeApiState *state = eventLoop->apidata;
- int retval, numevents = 0;
-
- retval = epoll_wait(state->epfd,state->events,AE_SETSIZE,
- tvp ? (tvp->tv_sec*1000 + tvp->tv_usec/1000) : -1);
- if (retval > 0) {
- int j;
-
- numevents = retval;
- for (j = 0; j < numevents; j++) {
- int mask = 0;
- struct epoll_event *e = state->events+j;
-
- if (e->events & EPOLLIN) mask |= AE_READABLE;
- if (e->events & EPOLLOUT) mask |= AE_WRITABLE;
- eventLoop->fired[j] = e->data.fd;
- }
- }
- return numevents;
-}
-
-
-/*
- be not referenced anywhere
-static char *aeApiName(void) {
- return "epoll";
-}
-*/
-
diff --git a/3rdparty/cuttdb/src/ae_kqueue.c b/3rdparty/cuttdb/src/ae_kqueue.c
deleted file mode 100644
index cd80a57be2d19d485f2a2ce3485b42494ba43640..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/ae_kqueue.c
+++ /dev/null
@@ -1,91 +0,0 @@
-/* Kqueue(2)-based ae.c module
- * Copyright (C) 2009 Harish Mallipeddi - harish.mallipeddi@gmail.com
- * Released under the BSD license. See the COPYING file for more info. */
-
-#include <sys/types.h>
-#include <sys/event.h>
-#include <sys/time.h>
-
-typedef struct aeApiState {
- int kqfd;
- struct kevent events[AE_SETSIZE];
-} aeApiState;
-
-static int aeApiCreate(EventLoop *eventLoop) {
- aeApiState *state = malloc(sizeof(aeApiState));
-
- if (!state) return -1;
- state->kqfd = kqueue();
- if (state->kqfd == -1) return -1;
- eventLoop->apidata = state;
-
- return 0;
-}
-
-static void aeApiFree(EventLoop *eventLoop) {
- aeApiState *state = eventLoop->apidata;
-
- close(state->kqfd);
- free(state);
-}
-
-static int aeApiAddEvent(EventLoop *eventLoop, int fd, int mask) {
- aeApiState *state = eventLoop->apidata;
- struct kevent ke;
-
- if (mask & AE_READABLE) {
- EV_SET(&ke, fd, EVFILT_READ, EV_ADD, 0, 0, NULL);
- if (kevent(state->kqfd, &ke, 1, NULL, 0, NULL) == -1) return -1;
- }
- if (mask & AE_WRITABLE) {
- EV_SET(&ke, fd, EVFILT_WRITE, EV_ADD, 0, 0, NULL);
- if (kevent(state->kqfd, &ke, 1, NULL, 0, NULL) == -1) return -1;
- }
- return 0;
-}
-
-static int aeApiUpdateEvent(EventLoop *eventLoop, int fd, int mask) {
- return aeApiAddEvent(eventLoop, fd, mask);
-}
-
-static int aeApiDelEvent(EventLoop *eventLoop, int fd) {
- aeApiState *state = eventLoop->apidata;
- struct kevent ke;
-
- EV_SET(&ke, fd, EVFILT_READ | EVFILT_WRITE, EV_DELETE, 0, 0, NULL);
- kevent(state->kqfd, &ke, 1, NULL, 0, NULL);
- return 0;
-}
-
-static int aeApiPoll(EventLoop *eventLoop, struct timeval *tvp) {
- aeApiState *state = eventLoop->apidata;
- int retval, numevents = 0;
-
- if (tvp != NULL) {
- struct timespec timeout;
- timeout.tv_sec = tvp->tv_sec;
- timeout.tv_nsec = tvp->tv_usec * 1000;
- retval = kevent(state->kqfd, NULL, 0, state->events, AE_SETSIZE, &timeout);
- } else {
- retval = kevent(state->kqfd, NULL, 0, state->events, AE_SETSIZE, NULL);
- }
-
- if (retval > 0) {
- int j;
-
- numevents = retval;
- for(j = 0; j < numevents; j++) {
- int mask = 0;
- struct kevent *e = state->events+j;
-
- if (e->filter == EVFILT_READ) mask |= AE_READABLE;
- if (e->filter == EVFILT_WRITE) mask |= AE_WRITABLE;
- eventLoop->fired[j] = e->ident;
- }
- }
- return numevents;
-}
-
-static char *aeApiName(void) {
- return "kqueue";
-}
diff --git a/3rdparty/cuttdb/src/ae_select.c b/3rdparty/cuttdb/src/ae_select.c
deleted file mode 100644
index 1e5d3ae91aa886a4b086ff07c28b9e10045ea292..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/ae_select.c
+++ /dev/null
@@ -1,72 +0,0 @@
-/* Select()-based ae.c module
- * Copyright (C) 2009-2010 Salvatore Sanfilippo - antirez@gmail.com
- * Released under the BSD license. See the COPYING file for more info. */
-
-#include <string.h>
-
-typedef struct aeApiState {
- fd_set rfds, wfds;
- /* We need to have a copy of the fd sets as it's not safe to reuse
- * FD sets after select(). */
- fd_set _rfds, _wfds;
-} aeApiState;
-
-static int aeApiCreate(EventLoop *eventLoop) {
- aeApiState *state = malloc(sizeof(aeApiState));
-
- if (!state) return -1;
- FD_ZERO(&state->rfds);
- FD_ZERO(&state->wfds);
- eventLoop->apidata = state;
- return 0;
-}
-
-static void aeApiFree(EventLoop *eventLoop) {
- free(eventLoop->apidata);
-}
-
-static int aeApiAddEvent(EventLoop *eventLoop, int fd, int mask) {
- aeApiState *state = eventLoop->apidata;
-
- if (mask & AE_READABLE) FD_SET(fd,&state->rfds);
- if (mask & AE_WRITABLE) FD_SET(fd,&state->wfds);
- return 0;
-}
-
-static void aeApiDelEvent(EventLoop *eventLoop, int fd, int mask) {
- aeApiState *state = eventLoop->apidata;
-
- if (mask & AE_READABLE) FD_CLR(fd,&state->rfds);
- if (mask & AE_WRITABLE) FD_CLR(fd,&state->wfds);
-}
-
-static int aeApiPoll(EventLoop *eventLoop, struct timeval *tvp) {
- aeApiState *state = eventLoop->apidata;
- int retval, j, numevents = 0;
-
- memcpy(&state->_rfds,&state->rfds,sizeof(fd_set));
- memcpy(&state->_wfds,&state->wfds,sizeof(fd_set));
-
- retval = select(eventLoop->maxfd+1,
- &state->_rfds,&state->_wfds,NULL,tvp);
- if (retval > 0) {
- for (j = 0; j <= eventLoop->maxfd; j++) {
- int mask = 0;
- aeFileEvent *fe = &eventLoop->events[j];
-
- if (fe->mask == AE_NONE) continue;
- if (fe->mask & AE_READABLE && FD_ISSET(j,&state->_rfds))
- mask |= AE_READABLE;
- if (fe->mask & AE_WRITABLE && FD_ISSET(j,&state->_wfds))
- mask |= AE_WRITABLE;
- eventLoop->fired[numevents].fd = j;
- eventLoop->fired[numevents].mask = mask;
- numevents++;
- }
- }
- return numevents;
-}
-
-static char *aeApiName(void) {
- return "select";
-}
diff --git a/3rdparty/cuttdb/src/cdb_bgtask.c b/3rdparty/cuttdb/src/cdb_bgtask.c
deleted file mode 100644
index d12bd26e6c3b35e61e97aab1f2be49b10ea9cd80..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_bgtask.c
+++ /dev/null
@@ -1,127 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cdb_bgtask.h"
-#include <stdlib.h>
-#include <errno.h>
-#include <signal.h>
-
-/* where thread begins */
-static void *_cdb_bgtask_func(void *arg);
-
-
-CDBBGTASK *cdb_bgtask_new()
-{
- CDBBGTASK *bt = (CDBBGTASK *)malloc(sizeof(CDBBGTASK));
-
- bt->tnum = 0;
- bt->run = 0;
- bt->tid = 0;
- pthread_cond_init(&bt->scond, NULL);
- pthread_mutex_init(&bt->smutex, NULL);
- return bt;
-}
-
-
-/* add a task into task list, must called before the thread run */
-int cdb_bgtask_add(CDBBGTASK *bt, TASKFUNC func, void *arg, int intval)
-{
- TASK *task = &bt->tasks[bt->tnum];
-
- if (bt->tid || bt->tnum > MAXTASKNUM)
- return -1;
-
- task->arg = arg;
- task->func = func;
- task->intval = intval;
- task->ltime = time(NULL);
- bt->tnum++;
- return 0;
-}
-
-
-static void *_cdb_bgtask_func(void *arg)
-{
- CDBBGTASK *bt = (CDBBGTASK *)arg;
-#ifndef _WIN32
- /* block all signals coming into current thread */
- sigset_t smask;
- sigfillset(&smask);
- pthread_sigmask(SIG_BLOCK, &smask, NULL);
-#endif
- /* loop */
- while(bt->run) {
- time_t now = time(NULL);
- struct timespec timeout;
- int l_cond_rc;
-
- /* check should run some tasks every 1 second */
- timeout.tv_sec = now + 1;
- timeout.tv_nsec = 0;
-
- /* iterate and run the tasks */
- for(int i = 0; i < bt->tnum; i++) {
- TASK *task = &bt->tasks[i];
- if (now >= task->ltime + task->intval) {
- task->func(task->arg);
- task->ltime = now;
- }
- }
- pthread_mutex_lock(&bt->smutex);
- l_cond_rc = pthread_cond_timedwait(&bt->scond, &bt->smutex, &timeout);
- pthread_mutex_unlock(&bt->smutex);
- }
-
- return NULL;
-}
-
-
-/* create a thread for tasks */
-void cdb_bgtask_start(CDBBGTASK *bt)
-{
- if (bt->run)
- return;
-
- bt->run = 1;
- pthread_create(&bt->tid, NULL, _cdb_bgtask_func, bt);
- return;
-}
-
-
-/* wait for the task thread exits */
-void cdb_bgtask_stop(CDBBGTASK *bt)
-{
- if (bt->run) {
- void **ret = NULL;
- bt->run = 0;
- pthread_cond_signal(&bt->scond);
- pthread_join(bt->tid, ret);
- }
-
- bt->tnum = 0;
-}
-
-
-void cdb_bgtask_destroy(CDBBGTASK *bt)
-{
- cdb_bgtask_stop(bt);
- pthread_cond_destroy(&bt->scond);
- pthread_mutex_destroy(&bt->smutex);
- free(bt);
-}
-
-
-
-
diff --git a/3rdparty/cuttdb/src/cdb_bgtask.h b/3rdparty/cuttdb/src/cdb_bgtask.h
deleted file mode 100644
index 6dee1b992d21bac8a496e7f8f89431c570b9358a..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_bgtask.h
+++ /dev/null
@@ -1,62 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CDB_BGTASK_H_
-#define _CDB_BGTASK_H_
-#include <time.h>
-#include <pthread.h>
-
-
-/* 16 tasks at most in a task thread */
-#define MAXTASKNUM 16
-
-typedef void (*TASKFUNC)(void *);
-
-/* struct for timer task */
-typedef struct {
- /* task function */
- TASKFUNC func;
- /* task argument */
- void *arg;
- /* task run interval(seconds) */
- int intval;
- /* time of last run */
- time_t ltime;
-} TASK;
-
-/* struct for a background task manager */
-typedef struct CDBBGTASK
-{
- TASK tasks[MAXTASKNUM];
- /* number of tasks */
- int tnum;
- /* is running? */
- int run;
- pthread_t tid;
- /* for wait the thread exit */
- pthread_mutex_t smutex;
- pthread_cond_t scond;
-} CDBBGTASK;
-
-
-
-CDBBGTASK *cdb_bgtask_new();
-int cdb_bgtask_add(CDBBGTASK *task, TASKFUNC func, void *arg, int intval);
-void cdb_bgtask_start(CDBBGTASK *bt);
-void cdb_bgtask_stop(CDBBGTASK *task);
-void cdb_bgtask_destroy(CDBBGTASK *task);
-
-
-#endif
diff --git a/3rdparty/cuttdb/src/cdb_bloomfilter.c b/3rdparty/cuttdb/src/cdb_bloomfilter.c
deleted file mode 100644
index ebf5e2d3e8b508a20985b76bd6a1e974a89c9daf..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_bloomfilter.c
+++ /dev/null
@@ -1,158 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cdb_bloomfilter.h"
-#include <stdlib.h>
-#include <string.h>
-
-#define CDBBFHASHNUM 16
-#define CDBBFSPLITPOW 6
-
-static uint64_t BFSEEDS[CDBBFHASHNUM] = {217636919,290182597,386910137,515880193,
- 687840301,917120411,1222827239,1610612741,
- 3300450239,3300450259,3300450281,3300450289,
- 3221225473ul,4294967291ul,163227661,122420729,};
-
-struct CDBBLOOMFILTER
-{
- uint8_t *bitmap[1<<CDBBFSPLITPOW];
- uint64_t rnum;
- uint64_t size;
- int hnum;
- int ratio;
-};
-
-
-CDBBLOOMFILTER *cdb_bf_new(uint64_t rnum, uint64_t size)
-{
- CDBBLOOMFILTER *bf = (CDBBLOOMFILTER *)malloc(sizeof(CDBBLOOMFILTER));
- bf->rnum = 0;
- bf->size = size;
- /* number of hash should be 0.7 * ratio */
- bf->hnum = size * 8 * 7 / (rnum * 10);
- /* number of hash is limit in [1, 16] */
- if (bf->hnum > CDBBFHASHNUM)
- bf->hnum = CDBBFHASHNUM;
- if (bf->hnum == 0)
- bf->hnum = 1;
- /* avoid malloc too much memory once */
- for(int i = 0; i < (1 << CDBBFSPLITPOW); i++) {
- bf->bitmap[i] = (uint8_t*)malloc(size >> CDBBFSPLITPOW);
- memset(bf->bitmap[i], 0, size >> CDBBFSPLITPOW);
- }
- return bf;
-}
-
-
-void cdb_bf_set(CDBBLOOMFILTER *bf, void *key, int ksize)
-{
- uint8_t *src = (uint8_t *)key, *end = src + ksize;
- uint64_t hval[CDBBFHASHNUM] = {0};
-
- for(;src < end; src++)
- for(int i = 0; i < bf->hnum; i++)
- hval[i] = hval[i] * BFSEEDS[i] + *src;
-
- for(int i = 0; i < bf->hnum; i++) {
- uint64_t p = (hval[i] >> CDBBFSPLITPOW) % ((bf->size >> CDBBFSPLITPOW) << 3);
- uint8_t *bitmap = bf->bitmap[hval[i] & ((1<<CDBBFSPLITPOW) - 1)];
- bitmap[p >> 3] |= (1 << (p & 0x07));
- }
-
- bf->rnum++;
-}
-
-
-bool cdb_bf_exist(CDBBLOOMFILTER *bf, void *key, int ksize)
-{
- uint8_t *src = (uint8_t *)key, *end = src + ksize;
- uint64_t hval[CDBBFHASHNUM] = {0};
- int exist = 0;
-
- for(;src < end; src++)
- for(int i = 0; i < bf->hnum; i++)
- hval[i] = hval[i] * BFSEEDS[i] + *src;
-
- for(int i = 0; i < bf->hnum; i++) {
- uint64_t p = (hval[i] >> CDBBFSPLITPOW) % ((bf->size >> CDBBFSPLITPOW) << 3);
- uint8_t *bitmap = bf->bitmap[hval[i] & ((1<<CDBBFSPLITPOW) - 1)];
- if (bitmap[p >> 3] & (1 << (p & 0x07)))
- exist++;
- else
- break;
- }
-
- return (exist == bf->hnum);
-}
-
-void cdb_bf_clean(CDBBLOOMFILTER *bf)
-{
- for(int i = 0; i < (1 << CDBBFSPLITPOW); i++)
- memset(bf->bitmap[i], 0, bf->size >> CDBBFSPLITPOW);
-
- bf->rnum = 0;
-}
-
-
-void cdb_bf_destroy(CDBBLOOMFILTER *bf)
-{
- for(int i = 0; i < (1 << CDBBFSPLITPOW); i++)
- free(bf->bitmap[i]);
- free(bf);
-}
-
-
-#ifdef _UT_CDBBF_
-#include <stdio.h>
-#include <stdlib.h>
-#include "cdb_bloomfilter.h"
-
-int main(int argc, char *argv[])
-{
- int size = 1048576;
- int rnum = 1048576;
- if (argc > 1)
- rnum = atoi(argv[1]);
- if (argc > 2)
- size = atoi(argv[2]);
-
- CDBBLOOMFILTER *bf = cdb_bf_new(rnum, size);
- for(int i = 0; i < rnum; i++) {
- int j = 2 * i;
- cdb_bf_set(bf, &j, 4);
- }
-
- int exist = 0;
- for(int i = 0; i < rnum; i++) {
- int j = 2 * i;
- if (cdb_bf_exist(bf, &j, 4))
- exist++;
- }
- printf("right positive: %.2f%%%%\n", (float)exist/(float)rnum*10000);
-
- exist = 0;
- for(int i = 0; i < rnum * 2; i++) {
- int j = 2 * i + 1;
- if (cdb_bf_exist(bf, &j, 4))
- exist++;
- }
-
- printf("false positive: %.2f%%%% %d/%d\n", (float)exist/(float)rnum*5000, exist, rnum * 2);
- printf("element num: %d\n", bf->rnum);
- cdb_bf_destroy(bf);
- return 0;
-}
-#endif
-
diff --git a/3rdparty/cuttdb/src/cdb_bloomfilter.h b/3rdparty/cuttdb/src/cdb_bloomfilter.h
deleted file mode 100644
index 6ccdab1fccc1d92d843f072046550741a5cdaf37..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_bloomfilter.h
+++ /dev/null
@@ -1,34 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-/*
-Bloom Filter is currently not used in cuttdb
-*/
-#ifndef _CDB_BLOOMFILTER_H_
-#define _CDB_BLOOMFILTER_H_
-#include <stdbool.h>
-#include <stdint.h>
-
-typedef struct CDBBLOOMFILTER CDBBLOOMFILTER;
-
-#define CDBBFRATIO 8
-
-CDBBLOOMFILTER *cdb_bf_new(uint64_t rnum, uint64_t size);
-void cdb_bf_set(CDBBLOOMFILTER *bf, void *key, int ksize);
-bool cdb_bf_exist(CDBBLOOMFILTER *bf, void *key, int ksize);
-void cdb_bf_clean(CDBBLOOMFILTER *bf);
-void cdb_bf_destroy(CDBBLOOMFILTER *bf);
-
-#endif
diff --git a/3rdparty/cuttdb/src/cdb_builddb.c b/3rdparty/cuttdb/src/cdb_builddb.c
deleted file mode 100644
index fc5f18dc68c9dbd78de89441d88ada791a8b97a5..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_builddb.c
+++ /dev/null
@@ -1,72 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-#include "cuttdb.h"
-#include <stdio.h>
-#include <stdlib.h>
-#include <stdint.h>
-#include <time.h>
-
-int main(int argc, char *argv[])
-{
- CDB *db = cdb_new();
- if (argc < 2) {
- fprintf(stderr, "Usage: %s db_path [hsize = 2000000]\n", argv[0]);
- return 0;
- }
-
- /* 1TB memory limit(unlimited) */
- cdb_option(db, argc >= 3? atoi(argv[2]):2000000 , 0, 1048576);
- cdb_seterrcb(db, cdb_deferrorcb, NULL);
- if (cdb_open(db, argv[1], CDB_CREAT | CDB_PAGEWARMUP) < 0) {
- return -1;
- }
- char *buf = NULL;
- long count = 0;
-
- size_t size, size2;
- while((size = getline(&buf, &size2, stdin)) != -1) {
- /* remove the delimiter*/
- buf[--size] = '\0';
- int klen = -1;
- int vlen = -1;
- uint32_t expire = 0;
- int parsenum = 0;
- for(int i = 0; i < size; i++) {
- if (buf[i] == '\t') {
- if (klen == -1)
- klen = i;
- else {
- vlen = i - klen - 1;
- parsenum = 1;
- }
- } else if (buf[i] >= '0' && buf[i] <= '9' && parsenum) {
- expire = expire * 10 + buf[i] - '0';
- }
- }
-
- if (klen > 0 && vlen > 0) {
- cdb_set2(db, buf, klen, buf + klen + 1, vlen,
- CDB_OVERWRITE, expire > 0? expire - time(NULL): 0);
- count++;
- }
- free(buf);
- buf = NULL;
- }
- cdb_destroy(db);
- fprintf(stderr, "imported %ld records\n", count);
- return 0;
-}
-
-
diff --git a/3rdparty/cuttdb/src/cdb_core.c b/3rdparty/cuttdb/src/cdb_core.c
deleted file mode 100644
index 75aabebe36dd7a12e0101d14a2229e0824d4ee64..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_core.c
+++ /dev/null
@@ -1,1457 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cuttdb.h"
-#include "cdb_crc64.h"
-#include "cdb_types.h"
-#include "cdb_hashtable.h"
-#include "cdb_bloomfilter.h"
-#include "cdb_lock.h"
-#include "cdb_bgtask.h"
-#include "cdb_errno.h"
-#include "cdb_vio.h"
-#include "cdb_core.h"
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-#include <pthread.h>
-
-static void _cdb_pageout(CDB *db);
-static void _cdb_defparam(CDB *db);
-static void _cdb_recout(CDB *db);
-static uint32_t _pagehash(const void *key, int len);
-static void _cdb_flushdpagetask(void *arg);
-static void _cdb_timerreset(struct timespec *ts);
-static uint32_t _cdb_timermicrosec(struct timespec *ts);
-static void _cdb_pagewarmup(CDB *db, bool loadbf);
-
-
-/* it isn't necessary to rehash bid in hash table cache */
-static uint32_t _pagehash(const void *key, int len)
-{
- uint32_t tst = 0;
- if (len < tst)
- return 0;
- (void) len;
- return *(uint32_t*)key;
-}
-
-
-/* used to get the duration of a procedure */
-static void _cdb_timerreset(struct timespec *ts)
-{
- clock_gettime(CLOCK_MONOTONIC, ts);
-}
-
-
-static uint32_t _cdb_timermicrosec(struct timespec *ts)
-{
- struct timespec ts2;
- uint32_t diff;
- clock_gettime(CLOCK_MONOTONIC, &ts2);
- diff = (ts2.tv_sec - ts->tv_sec) * 1000000;
- diff += ts2.tv_nsec / 1000;
- diff -= ts->tv_nsec / 1000;
- return diff;
-}
-
-
-/* reset the parameters */
-static void _cdb_defparam(CDB *db)
-{
- db->rnum = 0;
- db->bfsize = 0;
- db->rclimit = 128 * MB;
- db->pclimit = 1024 * MB;
- db->hsize = 1000000;
- db->rcache = db->pcache = db->dpcache = NULL;
- db->bf = NULL;
- db->opened = false;
- db->vio = NULL;
- db->mtable = NULL;
- db->oid = 0;
- db->roid = 0;
- db->errcbarg = NULL;
- db->errcb = NULL;
- db->areadsize = 4 * KB;
- return;
-}
-
-
-/* flush all dirty pages */
-void cdb_flushalldpage(CDB *db)
-{
- if (db->dpcache) {
- while (db->dpcache->num) {
- CDBHTITEM *item = cdb_ht_poptail(db->dpcache);
- uint32_t bid = *(uint32_t*)cdb_ht_itemkey(db->dpcache, item);
- FOFF off;
- db->vio->wpage(db->vio, (CDBPAGE*)cdb_ht_itemval(db->dpcache, item), &off);
- db->mtable[bid] = off;
- free(item);
- }
-
- db->roid = db->oid;
- db->vio->cleanpoint(db->vio);
- }
-}
-
-
-/* flush oldest dirty index page to disk, it runs in another thread and triggered by timer */
-static void _cdb_flushdpagetask(void *arg)
-{
- CDB *db = (CDB *)arg;
- CDBHTITEM *item;
- CDBPAGE *page;
- time_t now = time(NULL);
- bool cleandcache = false;
- uint32_t bid;
-
- if (!db->dpcache)
- /* no dirty page cache */
- return;
-
- /* if there isn't too much dirty page and some time passed since last clean,
- write out all dirty pages to make a recovery point(oid) */
- if (db->dpcache->num < 1024 && now > db->ndpltime + 120)
- cleandcache = true;
-
- while(db->dpcache->num) {
- FOFF off;
- cdb_lock_lock(db->dpclock);
- item = cdb_ht_gettail(db->dpcache);
- /* no item in dpcache after lock */
- if (item == NULL) {
- cdb_lock_unlock(db->dpclock);
- return;
- }
- page = (CDBPAGE *)cdb_ht_itemval(db->dpcache, item);
- /* bid = page->bid; also OK */
- bid = *(uint32_t*)cdb_ht_itemkey(db->dpcache, item);
- /* been dirty for too long? */
- if (now > page->mtime + DPAGETIMEOUT || cleandcache) {
- if (cdb_lock_trylock(db->mlock[page->bid % MLOCKNUM])) {
- /* avoid dead lock, since dpclock is holding */
- cdb_lock_unlock(db->dpclock);
- return;
- }
- /* remove it from dpcache */
- cdb_ht_poptail(db->dpcache);
- cdb_lock_unlock(db->dpclock);
-
- /* write to disk */
- struct timespec ts;
- _cdb_timerreset(&ts);
- db->vio->wpage(db->vio, page, &off);
- db->wcount++;
- db->wtime += _cdb_timermicrosec(&ts);
- db->mtable[bid] = off;
-
- /* move the clean page into pcache */
- cdb_lock_lock(db->pclock);
- cdb_ht_insert(db->pcache, item);
- cdb_lock_unlock(db->pclock);
- cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
- } else {
- /* tail in dpcache isn't expired */
- cdb_lock_unlock(db->dpclock);
- return;
- }
- }
-
- if (db->dpcache->num == 0 && cleandcache)
- db->ndpltime = now;
-
- if (cleandcache) {
- /* clean succeed if goes here, remember the recovery point */
- /* it's not necessary to lock */
- db->roid = db->oid;
- db->vio->cleanpoint(db->vio);
- }
-}
-
-
-/* fill the index page cache, and set the bloomfilter if necessary */
-static void _cdb_pagewarmup(CDB *db, bool loadbf)
-{
- char sbuf[SBUFSIZE];
- void *it = db->vio->pageitfirst(db->vio, 0);
-
- if (it == NULL)
- return;
-
- for(;;) {
- CDBPAGE *page = (CDBPAGE *)sbuf;
- if (db->vio->pageitnext(db->vio, &page, it) < 0)
- break;
-
- /* the page is the newest one because its offset matches the one in main table */
- if (OFFEQ(page->ooff, db->mtable[page->bid])) {
- if (loadbf) {
- /* iterate key hashes in page, set to the filter */
- cdb_lock_lock(db->bflock);
- for(uint32_t i = 0; i < page->num; i++) {
- uint64_t hash = (page->bid << 24) | (page->items[i].hash.i2 << 8)
- | (page->items[i].hash.i1);
- /* bloom filter use the combined record hash as key */
- cdb_bf_set(db->bf, &hash, SI8);
- }
- cdb_lock_unlock(db->bflock);
- }
-
- /* set the page to pcache if it doesn't exceed the limit size */
- if (db->pcache && db->pcache->size < db->pclimit) {
- cdb_lock_lock(db->pclock);
- cdb_ht_insert2(db->pcache, &page->bid, SI4, page, MPAGESIZE(page));
- cdb_lock_unlock(db->pclock);
- }
- }
- /* the page may not be still in stack */
- if (page != (CDBPAGE *)sbuf)
- free(page);
-
- if (!loadbf && (db->pcache && db->pcache->size > db->pclimit))
- break;
- }
-
- db->vio->pageitdestroy(db->vio, it);
-}
-
-
-/* generate an incremental global operation id */
-uint64_t cdb_genoid(CDB *db)
-{
- uint64_t oid;
- cdb_lock_lock(db->oidlock);
- oid = db->oid++;
- cdb_lock_unlock(db->oidlock);
- return oid;
-}
-
-
-/* get a new record iterator */
-void *cdb_iterate_new(CDB *db, uint64_t oid)
-{
- return db->vio->recitfirst(db->vio, oid);
-}
-
-
-
-/* iterate the database by callback */
-uint64_t cdb_iterate(CDB *db, CDB_ITERCALLBACK itcb, void *arg, void *iter)
-{
- char sbuf[SBUFSIZE];
- uint64_t cnt = 0;
-
- if (iter == NULL)
- return cnt;
- for(;;) {
- /* the rec is a copy from file, may in stack or allocated in heap */
- CDBREC *rec = (CDBREC *)sbuf;
- bool ret = true;
- if (db->vio->recitnext(db->vio, &rec, iter) < 0)
- break;
-
- if (cdb_checkoff(db, CDBHASH64(rec->key, rec->ksize), rec->ooff, CDB_NOTLOCKED)) {
- ret = itcb(arg, rec->key, rec->ksize, rec->val, rec->vsize, rec->expire, rec->oid);
- cnt++;
- }
- if (rec != (CDBREC *)sbuf)
- free(rec);
- if (!ret)
- break;
- }
- return cnt;
-}
-
-
-
-/* destroy the iterator */
-void cdb_iterate_destroy(CDB *db, void *iter)
-{
- db->vio->recitdestroy(db->vio, iter);
-}
-
-
-/* difficult to implement */
-/*
-static void _cdb_rcachewarmup(CDB *db)
-{
-}
-*/
-
-
-CDB *cdb_new()
-{
- CDB *db;
- db = (CDB *)malloc(sizeof(CDB));
- /* I assume all operation in this layer is 'fast', so no mutex used here */
- for(int i = 0; i < MLOCKNUM; i++)
- db->mlock[i] = cdb_lock_new(CDB_LOCKMUTEX);
- db->dpclock = cdb_lock_new(CDB_LOCKMUTEX);
- db->pclock = cdb_lock_new(CDB_LOCKMUTEX);
- db->rclock = cdb_lock_new(CDB_LOCKMUTEX);
- db->stlock = cdb_lock_new(CDB_LOCKMUTEX);
- db->oidlock = cdb_lock_new(CDB_LOCKMUTEX);
- db->bflock = cdb_lock_new(CDB_LOCKMUTEX);
- db->bgtask = cdb_bgtask_new();
- /* every thread should has its own errno */
- db->errkey = (pthread_key_t *)malloc(sizeof(pthread_key_t));
- pthread_key_create(db->errkey, NULL);
- /* set default parameter */
- _cdb_defparam(db);
- return db;
-}
-
-
-int cdb_option(CDB *db, int bnum, int rcacheMB, int pcacheMB)
-{
- /* too small bnum is not allowed */
- db->hsize = bnum > 4096? bnum : 4096;
-
- if (rcacheMB >= 0)
- db->rclimit = (uint64_t)rcacheMB * MB;
- if (pcacheMB >= 0)
- db->pclimit = (uint64_t)pcacheMB * MB;
- return 0;
-}
-
-
-void cdb_option_bloomfilter(CDB *db, uint64_t size)
-{
- db->bfsize = size;
-}
-
-void cdb_option_areadsize(CDB *db, uint32_t size)
-{
- db->areadsize = size;
- if (db->areadsize < 1 * KB)
- db->areadsize = 1 * KB;
-
- if (db->areadsize > SBUFSIZE - (sizeof(CDBREC) - RECHSIZE))
- db->areadsize = SBUFSIZE - (sizeof(CDBREC) - RECHSIZE);
-}
-
-int cdb_open(CDB *db, const char *file_name, int mode)
-{
- /* if will become into a hash table when file_name == CDB_MEMDB */
- int memdb = (strcmp(file_name, CDB_MEMDB) == 0);
-
- if (db->rclimit)
- /* record cache is enabled */
- db->rcache = cdb_ht_new(true, NULL);
- else if (memdb) {
- /* record cache is disabled, but in MEMDB mode */
- cdb_seterrno(db, CDB_MEMDBNOCACHE, __FILE__, __LINE__);
- goto ERRRET;
- }
-
- if (db->pclimit && !memdb) {
- /* page cache enabled. page cache is meaningless under MEMDB mode */
- db->dpcache = cdb_ht_new(true, _pagehash);
- db->pcache = cdb_ht_new(true, _pagehash);
- }
-
-
- if (!memdb) {
- if (db->bfsize) {
- /* bloom filter enabled */
- db->bf = cdb_bf_new(db->bfsize, db->bfsize);
- }
- /* now only one storage format is supported */
- db->vio = cdb_vio_new(CDBVIOAPND2);
- db->vio->db = db;
- if (db->vio->open(db->vio, file_name, mode) < 0)
- goto ERRRET;
- if (db->vio->rhead(db->vio) < 0) {
- db->mtable = (FOFF*)malloc(sizeof(FOFF) * db->hsize);
- memset(db->mtable, 0, sizeof(FOFF) * db->hsize);
- }
- /* dirty index page would be swap to disk by timer control */
- cdb_bgtask_add(db->bgtask, _cdb_flushdpagetask, db, 1);
- db->ndpltime = time(NULL);
- /* start background task thread */
- cdb_bgtask_start(db->bgtask);
- } else {
- /* no persistent storage under MEMDB mode */
- db->vio = NULL;
- db->bgtask = NULL;
- db->mtable = NULL;
- }
-
- if (db->bf || ((mode & CDB_PAGEWARMUP) && db->pcache)) {
- /* fill the bloom filter if it is enabled, and fill the page cache */
- _cdb_pagewarmup(db, !!db->bf);
- }
-
- /* reset the statistic info */
- cdb_stat(db, NULL);
- db->opened = true;
- return 0;
-
-ERRRET:
- if (db->rcache)
- cdb_ht_destroy(db->rcache);
- if (db->pcache)
- cdb_ht_destroy(db->pcache);
- if (db->dpcache)
- cdb_ht_destroy(db->dpcache);
- if (db->bf)
- cdb_bf_destroy(db->bf);
- cdb_bgtask_stop(db->bgtask);
- _cdb_defparam(db);
- return -1;
-}
-
-
-/* check if the page cache size exceed the limit. clean oldest page if necessary */
-static void _cdb_pageout(CDB *db)
-{
- while (PCOVERFLOW(db)) {
- if (db->pcache->num) {
- /* clean page cache is prior */
- cdb_lock_lock(db->pclock);
- cdb_ht_removetail(db->pcache);
- cdb_lock_unlock(db->pclock);
- } else if (db->dpcache->num) {
- CDBHTITEM *item;
- uint32_t bid;
- FOFF off;
- cdb_lock_lock(db->dpclock);
- item = cdb_ht_gettail(db->dpcache);
- if (item == NULL) {
- cdb_lock_unlock(db->dpclock);
- break;
- }
-
- bid = *(uint32_t*)cdb_ht_itemkey(db->dpcache, item);
- /* must lock the main table inside the dpclock protection */
- if (cdb_lock_trylock(db->mlock[bid % MLOCKNUM]) < 0) {
- /* avoid dead lock since dpclock is holding */
- cdb_lock_unlock(db->dpclock);
- /* do nothing this time */
- break;
- }
- cdb_ht_poptail(db->dpcache);
- cdb_lock_unlock(db->dpclock);
-
- /* write out dirty page */
- struct timespec ts;
- _cdb_timerreset(&ts);
- db->vio->wpage(db->vio, (CDBPAGE*)cdb_ht_itemval(db->dpcache, item), &off);
- db->wcount++;
- db->wtime += _cdb_timermicrosec(&ts);
- db->mtable[bid] = off;
- cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
- free(item);
- }
- }
-}
-
-
-/* check if the record cache size exceed the limit. clean oldest record if necessary */
-static void _cdb_recout(CDB *db)
-{
- while (RCOVERFLOW(db)) {
- cdb_lock_lock(db->rclock);
- if (db->rcache->num)
- cdb_ht_removetail(db->rcache);
- cdb_lock_unlock(db->rclock);
- }
-}
-
-
-/* get all offsets from index(page) by key, even if only one of them at most is valid.
- Others are due to the hash collision */
-int cdb_getoff(CDB *db, uint64_t hash, FOFF **offs, int locked)
-{
- char sbuf[SBUFSIZE];
- CDBPAGE *page = NULL;
- int rnum;
- bool incache = true;
- uint32_t bid = (hash >> 24) % db->hsize;
- PHASH phash;
-
- phash.i1 = hash & 0xff;
- phash.i2 = (hash >> 8) & 0xffff;
-
- if (db->bf) {
- uint64_t bfkey = (bid << 24) | (hash & 0xffffff);
- /* check the key-hash in bloom filter? return now if not exist */
- cdb_lock_lock(db->bflock);
- if (!cdb_bf_exist(db->bf, &bfkey, SI8)) {
- cdb_lock_unlock(db->bflock);
- return 0;
- }
- cdb_lock_unlock(db->bflock);
- }
-
- if (locked == CDB_NOTLOCKED) cdb_lock_lock(db->mlock[bid % MLOCKNUM]);
- /* page exists in clean page cache? */
- if (db->pcache) {
- cdb_lock_lock(db->pclock);
- page = cdb_ht_get2(db->pcache, &bid, SI4, true);
- cdb_lock_unlock(db->pclock);
- }
-
- /* not in pcache, exists in dirty page cache? */
- if (page == NULL && db->dpcache) {
- cdb_lock_lock(db->dpclock);
- page = cdb_ht_get2(db->dpcache, &bid, SI4, true);
- cdb_lock_unlock(db->dpclock);
- }
-
- if (page == NULL) {
- /* not in dpcache either, read from disk */
- incache = false;
- db->pcmiss++;
- /* page stays in stack by default */
- page = (CDBPAGE *)sbuf;
- if (OFFNOTNULL(db->mtable[bid])) {
- /* page offset not null in main table */
- int ret;
- struct timespec ts;
- _cdb_timerreset(&ts);
- ret = db->vio->rpage(db->vio, &page, db->mtable[bid]);
- db->rcount++;
- db->rtime += _cdb_timermicrosec(&ts);
-
- /* read page error, return */
- if (ret < 0) {
- if (locked == CDB_NOTLOCKED) cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
- if (page != (CDBPAGE *)sbuf)
- free(page);
- return -1;
- }
- } else {
- /* no page in this bucket */
- page->cap = page->num = 0;
- page->osize = 0;
- OFFZERO(page->ooff);
- }
- } else {
- db->pchit++;
- }
-
- rnum = 0;
- for(uint32_t i = 0; i < page->num; i++) {
- /* compare every hash in the page */
- if (PHASHEQ(page->items[i].hash, phash)) {
- (*offs)[rnum] = page->items[i].off;
- /* result offset list stays in stack by default. Allocate one in heap if
- it exceeds the limit */
- if (++rnum == SFOFFNUM) {
- /* very little possibility goes here */
- FOFF *tmp = (FOFF*)malloc((page->num - i + SFOFFNUM + 1) * sizeof(FOFF));
- memcpy(tmp, *offs, SFOFFNUM * sizeof(FOFF));
- *offs = tmp;
- }
- }
- }
-
- if (!incache) {
- /* set into clean page cache if not exists before */
- if (db->pcache) {
- cdb_lock_lock(db->pclock);
- cdb_ht_insert2(db->pcache, &bid, SI4, page, MPAGESIZE(page));
- cdb_lock_unlock(db->pclock);
- }
- /* if page now points to heap memory, free it */
- if (page != (CDBPAGE *)sbuf) {
- free(page);
- }
- }
- if (locked == CDB_NOTLOCKED) cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
-
- /* check page cache overflow */
- if (PCOVERFLOW(db))
- _cdb_pageout(db);
-
- return rnum;
-}
-
-
-/* replace a specified record's offset, may be used at disk space recycling
- off indicates its previous offset, noff is the new offset. return negative if not found */
-int cdb_replaceoff(CDB *db, uint64_t hash, FOFF off, FOFF noff, int locked)
-{
- char sbuf[SBUFSIZE];
- CDBPAGE *page = NULL;
- CDBHTITEM *pitem = NULL;
- bool indpcache = false;
- uint32_t bid = (hash >> 24) % db->hsize;
- PHASH phash;
- bool found = false;
-
- phash.i1 = hash & 0xff;
- phash.i2 = (hash >> 8) & 0xffff;
-
- if (locked == CDB_NOTLOCKED) cdb_lock_lock(db->mlock[bid % MLOCKNUM]);
- if (db->pcache) {
- /* in clean page cache, since it would be modified, it should be deleted from pcache */
- cdb_lock_lock(db->pclock);
- pitem = cdb_ht_del(db->pcache, &bid, SI4);
- cdb_lock_unlock(db->pclock);
- if (pitem)
- page = (CDBPAGE *)cdb_ht_itemval(db->pcache, pitem);
- }
- if (page == NULL && db->dpcache) {
- /* not in pcache, but in dirty page cache */
- cdb_lock_lock(db->dpclock);
- page = cdb_ht_get2(db->dpcache, &bid, SI4, true);
- cdb_lock_unlock(db->dpclock);
- if (page)
- indpcache = true;
- }
- if (page == NULL) {
- /* not exists either, read from disk */
- db->pcmiss++;
- page = (CDBPAGE *)sbuf;
- if (OFFNOTNULL(db->mtable[bid])) {
- int ret;
- struct timespec ts;
- _cdb_timerreset(&ts);
- ret = db->vio->rpage(db->vio, &page, db->mtable[bid]);
- db->rcount++;
- db->rtime += _cdb_timermicrosec(&ts);
-
- if (ret < 0) {
- if (locked == CDB_NOTLOCKED) cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
- if (page != (CDBPAGE *)sbuf)
- free(page);
- return -1;
- }
- } else {
- /* nullified the empty page */
- page->cap = page->num = 0;
- page->osize = 0;
- OFFZERO(page->ooff);
- }
- } else {
- db->pchit++;
- }
-
- /* check and modify */
- for(uint32_t i = 0; i < page->num; i++) {
- if (PHASHEQ(page->items[i].hash, phash)
- && OFFEQ(page->items[i].off, off)) {
- page->items[i].off = noff;
- found = true;
- break;
- }
- }
-
- if (db->dpcache && !indpcache) {
- /* if page already dirty in cache, need not do anything */
- /* dirty page cache is enabled but not exists before */
- if (pitem) {
- /* pitem not NULL indicates it belongs to pcache */
- if (found) {
- /* modified page */
- cdb_lock_lock(db->dpclock);
- cdb_ht_insert(db->dpcache, pitem);
- cdb_lock_unlock(db->dpclock);
- } else {
- /* got from pcache, but not modified */
- cdb_lock_lock(db->pclock);
- cdb_ht_insert(db->pcache, pitem);
- cdb_lock_unlock(db->pclock);
- }
- /* page belongs to memory in 'cache', must not free */
- } else if (page != NULL) {
- /* page read from disk, but not in cache */
- cdb_lock_lock(db->dpclock);
- cdb_ht_insert2(db->dpcache, &bid, SI4, page, MPAGESIZE(page));
- cdb_lock_unlock(db->dpclock);
- /* the 'page' won't be use anymore */
- if (page != (CDBPAGE *)sbuf)
- free(page);
- }
- } else if (!db->dpcache){
- /* no page cache. Write out dirty page immediately */
- FOFF poff;
- struct timespec ts;
- _cdb_timerreset(&ts);
- db->vio->wpage(db->vio, page, &poff);
- db->wcount++;
- db->wtime += _cdb_timermicrosec(&ts);
-
- db->mtable[bid] = poff;
- if (page != (CDBPAGE *)sbuf)
- free(page);
- }
- if (locked == CDB_NOTLOCKED) cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
-
- /* check page cache overflow */
- if (PCOVERFLOW(db))
- _cdb_pageout(db);
-
- return 0;
-}
-
-
-/* insert/delete a key-offset pair from index page */
-int cdb_updatepage(CDB *db, uint64_t hash, FOFF off, int opt, int locked)
-{
- char sbuf[SBUFSIZE], sbuf2[SBUFSIZE];
- CDBPAGE *page = NULL, *npage = NULL;
- CDBHTITEM *pitem = NULL, *nitem = NULL;
- CDBHASHTABLE *tmpcache = NULL;
- CDBLOCK *tmpclock = NULL;
- int npsize = 0;
- uint32_t bid = (hash >> 24) % db->hsize;
- PHASH phash;
-
- phash.i1 = hash & 0xff;
- phash.i2 = (hash >> 8) & 0xffff;
-
- if (locked == CDB_NOTLOCKED) cdb_lock_lock(db->mlock[bid % MLOCKNUM]);
- /* firstly, try move the page out of the cache if possible,
- it assumes that the page would be modified(pair exists) */
- if (db->pcache) {
- /* try clean page cache */
- cdb_lock_lock(db->pclock);
- pitem = cdb_ht_del(db->pcache, &bid, SI4);
- cdb_lock_unlock(db->pclock);
- if (pitem) {
- page = (CDBPAGE *)cdb_ht_itemval(db->pcache, pitem);
- tmpcache = db->pcache;
- tmpclock = db->pclock;
- }
- }
- if (page == NULL && db->dpcache) {
- /* try dirty page cache */
- cdb_lock_lock(db->dpclock);
- pitem = cdb_ht_del(db->dpcache, &bid, SI4);
- cdb_lock_unlock(db->dpclock);
- if (pitem) {
- page = (CDBPAGE *)cdb_ht_itemval(db->dpcache, pitem);
- tmpcache = db->dpcache;
- tmpclock = db->dpclock;
- }
- }
-
- if (page == NULL) {
- db->pcmiss++;
- page = (CDBPAGE *)sbuf;
- /* doesn't exist in cache, read from disk */
- if (OFFNOTNULL(db->mtable[bid])) {
- int ret;
- struct timespec ts;
- _cdb_timerreset(&ts);
- ret = db->vio->rpage(db->vio, &page, db->mtable[bid]);
- db->rcount++;
- db->rtime += _cdb_timermicrosec(&ts);
-
- if (ret < 0) {
- if (locked == CDB_NOTLOCKED) cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
- if (page != (CDBPAGE *)sbuf)
- free(page);
- return -1;
- }
- } else {
- page->cap = 0;
- page->num = 0;
- page->osize = 0;
- OFFZERO(page->ooff);
- }
- } else {
- db->pchit++;
- }
-
- npsize = MPAGESIZE(page);
-
- if (opt == CDB_PAGEDELETEOFF)
- ;// npsize = MPAGESIZE(page) - sizeof(PITEM);
- /* do not malloc new page on deletion */
-
- else if (opt == CDB_PAGEINSERTOFF && page->cap == page->num) {
- /* get a new page, from dirty page cache if possible */
- npsize = MPAGESIZE(page) + CDB_PAGEINCR * sizeof(PITEM);
- if (db->dpcache) {
- nitem = cdb_ht_newitem(db->dpcache, SI4, npsize);
- *(uint32_t*)cdb_ht_itemkey(db->dpcache, nitem) = bid;
- npage = (CDBPAGE *)cdb_ht_itemval(db->dpcache, nitem);
- } else {
- /* no dpcache, use stack if size fits */
- if (npsize > SBUFSIZE)
- npage = (CDBPAGE *)malloc(npsize);
- else
- npage = (CDBPAGE *)sbuf2;
- }
-
- /* initialize the new page */
-
- npage->bid = bid;
- npage->oid = cdb_genoid(db);
- npage->osize = page->osize;
- npage->ooff = page->ooff;
- npage->mtime = time(NULL);
- npage->cap = page->cap + CDB_PAGEINCR;
- npage->num = page->num;
- memcpy(npage->items, page->items, page->num * sizeof(PITEM));
- /* old page got from cache */
- if (pitem)
- free(pitem);
- /* old page read from disk, if in stack? */
- else if (page != (CDBPAGE *)sbuf)
- free(page);
-
- page = npage;
- pitem = nitem;
- }
-
- uint32_t onum = page->num;
-
- if (opt == CDB_PAGEDELETEOFF) {
- bool found = false;
- for(uint32_t i = 0; i < page->num; i++) {
- if (!found) {
- if (PHASHEQ(page->items[i].hash, phash)
- && OFFEQ(page->items[i].off, off))
- {
- found = true;
- /* records num is consistant with index */
- cdb_lock_lock(db->stlock);
- db->rnum--;
- cdb_lock_unlock(db->stlock);
- }
- }
- if (found && i + 1 < page->num)
- page->items[i] = page->items[i+1];
- }
- if (found)
- page->num--;
- } else if (opt == CDB_PAGEINSERTOFF) {
- bool found = false;
- /* check already exist? */
- for(uint32_t i = 0; i < page->num; i++) {
- if (PHASHEQ(page->items[i].hash, phash)
- && OFFEQ(page->items[i].off, off)) {
- /* avoid exceptional deduplicated item */
- found = true;
- break;
- }
- }
-
- /* append to the tail */
- if (!found) {
- page->items[page->num].hash = phash;
- page->items[page->num].off = off;
- page->num++;
- /* records num is consistant with index */
- cdb_lock_lock(db->stlock);
- db->rnum++;
- cdb_lock_unlock(db->stlock);
- if (db->bf) {
- uint64_t bfkey = (((hash >> 24) % db->hsize) << 24) | (hash & 0xffffff);
- cdb_lock_lock(db->bflock);
- cdb_bf_set(db->bf, &bfkey, SI8);
- cdb_lock_unlock(db->bflock);
- }
- }
- }
-
- if (page->num == onum) {
- /* nothing done */
- if (pitem) {
- /* insert the item back to the cache where it belongs */
- cdb_lock_lock(tmpclock);
- cdb_ht_insert(tmpcache, pitem);
- cdb_lock_unlock(tmpclock);
- } else {
- if (page != (CDBPAGE *)sbuf2
- && page != (CDBPAGE *)sbuf)
- free(page);
- }
- if (locked == CDB_NOTLOCKED) cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
- return -1;
- } else {
- if (pitem) {
- cdb_lock_lock(db->dpclock);
- cdb_ht_insert(db->dpcache, pitem);
- cdb_lock_unlock(db->dpclock);
- } else {
- struct timespec ts;
- _cdb_timerreset(&ts);
- db->vio->wpage(db->vio, page, &off);
- db->wcount++;
- db->wtime += _cdb_timermicrosec(&ts);
-
- db->mtable[bid] = off;
- if (page != (CDBPAGE *)sbuf2
- && page != (CDBPAGE *)sbuf)
- free(page);
- }
- }
-
- if (locked == CDB_NOTLOCKED) cdb_lock_unlock(db->mlock[bid % MLOCKNUM]);
-
- /* check page cache overflow */
- if (PCOVERFLOW(db))
- _cdb_pageout(db);
-
- return 0;
-}
-
-
-/* check if an record with specified key-offset exists in index */
-bool cdb_checkoff(CDB *db, uint64_t hash, FOFF off, int locked)
-{
- FOFF soffs[SFOFFNUM];
- FOFF *soff = (FOFF *)soffs;
- int dupnum;
- int ret = false;
-
- /* get all possible offsets */
- dupnum = cdb_getoff(db, hash, &soff, locked);
- for(int i = 0; i < dupnum; i++) {
- if (OFFEQ(soff[i], off)) {
- ret = true;
- break;
- }
- }
-
- if (soff != (FOFF *)soffs) {
- free(soff);
- }
-
- return ret;
-}
-
-
-/* wrapper and simplified of set operation */
-int cdb_set(CDB *db, const char *key, int ksize, const char *val, int vsize)
-{
- return cdb_set2(db, key, ksize, val, vsize, CDB_OVERWRITE, 0);
-}
-
-
-int cdb_set2(CDB *db, const char *key, int ksize, const char *val, int vsize, int opt, int expire)
-{
- CDBREC rec;
- FOFF ooff, noff;
- uint32_t now = time(NULL);
- uint64_t hash;
- uint32_t lockid;
- bool expired = false;
-
- if (db->vio == NULL) {
- /* if it is a memdb, just operate on the record cache and return */
- cdb_lock_lock(db->rclock);
- cdb_ht_insert2(db->rcache, key, ksize, val, vsize);
- cdb_lock_unlock(db->rclock);
- if (RCOVERFLOW(db))
- _cdb_recout(db);
- return 0;
- }
-
- hash = CDBHASH64(key, ksize);
- lockid = (hash >> 24) % db->hsize % MLOCKNUM;
- OFFZERO(rec.ooff);
- OFFZERO(ooff);
- rec.osize = 0;
- rec.key = (char*)key;
- rec.val = (char*)val;
- rec.ksize = ksize;
- rec.vsize = vsize;
- rec.oid = cdb_genoid(db);
- rec.expire = expire? now + expire : 0;
-
- cdb_lock_lock(db->mlock[lockid]);
- cdb_lock_lock(db->rclock);
- if (db->rcache) {
- /* if record already exists, get its old meta info */
- int item_vsize;
- char *cval;
- uint32_t old_expire = 0;
- cval = cdb_ht_get(db->rcache, key, ksize, &item_vsize, false);
- if (cval) {
- /* record already exists */
- ooff = rec.ooff = *(FOFF*)cval;
- rec.osize = item_vsize - SFOFF - SI4;
- old_expire = *(uint32_t*)(cval + SFOFF);
- }
- if (old_expire && old_expire <= now)
- /* once exist but expired? */
- expired = true;
- }
- cdb_lock_unlock(db->rclock);
-
- if (OFFNULL(ooff)) {
- FOFF soffs[SFOFFNUM];
- FOFF *soff = soffs;
- char sbuf[SBUFSIZE];
- CDBREC *rrec = (CDBREC*)sbuf;
-
- int retnum;
- if ((retnum = cdb_getoff(db, hash, &soff, CDB_LOCKED)) < 0) {
- cdb_lock_unlock(db->mlock[lockid]);
- return -1;
- }
-
- for(int i = 0; i < retnum; i++) {
- /* check for duplicate records/older version*/
- int cret;
- if (rrec != (CDBREC*)sbuf) {
- free(rrec);
- rrec = (CDBREC*)sbuf;
- }
-
- struct timespec ts;
- _cdb_timerreset(&ts);
- cret = db->vio->rrec(db->vio, &rrec, soff[i], false);
- db->rcount++;
- db->rtime += _cdb_timermicrosec(&ts);
-
- if (cret < 0)
- continue;
-
- if ((uint32_t) ksize == rrec->ksize && memcmp(rrec->key, key, ksize) == 0) {
- /* got its old meta info */
- rec.osize = rrec->osize;
- rec.ooff = rrec->ooff;
- ooff = rec.ooff;
- if (rrec->expire <= now)
- expired = true;
- break;
- }
- }
- if (soff != soffs)
- free(soff);
- if (rrec != (CDBREC*)sbuf)
- free(rrec);
- }
-
- if (OFFNOTNULL(ooff) && !expired) {
- /* record already exists*/
- if (opt & CDB_INSERTIFNOEXIST) {
- cdb_lock_unlock(db->mlock[lockid]);
- cdb_seterrno(db, CDB_EXIST, __FILE__, __LINE__);
- return -2;
- }
- } else {
- if (opt & CDB_INSERTIFEXIST) {
- cdb_lock_unlock(db->mlock[lockid]);
- cdb_seterrno(db, CDB_NOTFOUND, __FILE__, __LINE__);
- return -3;
- }
- }
-
- struct timespec ts;
- _cdb_timerreset(&ts);
- if (db->vio->wrec(db->vio, &rec, &noff) < 0) {
- cdb_lock_unlock(db->mlock[lockid]);
- return -1;
- }
- db->wcount++;
- db->wtime += _cdb_timermicrosec(&ts);
-
- if (OFFNOTNULL(ooff)) {
- cdb_replaceoff(db, hash, ooff, noff, CDB_LOCKED);
- } else {
- cdb_updatepage(db, hash, noff, CDB_PAGEINSERTOFF, CDB_LOCKED);
- }
-
- if (db->rcache) {
- if ((opt & CDB_INSERTCACHE) == CDB_INSERTCACHE) {
- char *cval;
- CDBHTITEM *item = cdb_ht_newitem(db->rcache, ksize, vsize + SI4 + SFOFF);
- memcpy(cdb_ht_itemkey(db->rcache, item), key, ksize);
- cval = cdb_ht_itemval(db->rcache, item);
- memcpy(cval + SI4 + SFOFF, val, vsize);
- *(FOFF*)(cval) = rec.ooff;
- *(uint32_t*)(cval + SFOFF) = rec.expire;
- cdb_lock_lock(db->rclock);
- cdb_ht_insert(db->rcache, item);
- cdb_lock_unlock(db->rclock);
- }
- }
- cdb_lock_unlock(db->mlock[lockid]);
-
- if (RCOVERFLOW(db))
- _cdb_recout(db);
-
- cdb_seterrno(db, CDB_SUCCESS, __FILE__, __LINE__);
- return 0;
-}
-
-
-int cdb_is(CDB *db, const char *key, int ksize)
-{
- FOFF soffs[SFOFFNUM];
- FOFF *offs;
- int dupnum, ret = -3;
- uint64_t hash;
- //uint32_t now = time(NULL);
- uint32_t lockid;
-
- if (db->rcache) {
- char *cval;
- cdb_lock_lock(db->rclock);
- cval = cdb_ht_get(db->rcache, key, ksize, 0, true);
- if (cval) {
- db->rchit++;
- cdb_lock_unlock(db->rclock);
- return 0;
- } else {
- db->rcmiss++;
- if (db->vio == NULL) {
- cdb_lock_unlock(db->rclock);
- return -3;
- }
- }
- cdb_lock_unlock(db->rclock);
- }
-
- offs = soffs;
- hash = CDBHASH64(key, ksize);
- lockid = (hash >> 24) % db->hsize % MLOCKNUM;
- cdb_lock_lock(db->mlock[lockid]);
- dupnum = cdb_getoff(db, hash, &offs, CDB_LOCKED);
- if (dupnum <= 0) {
- cdb_lock_unlock(db->mlock[lockid]);
- return -1;
- }
- else
- ret = 0;
- cdb_lock_unlock(db->mlock[lockid]);
-
- if (RCOVERFLOW(db))
- _cdb_recout(db);
-
- if (offs != soffs)
- free(offs);
-
- if (ret < 0)
- cdb_seterrno(db, CDB_NOTFOUND, __FILE__, __LINE__);
- else {
- db->rcmiss++;
- cdb_seterrno(db, CDB_SUCCESS, __FILE__, __LINE__);
- }
- return ret;
-}
-
-
-int cdb_get(CDB *db, const char *key, int ksize, void **val, int *vsize)
-{
- char sbuf[SBUFSIZE];
- CDBREC *rec = (CDBREC *)sbuf;
- FOFF soffs[SFOFFNUM];
- FOFF *offs;
- int dupnum, ret = -3;
- uint64_t hash;
- uint32_t now = time(NULL);
- uint32_t lockid;
-
- *vsize = 0;
- *val = NULL;
- cdb_lock_lock(db->rclock);
- if (db->rcache) {
- char *cval;
- cval = cdb_ht_get(db->rcache, key, ksize, vsize, true);
- if (cval) {
- db->rchit++;
- if (db->vio) {
- (*vsize) -= SI4 + SFOFF;
- if (*(uint32_t*)(cval + SFOFF)
- && *(uint32_t*)(cval + SFOFF) <= now) {
- cdb_lock_unlock(db->rclock);
- /* not found no not report error now */
- //cdb_seterrno(db, CDB_NOTFOUND, __FILE__, __LINE__);
- return -3;
- }
- cval = (void*)(cval + SI4 + SFOFF);
- }
- *val = malloc(*vsize);
- memcpy(*val, cval, *vsize);
- cdb_lock_unlock(db->rclock);
- return 0;
- } else {
- db->rcmiss++;
- if (db->vio == NULL) {
- cdb_lock_unlock(db->rclock);
- return -3;
- }
- }
- }
- cdb_lock_unlock(db->rclock);
-
- offs = soffs;
- hash = CDBHASH64(key, ksize);
- lockid = (hash >> 24) % db->hsize % MLOCKNUM;
- cdb_lock_lock(db->mlock[lockid]);
- dupnum = cdb_getoff(db, hash, &offs, CDB_LOCKED);
- if (dupnum < 0) {
- cdb_lock_unlock(db->mlock[lockid]);
- return -1;
- }
-
- for(int i = 0; i < dupnum; i++) {
- int cret;
- if (rec != (CDBREC*)sbuf) {
- free(rec);
- rec = (CDBREC*)sbuf;
- }
-
- struct timespec ts;
- _cdb_timerreset(&ts);
- cret = db->vio->rrec(db->vio, &rec, offs[i], true);
- db->rcount++;
- db->rtime += _cdb_timermicrosec(&ts);
-
- if (cret < 0)
- continue;
-
- if ((uint32_t) ksize == rec->ksize && memcmp(rec->key, key, ksize) == 0) {
- if (rec->expire && rec->expire <= now) {
- break;
- }
- *vsize = rec->vsize;
- *val = malloc(*vsize);
- memcpy(*val, rec->val, *vsize);
- ret = 0;
- break;
- }
- }
-
- if (ret == 0 && db->rcache) {
- char *cval;
- CDBHTITEM *item = cdb_ht_newitem(db->rcache, ksize, *vsize + SI4 + SFOFF);
- memcpy(cdb_ht_itemkey(db->rcache, item), key, ksize);
- cval = cdb_ht_itemval(db->rcache, item);
- memcpy(cval + SI4 + SFOFF, *val, *vsize);
- *(FOFF*)(cval) = rec->ooff;
- *(uint32_t*)(cval + SFOFF) = rec->expire;
- cdb_lock_lock(db->rclock);
- cdb_ht_insert(db->rcache, item);
- cdb_lock_unlock(db->rclock);
- }
- cdb_lock_unlock(db->mlock[lockid]);
-
- if (RCOVERFLOW(db))
- _cdb_recout(db);
-
- if (offs != soffs)
- free(offs);
-
- if (rec != (CDBREC*)sbuf)
- free(rec);
-
- if (ret < 0)
- cdb_seterrno(db, CDB_NOTFOUND, __FILE__, __LINE__);
- else {
- db->rcmiss++;
- cdb_seterrno(db, CDB_SUCCESS, __FILE__, __LINE__);
- }
- return ret;
-}
-
-
-void cdb_free_val(void **val)
-{
- if (*val)
- free(*val);
- *val = NULL;
-}
-
-
-int cdb_del(CDB *db, const char *key, int ksize)
-{
- FOFF ooff;
- CDBREC rec;
- uint32_t lockid;
- uint64_t hash;
-
- OFFZERO(rec.ooff);
- OFFZERO(ooff);
- rec.osize = 0;
- rec.key = (char*)key;
- rec.ksize = ksize;
- rec.val = NULL;
- rec.vsize = 0;
-
- if (db->vio == NULL) {
- /* if it is a memdb, just operate on the record cache and return */
- cdb_lock_lock(db->rclock);
- cdb_ht_del2(db->rcache, key, ksize);
- cdb_lock_unlock(db->rclock);
- if (RCOVERFLOW(db))
- _cdb_recout(db);
- return 0;
- }
-
- hash = CDBHASH64(key, ksize);
- lockid = (hash >> 24) % db->hsize % MLOCKNUM;
- cdb_lock_lock(db->mlock[lockid]);
- if (db->rcache) {
- /* if record already exists, get its old meta info */
- CDBHTITEM *item;
- cdb_lock_lock(db->rclock);
- item = cdb_ht_del(db->rcache, key, ksize);
- cdb_lock_unlock(db->rclock);
- if (item) {
- char *cval = cdb_ht_itemval(db->rcache, item);
- ooff = rec.ooff = *(FOFF*)cval;
- rec.osize = item->vsize - SFOFF - SI4;
- rec.expire = *(uint32_t*)(cval + SFOFF);
- free(item);
- }
- }
-
- if (OFFNULL(ooff)) {
- FOFF soffs[SFOFFNUM];
- FOFF *soff = soffs;
- char sbuf[SBUFSIZE];
- CDBREC *rrec = (CDBREC*)sbuf;
-
- int retnum;
- if ((retnum = cdb_getoff(db, hash, &soff, CDB_LOCKED)) < 0) {
- cdb_lock_unlock(db->mlock[lockid]);
- return -1;
- }
-
- for(int i = 0; i < retnum; i++) {
- /* check for duplicate records/older version*/
- int cret;
- if (rrec != (CDBREC*)sbuf) {
- free(rrec);
- rrec = (CDBREC*)sbuf;
- }
-
- struct timespec ts;
- _cdb_timerreset(&ts);
- cret = db->vio->rrec(db->vio, &rrec, soff[i], false);
- db->rcount++;
- db->rtime += _cdb_timermicrosec(&ts);
-
- if (cret < 0)
- continue;
-
- if ((uint32_t) ksize == rrec->ksize && memcmp(rrec->key, key, ksize) == 0) {
- /* got its old meta info */
- rec.osize = rrec->osize;
- rec.ooff = rrec->ooff;
- ooff = rec.ooff;
- break;
- }
- }
- if (soff != soffs)
- free(soff);
- if (rrec != (CDBREC*)sbuf)
- free(rrec);
- }
-
- if (OFFNOTNULL(ooff)) {
- cdb_updatepage(db, hash, ooff, CDB_PAGEDELETEOFF, CDB_LOCKED);
- cdb_lock_unlock(db->mlock[lockid]);
-
- struct timespec ts;
- _cdb_timerreset(&ts);
- db->vio->drec(db->vio, &rec, ooff);
- //if ( < 0)
- // return -1; succeed or not doesn't matter
- db->wcount++;
- db->wtime += _cdb_timermicrosec(&ts);
- cdb_seterrno(db, CDB_SUCCESS, __FILE__, __LINE__);
- return 0;
- } else {
- cdb_lock_unlock(db->mlock[lockid]);
- cdb_seterrno(db, CDB_NOTFOUND, __FILE__, __LINE__);
- return -3;
- }
-}
-
-
-void cdb_stat(CDB *db, CDBSTAT *stat)
-{
- if (stat == NULL) {
- db->rchit = db->rcmiss = 0;
- db->pchit = db->pcmiss = 0;
- db->rcount = db->rtime = 0;
- db->wcount = db->wtime = 0;
- } else {
- stat->rnum = db->rnum;
- stat->rcnum = db->rcache? db->rcache->num : 0;
- stat->pnum = db->hsize;
- stat->pcnum = (db->pcache? db->pcache->num : 0)
- + (db->dpcache? db->dpcache->num : 0);
- stat->rchit = db->rchit;
- stat->rcmiss = db->rcmiss;
- stat->pchit = db->pchit;
- stat->pcmiss = db->pcmiss;
- stat->rlatcy = db->rcount ? db->rtime / db->rcount : 0;
- stat->wlatcy = db->wcount ? db->wtime / db->wcount : 0;
- }
-}
-
-
-int cdb_close(CDB *db)
-{
- if (!db->opened)
- return -1;
-
- if (db->bgtask)
- cdb_bgtask_stop(db->bgtask);
- if (db->rcache)
- cdb_ht_destroy(db->rcache);
- if (db->pcache)
- cdb_ht_destroy(db->pcache);
- if (db->dpcache) {
- cdb_flushalldpage(db);
- cdb_ht_destroy(db->dpcache);
- }
-
- if (db->vio) {
- db->vio->whead(db->vio);
- db->vio->close(db->vio);
- cdb_vio_destroy(db->vio);
- }
- if (db->mtable)
- free(db->mtable);
- db->opened = false;
- _cdb_defparam(db);
- return 0;
-}
-
-
-void cdb_deferrorcb(void *arg, int errno, const char *file, int line)
-{
- (void) arg;
- fprintf(stderr, "DBERR: [%s:%d] %d - %s\n", file, line, errno, cdb_errmsg(errno));
-}
-
-
-int cdb_destroy(CDB *db)
-{
- if (db->opened)
- cdb_close(db);
- for(int i = 0; i < MLOCKNUM; i++)
- cdb_lock_destory(db->mlock[i]);
- cdb_lock_destory(db->dpclock);
- cdb_lock_destory(db->pclock);
- cdb_lock_destory(db->rclock);
- cdb_lock_destory(db->stlock);
- cdb_lock_destory(db->oidlock);
- cdb_lock_destory(db->bflock);
- cdb_bgtask_destroy(db->bgtask);
- pthread_key_delete(*(pthread_key_t*)db->errkey);
- free(db->errkey);
- free(db);
- return 0;
-}
-
-
-
diff --git a/3rdparty/cuttdb/src/cdb_core.h b/3rdparty/cuttdb/src/cdb_core.h
deleted file mode 100644
index bcd4ad9a6bef0d43993781f3707ed0f44935cae2..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_core.h
+++ /dev/null
@@ -1,122 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CDB_CORE_H_
-#define _CDB_CORE_H_
-#include "cuttdb.h"
-#include "cdb_types.h"
-#include "cdb_hashtable.h"
-#include "cdb_bloomfilter.h"
-#include "cdb_lock.h"
-#include "cdb_vio.h"
-#include "cdb_bgtask.h"
-#include <stdint.h>
-#include <stdbool.h>
-
-
-enum {
- CDB_PAGEDELETEOFF = 0,
- CDB_PAGEINSERTOFF = 1,
-};
-
-/* the DB object */
-struct CDB
-{
- /* size limit for record cache */
- uint64_t rclimit;
- /* size limit for index page cache */
- uint64_t pclimit;
- /* size of bloom filter */
- uint64_t bfsize;
- /* record number in db */
- uint64_t rnum;
- /* always increment operation id */
- uint64_t oid;
- /* recovery point oid */
- uint64_t roid;
- /* hash table size */
- uint32_t hsize;
- /* last timestamp of no dirty page state */
- uint32_t ndpltime;
- /* currently the database opened or not */
- bool opened;
- /* the size for a disk seek&read, should not greater than SBUFSIZE */
- uint32_t areadsize;
-
- /* record cache */
- CDBHASHTABLE *rcache;
- /* (clean) index page cache */
- CDBHASHTABLE *pcache;
- /* dirty index page cache */
- CDBHASHTABLE *dpcache;
- /* Bloom Filter */
- CDBBLOOMFILTER *bf;
-
- /* lock for rcache */
- CDBLOCK *rclock;
- /* lock for pcache */
- CDBLOCK *pclock;
- /* lock for dpcache */
- CDBLOCK *dpclock;
- /* lock for hash table operation, split to MLOCKNUM groups */
- CDBLOCK *mlock[MLOCKNUM];
- /* lock for statistic */
- CDBLOCK *stlock;
- /* lock for operation id */
- CDBLOCK *oidlock;
- /* lock for bloom filter */
- CDBLOCK *bflock;
- /* background tasks in another thread */
- CDBBGTASK *bgtask;
-
- /* main hash table, contains 'hsize' elements */
- FOFF *mtable;
- /* disk i/o layer object */
- CDBVIO *vio;
-
- /* callback function when error occurs */
- CDB_ERRCALLBACK errcb;
- /* argument for callback function */
- void *errcbarg;
- /* key to get error code in current thread */
- void *errkey;
-
- /* statistics below, this fields have no lock protection */
- /* record cache hit/miss */
- uint64_t rchit;
- uint64_t rcmiss;
- /* page cache hit/miss */
- uint64_t pchit;
- uint64_t pcmiss;
- /* cumulative disk read time */
- uint64_t rtime;
- /* number of disk read operation */
- uint64_t rcount;
- /* cumulative disk write time */
- uint64_t wtime;
- /* number of disk write operation */
- uint64_t wcount;
-};
-
-
-bool cdb_checkoff(CDB *db, uint64_t hash, FOFF off, int locked);
-int cdb_getoff(CDB *db, uint64_t hash, FOFF **offs, int locked);
-int cdb_replaceoff(CDB *db, uint64_t hash, FOFF off, FOFF noff, int locked);
-int cdb_updatepage(CDB *db, uint64_t hash, FOFF off, int opt, int locked);
-void cdb_flushalldpage(CDB *db);
-uint64_t cdb_genoid(CDB *db);
-
-#endif
-
diff --git a/3rdparty/cuttdb/src/cdb_crc64.c b/3rdparty/cuttdb/src/cdb_crc64.c
deleted file mode 100644
index 6c72eb73fb3b8aaf774cd0e87479fc0fe82c580b..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_crc64.c
+++ /dev/null
@@ -1,170 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-/**************************************************************
-* *
-* Fichier : crc64.c *
-* Fonction pour calculer le CRC64 *
-* *
-**************************************************************/
-#include "cdb_crc64.h"
-
-
-#define CONST64(n) (n##ULL)
-static uint64_t CRC64_Table[256] =
-{
- CONST64(0x0000000000000000), CONST64(0x42f0e1eba9ea3693),
- CONST64(0x85e1c3d753d46d26), CONST64(0xc711223cfa3e5bb5),
- CONST64(0x493366450e42ecdf), CONST64(0x0bc387aea7a8da4c),
- CONST64(0xccd2a5925d9681f9), CONST64(0x8e224479f47cb76a),
- CONST64(0x9266cc8a1c85d9be), CONST64(0xd0962d61b56fef2d),
- CONST64(0x17870f5d4f51b498), CONST64(0x5577eeb6e6bb820b),
- CONST64(0xdb55aacf12c73561), CONST64(0x99a54b24bb2d03f2),
- CONST64(0x5eb4691841135847), CONST64(0x1c4488f3e8f96ed4),
- CONST64(0x663d78ff90e185ef), CONST64(0x24cd9914390bb37c),
- CONST64(0xe3dcbb28c335e8c9), CONST64(0xa12c5ac36adfde5a),
- CONST64(0x2f0e1eba9ea36930), CONST64(0x6dfeff5137495fa3),
- CONST64(0xaaefdd6dcd770416), CONST64(0xe81f3c86649d3285),
- CONST64(0xf45bb4758c645c51), CONST64(0xb6ab559e258e6ac2),
- CONST64(0x71ba77a2dfb03177), CONST64(0x334a9649765a07e4),
- CONST64(0xbd68d2308226b08e), CONST64(0xff9833db2bcc861d),
- CONST64(0x388911e7d1f2dda8), CONST64(0x7a79f00c7818eb3b),
- CONST64(0xcc7af1ff21c30bde), CONST64(0x8e8a101488293d4d),
- CONST64(0x499b3228721766f8), CONST64(0x0b6bd3c3dbfd506b),
- CONST64(0x854997ba2f81e701), CONST64(0xc7b97651866bd192),
- CONST64(0x00a8546d7c558a27), CONST64(0x4258b586d5bfbcb4),
- CONST64(0x5e1c3d753d46d260), CONST64(0x1cecdc9e94ace4f3),
- CONST64(0xdbfdfea26e92bf46), CONST64(0x990d1f49c77889d5),
- CONST64(0x172f5b3033043ebf), CONST64(0x55dfbadb9aee082c),
- CONST64(0x92ce98e760d05399), CONST64(0xd03e790cc93a650a),
- CONST64(0xaa478900b1228e31), CONST64(0xe8b768eb18c8b8a2),
- CONST64(0x2fa64ad7e2f6e317), CONST64(0x6d56ab3c4b1cd584),
- CONST64(0xe374ef45bf6062ee), CONST64(0xa1840eae168a547d),
- CONST64(0x66952c92ecb40fc8), CONST64(0x2465cd79455e395b),
- CONST64(0x3821458aada7578f), CONST64(0x7ad1a461044d611c),
- CONST64(0xbdc0865dfe733aa9), CONST64(0xff3067b657990c3a),
- CONST64(0x711223cfa3e5bb50), CONST64(0x33e2c2240a0f8dc3),
- CONST64(0xf4f3e018f031d676), CONST64(0xb60301f359dbe0e5),
- CONST64(0xda050215ea6c212f), CONST64(0x98f5e3fe438617bc),
- CONST64(0x5fe4c1c2b9b84c09), CONST64(0x1d14202910527a9a),
- CONST64(0x93366450e42ecdf0), CONST64(0xd1c685bb4dc4fb63),
- CONST64(0x16d7a787b7faa0d6), CONST64(0x5427466c1e109645),
- CONST64(0x4863ce9ff6e9f891), CONST64(0x0a932f745f03ce02),
- CONST64(0xcd820d48a53d95b7), CONST64(0x8f72eca30cd7a324),
- CONST64(0x0150a8daf8ab144e), CONST64(0x43a04931514122dd),
- CONST64(0x84b16b0dab7f7968), CONST64(0xc6418ae602954ffb),
- CONST64(0xbc387aea7a8da4c0), CONST64(0xfec89b01d3679253),
- CONST64(0x39d9b93d2959c9e6), CONST64(0x7b2958d680b3ff75),
- CONST64(0xf50b1caf74cf481f), CONST64(0xb7fbfd44dd257e8c),
- CONST64(0x70eadf78271b2539), CONST64(0x321a3e938ef113aa),
- CONST64(0x2e5eb66066087d7e), CONST64(0x6cae578bcfe24bed),
- CONST64(0xabbf75b735dc1058), CONST64(0xe94f945c9c3626cb),
- CONST64(0x676dd025684a91a1), CONST64(0x259d31cec1a0a732),
- CONST64(0xe28c13f23b9efc87), CONST64(0xa07cf2199274ca14),
- CONST64(0x167ff3eacbaf2af1), CONST64(0x548f120162451c62),
- CONST64(0x939e303d987b47d7), CONST64(0xd16ed1d631917144),
- CONST64(0x5f4c95afc5edc62e), CONST64(0x1dbc74446c07f0bd),
- CONST64(0xdaad56789639ab08), CONST64(0x985db7933fd39d9b),
- CONST64(0x84193f60d72af34f), CONST64(0xc6e9de8b7ec0c5dc),
- CONST64(0x01f8fcb784fe9e69), CONST64(0x43081d5c2d14a8fa),
- CONST64(0xcd2a5925d9681f90), CONST64(0x8fdab8ce70822903),
- CONST64(0x48cb9af28abc72b6), CONST64(0x0a3b7b1923564425),
- CONST64(0x70428b155b4eaf1e), CONST64(0x32b26afef2a4998d),
- CONST64(0xf5a348c2089ac238), CONST64(0xb753a929a170f4ab),
- CONST64(0x3971ed50550c43c1), CONST64(0x7b810cbbfce67552),
- CONST64(0xbc902e8706d82ee7), CONST64(0xfe60cf6caf321874),
- CONST64(0xe224479f47cb76a0), CONST64(0xa0d4a674ee214033),
- CONST64(0x67c58448141f1b86), CONST64(0x253565a3bdf52d15),
- CONST64(0xab1721da49899a7f), CONST64(0xe9e7c031e063acec),
- CONST64(0x2ef6e20d1a5df759), CONST64(0x6c0603e6b3b7c1ca),
- CONST64(0xf6fae5c07d3274cd), CONST64(0xb40a042bd4d8425e),
- CONST64(0x731b26172ee619eb), CONST64(0x31ebc7fc870c2f78),
- CONST64(0xbfc9838573709812), CONST64(0xfd39626eda9aae81),
- CONST64(0x3a28405220a4f534), CONST64(0x78d8a1b9894ec3a7),
- CONST64(0x649c294a61b7ad73), CONST64(0x266cc8a1c85d9be0),
- CONST64(0xe17dea9d3263c055), CONST64(0xa38d0b769b89f6c6),
- CONST64(0x2daf4f0f6ff541ac), CONST64(0x6f5faee4c61f773f),
- CONST64(0xa84e8cd83c212c8a), CONST64(0xeabe6d3395cb1a19),
- CONST64(0x90c79d3fedd3f122), CONST64(0xd2377cd44439c7b1),
- CONST64(0x15265ee8be079c04), CONST64(0x57d6bf0317edaa97),
- CONST64(0xd9f4fb7ae3911dfd), CONST64(0x9b041a914a7b2b6e),
- CONST64(0x5c1538adb04570db), CONST64(0x1ee5d94619af4648),
- CONST64(0x02a151b5f156289c), CONST64(0x4051b05e58bc1e0f),
- CONST64(0x87409262a28245ba), CONST64(0xc5b073890b687329),
- CONST64(0x4b9237f0ff14c443), CONST64(0x0962d61b56fef2d0),
- CONST64(0xce73f427acc0a965), CONST64(0x8c8315cc052a9ff6),
- CONST64(0x3a80143f5cf17f13), CONST64(0x7870f5d4f51b4980),
- CONST64(0xbf61d7e80f251235), CONST64(0xfd913603a6cf24a6),
- CONST64(0x73b3727a52b393cc), CONST64(0x31439391fb59a55f),
- CONST64(0xf652b1ad0167feea), CONST64(0xb4a25046a88dc879),
- CONST64(0xa8e6d8b54074a6ad), CONST64(0xea16395ee99e903e),
- CONST64(0x2d071b6213a0cb8b), CONST64(0x6ff7fa89ba4afd18),
- CONST64(0xe1d5bef04e364a72), CONST64(0xa3255f1be7dc7ce1),
- CONST64(0x64347d271de22754), CONST64(0x26c49cccb40811c7),
- CONST64(0x5cbd6cc0cc10fafc), CONST64(0x1e4d8d2b65facc6f),
- CONST64(0xd95caf179fc497da), CONST64(0x9bac4efc362ea149),
- CONST64(0x158e0a85c2521623), CONST64(0x577eeb6e6bb820b0),
- CONST64(0x906fc95291867b05), CONST64(0xd29f28b9386c4d96),
- CONST64(0xcedba04ad0952342), CONST64(0x8c2b41a1797f15d1),
- CONST64(0x4b3a639d83414e64), CONST64(0x09ca82762aab78f7),
- CONST64(0x87e8c60fded7cf9d), CONST64(0xc51827e4773df90e),
- CONST64(0x020905d88d03a2bb), CONST64(0x40f9e43324e99428),
- CONST64(0x2cffe7d5975e55e2), CONST64(0x6e0f063e3eb46371),
- CONST64(0xa91e2402c48a38c4), CONST64(0xebeec5e96d600e57),
- CONST64(0x65cc8190991cb93d), CONST64(0x273c607b30f68fae),
- CONST64(0xe02d4247cac8d41b), CONST64(0xa2dda3ac6322e288),
- CONST64(0xbe992b5f8bdb8c5c), CONST64(0xfc69cab42231bacf),
- CONST64(0x3b78e888d80fe17a), CONST64(0x7988096371e5d7e9),
- CONST64(0xf7aa4d1a85996083), CONST64(0xb55aacf12c735610),
- CONST64(0x724b8ecdd64d0da5), CONST64(0x30bb6f267fa73b36),
- CONST64(0x4ac29f2a07bfd00d), CONST64(0x08327ec1ae55e69e),
- CONST64(0xcf235cfd546bbd2b), CONST64(0x8dd3bd16fd818bb8),
- CONST64(0x03f1f96f09fd3cd2), CONST64(0x41011884a0170a41),
- CONST64(0x86103ab85a2951f4), CONST64(0xc4e0db53f3c36767),
- CONST64(0xd8a453a01b3a09b3), CONST64(0x9a54b24bb2d03f20),
- CONST64(0x5d45907748ee6495), CONST64(0x1fb5719ce1045206),
- CONST64(0x919735e51578e56c), CONST64(0xd367d40ebc92d3ff),
- CONST64(0x1476f63246ac884a), CONST64(0x568617d9ef46bed9),
- CONST64(0xe085162ab69d5e3c), CONST64(0xa275f7c11f7768af),
- CONST64(0x6564d5fde549331a), CONST64(0x279434164ca30589),
- CONST64(0xa9b6706fb8dfb2e3), CONST64(0xeb46918411358470),
- CONST64(0x2c57b3b8eb0bdfc5), CONST64(0x6ea7525342e1e956),
- CONST64(0x72e3daa0aa188782), CONST64(0x30133b4b03f2b111),
- CONST64(0xf7021977f9cceaa4), CONST64(0xb5f2f89c5026dc37),
- CONST64(0x3bd0bce5a45a6b5d), CONST64(0x79205d0e0db05dce),
- CONST64(0xbe317f32f78e067b), CONST64(0xfcc19ed95e6430e8),
- CONST64(0x86b86ed5267cdbd3), CONST64(0xc4488f3e8f96ed40),
- CONST64(0x0359ad0275a8b6f5), CONST64(0x41a94ce9dc428066),
- CONST64(0xcf8b0890283e370c), CONST64(0x8d7be97b81d4019f),
- CONST64(0x4a6acb477bea5a2a), CONST64(0x089a2aacd2006cb9),
- CONST64(0x14dea25f3af9026d), CONST64(0x562e43b4931334fe),
- CONST64(0x913f6188692d6f4b), CONST64(0xd3cf8063c0c759d8),
- CONST64(0x5dedc41a34bbeeb2), CONST64(0x1f1d25f19d51d821),
- CONST64(0xd80c07cd676f8394), CONST64(0x9afce626ce85b507)
-};
-
-
-uint64_t cdb_crc64(const void *buf, uint32_t len)
-{
- uint32_t i;
- uint64_t crc = 0xFFFFFFFFFFFFFFFF;
- uint8_t *cbuf = (uint8_t *)buf;
-
- for (i = 0; i < len; i++) {
- crc = CRC64_Table[(uint8_t)(crc >> 56) ^ *cbuf++] ^ (crc << 8);
- }
- return crc;
-}
-
diff --git a/3rdparty/cuttdb/src/cdb_crc64.h b/3rdparty/cuttdb/src/cdb_crc64.h
deleted file mode 100644
index 50744fc844afe84cdcef8ddba5f6cff81ae5599a..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_crc64.h
+++ /dev/null
@@ -1,22 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CDB_CRC64_H_
-#define _CDB_CRC64_H_
-#include <stdint.h>
-
-uint64_t cdb_crc64(const void *buf, uint32_t len);
-
-#endif
diff --git a/3rdparty/cuttdb/src/cdb_dumpdb.c b/3rdparty/cuttdb/src/cdb_dumpdb.c
deleted file mode 100644
index 99cddbb5355cb44d235ef08b27234a7199c2fe67..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_dumpdb.c
+++ /dev/null
@@ -1,68 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-#include "cuttdb.h"
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <stdint.h>
-#include <stdbool.h>
-
-
-bool itcb(void *arg, const char *key, int ksize, const char *val, int vsize, uint32_t expire, uint64_t oid)
-{
-#define SBUFSIZE 4096
- char buf[SBUFSIZE];
- char *kvbuf = buf;
- if (ksize + vsize + 2 > SBUFSIZE)
- kvbuf = (char*)malloc(ksize + vsize + 2);
- memcpy(kvbuf, key, ksize);
- kvbuf[ksize] = '\t';
- memcpy(kvbuf + ksize + 1, val, vsize);
- kvbuf[ksize + vsize + 1] = '\0';
- printf("%s\t%u\n", kvbuf, expire);
- if (kvbuf != buf)
- free(kvbuf);
- return true;
-}
-
-int main(int argc, char *argv[])
-{
- /* 1TB */
- int cache_limit = 1048576;
-
- if (argc < 2) {
- fprintf(stderr, "Usage: %s dbpath [cachelimit(MB)].... \n", argv[0]);
- return -1;
- }
- if (argc > 2) {
- cache_limit = atoi(argv[2]);
- }
-
- CDB *db = cdb_new();
- cdb_option(db, 0, 0, cache_limit);
- if (cdb_open(db, argv[1], CDB_PAGEWARMUP) < 0) {
- fprintf(stderr, "Database open error, unable to recovery\n");
- return -1;
- }
- void *it = cdb_iterate_new(db, 0);
- cdb_iterate(db, itcb, NULL, it);
- cdb_iterate_destroy(db, it);
- cdb_destroy(db);
-}
-
-
-
-
-
diff --git a/3rdparty/cuttdb/src/cdb_dumpraw.c b/3rdparty/cuttdb/src/cdb_dumpraw.c
deleted file mode 100644
index 53bbe11c6e7bec1723c97fb951b63468889d555a..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_dumpraw.c
+++ /dev/null
@@ -1,115 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <sys/stat.h>
-#include <sys/types.h>
-#include <unistd.h>
-#include <fcntl.h>
-#include <sys/mman.h>
-#include <stdint.h>
-
-#define SI4 4
-#define SI8 8
-
-/* data record */
-typedef struct {
- /* disk store starts at following field */
- uint32_t magic;
- uint32_t ksize;
- uint32_t vsize;
- uint32_t expire;
- uint64_t oid;
- char buf[0];
-} __attribute__((packed)) CDBREC;
-
-/* real size of a record header when stored on disk */
-#define RECHSIZE (SI4 * 4 + SI8)
-/* real size of a record when stored on disk */
-#define RECSIZE(r) (RECHSIZE + (r)->ksize + (r)->vsize)
-
-#define FILEMETASIZE 64
-#define ALIGNBYTES 16
-#define RECMAGIC 0x19871022
-#define DELRECMAGIC 0x19871023
-#define FILEMAGICHEADER "CuTtDbFiLePaRtIaL"
-#define FILEMAGICLEN (strlen(FILEMAGICHEADER))
-#define OFFALIGNED(off) (((off) & (ALIGNBYTES - 1))? ((off) | (ALIGNBYTES - 1)) + 1: off)
-
-
-
-void process(const char *filename)
-{
-#define SBUFSIZE 4096
- int fd = open(filename, O_RDONLY, 0644);
- char buf[SBUFSIZE];
- if (fd < 0)
- fprintf(stderr, "%s Open failed\n", filename);
-
- long filesize = lseek(fd, 0, SEEK_END);
- long pos = FILEMETASIZE;
- char *map = (char*)mmap(NULL, filesize, PROT_READ, MAP_SHARED, fd, 0);
- if (memcmp(map, FILEMAGICHEADER, FILEMAGICLEN)) {
- fprintf(stderr, "%s is not a cuttdb file\n", filename);
- close(fd);
- return;
- }
-
- while(pos < filesize) {
- char *kvbuf = buf;
- CDBREC *rec = (CDBREC*)&map[pos];
- if (rec->magic != RECMAGIC && rec->magic != DELRECMAGIC) {
- pos += ALIGNBYTES;
- continue;
- }
-
- pos += OFFALIGNED(RECSIZE(rec));
- if (rec->magic != RECMAGIC)
- continue;
-
- if (rec->ksize + rec->vsize + 2 > SBUFSIZE) {
- kvbuf = (char*)malloc(rec->ksize + rec->vsize + 2);
- }
- memcpy(kvbuf, rec->buf, rec->ksize);
- kvbuf[rec->ksize] = '\t';
- memcpy(kvbuf + rec->ksize + 1, rec->buf + rec->ksize, rec->vsize);
- kvbuf[rec->ksize + rec->vsize + 1] = '\0';
- printf("%s\t%u\n", kvbuf, rec->expire);
- if (kvbuf != buf)
- free(kvbuf);
- }
-
- munmap(map, filesize);
- close(fd);
-}
-
-
-
-
-int main(int argc, char *argv[])
-{
- if (argc < 2) {
- fprintf(stderr, "Usage: %s dat########.cdb dat########.cdb .... \n", argv[0]);
- return 0;
- }
- for(int i = 1; i < argc; i++)
- process(argv[i]);
- return 0;
-}
-
-
-
-
diff --git a/3rdparty/cuttdb/src/cdb_errno.c b/3rdparty/cuttdb/src/cdb_errno.c
deleted file mode 100644
index 0e9765acce186a2a0dc3c83260e6c10d884b613a..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_errno.c
+++ /dev/null
@@ -1,78 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cuttdb.h"
-#include "cdb_errno.h"
-#include "cdb_types.h"
-#include "cdb_core.h"
-#include <pthread.h>
-
-
-int cdb_errno(CDB *db)
-{
- return (size_t)pthread_getspecific(*(pthread_key_t*)db->errkey);
-}
-
-const char *cdb_errmsg(int ecode)
-{
- switch(ecode) {
- case CDB_SUCCESS:
- return "Success";
- case CDB_NOTFOUND:
- return "Key Not Found";
- case CDB_EXIST:
- return "Item Already Exists";
- case CDB_DIRNOEXIST:
- return "Path Open Failed";
- case CDB_OPENERR:
- return "File Open Failed";
- case CDB_PIDEXIST:
- return "Opened By Another Process";
- case CDB_DATAERRDAT:
- return "Data File Content Error";
- case CDB_DATAERRIDX:
- return "Index File Content Error";
- case CDB_WRITEERR:
- return "Write To File Error";
- case CDB_READERR:
- return "Read From File Error";
- case CDB_NOFID:
- return "Internal File Lost";
- case CDB_INTERNALERR:
- return "Internal Error";
- case CDB_DATAERRMETA:
- return "File Header Error";
- case CDB_MEMDBNOCACHE:
- return "MemDB Mode With Zero Record Cache Size";
- default:
- return "Error For Errno";
- }
-}
-
-
-void cdb_seterrcb(CDB *db, CDB_ERRCALLBACK errcb, void *arg)
-{
- db->errcb = errcb;
- db->errcbarg = arg;
-}
-
-
-void cdb_seterrno(CDB *db, int ecode, const char *source, int line)
-{
- pthread_setspecific(*(pthread_key_t*)db->errkey, (void*)(size_t)ecode);
- if (ecode != CDB_SUCCESS && db->errcb) {
- db->errcb(db->errcbarg, ecode, source, line);
- }
-}
diff --git a/3rdparty/cuttdb/src/cdb_errno.h b/3rdparty/cuttdb/src/cdb_errno.h
deleted file mode 100644
index f274819de73b2133d2648aa6490ea8f5cf66b41c..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_errno.h
+++ /dev/null
@@ -1,22 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CDB_ERRNO_H_
-#define _CDB_ERRNO_H_
-
-void cdb_seterrno(CDB *db, int ecode, const char *source, int line);
-
-#endif
-
diff --git a/3rdparty/cuttdb/src/cdb_hashtable.c b/3rdparty/cuttdb/src/cdb_hashtable.c
deleted file mode 100644
index 00d1a640c7657ddc8569c8071d14b90c7e736504..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_hashtable.c
+++ /dev/null
@@ -1,541 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cdb_hashtable.h"
-#include <stdlib.h>
-#include <string.h>
-
-/*
-#define LRUPREV(i) (*(CDBHTITEM**)&((i)->buf[0]))
-#define LRUNEXT(i) (*(CDBHTITEM**)&((i)->buf[sizeof(void*)]))
-*/
-
-#define LRUPREV(i) ((i)->lruptr[0])
-#define LRUNEXT(i) ((i)->lruptr[1])
-
-static uint32_t MurmurHash1( const void * key, int len)
-{
- const unsigned int m = 0xc6a4a793;
- const int r = 16;
- unsigned int h = 0x19900917 ^ (len * m);
- const unsigned char * data = (const unsigned char *)key;
-
- while(len >= 4)
- {
- unsigned int k = *(unsigned int *)data;
- h += k; h *= m; h ^= h >> 16;
- data += 4; len -= 4;
- }
-
- switch(len)
- {
- case 3:
- h += data[2] << 16;
- case 2:
- h += data[1] << 8;
- case 1:
- h += data[0];
- h *= m;
- h ^= h >> r;
- };
-
- h *= m; h ^= h >> 10;
- h *= m; h ^= h >> 17;
- return h;
-}
-
-void *cdb_ht_itemkey(CDBHASHTABLE *ht, CDBHTITEM *item)
-{
- return (void *)(item->buf + ht->lru * 2 * sizeof(void*));
-}
-
-void *cdb_ht_itemval(CDBHASHTABLE *ht, CDBHTITEM *item)
-{
- return (void *)(item->buf + ht->lru * 2 * sizeof(void*) + item->ksize);
-}
-
-CDBHASHTABLE *cdb_ht_new(bool lru, CDBHASHFUNC hashfunc)
-{
- CDBHASHTABLE *ht;
-
- ht = (CDBHASHTABLE*)malloc(sizeof(CDBHASHTABLE));
- ht->hash = NULL;
- ht->lru = lru;
- ht->num = ht->size = 0;
- ht->tail = ht->head = NULL;
- for(uint32_t i = 0; i < (1<<CDBHTBNUMPOW); i++) {
- CDBHTBUCKET *bucket = &(ht->buckets[i]);
- bucket->bnum = 2;
- uint32_t lsize = sizeof(CDBHTITEM *) * bucket->bnum;
- bucket->rnum = 0;
- bucket->items = (CDBHTITEM **)malloc(lsize);
- ht->size += lsize;
- memset(bucket->items, 0, lsize);
- }
- ht->hash = hashfunc;
- if (ht->hash == NULL)
- ht->hash = MurmurHash1;
-
- ht->size += sizeof(CDBHASHTABLE);
-
- return ht;
-}
-
-CDBHTITEM *cdb_ht_newitem(CDBHASHTABLE *ht, int ksize, int vsize)
-{
- CDBHTITEM *item;
- int hsize;
-
- if (ht->lru)
- hsize = sizeof(CDBHTITEM) + 2 * sizeof(void*);
- else
- hsize = sizeof(CDBHTITEM);
-
- item = (CDBHTITEM*)malloc(hsize + ksize + vsize);
- item->ksize = ksize;
- item->vsize = vsize;
- if (ht->lru) {
- LRUPREV(item) = NULL;
- LRUNEXT(item) = NULL;
- }
- return item;
-}
-
-
-
-
-void cdb_ht_insert(CDBHASHTABLE *ht, CDBHTITEM *item)
-{
- uint32_t bid, hid;
- CDBHTBUCKET *bucket;
-
- item->hash = ht->hash(cdb_ht_itemkey(ht, item), item->ksize);
- bid = item->hash & ((1<<CDBHTBNUMPOW)-1);
- bucket = &(ht->buckets[bid]);
- hid = (item->hash >> CDBHTBNUMPOW) & (bucket->bnum-1);
-
- if (bucket->rnum > bucket->bnum * 2) {
- CDBHTITEM **ilist;
- uint32_t exp = 2;
- if (bucket->bnum < 512)
- exp = 4;
- int listsize = (bucket->bnum * exp) * sizeof(CDBHTITEM*);
- ilist = (CDBHTITEM**)malloc(listsize);
- memset(ilist, 0, listsize);
- for(uint32_t i = 0; i < bucket->bnum; i++) {
- CDBHTITEM *curitem = bucket->items[i];
- while(curitem != NULL) {
- CDBHTITEM *nextitem = curitem->hnext;
- uint32_t hid = (curitem->hash>>CDBHTBNUMPOW)
- & (bucket->bnum * exp - 1);
- curitem->hnext = ilist[hid];
- ilist[hid] = curitem;
- curitem = nextitem;
- }
- }
- free(bucket->items);
- bucket->items = ilist;
- ht->size += listsize - bucket->bnum * sizeof(CDBHTITEM *);
- bucket->bnum *= exp;
- hid = (item->hash >> CDBHTBNUMPOW) & (bucket->bnum - 1);
- }
-
- {
- CDBHTITEM *curitem = bucket->items[hid];
- CDBHTITEM *preitem = NULL;
- while(curitem != NULL) {
- if (curitem->hash == item->hash
- && curitem->ksize == item->ksize
- && memcmp(cdb_ht_itemkey(ht, curitem),
- cdb_ht_itemkey(ht, item) ,curitem->ksize) == 0) {
- CDBHTITEM *tmp;
- if (ht->lru) {
- if (LRUPREV(curitem))
- LRUNEXT(LRUPREV(curitem)) = LRUNEXT(curitem);
- if (LRUNEXT(curitem))
- LRUPREV(LRUNEXT(curitem)) = LRUPREV(curitem);
- if (ht->head == curitem)
- ht->head = LRUNEXT(curitem);
- if (ht->tail == curitem)
- ht->tail = LRUPREV(curitem);
- }
- if (preitem)
- preitem->hnext = curitem->hnext;
- else
- bucket->items[hid] = curitem->hnext;
- tmp = curitem->hnext;
- ht->size -= sizeof(CDBHTITEM) + curitem->ksize + curitem->vsize
- + (ht->lru > 0) * sizeof(CDBHTITEM*) * 2;
- ht->num--;
- bucket->rnum--;
- free(curitem);
- curitem = tmp;
- break;
- }
- preitem = curitem;
- curitem = curitem->hnext;
- }
- }
-
- item->hnext = bucket->items[hid];
- bucket->items[hid] = item;
-
- if (ht->lru) {
- if (ht->head) LRUPREV(ht->head) = item;
- LRUPREV(item) = NULL;
- LRUNEXT(item) = ht->head;
- ht->head = item;
- if (ht->tail == NULL)
- ht->tail = item;
- }
-
- bucket->rnum++;
- ht->num++;
- ht->size += sizeof(CDBHTITEM) + item->ksize + item->vsize
- + ht->lru * sizeof(CDBHTITEM*) * 2;
-}
-
-
-void *cdb_ht_insert2(CDBHASHTABLE *ht, const void *key, int ksize, const void *val, int vsize)
-{
- CDBHTITEM *item;
-
- item = cdb_ht_newitem(ht, ksize, vsize);
- memcpy(cdb_ht_itemkey(ht, item), key, ksize);
- memcpy(cdb_ht_itemval(ht, item), val, vsize);
- cdb_ht_insert(ht, item);
- return cdb_ht_itemval(ht, item);
-}
-
-void *cdb_ht_get(CDBHASHTABLE *ht, const void *key, int ksize, int *vsize, bool mtf)
-{
- CDBHTITEM *res;
-
- res = cdb_ht_get3(ht, key, ksize, mtf);
- if (res) {
- if(vsize)
- *vsize = res->vsize;
- return cdb_ht_itemval(ht, res);
- } else {
- if(vsize)
- *vsize = 0;
- return NULL;
- }
-}
-
-
-void *cdb_ht_get2(CDBHASHTABLE *ht, const void *key, int ksize, bool mtf)
-{
- CDBHTITEM *res;
-
- res = cdb_ht_get3(ht, key, ksize, mtf);
- if (res)
- return cdb_ht_itemval(ht, res);
- else
- return NULL;
-}
-
-
-CDBHTITEM *cdb_ht_get3(CDBHASHTABLE *ht, const void *key, int ksize, bool mtf)
-{
- uint32_t hash, bid, hid;
- CDBHTBUCKET *bucket;
- CDBHTITEM *curitem;
-
- hash = ht->hash(key, ksize);
- bid = hash & ((1<<CDBHTBNUMPOW)-1);
- bucket = &(ht->buckets[bid]);
- hid = (hash >> CDBHTBNUMPOW) & (bucket->bnum - 1);
-
- curitem = bucket->items[hid];
- while (curitem != NULL) {
- if (curitem->hash == hash
- && curitem->ksize == ksize
- && memcmp(cdb_ht_itemkey(ht, curitem), key , ksize) == 0) {
- if (ht->lru && mtf && ht->head != curitem) {
- if (LRUPREV(curitem))
- LRUNEXT(LRUPREV(curitem)) = LRUNEXT(curitem);
- if (LRUNEXT(curitem))
- LRUPREV(LRUNEXT(curitem)) = LRUPREV(curitem);
- if (ht->tail == curitem)
- ht->tail = LRUPREV(curitem);
-
- LRUNEXT(curitem) = ht->head;
- LRUPREV(ht->head) = curitem;
- ht->head = curitem;
- LRUPREV(curitem) = NULL;
- }
- return curitem;
- }
- curitem = curitem->hnext;
- }
- return NULL;
-}
-
-
-bool cdb_ht_exist(CDBHASHTABLE *ht, const void *key, int ksize)
-{
- int vsize;
- return (cdb_ht_get(ht, key, ksize, &vsize, false) != NULL);
-}
-
-
-int cdb_ht_del2(CDBHASHTABLE *ht, const void *key, int ksize)
-{
- CDBHTITEM *res = NULL;
- res = cdb_ht_del(ht, key, ksize);
- if (res) {
- free(res);
- return 0;
- }
- return -1;
-}
-
-
-CDBHTITEM *cdb_ht_del(CDBHASHTABLE *ht, const void *key, int ksize)
-{
- uint32_t hash, bid, hid;
- CDBHTBUCKET *bucket;
- CDBHTITEM *curitem, *preitem;
- CDBHTITEM *res = NULL;
-
- hash = ht->hash(key, ksize);
- bid = hash & ((1<<CDBHTBNUMPOW)-1);
- bucket = &(ht->buckets[bid]);
- hid = (hash >> CDBHTBNUMPOW) & (bucket->bnum - 1);
-
- curitem = bucket->items[hid];
- preitem = NULL;
- while(curitem != NULL) {
- if (curitem->hash == hash
- && curitem->ksize == ksize
- && memcmp(cdb_ht_itemkey(ht, curitem),
- key, ksize) == 0) {
- if (ht->lru) {
- if (LRUPREV(curitem))
- LRUNEXT(LRUPREV(curitem)) = LRUNEXT(curitem);
- if (LRUNEXT(curitem))
- LRUPREV(LRUNEXT(curitem)) = LRUPREV(curitem);
- if (ht->head == curitem)
- ht->head = LRUNEXT(curitem);
- if (ht->tail == curitem)
- ht->tail = LRUPREV(curitem);
- }
- if (preitem)
- preitem->hnext = curitem->hnext;
- else
- bucket->items[hid] = curitem->hnext;
- ht->size -= sizeof(CDBHTITEM) + curitem->ksize + curitem->vsize
- + (ht->lru > 0) * sizeof(CDBHTITEM*) * 2;
- ht->num--;
- bucket->rnum--;
- res = curitem;
- curitem = curitem->hnext;
- break;
- }
- preitem = curitem;
- curitem = curitem->hnext;
- }
-
- return res;
-}
-
-
-void cdb_ht_removetail(CDBHASHTABLE *ht)
-{
- CDBHTITEM *item;
-
- item = cdb_ht_poptail(ht);
- if (item)
- free(item);
- return;
-}
-
-
-CDBHTITEM *cdb_ht_gettail(CDBHASHTABLE *ht)
-{
- return ht->tail;
-}
-
-
-CDBHTITEM *cdb_ht_poptail(CDBHASHTABLE *ht)
-{
- CDBHTITEM *item = ht->tail, *curitem, *preitem;;
- CDBHTBUCKET *bucket;
- uint32_t bid, hid;
-
- if (!(ht->lru) || item == NULL)
- return NULL;
-
- bid = item->hash & ((1<<CDBHTBNUMPOW)-1);
- bucket = &(ht->buckets[bid]);
- hid = (item->hash >> CDBHTBNUMPOW) & (bucket->bnum - 1);
-
- curitem = bucket->items[hid];
- preitem = NULL;
- while (curitem != NULL) {
- if (curitem->hash == item->hash
- && curitem->ksize == item->ksize
- && memcmp(cdb_ht_itemkey(ht, curitem),
- cdb_ht_itemkey(ht, item), item->ksize) == 0) {
- if (preitem) {
- preitem->hnext = curitem->hnext;
- } else {
- bucket->items[hid] = curitem->hnext;
- }
- break;
- }
- preitem = curitem;
- curitem = curitem->hnext;
- }
-
- if (LRUPREV(item))
- LRUNEXT(LRUPREV(item)) = NULL;
- if (ht->head == item)
- ht->head = NULL;
- ht->tail = LRUPREV(item);
- bucket->rnum--;
- ht->num--;
- ht->size -= sizeof(CDBHTITEM) + item->ksize + item->vsize
- + sizeof(CDBHTITEM*) * 2;
- return item;
-}
-
-void cdb_ht_clean(CDBHASHTABLE *ht)
-{
- for(uint32_t i = 0; i < (1<<CDBHTBNUMPOW); i++) {
- CDBHTBUCKET *bucket = &(ht->buckets[i]);
- for(uint32_t j = 0; j < bucket->bnum; j++) {
- CDBHTITEM *curitem = bucket->items[j];
- while(curitem != NULL) {
- CDBHTITEM *tmp = curitem->hnext;
- free(curitem);
- curitem = tmp;
- }
- bucket->items[j] = NULL;
- }
- bucket->rnum = 0;
- }
- ht->num = 0;
-}
-
-
-void cdb_ht_destroy(CDBHASHTABLE *ht)
-{
- if (ht->lru) {
- CDBHTITEM *curitem = ht->head;
- while(curitem) {
- CDBHTITEM *nextitem = LRUNEXT(curitem);
- free(curitem);
- curitem = nextitem;
- }
- }
-
- for(uint32_t i = 0; i < (1<<CDBHTBNUMPOW); i++) {
- CDBHTBUCKET *bucket = &(ht->buckets[i]);
-
- for(uint32_t j = 0; j < bucket->bnum && (!ht->lru); j++) {
- CDBHTITEM *curitem = bucket->items[j];
- while(curitem != NULL) {
- CDBHTITEM *tmp = curitem->hnext;
- free(curitem);
- curitem = tmp;
- }
- }
- free(bucket->items);
- }
- free(ht);
-}
-
-
-CDBHTITEM *cdb_ht_iterbegin(CDBHASHTABLE *ht)
-{
- for(uint32_t i = 0; i < (1<<CDBHTBNUMPOW); i++) {
- CDBHTBUCKET *bucket = &(ht->buckets[i]);
- if (!bucket->rnum)
- continue;
- for(uint32_t j = 0; j < bucket->bnum; j++)
- if (bucket->items[j])
- return bucket->items[j];
- }
-
- return NULL;
-}
-
-
-CDBHTITEM *cdb_ht_iternext(CDBHASHTABLE *ht, CDBHTITEM *cur)
-{
- if (cur == NULL)
- return NULL;
-
- if (cur->hnext)
- return cur->hnext;
-
- uint32_t bid = cur->hash & ((1<<CDBHTBNUMPOW)-1);
- CDBHTBUCKET *bucket = &(ht->buckets[bid]);
- uint32_t hid = (cur->hash >> CDBHTBNUMPOW) & (bucket->bnum - 1);
-
- for(uint32_t i = hid + 1; i < bucket->bnum; i++) {
- if (bucket->items[i])
- return bucket->items[i];
- }
-
- for(uint32_t i = bid + 1; i < (1<<CDBHTBNUMPOW); i++) {
- CDBHTBUCKET *bucket = &(ht->buckets[i]);
- if (!bucket->rnum)
- continue;
- for(uint32_t j = 0; j < bucket->bnum; j++)
- if (bucket->items[j])
- return bucket->items[j];
- }
-
- return NULL;
-}
-
-
-#ifdef _UT_
-#include <stdio.h>
-#include <time.h>
-int main(int argc, char *argv[])
-{
- CDBHASHTABLE *ht;
- long k, v;
- ht = cdb_ht_new(true, NULL);
- for(int i = 0; i < 1000; i++) {
- k = i;
- v = i * 1000;
- cdb_ht_insert2(ht, &k, sizeof(long), &v, sizeof(long));
- }
-
- srand(time(NULL));
-
- for(int i = 0; i < 1000; i++) {
- long *v, k = rand() % 1000;
- int vsize;
- v = (long*)cdb_ht_get(ht, &k, sizeof(long), &vsize, true);
- printf("get: %ld -> %ld (%d)\n", k, *v, vsize);
- }
-
- printf("total size: %d num: %d\n", ht->size, ht->num);
-
- CDBHTITEM *item;
- item = cdb_ht_poptail(ht);
- printf("tail: %ld - %ld\n", *(long*)cdb_ht_itemkey(ht, item), *(long*)cdb_ht_itemval(ht, item));
- free(item);
- item = cdb_ht_poptail(ht);
- printf("tail: %ld - %ld\n", *(long*)cdb_ht_itemkey(ht, item), *(long*)cdb_ht_itemval(ht, item));
- free(item);
-}
-#endif
diff --git a/3rdparty/cuttdb/src/cdb_hashtable.h b/3rdparty/cuttdb/src/cdb_hashtable.h
deleted file mode 100644
index 1f35b376dae7dd7618c24500dc729ab71577ad45..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_hashtable.h
+++ /dev/null
@@ -1,139 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CDB_HASHTABLE_H_
-#define _CDB_HASHTABLE_H_
-#include <stdint.h>
-#include <stdlib.h>
-#include <stdbool.h>
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-typedef uint32_t (*CDBHASHFUNC)(const void *, int);
-
-/* default 1<<8 level-1 buckets, which makes the table expanding more smoothly */
-#define CDBHTBNUMPOW 8
-
-
-typedef struct CDBHTITEM
-{
- int ksize;
- int vsize;
- uint32_t hash;
- /* next element with the same hash */
- struct CDBHTITEM *hnext;
- /* if LRU is true, the first several bytes are two pointers of prev/next element */
- struct CDBHTITEM *lruptr[0];
- char buf[0];
-} __attribute__((packed)) CDBHTITEM;
-
-
-typedef struct {
- /* array for items */
- CDBHTITEM **items;
- /* number of allocated slots in the bucket */
- uint32_t bnum;
- /* number of items exist in the bucket */
- uint32_t rnum;
-} CDBHTBUCKET;
-
-
-typedef struct CDBHASHTABLE {
- /* is in LRU mode? */
- bool lru;
- /* user specified hash function */
- CDBHASHFUNC hash;
- /* fixed number for level-1 buckets */
- CDBHTBUCKET buckets[1<<CDBHTBNUMPOW];
- /* memory usage */
- uint64_t size;
- /* number of items */
- uint64_t num;
- /* in LRU mode, the newest item */
- CDBHTITEM *head;
- /* in LRU mode, the oldest item */
- CDBHTITEM *tail;
-} CDBHASHTABLE;
-
-
-/* get the pointer of key in current item */
-/* #define cdb_ht_itemkey(ht, item) (item->buf + ht->lru * 2 * sizeof(void*)) */
-void *cdb_ht_itemkey(CDBHASHTABLE *ht, CDBHTITEM *item);
-
-/* get the pointer of value in current item */
-/* #define cdb_ht_itemval(ht, item) (item->buf + ht->lru * 2 * sizeof(void*) + item->ksize) */
-void *cdb_ht_itemval(CDBHASHTABLE *ht, CDBHTITEM *item);
-
-/* create an hashtable, it can be a simple hashtable or with LeastRecentUse
- The LRU mode needs extra two pointer space for every element
- hash function can by specified by user */
-CDBHASHTABLE *cdb_ht_new(bool lru, CDBHASHFUNC hashfunc);
-
-/* clean and free the hastable */
-void cdb_ht_destroy(CDBHASHTABLE *ht);
-
-/* allocate a new item with specified size, but do not insert it into table */
-CDBHTITEM *cdb_ht_newitem(CDBHASHTABLE *ht, int ksize, int vsize);
-
-/* insert an item which already exists into table */
-void cdb_ht_insert(CDBHASHTABLE *ht, CDBHTITEM *item);
-
-/* allocate and insert an item into table by key and value, return the pointer of value in table */
-void *cdb_ht_insert2(CDBHASHTABLE *ht, const void *key, int ksize, const void *val, int vsize);
-
-/* get the value of an item and its size in table, move the item to front if mtf == true */
-void *cdb_ht_get(CDBHASHTABLE *ht, const void *key, int ksize, int *vsize, bool mtf);
-
-/* get the value of an item, assume the size is known, move the item to front if mtf == true */
-void *cdb_ht_get2(CDBHASHTABLE *ht, const void *key, int ksize, bool mtf);
-
-/* get the pointer of an item, it hasn't been copied */
-CDBHTITEM *cdb_ht_get3(CDBHASHTABLE *ht, const void *key, int ksize, bool mtf);
-
-/* check if an item with the key exists */
-bool cdb_ht_exist(CDBHASHTABLE *ht, const void *key, int ksize);
-
-/* delete and free an item from table by its key */
-int cdb_ht_del2(CDBHASHTABLE *ht, const void *key, int ksize);
-
-/* return and delete an item from table, the item should be freed by user */
-CDBHTITEM *cdb_ht_del(CDBHASHTABLE *ht, const void *key, int ksize);
-
-/* delete and free the last item in table */
-void cdb_ht_removetail(CDBHASHTABLE *ht);
-
-/* return last item in table, do not delete nor free */
-CDBHTITEM *cdb_ht_gettail(CDBHASHTABLE *ht);
-
-/* return last item in table, delete but should be freed by user */
-CDBHTITEM *cdb_ht_poptail(CDBHASHTABLE *ht);
-
-/* clean and free all elements in the table*/
-void cdb_ht_clean(CDBHASHTABLE *ht);
-
-/* iterate the table by get the front one firstly */
-CDBHTITEM *cdb_ht_iterbegin(CDBHASHTABLE *ht);
-
-/* get the next item of current element */
-CDBHTITEM *cdb_ht_iternext(CDBHASHTABLE *ht, CDBHTITEM *cur);
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif
-
diff --git a/3rdparty/cuttdb/src/cdb_lock.c b/3rdparty/cuttdb/src/cdb_lock.c
deleted file mode 100644
index 92aab58be4ec916aaa1037e48b62f641176ee9c6..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_lock.c
+++ /dev/null
@@ -1,91 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cdb_lock.h"
-#include <stdlib.h>
-#include <pthread.h>
-#include <sched.h>
-
-
-CDBLOCK *cdb_lock_new(int ltype)
-{
- CDBLOCK *lock = NULL;
-#ifndef DAP_OS_DARWIN
- if (ltype == CDB_LOCKSPIN) {
- lock = (CDBLOCK *)malloc(sizeof(CDBLOCK) + sizeof(pthread_spinlock_t));
- pthread_spin_init((pthread_spinlock_t*)&lock->lock, PTHREAD_PROCESS_PRIVATE);
- } else
-#endif
- if (ltype == CDB_LOCKMUTEX) {
- lock = (CDBLOCK *)malloc(sizeof(CDBLOCK) + sizeof(pthread_mutex_t));
- pthread_mutex_init((pthread_mutex_t*)&lock->lock, NULL);
- }
- if( lock)
- lock->ltype = ltype;
-
- return lock;
-}
-
-
-void cdb_lock_lock(CDBLOCK *lock)
-{
-#ifndef DAP_OS_DARWIN
- if (lock->ltype == CDB_LOCKSPIN)
- pthread_spin_lock((pthread_spinlock_t*)&lock->lock);
- else
-#endif
- if (lock->ltype == CDB_LOCKMUTEX)
- pthread_mutex_lock((pthread_mutex_t*)&lock->lock);
-}
-
-
-void cdb_lock_unlock(CDBLOCK *lock)
-{
-#ifndef DAP_OS_DARWIN
- if (lock->ltype == CDB_LOCKSPIN)
- pthread_spin_unlock((pthread_spinlock_t*)&lock->lock);
- else
-#endif
- if (lock->ltype == CDB_LOCKMUTEX)
- pthread_mutex_unlock((pthread_mutex_t*)&lock->lock);
-}
-
-
-void cdb_lock_destory(CDBLOCK *lock)
-{
-#ifndef DAP_OS_DARWIN
- if (lock->ltype == CDB_LOCKSPIN)
- pthread_spin_destroy((pthread_spinlock_t*)&lock->lock);
- else
-#endif
- if (lock->ltype == CDB_LOCKMUTEX)
- pthread_mutex_destroy((pthread_mutex_t*)&lock->lock);
-
- free(lock);
-}
-
-
-int cdb_lock_trylock(CDBLOCK *lock)
-{
-#ifndef DAP_OS_DARWIN
- if (lock->ltype == CDB_LOCKSPIN)
- return pthread_spin_trylock((pthread_spinlock_t*)&lock->lock);
- else
-#endif
- if (lock->ltype == CDB_LOCKMUTEX)
- return pthread_mutex_trylock((pthread_mutex_t*)&lock->lock);
- return 0;
-}
-
diff --git a/3rdparty/cuttdb/src/cdb_lock.h b/3rdparty/cuttdb/src/cdb_lock.h
deleted file mode 100644
index bbd13098603bc0e0048d9e752b646ea7fea7d53a..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_lock.h
+++ /dev/null
@@ -1,51 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CDB_LOCK_H_
-#define _CDB_LOCK_H_
-
-
-enum {
- /* spinlock */
-#ifndef DAP_OS_DARWIN
- CDB_LOCKSPIN,
-#endif
- /* mutex, which may cause OS context switch, mainly used in where Disk IO happens */
- CDB_LOCKMUTEX,
-};
-
-/* may be used to indicated whether the area is protected */
-enum {
- CDB_LOCKED,
- CDB_NOTLOCKED,
-};
-
-typedef struct CDBLOCK
-{
- int ltype;
- char lock[0];
-} CDBLOCK;
-
-
-CDBLOCK *cdb_lock_new(int ltype);
-void cdb_lock_lock(CDBLOCK *lock);
-void cdb_lock_unlock(CDBLOCK *lock);
-void cdb_lock_destory(CDBLOCK *lock);
-int cdb_lock_trylock(CDBLOCK *lock);
-
-
-
-#endif
-
diff --git a/3rdparty/cuttdb/src/cdb_types.h b/3rdparty/cuttdb/src/cdb_types.h
deleted file mode 100644
index c32676d1bbc16ddef0b39c44b91df51cead61ba9..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_types.h
+++ /dev/null
@@ -1,144 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CDB_TYPES_H_
-#define _CDB_TYPES_H_
-#include <stdint.h>
-
-#define KB 1024
-#define MB 1048576
-#define CDBMIN(a, b) ((a)<(b)?(a):(b))
-#define CDBMAX(a, b) ((a)>(b)?(a):(b))
-
-#define SI8 8
-#define SI4 4
-/* space reserved in stack for i/o, avoid some malloc/free */
-#define SBUFSIZE (64 * KB)
-
-/* a default disk read size for index page, 3KB is enough(a page with 300 items) */
-#define PAGEAREADSIZE (3 * KB)
-
-/* reserved in stack for matched items in a hash index page */
-#define SFOFFNUM 8
-
-/* a valid virtual offset */
-#define OFFNOTNULL(o) (((o).i4)||((o).i2))
-/* a null virtual offset */
-#define OFFNULL(o) (((o).i4==0)&&((o).i2==0))
-/* nullify an offset */
-#define OFFZERO(o) do{(o).i4=0;(o).i2=0;}while(0)
-/* offset is equal ? */
-#define OFFEQ(a,b) (((a).i4==(b).i4)&&((a).i2==(b).i2))
-/* hash in page is equal ? */
-#define PHASHEQ(a,b) (((a).i2==(b).i2)&&((a).i1==(b).i1))
-/* page size increment */
-#define CDB_PAGEINCR 4
-
-
-/* if page cache size exceeds the limit */
-#define PCOVERFLOW(db) ( (db) && (db)->dpcache && (db)->pcache && (db)->dpcache->size + (db)->pcache->size > (db)->pclimit)
-/* if record cache size exceeds the limit */
-#define RCOVERFLOW(db) ((db) && (db)->rcache && (db)->rcache->size > (db)->rclimit)
-
-/* timeout for a dirty index page stays since last modify */
-#define DPAGETIMEOUT 40
-/* operation on main table are isolated by these locks */
-#define MLOCKNUM 256
-
-#define CDBHASH64(a, b) cdb_crc64(a, b)
-
-/* all virtual offsets are 48-bits */
-typedef struct FOFF
-{
- uint32_t i4;
- uint16_t i2;
-} __attribute__((packed)) FOFF;
-
-
-
-#define SFOFF (sizeof(FOFF))
-
-
-/* all hash value in index page are 24-bits
- range 0..16M guarantee very low collision
- with less than a hundred records in a page */
-typedef struct PHASH
-{
- uint16_t i2;
- uint8_t i1;
-} __attribute__((packed)) PHASH;
-
-
-/* an item in index page contains a hash and an offset */
-typedef struct PITEM
-{
- FOFF off;
- PHASH hash;
-} __attribute__((packed)) PITEM;
-
-
-/* data record */
-typedef struct CDBREC{
- /* where the data come from */
- FOFF ooff;
- uint32_t osize;
-
- /* access convenient*/
- void *key;
- void *val;
-
- /* disk store starts at following field */
- uint32_t magic;
- uint32_t ksize;
- uint32_t vsize;
- uint32_t expire;
- uint64_t oid;
- char buf[0];
-} __attribute__((packed)) CDBREC;
-
-/* real size of a record header when stored on disk */
-#define RECHSIZE (SI4 * 4 + SI8)
-/* real size of a record when stored on disk */
-#define RECSIZE(r) (RECHSIZE + (r)->ksize + (r)->vsize)
-
-
-/* index page */
-typedef struct CDBPAGE{
- FOFF ooff;
- uint32_t osize;
- uint32_t cap;
-
- union {
- /* what it be on disk */
- uint32_t magic;
- /* what it be in memory */
- uint32_t mtime;
- };
- /* which bucket it belongs to */
- uint32_t bid;
- uint32_t num;
- uint64_t oid;
- PITEM items[0];
-} __attribute__((packed)) CDBPAGE;
-
-/* real size of a page header when stored on disk */
-#define PAGEHSIZE (SI4 * 3 + SI8)
-/* real size of a page when stored on disk */
-#define PAGESIZE(p) (PAGEHSIZE + sizeof(PITEM) * (p)->num)
-/* in-memory size of an record structure */
-#define MPAGESIZE(p) (sizeof(CDBPAGE) + sizeof(PITEM) * (p)->cap)
-
-#endif
-
diff --git a/3rdparty/cuttdb/src/cdb_vio.c b/3rdparty/cuttdb/src/cdb_vio.c
deleted file mode 100644
index c0da6d1572812cbda4afe795cf24e25a733e375b..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_vio.c
+++ /dev/null
@@ -1,42 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cdb_vio.h"
-#include "cdb_types.h"
-#include "vio_apnd2.h"
-#include "stdlib.h"
-
-
-CDBVIO *cdb_vio_new(int type)
-{
- CDBVIO *res;
- res = (CDBVIO *)malloc(sizeof(CDBVIO));
- switch(type) {
- case CDBVIOAPND2:
- vio_apnd2_init(res);
- break;
- default:
- vio_apnd2_init(res);
- break;
- }
- return res;
-}
-
-int cdb_vio_destroy(CDBVIO *vio)
-{
- free(vio);
- return 0;
-}
-
diff --git a/3rdparty/cuttdb/src/cdb_vio.h b/3rdparty/cuttdb/src/cdb_vio.h
deleted file mode 100644
index 5c6e7e205813f11d84d45f0fa8916cff1edc3bbe..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cdb_vio.h
+++ /dev/null
@@ -1,101 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CDB_VIO_H_
-#define _CDB_VIO_H_
-#include "cdb_types.h"
-#include "cuttdb.h"
-#include <stdint.h>
-#include <stdbool.h>
-
-enum {
- /* obsoleted */
- CDBVIOAPPEND,
- /* append only format storage */
- CDBVIOAPND2,
-};
-
-typedef struct CDBVIO CDBVIO;
-
-/* write a record, returns virtual offset at 3rd parameter */
-typedef int (*VIOWRITEREC)(CDBVIO*, CDBREC*, FOFF*);
-/* delete a record, pass in the current offset at 3rd parameter */
-typedef int (*VIODELETEREC)(CDBVIO*, CDBREC*, FOFF);
-/* read a record, 2nd parameter default points to stack buffer, if its real size
-greater than the stack buffer size, it will be changed to points to a space in heap,
-the last parameter decides whether read the whole record or just read key for comparsion */
-typedef int (*VIOREADREC)(CDBVIO*, CDBREC**, FOFF, bool);
-/* close the storage */
-typedef int (*VIOCLOSE)(CDBVIO*);
-/* open the storage, pass in the storage path and open mode */
-typedef int (*VIOOPEN)(CDBVIO*, const char*, int);
-/* write an index page, return its virtual offset at 3rd parameter */
-typedef int (*VIOWRITEPAGE)(CDBVIO*, CDBPAGE *, FOFF*);
-/* read an index page, 2nd parameter default points to stack buffer, if its real size
-greater than the stack buffer size, it will be changed to points to a space in heap */
-typedef int (*VIOREADPAGE)(CDBVIO*, CDBPAGE **, FOFF);
-/* make the storage do an sync operation */
-typedef int (*VIOSYNC)(CDBVIO*);
-/* write db header, which contains main-index */
-typedef int (*VIOWRITEHEAD)(CDBVIO*);
-/* read db header, which contains main-index */
-typedef int (*VIOREADHEAD)(CDBVIO*);
-/* tell that no dirty page exists */
-typedef void (*VIOCLEANPOINT)(CDBVIO*);
-/* get the record/page iterator at oid */
-typedef void* (*VIOITFIRST)(CDBVIO *, uint64_t oid);
-/* get the next index page by iterator */
-typedef int (*VIOPAGEITNEXT)(CDBVIO *, CDBPAGE **, void *);
-/* get the next record by iterator */
-typedef int (*VIORECITNEXT)(CDBVIO *, CDBREC **, void *);
-/* destroy and free the iterator */
-typedef void (*VIOITDESTROY)(CDBVIO *, void *);
-
-struct CDBVIO
-{
- VIOOPEN open;
- VIOCLOSE close;
-
- VIOWRITEREC wrec;
- VIODELETEREC drec;
- VIOREADREC rrec;
-
- VIOWRITEPAGE wpage;
- VIOREADPAGE rpage;
-
- VIOSYNC sync;
- VIOWRITEHEAD whead;
- VIOREADHEAD rhead;
-
- VIOCLEANPOINT cleanpoint;
-
- VIOITFIRST pageitfirst;
- VIOPAGEITNEXT pageitnext;
- VIOITDESTROY pageitdestroy;
-
- VIOITFIRST recitfirst;
- VIORECITNEXT recitnext;
- VIOITDESTROY recitdestroy;
-
- CDB *db;
- void *iometa;
-};
-
-
-CDBVIO *cdb_vio_new(int type);
-int cdb_vio_destroy(CDBVIO *vio);
-
-
-#endif
diff --git a/3rdparty/cuttdb/src/cuttdb-server.c b/3rdparty/cuttdb/src/cuttdb-server.c
deleted file mode 100644
index 9b09a2863a28eeb88b74eebd1f23ebf6f6bfaa71..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cuttdb-server.c
+++ /dev/null
@@ -1,2152 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * The server&network part of CuttDB is based on Beansdb:
- *
- * http://beansdb.googlecode.com
- *
- * Beansdb is most based on Memcachedb and Memcached:
- *
- * http://memcachedb.org/
- * http://danga.com/memcached/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cuttdb-server.h"
-#include "cuttdb.h"
-#include <sys/stat.h>
-#include <sys/socket.h>
-#include <sys/un.h>
-#include <signal.h>
-#include <sys/resource.h>
-#include <sys/uio.h>
-#include <unistd.h>
-
-/* need this to get IOV_MAX on some platforms. */
-#ifndef __need_IOV_MAX
-#define __need_IOV_MAX
-#endif
-#include <pwd.h>
-#include <sys/mman.h>
-#include <fcntl.h>
-#include <netinet/tcp.h>
-#include <arpa/inet.h>
-#include <errno.h>
-#include <stdlib.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-#include <assert.h>
-#include <limits.h>
-#include <inttypes.h>
-#include <ctype.h>
-
-
-#ifdef HAVE_READPROC
-#include <proc/readproc.h>
-#endif
-
-#ifdef HAVE_MALLOC_H
-/* OpenBSD has a malloc.h, but warns to use stdlib.h instead */
-#ifndef __OpenBSD__
-#include <malloc.h>
-#endif
-#endif
-
-/* FreeBSD 4.x doesn't have IOV_MAX exposed. */
-#ifndef IOV_MAX
-#if defined(__FreeBSD__) || defined(__APPLE__)
-# define IOV_MAX 1024
-#endif
-#endif
-
-#ifndef IOV_MAX
-# define IOV_MAX 1024
-#endif
-
-#ifndef CLOCK_MONOTONIC
-#include "clock_gettime_stub.c"
-#endif
-
-/*
- * forward declarations
- */
-static int new_socket(struct addrinfo *ai);
-static int server_socket(const int port, const bool is_udp);
-static int try_read_command(conn *c);
-static int try_read_network(conn *c);
-
-/* stats */
-static void stats_reset(void);
-static void stats_init(void);
-
-/* defaults */
-static void settings_init(void);
-
-/* event handling, network IO */
-static void conn_close(conn *c);
-static void conn_init(void);
-static bool update_event(conn *c, const int new_flags);
-int delete_event(int fd);
-static void complete_nread(conn *c);
-static void process_command(conn *c, char *command);
-static int transmit(conn *c);
-static int ensure_iov_space(conn *c);
-static int add_iov(conn *c, const void *buf, int len);
-static int add_msghdr(conn *c);
-static void conn_free(conn *c);
-
-
-static size_t item_make_header(const uint8_t nkey, const int flags, const int nbytes,
- char *suffix, uint8_t *nsuffix);
-static int item_free(item *it);
-static item *item_get(char *key, size_t nkey);
-static item *item_alloc1(char *key, const size_t nkey, const int flags, const int nbytes);
-
-/** exported globals **/
-struct stats stats;
-struct settings settings;
-
-CDB *db = NULL;
-FILE *access_log = NULL;
-int daemon_quit = 0;
-
-/** file scope variables **/
-static int stub_fd = 0;
-
-#define TRANSMIT_COMPLETE 0
-#define TRANSMIT_INCOMPLETE 1
-#define TRANSMIT_SOFT_ERROR 2
-#define TRANSMIT_HARD_ERROR 3
-
-
-void item_init(void) {
- /*freeitemtotal = INIT_ITEM_FREELIST_LENGTH;
- freeitemcurr = 0;
-
- freeitem = (item **)malloc( sizeof(item *) * freeitemtotal );
- if (freeitem == NULL) {
- perror("malloc()");
- }*/
- return;
-}
-
-static size_t item_make_header(const uint8_t nkey, const int flags, const int nbytes,
- char *suffix, uint8_t *nsuffix) {
- /* suffix is defined at 40 chars elsewhere.. */
- *nsuffix = (uint8_t) snprintf(suffix, 40, " %d %d\r\n", flags, nbytes - 2);
- return sizeof(item) + nkey + *nsuffix + nbytes;
-}
-
-static int item_free(item *it)
-{
- free(it);
- return 0;
-}
-
-static item *item_get(char *key, size_t nkey)
-{
- item *it = NULL;
- int vlen;
- uint32_t flag;
- void *value;
- int ret = cdb_get(db, key, nkey, &value, &vlen);
- flag = 0;
- if (ret == 0){
- it = item_alloc1(key, nkey, flag, vlen + 2);
- if (it){
- memcpy(ITEM_data(it), value, vlen);
- memcpy(ITEM_data(it) + vlen, "\r\n", 2);
- }
- cdb_free_val(&value);
- }
- return it;
-
-}
-
-static item *item_alloc1(char *key, const size_t nkey, const int flags, const int nbytes)
-{
- uint8_t nsuffix;
- item *it;
- char suffix[40];
- size_t ntotal = item_make_header(nkey + 1, flags, nbytes, suffix, &nsuffix);
-
- it = (item *)malloc(ntotal);
- if (it == NULL){
- return NULL;
- }
- memset(it, 0, ntotal);
-
- it->nkey = nkey;
- it->nbytes = nbytes;
- strcpy(ITEM_key(it), key);
- memcpy(ITEM_suffix(it), suffix, (size_t)nsuffix);
- it->nsuffix = nsuffix;
- return it;
-}
-
-
-static void stats_init(void) {
- stats.curr_conns = stats.total_conns = stats.conn_structs = 0;
- stats.get_cmds = stats.set_cmds = stats.delete_cmds = 0;
- stats.slow_cmds = stats.get_hits = stats.get_misses = 0;
- stats.bytes_read = stats.bytes_written = 0;
-
- /* make the time we started always be 2 seconds before we really
- did, so time(0) - time.started is never zero. if so, things
- like 'settings.oldest_live' which act as booleans as well as
- values are now false in boolean context... */
- stats.started = time(0) - 2;
-}
-
-static void stats_reset(void) {
- STATS_LOCK();
- stats.total_conns = 0;
- stats.get_cmds = stats.set_cmds = stats.delete_cmds = 0;
- stats.slow_cmds = stats.get_hits = stats.get_misses = 0;
- stats.bytes_read = stats.bytes_written = 0;
- STATS_UNLOCK();
-}
-
-static void settings_init(void) {
- settings.port = 8964;
- /* By default this string should be NULL for getaddrinfo() */
- settings.inter = NULL;
- settings.item_buf_size = 4 * 1024; /* default is 4KB */
- settings.maxconns = 1024; /* to limit connections-related memory to about 5MB */
- settings.verbose = 0;
- settings.num_threads = 16;
- settings.flush_period = 1; // 1 secs
- settings.slow_cmd_time = 0.1; // 100ms
-}
-
-/*
- * Adds a message header to a connection.
- *
- * Returns 0 on success, -1 on out-of-memory.
- */
-static int add_msghdr(conn *c)
-{
- struct msghdr *msg;
-
- assert(c != NULL);
-
- if (c->msgsize == c->msgused) {
- msg = realloc(c->msglist, c->msgsize * 2 * sizeof(struct msghdr));
- if (! msg)
- return -1;
- c->msglist = msg;
- c->msgsize *= 2;
- }
-
- msg = c->msglist + c->msgused;
-
- /* this wipes msg_iovlen, msg_control, msg_controllen, and
- msg_flags, the last 3 of which aren't defined on solaris: */
- memset(msg, 0, sizeof(struct msghdr));
-
- msg->msg_iov = &c->iov[c->iovused];
-
- c->msgbytes = 0;
- c->msgused++;
-
- return 0;
-}
-
-
-/*
- * Free list management for connections.
- */
-
-static conn **freeconns;
-static int freetotal;
-static int freecurr;
-
-
-static void conn_init(void) {
- freetotal = 200;
- freecurr = 0;
- if ((freeconns = (conn **)malloc(sizeof(conn *) * freetotal)) == NULL) {
- fprintf(stderr, "malloc()\n");
- }
- return;
-}
-
-/*
- * Returns a connection from the freelist, if any. Should call this using
- * conn_from_freelist() for thread safety.
- */
-conn *do_conn_from_freelist() {
- conn *c;
-
- if (freecurr > 0) {
- c = freeconns[--freecurr];
- } else {
- c = NULL;
- }
-
- return c;
-}
-
-/*
- * Adds a connection to the freelist. 0 = success. Should call this using
- * conn_add_to_freelist() for thread safety.
- */
-bool do_conn_add_to_freelist(conn *c) {
- if (freecurr < freetotal) {
- freeconns[freecurr++] = c;
- return false;
- } else {
- /* try to enlarge free connections array */
- conn **new_freeconns = realloc(freeconns, sizeof(conn *) * freetotal * 2);
- if (new_freeconns) {
- freetotal *= 2;
- freeconns = new_freeconns;
- freeconns[freecurr++] = c;
- return false;
- }
- }
- return true;
-}
-
-conn *conn_new(const int sfd, const int init_state, const int read_buffer_size) {
- conn *c = conn_from_freelist();
-
- if (NULL == c) {
- if (!(c = (conn *)calloc(1, sizeof(conn)))) {
- fprintf(stderr, "calloc()\n");
- return NULL;
- }
- c->rbuf = c->wbuf = 0;
- c->ilist = 0;
- c->iov = 0;
- c->msglist = 0;
-
- c->rsize = read_buffer_size;
- c->wsize = DATA_BUFFER_SIZE;
- c->isize = ITEM_LIST_INITIAL;
- c->iovsize = IOV_LIST_INITIAL;
- c->msgsize = MSG_LIST_INITIAL;
-
- c->rbuf = (char *)malloc((size_t)c->rsize);
- c->wbuf = (char *)malloc((size_t)c->wsize);
- c->ilist = (item **)malloc(sizeof(item *) * c->isize);
- c->iov = (struct iovec *)malloc(sizeof(struct iovec) * c->iovsize);
- c->msglist = (struct msghdr *)malloc(sizeof(struct msghdr) * c->msgsize);
-
- if (c->rbuf == 0 || c->wbuf == 0 || c->ilist == 0 || c->iov == 0 ||
- c->msglist == 0) {
- conn_free(c);
- fprintf(stderr, "malloc()\n");
- return NULL;
- }
-
- STATS_LOCK();
- stats.conn_structs++;
- STATS_UNLOCK();
- }
-
- if (settings.verbose > 1) {
- if (init_state == conn_listening)
- fprintf(stderr, "<%d server listening\n", sfd);
- else
- fprintf(stderr, "<%d new client connection\n", sfd);
- }
-
- c->sfd = sfd;
- c->state = init_state;
- c->rlbytes = 0;
- c->rbytes = c->wbytes = 0;
- c->wcurr = c->wbuf;
- c->rcurr = c->rbuf;
- c->ritem = 0;
- c->icurr = c->ilist;
- c->ileft = 0;
- c->iovused = 0;
- c->msgcurr = 0;
- c->msgused = 0;
-
- c->write_and_go = conn_read;
- c->write_and_free = 0;
- c->item = 0;
- c->noreply = false;
-
- update_event(c, AE_READABLE);
- if (add_event(sfd, AE_READABLE, c) == -1) {
- if (conn_add_to_freelist(c)) {
- conn_free(c);
- }
- perror("event_add");
- return NULL;
- }
-
- STATS_LOCK();
- stats.curr_conns++;
- stats.total_conns++;
- STATS_UNLOCK();
-
- return c;
-}
-
-static void conn_cleanup(conn *c) {
- assert(c != NULL);
-
- if (c->item) {
- item_free(c->item);
- c->item = 0;
- }
-
- if (c->ileft != 0) {
- for (; c->ileft > 0; c->ileft--,c->icurr++) {
- item_free(*(c->icurr));
- }
- }
-
- if (c->write_and_free) {
- free(c->write_and_free);
- c->write_and_free = 0;
- }
-}
-
-/*
- * Frees a connection.
- */
-void conn_free(conn *c) {
- if (c) {
- if (c->msglist)
- free(c->msglist);
- if (c->rbuf)
- free(c->rbuf);
- if (c->wbuf)
- free(c->wbuf);
- if (c->ilist)
- free(c->ilist);
- if (c->iov)
- free(c->iov);
- free(c);
- }
-}
-
-static void conn_close(conn *c) {
- assert(c != NULL);
-
- if (settings.verbose > 1)
- fprintf(stderr, "<%d connection closed.\n", c->sfd);
-
- delete_event(c->sfd);
- close(c->sfd);
- c->sfd = -1;
- update_event(c, 0);
- conn_cleanup(c);
-
- /* if the connection has big buffers, just free it */
- if (c->rsize > READ_BUFFER_HIGHWAT || conn_add_to_freelist(c)) {
- conn_free(c);
- }
-
- STATS_LOCK();
- stats.curr_conns--;
- STATS_UNLOCK();
-
- return;
-}
-
-
-/*
- * Shrinks a connection's buffers if they're too big. This prevents
- * periodic large "get" requests from permanently chewing lots of server
- * memory.
- *
- * This should only be called in between requests since it can wipe output
- * buffers!
- */
-static void conn_shrink(conn *c) {
- assert(c != NULL);
-
- if (c->rsize > READ_BUFFER_HIGHWAT && c->rbytes < DATA_BUFFER_SIZE) {
- char *newbuf;
-
- if (c->rcurr != c->rbuf)
- memmove(c->rbuf, c->rcurr, (size_t)c->rbytes);
-
- newbuf = (char *)realloc((void *)c->rbuf, DATA_BUFFER_SIZE);
-
- if (newbuf) {
- c->rbuf = newbuf;
- c->rsize = DATA_BUFFER_SIZE;
- }
- /* TODO check other branch... */
- c->rcurr = c->rbuf;
- }
-
- if (c->isize > ITEM_LIST_HIGHWAT) {
- item **newbuf = (item**) realloc((void *)c->ilist, ITEM_LIST_INITIAL * sizeof(c->ilist[0]));
- if (newbuf) {
- c->ilist = newbuf;
- c->isize = ITEM_LIST_INITIAL;
- }
- /* TODO check error condition? */
- }
-
- if (c->msgsize > MSG_LIST_HIGHWAT) {
- struct msghdr *newbuf = (struct msghdr *) realloc((void *)c->msglist, MSG_LIST_INITIAL * sizeof(c->msglist[0]));
- if (newbuf) {
- c->msglist = newbuf;
- c->msgsize = MSG_LIST_INITIAL;
- }
- /* TODO check error condition? */
- }
-
- if (c->iovsize > IOV_LIST_HIGHWAT) {
- struct iovec *newbuf = (struct iovec *) realloc((void *)c->iov, IOV_LIST_INITIAL * sizeof(c->iov[0]));
- if (newbuf) {
- c->iov = newbuf;
- c->iovsize = IOV_LIST_INITIAL;
- }
- /* TODO check return value */
- }
-}
-
-/*
- * Sets a connection's current state in the state machine. Any special
- * processing that needs to happen on certain state transitions can
- * happen here.
- */
-static void conn_set_state(conn *c, int state) {
- assert(c != NULL);
-
- if (state != c->state) {
- if (state == conn_read) {
- conn_shrink(c);
- }
- c->state = state;
- }
-}
-
-
-/*
- * Ensures that there is room for another struct iovec in a connection's
- * iov list.
- *
- * Returns 0 on success, -1 on out-of-memory.
- */
-static int ensure_iov_space(conn *c) {
- assert(c != NULL);
-
- if (c->iovused >= c->iovsize) {
- int i, iovnum;
- struct iovec *new_iov = (struct iovec *)realloc(c->iov,
- (c->iovsize * 2) * sizeof(struct iovec));
- if (! new_iov)
- return -1;
- c->iov = new_iov;
- c->iovsize *= 2;
-
- /* Point all the msghdr structures at the new list. */
- for (i = 0, iovnum = 0; i < c->msgused; i++) {
- c->msglist[i].msg_iov = &c->iov[iovnum];
- iovnum += c->msglist[i].msg_iovlen;
- }
- }
-
- return 0;
-}
-
-
-/*
- * Adds data to the list of pending data that will be written out to a
- * connection.
- *
- * Returns 0 on success, -1 on out-of-memory.
- */
-
-static int add_iov(conn *c, const void *buf, int len) {
- struct msghdr *m;
- int leftover;
- bool limit_to_mtu;
-
- assert(c != NULL);
-
- do {
- m = &c->msglist[c->msgused - 1];
-
- /*
- * Limit the first payloads of TCP replies, to
- * MAX_PAYLOAD_SIZE bytes.
- */
- limit_to_mtu = (1 == c->msgused);
-
- /* We may need to start a new msghdr if this one is full. */
- if (m->msg_iovlen == IOV_MAX ||
- (limit_to_mtu && c->msgbytes >= MAX_PAYLOAD_SIZE)) {
- add_msghdr(c);
- m = &c->msglist[c->msgused - 1];
- }
-
- if (ensure_iov_space(c) != 0)
- return -1;
-
- /* If the fragment is too big to fit in the datagram, split it up */
- if (limit_to_mtu && len + c->msgbytes > MAX_PAYLOAD_SIZE) {
- leftover = len + c->msgbytes - MAX_PAYLOAD_SIZE;
- len -= leftover;
- } else {
- leftover = 0;
- }
-
- m = &c->msglist[c->msgused - 1];
- m->msg_iov[m->msg_iovlen].iov_base = (void *)buf;
- m->msg_iov[m->msg_iovlen].iov_len = len;
-
- c->msgbytes += len;
- c->iovused++;
- m->msg_iovlen++;
-
- buf = ((char *)buf) + len;
- len = leftover;
- } while (leftover > 0);
-
- return 0;
-}
-
-
-static void out_string(conn *c, const char *str) {
- size_t len;
-
- assert(c != NULL);
-
- if (c->noreply) {
- if (settings.verbose > 1)
- fprintf(stderr, ">%d %s\n", c->sfd, str);
- c->noreply = false;
- conn_set_state(c, conn_read);
- return;
- }
-
- len = strlen(str);
- if ((len + 2) > c->wsize) {
- /* ought to be always enough. just fail for simplicity */
- str = "SERVER_ERROR output line too long";
- len = strlen(str);
- }
-
- memcpy(c->wbuf, str, len);
- memcpy(c->wbuf + len, "\r\n", 2);
- c->wbytes = len + 2;
- c->wcurr = c->wbuf;
-
- conn_set_state(c, conn_write);
- c->write_and_go = conn_read;
- return;
-}
-
-/*
- * we get here after reading the value in set/add/replace commands. The command
- * has been stored in c->item_comm, and the item is ready in c->item.
- */
-
-static void complete_nread(conn *c) {
- assert(c != NULL);
-
- item *it = c->item;
- int comm = c->item_comm;
- int ret;
-
- STATS_LOCK();
- stats.set_cmds++;
- STATS_UNLOCK();
-
- if (strncmp(ITEM_data(it) + it->nbytes - 2, "\r\n", 2) != 0) {
- out_string(c, "CLIENT_ERROR bad data chunk");
- } else {
- ret = store_item(it, comm);
- if (ret == 0)
- out_string(c, "STORED");
- else if(ret == -2)
- out_string(c, "EXISTS");
- else if(ret == -3)
- out_string(c, "NOT_FOUND");
- else
- out_string(c, "NOT_STORED");
- }
-
- item_free(c->item);
- c->item = 0;
-}
-
-/*
- * Stores an item in the cache according to the semantics of one of the set
- * commands. In threaded mode, this is protected by the cache lock.
- *
- * Returns true if the item was stored.
- */
-int store_item(item *it, int comm) {
- char *key = ITEM_key(it);
-
- switch (comm) {
- case NREAD_SET:
- return cdb_set2(db, key, it->nkey, ITEM_data(it), it->nbytes - 2, CDB_INSERTCACHE | CDB_OVERWRITE, it->expire);
- case NREAD_ADD:
- return cdb_set2(db, key, it->nkey, ITEM_data(it), it->nbytes - 2, CDB_INSERTCACHE | CDB_INSERTIFNOEXIST, it->expire);
- case NREAD_REPLACE:
- return cdb_set2(db, key, it->nkey, ITEM_data(it), it->nbytes - 2, CDB_INSERTCACHE | CDB_INSERTIFEXIST, it->expire);
- }
- return 0;
-}
-
-/*
- * adds a delta value to a numeric item.
- */
-/*
-
-int add_delta(char *key, size_t nkey, int64_t delta, char *buf);
-int add_delta(char* key, size_t nkey, int64_t delta, char *buf) {
- uint64_t value = hs_incr(db, key, delta);
- snprintf(buf, INCR_MAX_STORAGE_LEN, "%llu", (unsigned long long)value);
- return 0;
-}
-*/
-
-typedef struct token_s {
- char *value;
- size_t length;
-} token_t;
-
-#define COMMAND_TOKEN 0
-#define SUBCOMMAND_TOKEN 1
-#define KEY_TOKEN 1
-#define KEY_MAX_LENGTH 250
-
-#define MAX_TOKENS 8
-
-/*
- * Tokenize the command string by replacing whitespace with '\0' and update
- * the token array tokens with pointer to start of each token and length.
- * Returns total number of tokens. The last valid token is the terminal
- * token (value points to the first unprocessed character of the string and
- * length zero).
- *
- * Usage example:
- *
- * while(tokenize_command(command, ncommand, tokens, max_tokens) > 0) {
- * for(int ix = 0; tokens[ix].length != 0; ix++) {
- * ...
- * }
- * ncommand = tokens[ix].value - command;
- * command = tokens[ix].value;
- * }
- */
-static size_t tokenize_command(char *command, token_t *tokens, const size_t max_tokens) {
- char *s, *e;
- size_t ntokens = 0;
-
- assert(command != NULL && tokens != NULL && max_tokens > 1);
-
- for (s = e = command; ntokens < max_tokens - 1; ++e) {
- if (*e == ' ') {
- if (s != e) {
- tokens[ntokens].value = s;
- tokens[ntokens].length = e - s;
- ntokens++;
- *e = '\0';
- }
- s = e + 1;
- }
- else if (*e == '\0') {
- if (s != e) {
- tokens[ntokens].value = s;
- tokens[ntokens].length = e - s;
- ntokens++;
- }
-
- break; /* string end */
- }
- }
-
- /*
- * If we scanned the whole string, the terminal value pointer is null,
- * otherwise it is the first unprocessed character.
- */
- tokens[ntokens].value = *e == '\0' ? NULL : e;
- tokens[ntokens].length = 0;
- ntokens++;
-
- return ntokens;
-}
-
-static inline bool set_noreply_maybe(conn *c, token_t *tokens, size_t ntokens)
-{
- int noreply_index = ntokens - 2;
-
- /*
- NOTE: this function is not the first place where we are going to
- send the reply. We could send it instead from process_command()
- if the request line has wrong number of tokens. However parsing
- malformed line for "noreply" option is not reliable anyway, so
- it can't be helped.
- */
- if (tokens[noreply_index].value
- && strcmp(tokens[noreply_index].value, "noreply") == 0) {
- c->noreply = true;
- }
- return c->noreply;
-}
-
-static void process_stat(conn *c, token_t *tokens, const size_t ntokens) {
- time_t now = time(0);
- char *command;
- char *subcommand;
-
- assert(c != NULL);
-
- if(ntokens < 2) {
- out_string(c, "CLIENT_ERROR bad command line");
- return;
- }
-
- command = tokens[COMMAND_TOKEN].value;
-
- if (ntokens == 2 && strcmp(command, "stats") == 0) {
- char temp[1024];
- pid_t pid = getpid();
- uint64_t total = 0, curr = 0;
- CDBSTAT db_stat;
- cdb_stat(db, &db_stat);
- total = db_stat.rnum;
- char *pos = temp;
-
-#ifndef WIN32
- struct rusage usage;
- getrusage(RUSAGE_SELF, &usage);
-#endif /* !WIN32 */
-
- STATS_LOCK();
- pos += sprintf(pos, "STAT pid %ld\r\n", (long)pid);
- pos += sprintf(pos, "STAT uptime %"PRIuS"\r\n", now - stats.started);
- pos += sprintf(pos, "STAT time %"PRIuS"\r\n", now);
- pos += sprintf(pos, "STAT version " VERSION "\r\n");
- pos += sprintf(pos, "STAT pointer_size %"PRIuS"\r\n", 8 * sizeof(void *));
-#ifndef WIN32
- pos += sprintf(pos, "STAT rusage_user %ld.%06ld\r\n", usage.ru_utime.tv_sec, usage.ru_utime.tv_usec);
- pos += sprintf(pos, "STAT rusage_system %ld.%06ld\r\n", usage.ru_stime.tv_sec, usage.ru_stime.tv_usec);
- pos += sprintf(pos, "STAT rusage_minflt %"PRIu64"\r\n", usage.ru_minflt);
- pos += sprintf(pos, "STAT rusage_majflt %"PRIu64"\r\n", usage.ru_majflt);
- pos += sprintf(pos, "STAT rusage_nswap %"PRIu64"\r\n", usage.ru_nswap);
- pos += sprintf(pos, "STAT rusage_inblock %"PRIu64"\r\n", usage.ru_inblock);
- pos += sprintf(pos, "STAT rusage_oublock %"PRIu64"\r\n", usage.ru_oublock);
- pos += sprintf(pos, "STAT rusage_nvcsw %"PRIu64"\r\n", usage.ru_nvcsw);
- pos += sprintf(pos, "STAT rusage_nivcsw %"PRIu64"\r\n", usage.ru_nivcsw);
-#endif /* !WIN32 */
-#ifdef HAVE_READPROC
- proc_t p;
- get_proc_stats(getpid(), &p);
- pos += sprintf(pos, "STAT rusage_maxrss %"PRIu64"\r\n", p.vm_rss);
-#endif
- pos += sprintf(pos, "STAT item_buf_size %"PRIuS"\r\n", settings.item_buf_size);
- pos += sprintf(pos, "STAT curr_connections %"PRIu32"\r\n", stats.curr_conns - 1); /* ignore listening conn */
- pos += sprintf(pos, "STAT total_connections %"PRIu32"\r\n", stats.total_conns);
- pos += sprintf(pos, "STAT connection_structures %"PRIu32"\r\n", stats.conn_structs);
- pos += sprintf(pos, "STAT cmd_get %"PRIu64"\r\n", stats.get_cmds);
- pos += sprintf(pos, "STAT cmd_set %"PRIu64"\r\n", stats.set_cmds);
- pos += sprintf(pos, "STAT cmd_delete %"PRIu64"\r\n", stats.delete_cmds);
- pos += sprintf(pos, "STAT slow_cmd %"PRIu64"\r\n", stats.slow_cmds);
- pos += sprintf(pos, "STAT get_hits %"PRIu64"\r\n", stats.get_hits);
- pos += sprintf(pos, "STAT get_misses %"PRIu64"\r\n", stats.get_misses);
- pos += sprintf(pos, "STAT curr_items %"PRIu64"\r\n", curr);
- pos += sprintf(pos, "STAT total_items %"PRIu64"\r\n", total);
- pos += sprintf(pos, "STAT bytes_read %"PRIu64"\r\n", stats.bytes_read);
- pos += sprintf(pos, "STAT bytes_written %"PRIu64"\r\n", stats.bytes_written);
- pos += sprintf(pos, "STAT threads %d\r\n", settings.num_threads);
- pos += sprintf(pos, "STAT records_in_cache %lu\r\n", db_stat.rcnum);
- pos += sprintf(pos, "STAT pages_total %lu\r\n", db_stat.pnum);
- pos += sprintf(pos, "STAT pages_in_cache %lu\r\n", db_stat.pcnum);
- pos += sprintf(pos, "STAT record_cache_hits %lu\r\n", db_stat.rchit);
- pos += sprintf(pos, "STAT record_cache_misses %lu\r\n", db_stat.rcmiss);
- pos += sprintf(pos, "STAT page_cache_hits %lu\r\n", db_stat.pchit);
- pos += sprintf(pos, "STAT page_cache_misses %lu\r\n", db_stat.pcmiss);
- pos += sprintf(pos, "STAT read_latency_avg %u\r\n", db_stat.rlatcy);
- pos += sprintf(pos, "STAT write_latency_avg %u\r\n", db_stat.wlatcy);
- pos += sprintf(pos, "END");
- STATS_UNLOCK();
- out_string(c, temp);
- return;
- }
-
- subcommand = tokens[SUBCOMMAND_TOKEN].value;
-
- if (strcmp(subcommand, "reset") == 0) {
- stats_reset();
- out_string(c, "RESET");
- return;
- }
-
- out_string(c, "ERROR");
-}
-
-/* ntokens is overwritten here... shrug.. */
-static inline void process_get_command(conn *c, token_t *tokens, size_t ntokens) {
- char *key;
- size_t nkey;
- int i = 0;
- item *it = NULL;
- token_t *key_token = &tokens[KEY_TOKEN];
- int stats_get_cmds = 0;
- int stats_get_hits = 0;
- int stats_get_misses = 0;
- assert(c != NULL);
-
- do {
- while(key_token->length != 0) {
-
- key = key_token->value;
- nkey = key_token->length;
-
- if(nkey > KEY_MAX_LENGTH) {
- STATS_LOCK();
- stats.get_cmds += stats_get_cmds;
- stats.get_hits += stats_get_hits;
- stats.get_misses += stats_get_misses;
- STATS_UNLOCK();
- out_string(c, "CLIENT_ERROR bad command line format");
- return;
- }
-
- stats_get_cmds++;
-
- it = item_get(key, nkey);
-
- if (it) {
- if (i >= c->isize) {
- item **new_list = realloc(c->ilist, sizeof(item *) * c->isize * 2);
- if (new_list) {
- c->isize *= 2;
- c->ilist = new_list;
- } else {
- item_free(it);
- it = NULL;
- break;
- }
- }
-
- /*
- * Construct the response. Each hit adds three elements to the
- * outgoing data list:
- * "VALUE "
- * key
- * " " + flags + " " + data length + "\r\n" + data (with \r\n)
- */
-
- if (add_iov(c, "VALUE ", 6) != 0 ||
- add_iov(c, ITEM_key(it), it->nkey) != 0 ||
- add_iov(c, ITEM_suffix(it), it->nsuffix + it->nbytes) != 0)
- {
- item_free(it);
- it = NULL;
- break;
- }
-
- if (settings.verbose > 1)
- fprintf(stderr, ">%d sending key %s\n", c->sfd, ITEM_key(it));
-
- stats_get_hits++;
- *(c->ilist + i) = it;
- i++;
-
- } else {
- stats_get_misses++;
- }
-
- key_token++;
- }
-
- /*
- * If the command string hasn't been fully processed, get the next set
- * of tokens.
- */
- if(key_token->value != NULL) {
- ntokens = tokenize_command(key_token->value, tokens, MAX_TOKENS);
- key_token = tokens;
- }
-
- } while(key_token->value != NULL);
-
- c->icurr = c->ilist;
- c->ileft = i;
-
- if (settings.verbose > 1)
- fprintf(stderr, ">%d END\n", c->sfd);
-
- /*
- If the loop was terminated because of out-of-memory, it is not
- reliable to add END\r\n to the buffer, because it might not end
- in \r\n. So we send SERVER_ERROR instead.
- */
- if (key_token->value != NULL || add_iov(c, "END\r\n", 5) != 0) {
- out_string(c, "SERVER_ERROR out of memory writing get response");
- }
- else {
- conn_set_state(c, conn_mwrite);
- c->msgcurr = 0;
- }
-
- STATS_LOCK();
- stats.get_cmds += stats_get_cmds;
- stats.get_hits += stats_get_hits;
- stats.get_misses += stats_get_misses;
- STATS_UNLOCK();
-
- return;
-}
-
-static void process_update_command(conn *c, token_t *tokens, const size_t ntokens, int comm) {
- char *key;
- size_t nkey;
- int flags;
- time_t exptime;
- int vlen;
- item *it = NULL;
-
- assert(c != NULL);
-
- set_noreply_maybe(c, tokens, ntokens);
-
- if (tokens[KEY_TOKEN].length > KEY_MAX_LENGTH) {
- out_string(c, "CLIENT_ERROR bad command line format");
- return;
- }
-
- key = tokens[KEY_TOKEN].value;
- nkey = tokens[KEY_TOKEN].length;
-
- flags = strtoul(tokens[2].value, NULL, 10);
- exptime = strtol(tokens[3].value, NULL, 10);
- vlen = strtol(tokens[4].value, NULL, 10);
-
- if(errno == ERANGE || ((flags == 0 || exptime == 0) && errno == EINVAL)
- || vlen < 0) {
- out_string(c, "CLIENT_ERROR bad command line format");
- return;
- }
-
- it = item_alloc1(key, nkey, flags, vlen+2);
- it->expire = exptime;
- it->flag = flags;
-
- if (it == NULL) {
- out_string(c, "SERVER_ERROR out of memory storing object");
- /* swallow the data line */
- c->write_and_go = conn_swallow;
- c->sbytes = vlen + 2;
- return;
- }
-
- c->item = it;
- c->ritem = ITEM_data(it);
- c->rlbytes = it->nbytes;
- c->item_comm = comm;
- conn_set_state(c, conn_nread);
-}
-
-bool safe_strtoull(const char *str, uint64_t *out) {
- assert(out != NULL);
- errno = 0;
- *out = 0;
- char *endptr;
- unsigned long long ull = strtoull(str, &endptr, 10);
- if (errno == ERANGE)
- return false;
- if (isspace(*endptr) || (*endptr == '\0' && endptr != str)) {
- *out = ull;
- return true;
- }
- return false;
-}
-
-/*
-
-
-static void process_arithmetic_command(conn *c, token_t *tokens, const size_t ntokens, const bool incr) {
- char temp[INCR_MAX_STORAGE_LEN];
- uint64_t delta;
- char *key;
- size_t nkey;
-
- assert(c != NULL);
-
- set_noreply_maybe(c, tokens, ntokens);
-
- STATS_LOCK();
- stats.set_cmds++;
- STATS_UNLOCK();
-
- if (tokens[KEY_TOKEN].length > KEY_MAX_LENGTH) {
- out_string(c, "CLIENT_ERROR bad command line format");
- return;
- }
-
- key = tokens[KEY_TOKEN].value;
- nkey = tokens[KEY_TOKEN].length;
-
- if (!safe_strtoull(tokens[2].value, &delta)) {
- out_string(c, "CLIENT_ERROR invalid numeric delta argument");
- return;
- }
-
- switch(add_delta(key, nkey, delta, temp)) {
- case 0:
- out_string(c, temp);
- break;
-// case NON_NUMERIC:
-// out_string(c, "CLIENT_ERROR cannot increment or decrement non-numeric value");
-// break;
-// case EOM:
-// out_string(c, "SERVER_ERROR out of memory");
-// break;
- }
-}
-*/
-
-
-static void process_delete_command(conn *c, token_t *tokens, const size_t ntokens) {
- char *key;
- size_t nkey;
- assert(c != NULL);
-
- set_noreply_maybe(c, tokens, ntokens);
-
- STATS_LOCK();
- stats.delete_cmds++;
- STATS_UNLOCK();
-
- key = tokens[KEY_TOKEN].value;
- nkey = tokens[KEY_TOKEN].length;
- if(nkey > KEY_MAX_LENGTH) {
- out_string(c, "CLIENT_ERROR bad command line format");
- return;
- }
-
- switch (cdb_del(db, key, nkey)) {
- case 0:
- out_string(c, "DELETED");
- break;
- case -3:
- out_string(c, "NOT_FOUND");
- break;
-// case -1:
-// out_string(c, "SERVER_ERROR while delete a item");
-// break;
-// default:
-// out_string(c, "SERVER_ERROR nothing to do");
- }
- return;
-}
-
-static void process_verbosity_command(conn *c, token_t *tokens, const size_t ntokens) {
- unsigned int level;
-
- assert(c != NULL);
-
- set_noreply_maybe(c, tokens, ntokens);
-
- level = strtoul(tokens[1].value, NULL, 10);
- if(errno == ERANGE) {
- out_string(c, "CLIENT_ERROR bad command line format");
- return;
- }
- settings.verbose = level > MAX_VERBOSITY_LEVEL ? MAX_VERBOSITY_LEVEL : level;
- out_string(c, "OK");
- return;
-}
-
-static void process_command(conn *c, char *command) {
-
- token_t tokens[MAX_TOKENS];
- size_t ntokens;
- int comm;
- struct timespec start, end;
-
- assert(c != NULL);
-
- if (settings.verbose > 1)
- fprintf(stderr, "<%d %s\n", c->sfd, command);
-
- /*
- * for commands set/add/replace, we build an item and read the data
- * directly into it, then continue in nread_complete().
- */
-
- c->msgcurr = 0;
- c->msgused = 0;
- c->iovused = 0;
- if (add_msghdr(c) != 0) {
- out_string(c, "SERVER_ERROR out of memory preparing response");
- return;
- }
-
- clock_gettime(CLOCK_MONOTONIC, &start);
-
- ntokens = tokenize_command(command, tokens, MAX_TOKENS);
- if (ntokens >= 3 &&
- (strcmp(tokens[COMMAND_TOKEN].value, "get") == 0) ) {
-
- process_get_command(c, tokens, ntokens);
-
- } else if ((ntokens == 6 || ntokens == 7) &&
- ((strcmp(tokens[COMMAND_TOKEN].value, "set") == 0 && (comm = NREAD_SET)) ||
- (strcmp(tokens[COMMAND_TOKEN].value, "add") == 0 && (comm = NREAD_ADD)) ||
- (strcmp(tokens[COMMAND_TOKEN].value, "replace") == 0 && (comm = NREAD_REPLACE)))) {
-
- process_update_command(c, tokens, ntokens, comm);
-
-// } else if ((ntokens == 4 || ntokens == 5) && (strcmp(tokens[COMMAND_TOKEN].value, "incr") == 0)) {
-
-// process_arithmetic_command(c, tokens, ntokens, 1);
-
- } else if (ntokens >= 3 && ntokens <= 4 && (strcmp(tokens[COMMAND_TOKEN].value, "delete") == 0)) {
-
- process_delete_command(c, tokens, ntokens);
-
- } else if (ntokens >= 2 && (strcmp(tokens[COMMAND_TOKEN].value, "stats") == 0)) {
-
- process_stat(c, tokens, ntokens);
-
- } else if (ntokens == 2 && (strcmp(tokens[COMMAND_TOKEN].value, "version") == 0)) {
-
- out_string(c, "VERSION " VERSION);
-
- } else if (ntokens == 2 && (strcmp(tokens[COMMAND_TOKEN].value, "quit") == 0)) {
-
- conn_set_state(c, conn_closing);
-
- } else if (ntokens == 3 && (strcmp(tokens[COMMAND_TOKEN].value, "verbosity") == 0)) {
-
- process_verbosity_command(c, tokens, ntokens);
-
-/* } else if (ntokens >= 2 && ntokens <= 4 && (strcmp(tokens[COMMAND_TOKEN].value, "flush_all") == 0)) {
-
- set_noreply_maybe(c, tokens, ntokens);
-
- int limit = 10000;
- if (ntokens == (c->noreply ? 4 : 3)) {
- limit = strtol(tokens[1].value, NULL, 10);
- if(errno == ERANGE) {
- out_string(c, "CLIENT_ERROR bad command line format");
- return;
- }
- }
-
- hs_optimize(db, limit);
- out_string(c, "OK");
- return;
-*/
- } else {
- out_string(c, "ERROR");
- return;
- }
-
- clock_gettime(CLOCK_MONOTONIC, &end);
- float secs = (end.tv_sec - start.tv_sec) + (end.tv_nsec - start.tv_nsec) / 1e9;
- if (secs > settings.slow_cmd_time) {
- STATS_LOCK();
- stats.slow_cmds ++;
- STATS_UNLOCK();
- }
-
- // access logging
- if (NULL != access_log && ntokens >= 3) {
- char now[255];
- time_t t = time(NULL);
- strftime(now, 200, "%Y-%m-%d %H:%M:%S", localtime(&t));
- struct sockaddr_storage addr;
- socklen_t addrlen = sizeof(addr);
- getpeername(c->sfd, (struct sockaddr*)&addr, &addrlen);
- char host[NI_MAXHOST], serv[NI_MAXSERV];
- getnameinfo((struct sockaddr*)&addr, addrlen, host, sizeof(host), serv, sizeof(serv),
- NI_NUMERICSERV);
- fprintf(access_log, "%s %s:%s %s %s %.3f\n", now, host, serv,
- command, tokens[1].value, secs*1000);
- }
-
- return;
-}
-
-/*
- * if we have a complete line in the buffer, process it.
- */
-static int try_read_command(conn *c) {
- char *el, *cont;
-
- assert(c != NULL);
- assert(c->rcurr <= (c->rbuf + c->rsize));
-
- if (c->rbytes == 0)
- return 0;
- el = memchr(c->rcurr, '\n', c->rbytes);
- if (!el)
- return 0;
- cont = el + 1;
- if ((el - c->rcurr) > 1 && *(el - 1) == '\r') {
- el--;
- }
- *el = '\0';
-
- assert(cont <= (c->rcurr + c->rbytes));
-
- process_command(c, c->rcurr);
-
- c->rbytes -= (cont - c->rcurr);
- c->rcurr = cont;
-
- assert(c->rcurr <= (c->rbuf + c->rsize));
-
- return 1;
-}
-
-/*
- * read from network as much as we can, handle buffer overflow and connection
- * close.
- * before reading, move the remaining incomplete fragment of a command
- * (if any) to the beginning of the buffer.
- * return 0 if there's nothing to read on the first read.
- */
-static int try_read_network(conn *c) {
- int gotdata = 0;
- int res;
-
- assert(c != NULL);
-
- if (c->rcurr != c->rbuf) {
- if (c->rbytes != 0) /* otherwise there's nothing to copy */
- memmove(c->rbuf, c->rcurr, c->rbytes);
- c->rcurr = c->rbuf;
- }
-
- while (1) {
- if (c->rbytes >= c->rsize) {
- char *new_rbuf = realloc(c->rbuf, c->rsize * 2);
- if (!new_rbuf) {
- if (settings.verbose > 0)
- fprintf(stderr, "Couldn't realloc input buffer\n");
- c->rbytes = 0; /* ignore what we read */
- out_string(c, "SERVER_ERROR out of memory reading request");
- c->write_and_go = conn_closing;
- return 1;
- }
- c->rcurr = c->rbuf = new_rbuf;
- c->rsize *= 2;
- }
-
-
- int avail = c->rsize - c->rbytes;
- res = read(c->sfd, c->rbuf + c->rbytes, avail);
- if (res > 0) {
- STATS_LOCK();
- stats.bytes_read += res;
- STATS_UNLOCK();
- gotdata = 1;
- c->rbytes += res;
- if (res == avail) {
- continue;
- } else {
- break;
- }
- }
- if (res == 0) {
- /* connection closed */
- conn_set_state(c, conn_closing);
- return 1;
- }
- if (res == -1) {
- if (errno == EAGAIN || errno == EWOULDBLOCK) break;
- /* Should close on unhandled errors. */
- conn_set_state(c, conn_closing);
- return 1;
- }
- }
- return gotdata;
-}
-
-static bool update_event(conn *c, const int new_flags) {
- c->ev_flags = new_flags;
- return true;
-}
-
-/*
- * Transmit the next chunk of data from our list of msgbuf structures.
- *
- * Returns:
- * TRANSMIT_COMPLETE All done writing.
- * TRANSMIT_INCOMPLETE More data remaining to write.
- * TRANSMIT_SOFT_ERROR Can't write any more right now.
- * TRANSMIT_HARD_ERROR Can't write (c->state is set to conn_closing)
- */
-static int transmit(conn *c) {
- assert(c != NULL);
-
- if (c->msgcurr < c->msgused &&
- c->msglist[c->msgcurr].msg_iovlen == 0) {
- /* Finished writing the current msg; advance to the next. */
- c->msgcurr++;
- }
- if (c->msgcurr < c->msgused) {
- ssize_t res;
- struct msghdr *m = &c->msglist[c->msgcurr];
-
- res = sendmsg(c->sfd, m, 0);
- if (res > 0) {
- STATS_LOCK();
- stats.bytes_written += res;
- STATS_UNLOCK();
-
- /* We've written some of the data. Remove the completed
- iovec entries from the list of pending writes. */
- while (m->msg_iovlen > 0 && res >= m->msg_iov->iov_len) {
- res -= m->msg_iov->iov_len;
- m->msg_iovlen--;
- m->msg_iov++;
- }
-
- /* Might have written just part of the last iovec entry;
- adjust it so the next write will do the rest. */
- if (res > 0) {
- m->msg_iov->iov_base += res;
- m->msg_iov->iov_len -= res;
- }
- return TRANSMIT_INCOMPLETE;
- }
- if (res == -1 && (errno == EAGAIN || errno == EWOULDBLOCK)) {
- update_event(c, AE_WRITABLE);
- return TRANSMIT_SOFT_ERROR;
- }
- /* if res==0 or res==-1 and error is not EAGAIN or EWOULDBLOCK,
- we have a real error, on which we close the connection */
- if (settings.verbose > 0)
- perror("Failed to write, and not due to blocking");
-
- conn_set_state(c, conn_closing);
- return TRANSMIT_HARD_ERROR;
- } else {
- return TRANSMIT_COMPLETE;
- }
-}
-
-void drive_machine(conn *c) {
- bool stop = false;
- int sfd, flags = 1;
- socklen_t addrlen;
- struct sockaddr_storage addr;
- int res;
-
- assert(c != NULL);
-
- while (!stop) {
-
- switch(c->state) {
- case conn_listening:
- addrlen = sizeof(addr);
- if ((sfd = accept(c->sfd, (struct sockaddr *)&addr, &addrlen)) == -1) {
- stop = true;
- if (errno == EAGAIN || errno == EWOULDBLOCK) {
- /* these are transient, so don't log anything */
- } else if (errno == EMFILE) {
- if (settings.verbose > 0)
- fprintf(stderr, "Too many open connections\n");
- if (stub_fd > 0){
- close(stub_fd);
- if ((sfd = accept(c->sfd, (struct sockaddr *)&addr, &addrlen)) != -1) {
- close(sfd);
- stub_fd = open("/dev/null", O_RDONLY);
- stop = false;
- }else{
- if (settings.verbose > 0)
- fprintf(stderr, "Too many open connections 2\n");
- }
- }
- } else {
- perror("accept()");
- }
- if (stop) break;
- }
- if ((flags = fcntl(sfd, F_GETFL, 0)) < 0 ||
- fcntl(sfd, F_SETFL, flags | O_NONBLOCK) < 0) {
- perror("setting O_NONBLOCK");
- close(sfd);
- break;
- }
- if (NULL == conn_new(sfd, conn_read, DATA_BUFFER_SIZE)) {
- if (settings.verbose > 0) {
- fprintf(stderr, "Can't listen for events on fd %d\n", sfd);
- }
- close(sfd);
- }
- break;
-
- case conn_read:
- if (try_read_command(c) != 0) {
- continue;
- }
- if (try_read_network(c) != 0) {
- continue;
- }
- /* we have no command line and no data to read from network */
- update_event(c, AE_READABLE);
- stop = true;
- break;
-
- case conn_nread:
- /* we are reading rlbytes into ritem; */
- if (c->rlbytes == 0) {
- complete_nread(c);
- break;
- }
- /* first check if we have leftovers in the conn_read buffer */
- if (c->rbytes > 0) {
- int tocopy = c->rbytes > c->rlbytes ? c->rlbytes : c->rbytes;
- memcpy(c->ritem, c->rcurr, tocopy);
- c->ritem += tocopy;
- c->rlbytes -= tocopy;
- c->rcurr += tocopy;
- c->rbytes -= tocopy;
- break;
- }
-
- /* now try reading from the socket */
- res = read(c->sfd, c->ritem, c->rlbytes);
- if (res > 0) {
- STATS_LOCK();
- stats.bytes_read += res;
- STATS_UNLOCK();
- c->ritem += res;
- c->rlbytes -= res;
- break;
- }
- if (res == 0) { /* end of stream */
- conn_set_state(c, conn_closing);
- break;
- }
- if (res == -1 && (errno == EAGAIN || errno == EWOULDBLOCK)) {
- update_event(c, AE_READABLE);
- stop = true;
- break;
- }
- /* otherwise we have a real error, on which we close the connection */
- if (settings.verbose > 0)
- fprintf(stderr, "Failed to read, and not due to blocking\n");
- conn_set_state(c, conn_closing);
- break;
-
- case conn_swallow:
- /* we are reading sbytes and throwing them away */
- if (c->sbytes == 0) {
- conn_set_state(c, conn_read);
- break;
- }
-
- /* first check if we have leftovers in the conn_read buffer */
- if (c->rbytes > 0) {
- int tocopy = c->rbytes > c->sbytes ? c->sbytes : c->rbytes;
- c->sbytes -= tocopy;
- c->rcurr += tocopy;
- c->rbytes -= tocopy;
- break;
- }
-
- /* now try reading from the socket */
- res = read(c->sfd, c->rbuf, c->rsize > c->sbytes ? c->sbytes : c->rsize);
- if (res > 0) {
- STATS_LOCK();
- stats.bytes_read += res;
- STATS_UNLOCK();
- c->sbytes -= res;
- break;
- }
- if (res == 0) { /* end of stream */
- conn_set_state(c, conn_closing);
- break;
- }
- if (res == -1 && (errno == EAGAIN || errno == EWOULDBLOCK)) {
- update_event(c, AE_READABLE);
- stop = true;
- break;
- }
- /* otherwise we have a real error, on which we close the connection */
- if (settings.verbose > 0)
- fprintf(stderr, "Failed to read, and not due to blocking\n");
- conn_set_state(c, conn_closing);
- break;
-
- case conn_write:
- /*
- * We want to write out a simple response. If we haven't already,
- * assemble it into a msgbuf list (this will be a single-entry
- * list for TCP or a two-entry list for UDP).
- */
- if (c->iovused == 0) {
- if (add_iov(c, c->wcurr, c->wbytes) != 0) {
- if (settings.verbose > 0)
- fprintf(stderr, "Couldn't build response\n");
- conn_set_state(c, conn_closing);
- break;
- }
- }
-
- /* fall through... */
-
- case conn_mwrite:
- switch (transmit(c)) {
- case TRANSMIT_COMPLETE:
- if (c->state == conn_mwrite) {
- while (c->ileft > 0) {
- item *it = *(c->icurr);
- item_free(it);
- c->icurr++;
- c->ileft--;
- }
- conn_set_state(c, conn_read);
- } else if (c->state == conn_write) {
- if (c->write_and_free) {
- free(c->write_and_free);
- c->write_and_free = 0;
- }
- conn_set_state(c, c->write_and_go);
- } else {
- if (settings.verbose > 0)
- fprintf(stderr, "Unexpected state %d\n", c->state);
- conn_set_state(c, conn_closing);
- }
- break;
-
- case TRANSMIT_INCOMPLETE:
- case TRANSMIT_HARD_ERROR:
- break; /* Continue in state machine. */
-
- case TRANSMIT_SOFT_ERROR:
- stop = true;
- break;
- }
- break;
-
- case conn_closing:
- conn_close(c);
- stop = true;
- break;
- }
- }
-
- return;
-}
-
-static int new_socket(struct addrinfo *ai) {
- int sfd;
- int flags;
-
- if ((sfd = socket(ai->ai_family, ai->ai_socktype, ai->ai_protocol)) == -1) {
- perror("socket()");
- return -1;
- }
-
- if ((flags = fcntl(sfd, F_GETFL, 0)) < 0 ||
- fcntl(sfd, F_SETFL, flags | O_NONBLOCK) < 0) {
- perror("setting O_NONBLOCK");
- close(sfd);
- return -1;
- }
- return sfd;
-}
-
-static int server_socket(const int port, const bool is_udp) {
- int sfd;
- struct linger ling = {0, 0};
- struct addrinfo *ai;
- struct addrinfo *next;
- struct addrinfo hints;
- char port_buf[NI_MAXSERV];
- int error;
- int success = 0;
-
- int flags =1;
-
- /*
- * the memset call clears nonstandard fields in some impementations
- * that otherwise mess things up.
- */
- memset(&hints, 0, sizeof (hints));
- hints.ai_flags = AI_PASSIVE|AI_ADDRCONFIG;
- hints.ai_family = AF_UNSPEC;
- hints.ai_protocol = IPPROTO_TCP;
- hints.ai_socktype = SOCK_STREAM;
-
- snprintf(port_buf, NI_MAXSERV, "%d", port);
- error= getaddrinfo(settings.inter, port_buf, &hints, &ai);
- if (error != 0) {
- if (error != EAI_SYSTEM)
- fprintf(stderr, "getaddrinfo(): %s\n", gai_strerror(error));
- else
- perror("getaddrinfo()");
-
- return 1;
- }
-
- for (next= ai; next; next= next->ai_next) {
- conn *listen_conn_add;
- if ((sfd = new_socket(next)) == -1) {
- freeaddrinfo(ai);
- return 1;
- }
-
- setsockopt(sfd, SOL_SOCKET, SO_REUSEADDR, (void *)&flags, sizeof(flags));
- setsockopt(sfd, SOL_SOCKET, SO_KEEPALIVE, (void *)&flags, sizeof(flags));
- setsockopt(sfd, SOL_SOCKET, SO_LINGER, (void *)&ling, sizeof(ling));
- setsockopt(sfd, IPPROTO_TCP, TCP_NODELAY, (void *)&flags, sizeof(flags));
-
- if (bind(sfd, next->ai_addr, next->ai_addrlen) == -1) {
- if (errno != EADDRINUSE) {
- perror("bind()");
- close(sfd);
- freeaddrinfo(ai);
- return 1;
- }
- close(sfd);
- continue;
- } else {
- success++;
- if (listen(sfd, 1024) == -1) {
- perror("listen()");
- close(sfd);
- freeaddrinfo(ai);
- return 1;
- }
- }
-
- if (!(listen_conn_add = conn_new(sfd, conn_listening, 1))) {
- fprintf(stderr, "failed to create listening connection\n");
- exit(EXIT_FAILURE);
- }
- }
-
- freeaddrinfo(ai);
-
- /* Return zero iff we detected no errors in starting up connections */
- return success == 0;
-}
-
-static void usage(void) {
- printf(PACKAGE " " VERSION "\n");
- printf("-p <num> TCP port number to listen on (default: 8964)\n"
- "-l <ip_addr> interface to listen on, default is INDRR_ANY\n"
- "-d run as a daemon\n"
- "-P <num> page cache limit(MB), default 256(MB)\n"
- "-r <num> record cache limit(MB), default 256(MB)\n"
- "-R <num> bytes for a disk read operation, must be between[1024,65535), recommend to be larger than most small records, default is 4096(Bytes)\n"
- "-L <file> log file\n"
- "-u <username> assume identity of <username> (only when run as root)\n"
- "-c <num> max simultaneous connections, default is 1024\n"
- "-t <num> number of threads to use, default 16\n"
- "-H <dir> home of database, default is 'testdb', keep sure the directory exists\n"
- "-s <num> slow command time limit, in ms, default is 100ms\n"
- "-n main hash table size, recommend to be 1%% - 10%% of maximum record num, default is 1000000\n"
- "-v verbose (print errors/warnings while in event loop)\n"
- "-vv very verbose (also print client commands/reponses)\n"
- "-h print this help and exit\n"
- "-i print license info\n"
- );
-
- return;
-}
-
-static void usage_license(void) {
- printf(PACKAGE " " VERSION "\n\n");
- printf(
- "Copyright (c) 2012, Siyuan Fu. <fusiyuan2010@gmail.com>\n"
- "All rights reserved.\n"
- "\n"
- "\n"
- "This product includes software developed by Douban Inc.\n"
- "\n"
- "[ Beansdb ]\n"
- "\n"
- "Copyright (c) 2009, Douban Inc. <http://www.douban.com/>\n"
- "All rights reserved.\n"
- "\n"
- "Redistribution and use in source and binary forms, with or without\n"
- "modification, are permitted provided that the following conditions are\n"
- "met:\n"
- "\n"
- " * Redistributions of source code must retain the above copyright\n"
- "notice, this list of conditions and the following disclaimer.\n"
- "\n"
- " * Redistributions in binary form must reproduce the above\n"
- "copyright notice, this list of conditions and the following disclaimer\n"
- "in the documentation and/or other materials provided with the\n"
- "distribution.\n"
- "\n"
- " * Neither the name of the Douban Inc. nor the names of its\n"
- "contributors may be used to endorse or promote products derived from\n"
- "this software without specific prior written permission.\n"
- "\n"
- "THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS\n"
- "\"AS IS\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT\n"
- "LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR\n"
- "A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT\n"
- "OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,\n"
- "SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT\n"
- "LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,\n"
- "DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY\n"
- "THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT\n"
- "(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE\n"
- "OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n"
- "\n"
- "\n"
- "This product includes software developed by Douban Inc.\n"
- "\n"
- "[ memcachedb ]\n"
- "\n"
- "Copyright (c) 2008, Steve Chu. <stvchu@gmail.com>\n"
- "All rights reserved.\n"
- "\n"
- "Redistribution and use in source and binary forms, with or without\n"
- "modification, are permitted provided that the following conditions are\n"
- "met:\n"
- "\n"
- " * Redistributions of source code must retain the above copyright\n"
- "notice, this list of conditions and the following disclaimer.\n"
- "\n"
- " * Redistributions in binary form must reproduce the above\n"
- "copyright notice, this list of conditions and the following disclaimer\n"
- "in the documentation and/or other materials provided with the\n"
- "distribution.\n"
- "\n"
- " * Neither the name of the Danga Interactive nor the names of its\n"
- "contributors may be used to endorse or promote products derived from\n"
- "this software without specific prior written permission.\n"
- "\n"
- "THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS\n"
- "\"AS IS\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT\n"
- "LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR\n"
- "A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT\n"
- "OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,\n"
- "SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT\n"
- "LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,\n"
- "DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY\n"
- "THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT\n"
- "(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE\n"
- "OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n"
- "\n"
- "\n"
- "This product includes software developed by Danga Interactive, Inc.\n"
- "\n"
- "[ memcached ]\n"
- "\n"
- "Copyright (c) 2003, Danga Interactive, Inc. <http://www.danga.com/>\n"
- "All rights reserved.\n"
- "\n"
- "Redistribution and use in source and binary forms, with or without\n"
- "modification, are permitted provided that the following conditions are\n"
- "met:\n"
- "\n"
- " * Redistributions of source code must retain the above copyright\n"
- "notice, this list of conditions and the following disclaimer.\n"
- "\n"
- " * Redistributions in binary form must reproduce the above\n"
- "copyright notice, this list of conditions and the following disclaimer\n"
- "in the documentation and/or other materials provided with the\n"
- "distribution.\n"
- "\n"
- " * Neither the name of the Danga Interactive nor the names of its\n"
- "contributors may be used to endorse or promote products derived from\n"
- "this software without specific prior written permission.\n"
- "\n"
- "THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS\n"
- "\"AS IS\" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT\n"
- "LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR\n"
- "A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT\n"
- "OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,\n"
- "SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT\n"
- "LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,\n"
- "DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY\n"
- "THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT\n"
- "(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE\n"
- "OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n"
- "\n"
- "\n"
- "This product includes software developed by Niels Provos.\n"
- "\n"
- "[ libevent ]\n"
- "\n"
- "Copyright 2000-2003 Niels Provos <provos@citi.umich.edu>\n"
- "All rights reserved.\n"
- "\n"
- "Redistribution and use in source and binary forms, with or without\n"
- "modification, are permitted provided that the following conditions\n"
- "are met:\n"
- "1. Redistributions of source code must retain the above copyright\n"
- " notice, this list of conditions and the following disclaimer.\n"
- "2. Redistributions in binary form must reproduce the above copyright\n"
- " notice, this list of conditions and the following disclaimer in the\n"
- " documentation and/or other materials provided with the distribution.\n"
- "3. All advertising materials mentioning features or use of this software\n"
- " must display the following acknowledgement:\n"
- " This product includes software developed by Niels Provos.\n"
- "4. The name of the author may not be used to endorse or promote products\n"
- " derived from this software without specific prior written permission.\n"
- "\n"
- "THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR\n"
- "IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES\n"
- "OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.\n"
- "IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,\n"
- "INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT\n"
- "NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,\n"
- "DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY\n"
- "THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT\n"
- "(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF\n"
- "THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.\n"
- );
-
- return;
-}
-
-
-/* for safely exit, make sure to do checkpoint*/
-static void sig_handler(const int sig)
-{
- if (sig != SIGTERM && sig != SIGQUIT && sig != SIGINT) {
- return;
- }
- if (daemon_quit == 1){
- return;
- }
- daemon_quit = 1;
- fprintf(stderr, "Signal(%d) received, try to exit daemon gracefully..\n", sig);
-}
-
-
-int main (int argc, char **argv) {
- int c;
- //struct in_addr addr;
- char *dbhome = "testdb";
- bool daemonize = false;
- char *username = NULL;
- FILE *log_file = NULL;
- struct passwd *pw;
- struct sigaction sa;
- struct rlimit rlim;
- int rcache = 256, pcache = 256;
- /* recommend for 100,000,000 records*/
- int db_hsize = 1000000;
- int areadsize = 4096;
-
- /* init settings */
- settings_init();
-
- /* set stderr non-buffering (for running under, say, daemontools) */
- setbuf(stderr, NULL);
-
- /* process arguments */
- while ((c = getopt(argc, argv, "a:p:c:hivl:dr:u:P:L:t:b:H:s:n:R:")) != -1) {
- switch (c) {
- case 'a':
- if (strcmp(optarg, "-") == 0) {
- access_log = stdout;
- }else{
- access_log = fopen(optarg, "a");
- if (NULL == access_log) {
- fprintf(stderr, "open access_log %s failed\n", optarg);
- exit(1);
- }
- }
- break;
- case 'p':
- settings.port = atoi(optarg);
- break;
- case 'c':
- settings.maxconns = atoi(optarg);
- break;
- case 'h':
- usage();
- exit(EXIT_SUCCESS);
- case 'i':
- usage_license();
- exit(EXIT_SUCCESS);
- case 'v':
- settings.verbose++;
- break;
- case 'l':
- settings.inter= strdup(optarg);
- break;
- case 'd':
- daemonize = true;
- break;
- case 'r':
- rcache = atoi(optarg);
- break;
- case 'R':
- areadsize = atoi(optarg);
- break;
- case 'u':
- username = optarg;
- break;
- case 'P':
- pcache = atoi(optarg);
- break;
- case 'L':
- if ((log_file = fopen(optarg, "a")) != NULL){
- setlinebuf(log_file);
- fclose(stdout);
- fclose(stderr);
- stdout = stderr = log_file;
- }else{
- fprintf(stderr, "open log file %s failed\n", optarg);
- }
- break;
- case 't':
- settings.num_threads = atoi(optarg);
- if (settings.num_threads == 0) {
- fprintf(stderr, "Number of threads must be greater than 0\n");
- exit(EXIT_FAILURE);
- }
- break;
- case 'b':
- settings.item_buf_size = atoi(optarg);
- if(settings.item_buf_size < 512){
- fprintf(stderr, "item buf size must be larger than 512 bytes\n");
- exit(EXIT_FAILURE);
- }
- if(settings.item_buf_size > 256 * 1024){
- fprintf(stderr, "Warning: item buffer size(-b) larger than 256KB may cause performance issue\n");
- }
- break;
- case 'H':
- dbhome = optarg;
- break;
- case 's':
- settings.slow_cmd_time = atoi(optarg) / 1000.0;
- break;
- case 'n':
- db_hsize = atoi(optarg);
- break;
- default:
- fprintf(stderr, "Illegal argument \"%c\"\n", c);
- exit(EXIT_FAILURE);
- }
- }
-
- /*
- * If needed, increase rlimits to allow as many connections
- * as needed.
- */
-
- if (getrlimit(RLIMIT_NOFILE, &rlim) != 0) {
- fprintf(stderr, "failed to getrlimit number of files\n");
- exit(EXIT_FAILURE);
- } else {
- int maxfiles = settings.maxconns;
- if (rlim.rlim_cur < maxfiles)
- rlim.rlim_cur = maxfiles + 3;
- if (rlim.rlim_max < rlim.rlim_cur)
- rlim.rlim_max = rlim.rlim_cur;
- if (setrlimit(RLIMIT_NOFILE, &rlim) != 0) {
- fprintf(stderr, "failed to set rlimit for open files. Try running as root or requesting smaller maxconns value.\n");
- exit(EXIT_FAILURE);
- }
- }
-
- /* daemonize if requested */
- /* if we want to ensure our ability to dump core, don't chdir to / */
- if (daemonize) {
- int res;
- res = daemon(1, settings.verbose || log_file);
- if (res == -1) {
- fprintf(stderr, "failed to daemon() in order to daemonize\n");
- return 1;
- }
- }
-
- /* lose root privileges if we have them */
- if (getuid() == 0 || geteuid() == 0) {
- if (username == 0 || *username == '\0') {
- fprintf(stderr, "can't run as root without the -u switch\n");
- return 1;
- }
- if ((pw = getpwnam(username)) == 0) {
- fprintf(stderr, "can't find the user %s to switch to\n", username);
- return 1;
- }
- if (setgid(pw->pw_gid) < 0 || setuid(pw->pw_uid) < 0) {
- fprintf(stderr, "failed to assume identity of user %s\n", username);
- return 1;
- }
- }
-
- /* initialize other stuff */
- item_init();
- stats_init();
- conn_init();
-
- /*
- * ignore SIGPIPE signals; we can use errno==EPIPE if we
- * need that information
- */
- sa.sa_handler = SIG_IGN;
- sa.sa_flags = 0;
- if (sigemptyset(&sa.sa_mask) == -1 ||
- sigaction(SIGPIPE, &sa, 0) == -1) {
- perror("failed to ignore SIGPIPE; sigaction");
- exit(EXIT_FAILURE);
- }
-
- /* open db */
- db = cdb_new();
- cdb_option(db, db_hsize, rcache, pcache);
- cdb_option_areadsize(db, areadsize);
-
- if (cdb_open(db, dbhome, CDB_CREAT | CDB_PAGEWARMUP) < 0) {
- fprintf(stderr, "failed to open db %s\n", dbhome);
- exit(1);
- }
-
- if ((stub_fd = open("/dev/null", O_RDONLY)) == -1) {
- perror("open stub file failed");
- exit(1);
- }
- thread_init(settings.num_threads);
-
- /* create the listening socket, bind it, and init */
- if (server_socket(settings.port, false)) {
- fprintf(stderr, "failed to listen\n");
- exit(EXIT_FAILURE);
- }
-
- /* register signal callback */
- if (signal(SIGTERM, sig_handler) == SIG_ERR)
- fprintf(stderr, "can not catch SIGTERM\n");
- if (signal(SIGQUIT, sig_handler) == SIG_ERR)
- fprintf(stderr, "can not catch SIGQUIT\n");
- if (signal(SIGINT, sig_handler) == SIG_ERR)
- fprintf(stderr, "can not catch SIGINT\n");
-
- /* enter the event loop */
- printf("all ready.\n");
- loop_run(settings.num_threads);
-
- /* wait other thread to ends */
- fprintf(stderr, "waiting for close ... \n");
- cdb_destroy(db);
- fprintf(stderr, "done.\n");
-
- if (log_file) {
- fclose(log_file);
- }
-
- return 0;
-}
-
diff --git a/3rdparty/cuttdb/src/cuttdb-server.h b/3rdparty/cuttdb/src/cuttdb-server.h
deleted file mode 100644
index 90cc9b6271683c058dfacd007fad43ffef239c55..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cuttdb-server.h
+++ /dev/null
@@ -1,270 +0,0 @@
-/*
- * Beansdb - A high available distributed key-value storage system:
- *
- * http://beansdb.googlecode.com
- *
- * The source code of Beansdb is most based on Memcachedb and Memcached:
- *
- * http://memcachedb.org/
- * http://danga.com/memcached/
- *
- * Copyright 2009 Douban Inc. All rights reserved.
- *
- * Use and distribution licensed under the BSD license. See
- * the LICENSE file for full text.
- *
- * Authors:
- * Davies Liu <davies.liu@gmail.com>
- *
- */
-
-#ifdef HAVE_CONFIG_H
-#include "config.h"
-#endif
-
-#include <sys/types.h>
-#include <sys/socket.h>
-#include <sys/time.h>
-#include <netinet/in.h>
-#include <netdb.h>
-
-#define DATA_BUFFER_SIZE 2048
-#define MAX_PAYLOAD_SIZE 1400
-#define MAX_SENDBUF_SIZE (256 * 1024 * 1024)
-/* I'm told the max legnth of a 64-bit num converted to string is 20 bytes.
- * Plus a few for spaces, \r\n, \0 */
-#define SUFFIX_SIZE 24
-#define INCR_MAX_STORAGE_LEN 24
-
-/** Initial size of list of items being returned by "get". */
-#define ITEM_LIST_INITIAL 200
-
-/** Initial size of the sendmsg() scatter/gather array. */
-#define IOV_LIST_INITIAL 400
-
-/** Initial number of sendmsg() argument structures to allocate. */
-#define MSG_LIST_INITIAL 10
-
-/** High water marks for buffer shrinking */
-#define READ_BUFFER_HIGHWAT 8192
-#define ITEM_LIST_HIGHWAT 400
-#define IOV_LIST_HIGHWAT 600
-#define MSG_LIST_HIGHWAT 100
-
-#define MAX_REP_PRIORITY 1000000
-#define MAX_REP_ACK_POLICY 6
-#define MAX_REP_NSITES 1000
-
-
-#define RGET_MAX_ITEMS 100
-#define PACKAGE "CuttDB"
-#define VERSION "0.1.0"
-
-/* Get a consistent bool type */
-#include <stdbool.h>
-
-#if HAVE_STDINT_H
-# include <stdint.h>
-#else
- typedef unsigned char uint8_t;
-#endif
-
-/* unistd.h is here */
-#if HAVE_UNISTD_H
-# include <unistd.h>
-#endif
-
-/* 64-bit Portable printf */
-/* printf macros for size_t, in the style of inttypes.h */
-#ifdef _LP64
-#define __PRIS_PREFIX "z"
-#else
-#define __PRIS_PREFIX
-#endif
-
-#define AE_SETSIZE (1024*60) /* Max number of fd supported */
-
-#define AE_OK 0
-#define AE_ERR -1
-
-#define AE_NONE 0
-#define AE_READABLE 1
-#define AE_WRITABLE 2
-
-/* Use these macros after a % in a printf format string
- to get correct 32/64 bit behavior, like this:
- size_t size = records.size();
- printf("%"PRIuS"\n", size); */
-
-#define PRIdS __PRIS_PREFIX "d"
-#define PRIxS __PRIS_PREFIX "x"
-#define PRIuS __PRIS_PREFIX "u"
-#define PRIXS __PRIS_PREFIX "X"
-#define PRIoS __PRIS_PREFIX "o"
-
-struct stats {
- uint32_t curr_conns;
- uint32_t total_conns;
- uint32_t conn_structs;
- uint64_t get_cmds;
- uint64_t set_cmds;
- uint64_t delete_cmds;
- uint64_t slow_cmds;
- uint64_t get_hits;
- uint64_t get_misses;
- time_t started; /* when the process was started */
- uint64_t bytes_read;
- uint64_t bytes_written;
-};
-
-#define MAX_VERBOSITY_LEVEL 2
-
-struct settings {
- size_t item_buf_size;
- int maxconns;
- int port;
- char *inter;
- int verbose;
- float slow_cmd_time;
- int flush_period;
- int flush_limit;
- int num_threads; /* number of libevent threads to run */
-};
-
-extern struct stats stats;
-extern struct settings settings;
-
-typedef struct _stritem {
- int expire; /* expire time */
- uint32_t flag; /* flag of item */
- int nbytes; /* size of data */
- uint8_t nsuffix; /* length of flags-and-length string */
- uint8_t nkey; /* key length, w/terminating null and padding */
- void * end[];
- /* then null-terminated key */
- /* then " flags length\r\n" (no terminating null) */
- /* then data with terminating \r\n (no terminating null; it's binary!) */
-} item;
-
-#define ITEM_key(item) ((char*)&((item)->end[0]))
-
-/* warning: don't use these macros with a function, as it evals its arg twice */
-#define ITEM_suffix(item) ((char*) &((item)->end[0]) + (item)->nkey + 1)
-#define ITEM_data(item) ((char*) &((item)->end[0]) + (item)->nkey + 1 + (item)->nsuffix)
-#define ITEM_ntotal(item) (sizeof(struct _stritem) + (item)->nkey + 1 + (item)->nsuffix + (item)->nbytes)
-
-enum conn_states {
- conn_listening, /** the socket which listens for connections */
- conn_read, /** reading in a command line */
- conn_write, /** writing out a simple response */
- conn_nread, /** reading in a fixed number of bytes */
- conn_swallow, /** swallowing unnecessary bytes w/o storing */
- conn_closing, /** closing this connection */
- conn_mwrite, /** writing out many items sequentially */
-};
-
-#define NREAD_ADD 1
-#define NREAD_SET 2
-#define NREAD_REPLACE 3
-#define NREAD_APPEND 4
-#define NREAD_PREPEND 5
-
-typedef struct conn conn;
-struct conn {
- int sfd;
- int state;
- short ev_flags;
-
- char *rbuf; /** buffer to read commands into */
- char *rcurr; /** but if we parsed some already, this is where we stopped */
- int rsize; /** total allocated size of rbuf */
- int rbytes; /** how much data, starting from rcur, do we have unparsed */
-
- char *wbuf;
- char *wcurr;
- int wsize;
- int wbytes;
- int write_and_go; /** which state to go into after finishing current write */
- void *write_and_free; /** free this memory after finishing writing */
- bool noreply; /* True if the reply should not be sent. */
-
- char *ritem; /** when we read in an item's value, it goes here */
- int rlbytes;
-
- /* data for the nread state */
-
- /**
- * item is used to hold an item structure created after reading the command
- * line of set/add/replace commands, but before we finished reading the actual
- * data. The data is read into ITEM_data(item) to avoid extra copying.
- */
-
- void *item; /* for commands set/add/replace */
- int item_comm; /* which one is it: set/add/replace */
-
- /* data for the swallow state */
- int sbytes; /* how many bytes to swallow */
-
- /* data for the mwrite state */
- struct iovec *iov;
- int iovsize; /* number of elements allocated in iov[] */
- int iovused; /* number of elements used in iov[] */
-
- struct msghdr *msglist;
- int msgsize; /* number of elements allocated in msglist[] */
- int msgused; /* number of elements used in msglist[] */
- int msgcurr; /* element in msglist[] being transmitted now */
- int msgbytes; /* number of bytes in current msg */
-
- item **ilist; /* list of items to write out */
- int isize;
- item **icurr;
- int ileft;
-
- conn *next; /* Used for generating a list of conn structures */
-};
-
-/*
- * Functions
- */
-
-/* item management */
-/*
-void item_init(void);
-item *do_item_from_freelist(void);
-int do_item_add_to_freelist(item *it);
-item *item_alloc1(char *key, const size_t nkey, const int flags, const int nbytes);
-int item_free(item *it);
-item *item_get(char *key, size_t nkey);
-*/
-
-/* conn management */
-conn *do_conn_from_freelist();
-bool do_conn_add_to_freelist(conn *c);
-conn *conn_new(const int sfd, const int init_state, const int read_buffer_size);
-
-int store_item(item *item, int comm);
-
-void thread_init(int nthreads);
-int add_event(int fd, int mask, conn *c);
-void loop_run(int nthreads);
-
-void drive_machine(conn *c);
-
-/* Lock wrappers for cache functions that are called from main loop. */
-conn *mt_conn_from_freelist(void);
-bool mt_conn_add_to_freelist(conn *c);
-item *mt_item_from_freelist(void);
-int mt_item_add_to_freelist(item *it);
-void mt_stats_lock(void);
-void mt_stats_unlock(void);
-
-#define conn_from_freelist() mt_conn_from_freelist()
-#define conn_add_to_freelist(x) mt_conn_add_to_freelist(x)
-#define item_from_freelist() mt_item_from_freelist()
-#define item_add_to_freelist(x) mt_item_add_to_freelist(x)
-#define STATS_LOCK() mt_stats_lock()
-#define STATS_UNLOCK() mt_stats_unlock()
-
-extern int daemon_quit;
-
diff --git a/3rdparty/cuttdb/src/cuttdb.c b/3rdparty/cuttdb/src/cuttdb.c
deleted file mode 100644
index 74e342623a5308fd4275868263eaac87c1c726e7..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cuttdb.c
+++ /dev/null
@@ -1,21 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "cuttdb.h"
-#include "cdb_types.h"
-#include "cdb_vio.h"
-
-
-/* nothing here */
diff --git a/3rdparty/cuttdb/src/cuttdb.h b/3rdparty/cuttdb/src/cuttdb.h
deleted file mode 100644
index 1814028f51f0ebd8300adc9e2eb34ae684e7218e..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/cuttdb.h
+++ /dev/null
@@ -1,217 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _CUTTDB_H_
-#define _CUTTDB_H_
-#include <stdint.h>
-#include <stdbool.h>
-
-#if defined(__cplusplus)
-extern "C" {
-#endif
-
-typedef struct CDB CDB;
-typedef void (*CDB_ERRCALLBACK)(void *, int, const char *, int);
-typedef bool (*CDB_ITERCALLBACK)(void *, const char *, int, const char *, int, uint32_t, uint64_t);
-
-/* performance statistical information of an database instance */
-typedef struct {
- /* number of records in db */
- uint64_t rnum;
- /* number of records in cache */
- uint64_t rcnum;
- /* number of pages in db */
- uint64_t pnum;
- /* number of pages in cache */
- uint64_t pcnum;
- /* cache hit of record cache */
- uint64_t rchit;
- /* cache miss of record cache */
- uint64_t rcmiss;
- /* cache hit of page cache */
- uint64_t pchit;
- /* cache miss of page cache */
- uint64_t pcmiss;
- /* average disk read latency */
- uint32_t rlatcy;
- /* average disk write latency */
- uint32_t wlatcy;
-} CDBSTAT;
-
-/* options to open a database*/
-enum {
- /* create an database if not exist */
- CDB_CREAT = 0x1,
- /* clean the database if already exist */
- CDB_TRUNC = 0x2,
- /* fill the cache when start up */
- CDB_PAGEWARMUP = 0x4,
-};
-
-/* error codes */
-enum {
- CDB_SUCCESS = 0,
- CDB_NOTFOUND,
- CDB_EXIST,
- CDB_DIRNOEXIST,
- CDB_OPENERR,
- CDB_PIDEXIST,
- CDB_DATAERRDAT,
- CDB_DATAERRIDX,
- CDB_WRITEERR,
- CDB_READERR,
- CDB_NOFID,
- CDB_INTERNALERR,
- CDB_DATAERRMETA,
- CDB_MEMDBNOCACHE,
-};
-
-/* record insertion options */
-enum {
- CDB_OVERWRITE = 0,
- CDB_INSERTIFEXIST = 0x1,
- CDB_INSERTIFNOEXIST = 0x2,
- CDB_INSERTCACHE = 0x8,
-};
-
-/* if database path is CDB_MEMDB, records are never written to disk, they stay in cache only */
-#define CDB_MEMDB ":memory:"
-
-/*
- WARNING:
-
- the library has auxiliary thread, which means do fork() after open a database will cause
- unpredictable situation.
-*/
-
-/* create an cuttdb object, which should be freed by cdb_destory() */
-CDB *cdb_new();
-
-/* cdb_option() must be called before cdb_open()
-
- the second parameter 'hsize' indicates the size of main hash table, which can't be
- modified after the database be created. To get better performance, it is suggest to
- set the 'hsize' to 10% - 1% of the total number of records. The default value 1 million
- should be proper for about 100 million records. Too large or small of the value would
- lead to drop in speed or waste of memory
-
- the third parameter 'rcacheMB' indicates the size limit of record cache (measured by
- MegaBytes), every record in cache would have about 40 bytes overhead.
-
- the fourth parameter 'pcacheMB' indicates the size limit of index page cache (measured
- by MegaBytes). If a record is not in record cache, it will be read by only 1 disk seek
- with enough page cache, or it have to make an extra disk seek to load the page.
- cuttdb will use about {10 * number of records} bytes to cache all index pages, which
- ensures fastest 'set' operation.
-
- the default parameter is (_db, 1000000, 128, 1024)
-
- return 0 if success, or -1 at failure. */
-int cdb_option(CDB *db, int hsize, int rcacheMB, int pcacheMB);
-
-/* Enable bloomfilter, size should be the estimated number of records in database
- must be called before cdb_open(),
- The value is 100000 at minimum. Memory cost of bloomfilter is size/8 bytes */
-void cdb_option_bloomfilter(CDB *db, uint64_t size);
-
-/* this is an advanced parameter. It is the size for cuttdb making a read from disk.
- CuttDB do not know the record size even if the index is in memory,
- so at least a read with default size will performed while in cdb_get().
- The value is recommended to be larger than the size of most records in database,
- unless the records are mostly larger than tens of KB.
- If the value is much larger than recommended, it will be a waste of computing.
- The value can only be 65536 at maximum, 1024 at minimum */
-void cdb_option_areadsize(CDB *db, uint32_t size);
-
-/* open an database, 'file' should be an existing directory, or CDB_MEMDB for temporary store,
- 'mode' should be combination of CDB_CREAT / CDB_TRUNC / CDB_PAGEWARMUP
- CDB_PAGEWARMUP means to warm up page cache while opening
- If there is a file called 'force_recovery' in the data directory, even if it might be made by 'touch force_recovery',
- a force recovery will happen to rebuild the index (be aware that some deleted records would reappear after this)
- */
-int cdb_open(CDB *db, const char *file, int mode);
-
-
-/* simplified cdb_set2, insert a record with CDB_OVERWRITE and never expire */
-int cdb_set(CDB *db, const char *key, int ksize, const char *val, int vsize);
-
-/* set a record by 'key' and 'value',
- opt could be bit combination of CDB_INSERTCACHE and one in {CDB_INSERTIFEXIST, CDB_INSERTNOEXIST,
- CDB_OVERWRITE}
- expire is the time for the record be valid, measured by second. 0 means never expire.
- return 0 if success, or -1 at failure. */
-int cdb_set2(CDB *db, const char *key, int ksize, const char *val, int vsize, int opt, int expire);
-
-
-int cdb_is(CDB *db, const char *key, int ksize);
-
-/* get an record by 'key', the value will be allocated and passed out by 'val', its size is
- 'vsize'. return 0 if success, or -1 at failure. */
-int cdb_get(CDB *db, const char *key, int ksize, void **val, int *vsize);
-
-
-/* the val got by cdb_get should be freed by this for safety.
- If there is more than one memory allocator */
-void cdb_free_val(void **val);
-
-
-/* delete an record by 'key'. However ,the space of the record would not be recycled.
- 'vsize'. return 0 if success, or -1 at failure. */
-int cdb_del(CDB *db, const char *key, int ksize);
-
-
-/* create a new iterator begins at given operation id */
-void *cdb_iterate_new(CDB *db, uint64_t oid);
-
-/* iterate through the database with a callback, the function would stop if callback returned false
- The callback should accept key, ksize, value, vsize, expire time, oid
- Returns the number of records have been visited */
-uint64_t cdb_iterate(CDB *db, CDB_ITERCALLBACK itcb, void *arg, void *iter);
-
-/* destroy the iterator */
-void cdb_iterate_destroy(CDB *db, void *iter);
-
-/* get the current statistic information of db. 'stat' should be the struct already allocated.
- if 'stat' is NULL, the statistic will be reset to zero. */
-void cdb_stat(CDB *db, CDBSTAT *stat);
-
-
-/* close the database. IT MUST BE CALLED BEFORE PROGRAM EXITS TO ENSURE DATA COMPLETION */
-int cdb_close(CDB *db);
-
-
-/* close the database if it opened, and free the object */
-int cdb_destroy(CDB *db);
-
-
-/* get last error number in current thread */
-int cdb_errno(CDB *db);
-
-
-/* get the description of an error number */
-const char *cdb_errmsg(int ecode);
-
-
-/* set callback when error happened, 'cdb_deferrorcb' is optional, which shows the error to stderr */
-void cdb_seterrcb(CDB *db, CDB_ERRCALLBACK errcb, void *arg);
-
-/* a possible error callback */
-void cdb_deferrorcb(void *arg, int errno, const char *file, int line);
-
-#if defined(__cplusplus)
-}
-#endif
-
-#endif
diff --git a/3rdparty/cuttdb/src/mman.c b/3rdparty/cuttdb/src/mman.c
deleted file mode 100644
index ea5d358adc7768acd227507d7ac9e7853823729d..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/mman.c
+++ /dev/null
@@ -1,172 +0,0 @@
-/*
- * mman-win32 library
- * https://code.google.com/p/mman-win32/
- * reinterpreted by Konstantin Papizh <konstantin.papizh@demlabs.net>
- * DeM Labs Inc. https://demlabs.net
- */
-
-#include <windows.h>
-#include <errno.h>
-#include <stdio.h>
-#include "mman.h"
-
-static DWORD __map_mmap_prot_page(const int prot) {
- DWORD protect = 0;
-
- if (prot == PROT_NONE)
- return protect;
-
- if ((prot & PROT_EXEC) != 0) {
- protect = ((prot & PROT_WRITE) != 0) ?
- PAGE_EXECUTE_READWRITE : PAGE_EXECUTE_READ;
- } else {
- protect = ((prot & PROT_WRITE) != 0) ?
- PAGE_READWRITE : PAGE_READONLY;
- }
- return protect;
-}
-
-static DWORD __map_mmap_prot_file(const int prot) {
-
- DWORD desiredAccess = 0;
- if (prot == PROT_NONE)
- return desiredAccess;
-
- if ((prot & PROT_READ) != 0)
- desiredAccess |= FILE_MAP_READ;
- if ((prot & PROT_WRITE) != 0)
- desiredAccess |= FILE_MAP_WRITE;
- if ((prot & PROT_EXEC) != 0)
- desiredAccess |= FILE_MAP_EXECUTE;
-
- return desiredAccess;
-}
-
-void* mmap(void *addr, size_t len, int prot, int flags, int fildes, offset_t off)
-{
- HANDLE fm, h;
- void *map = MAP_FAILED;
-
- const DWORD dwFileOffsetLow = (sizeof(offset_t) <= sizeof(DWORD)) ?
- (DWORD)off : (DWORD)(off & 0xFFFFFFFFL);
- const DWORD dwFileOffsetHigh = (sizeof(offset_t) <= sizeof(DWORD)) ?
- (DWORD)0 : (DWORD)((off >> 32) & 0xFFFFFFFF00000000L);
- const DWORD protect = __map_mmap_prot_page(prot);
- const DWORD desiredAccess = __map_mmap_prot_file(prot);
-
- const offset_t maxSize = off + (offset_t)len;
-
- const DWORD dwMaxSizeLow = (sizeof(offset_t) <= sizeof(DWORD)) ?
- (DWORD)maxSize : (DWORD)(maxSize & 0xFFFFFFFFL);
- const DWORD dwMaxSizeHigh = (sizeof(offset_t) <= sizeof(DWORD)) ?
- (DWORD)0 : (DWORD)((maxSize >> 32) & 0xFFFFFFFF00000000L);
- _set_errno(0);
-
- if (len == 0 || prot == PROT_EXEC) {
- _set_errno(EINVAL);
- return MAP_FAILED;
- }
-
- h = ((flags & MAP_ANONYMOUS) == 0) ?
- (HANDLE)_get_osfhandle(fildes) : INVALID_HANDLE_VALUE;
-
- if ((flags & MAP_ANONYMOUS) == 0 && h == INVALID_HANDLE_VALUE) {
- _set_errno(EBADF);
- return MAP_FAILED;
- }
-
- fm = CreateFileMapping(h, NULL, protect, dwMaxSizeHigh, dwMaxSizeLow, NULL);
-
- if (fm == NULL) {
- int a = errno;
- _set_errno(GetLastError());
- a = errno;
- printf("%d", a);
- return MAP_FAILED;
- }
-
- if ((flags & MAP_FIXED) == 0) {
- map = MapViewOfFile(fm, desiredAccess, dwFileOffsetHigh, dwFileOffsetLow, len);
- }
- else {
- map = MapViewOfFileEx(fm, desiredAccess, dwFileOffsetHigh, dwFileOffsetLow, len, addr);
- }
- CloseHandle(fm);
-
- if (map == NULL) {
- _set_errno(GetLastError());
- return MAP_FAILED;
- }
- return map;
-}
-
-int munmap(void *addr, size_t len) {
- if (UnmapViewOfFile(addr))
- return 0;
-
- _set_errno(GetLastError());
- return -1;
-}
-
-int _mprotect(void *addr, size_t len, int prot) {
- DWORD newProtect = __map_mmap_prot_page(prot);
- DWORD oldProtect = 0;
-
- if (VirtualProtect(addr, len, newProtect, &oldProtect))
- return 0;
- _set_errno(GetLastError());
- return -1;
-}
-
-int msync(void *addr, size_t len, int flags) {
- if (FlushViewOfFile(addr, len))
- return 0;
- _set_errno(GetLastError());
- return -1;
-}
-
-int mlock(const void *addr, size_t len) {
- if (VirtualLock((LPVOID)addr, len))
- return 0;
- _set_errno(GetLastError());
- return -1;
-}
-
-int munlock(const void *addr, size_t len) {
- if (VirtualUnlock((LPVOID)addr, len))
- return 0;
- _set_errno(GetLastError());
- return -1;
-}
-
-ssize_t pread(int fd, void *buf, unsigned long count, offset_t offset) {
- unsigned long len = 0;
-
- OVERLAPPED overlapped;
- memset(&overlapped, 0, sizeof(OVERLAPPED));
- overlapped.OffsetHigh = (uint32_t)((offset & 0xFFFFFFFF00000000LL) >> 32);
- overlapped.Offset = (uint32_t)(offset & 0xFFFFFFFFLL);
-
- HANDLE file = (HANDLE)_get_osfhandle(fd);
- if ((!ReadFile(file, buf, count, &len, &overlapped)) && GetLastError() != ERROR_HANDLE_EOF) {
- _set_errno(GetLastError());
- return -1;
- }
- return len;
-}
-
-ssize_t pwrite(int fd, const void *buf, unsigned long count, offset_t offset) {
- long unsigned int len = 0;
-
- OVERLAPPED overlapped;
- memset(&overlapped, 0, sizeof(OVERLAPPED));
- overlapped.OffsetHigh = (uint32_t)((offset & 0xFFFFFFFF00000000LL) >> 32);
- overlapped.Offset = (uint32_t)(offset & 0xFFFFFFFFLL);
-
- HANDLE file = (HANDLE)_get_osfhandle(fd);
- if (!WriteFile(file, buf, count, &len, &overlapped)) {
- _set_errno(GetLastError());
- return -1;
- }
- return len;
-}
diff --git a/3rdparty/cuttdb/src/mman.h b/3rdparty/cuttdb/src/mman.h
deleted file mode 100644
index 7367f3d4cfa4a37ff423e6202cfb4d79d6058c8e..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/mman.h
+++ /dev/null
@@ -1,55 +0,0 @@
-#ifndef _MMAN_H_
-#define _MMAN_H_
-
-#include <_mingw.h>
-#include <stdint.h>
-#include <io.h>
-
-#if defined(_WIN64)
-typedef int64_t offset_t;
-#else
-typedef uint32_t offset_t;
-#endif
-
-#include <sys/types.h>
-#include <stdbool.h>
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-#define PROT_NONE 0
-#define PROT_READ 1
-#define PROT_WRITE 2
-#define PROT_EXEC 4
-
-#define MAP_FILE 0
-#define MAP_SHARED 1
-#define MAP_PRIVATE 2
-#define MAP_TYPE 0xf
-#define MAP_FIXED 0x10
-#define MAP_ANONYMOUS 0x20
-#define MAP_ANON MAP_ANONYMOUS
-
-#define MAP_FAILED ((void *)-1)
-
-#define MS_ASYNC 1
-#define MS_SYNC 2
-#define MS_INVALIDATE 4
-
-#define fdatasync(fd) _commit(fd)
-
-void* mmap(void *addr, size_t len, int prot, int flags, int fildes, offset_t offset);
-int munmap(void *addr, size_t len);
-int _mprotect(void *addr, size_t len, int prot);
-int msync(void *addr, size_t len, int flags);
-int mlock(const void *addr, size_t len);
-int munlock(const void *addr, size_t len);
-
-ssize_t pread(int fd, void *buf, unsigned long count, offset_t offset);
-ssize_t pwrite(int fd, const void *buf, unsigned long count, offset_t offset);
-
-#ifdef __cplusplus
-}
-#endif
-
-#endif /* _MMAN_H_ */
diff --git a/3rdparty/cuttdb/src/server-thread.c b/3rdparty/cuttdb/src/server-thread.c
deleted file mode 100644
index c7a05c30319e63a177178ab43a84bc7b5435fb11..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/server-thread.c
+++ /dev/null
@@ -1,217 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * The server&network part of CuttDB is based on Beansdb:
- *
- * http://beansdb.googlecode.com
- *
- * Beansdb is most based on Memcachedb and Memcached:
- *
- * http://memcachedb.org/
- * http://danga.com/memcached/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-#include "cuttdb-server.h"
-#include <stdio.h>
-#include <errno.h>
-#include <stdlib.h>
-#include <errno.h>
-#include <assert.h>
-#include <string.h>
-#include <unistd.h>
-
-#ifdef HAVE_MALLOC_H
-#include <malloc.h>
-#endif
-
-#ifdef HAVE_STRING_H
-#include <string.h>
-#endif
-
-#include <pthread.h>
-
-typedef struct EventLoop {
-// int maxfd;
- conn* conns[AE_SETSIZE];
- int fired[AE_SETSIZE];
- int nready;
- void *apidata;
-} EventLoop;
-
-/* Lock for connection freelist */
-static pthread_mutex_t conn_lock;
-
-/* Lock for item buffer freelist */
-static pthread_mutex_t ibuffer_lock;
-
-static EventLoop loop;
-static pthread_mutex_t leader;
-
-/*
- * Pulls a conn structure from the freelist, if one is available.
- */
-conn *mt_conn_from_freelist() {
- conn *c;
- pthread_mutex_lock(&conn_lock);
- c = do_conn_from_freelist();
- pthread_mutex_unlock(&conn_lock);
- return c;
-}
-
-/*
- * Adds a conn structure to the freelist.
- *
- * Returns 0 on success, 1 if the structure couldn't be added.
- */
-bool mt_conn_add_to_freelist(conn *c) {
- bool result;
-
- pthread_mutex_lock(&conn_lock);
- result = do_conn_add_to_freelist(c);
- pthread_mutex_unlock(&conn_lock);
-
- return result;
-}
-
-
-/******************************* GLOBAL STATS ******************************/
-
-void mt_stats_lock() {
-}
-
-void mt_stats_unlock() {
-}
-
-/* Include the best multiplexing layer supported by this system.
- * The following should be ordered by performances, descending. */
-#ifdef HAVE_EPOLL
-#include "ae_epoll.c"
-#else
- #ifdef HAVE_KQUEUE
- #include "ae_kqueue.c"
- #else
- #include "ae_select.c"
- #endif
-#endif
-
-/*
- * Initializes the thread subsystem, creating various worker threads.
- *
- * nthreads Number of event handler threads to spawn
- */
-void thread_init(int nthreads) {
- pthread_mutex_init(&ibuffer_lock, NULL);
- pthread_mutex_init(&conn_lock, NULL);
- pthread_mutex_init(&leader, NULL);
-
- memset(&loop, 0, sizeof(loop));
- if (aeApiCreate(&loop) == -1) {
- exit(1);
- }
-}
-
-int add_event(int fd, int mask, conn *c)
-{
- if (fd >= AE_SETSIZE) {
- fprintf(stderr, "fd is too large: %d\n", fd);
- return AE_ERR;
- }
- assert(loop.conns[fd] == NULL);
- loop.conns[fd] = c;
- if (aeApiAddEvent(&loop, fd, mask) == -1){
- loop.conns[fd] = NULL;
- return AE_ERR;
- }
-// if (fd > loop.maxfd)
-// loop.maxfd = fd;
- return AE_OK;
-}
-
-int update_event(int fd, int mask, conn *c)
-{
- loop.conns[fd] = c;
- if (aeApiUpdateEvent(&loop, fd, mask) == -1){
- loop.conns[fd] = NULL;
- return AE_ERR;
- }
- return AE_OK;
-}
-
-int delete_event(int fd)
-{
- if (fd >= AE_SETSIZE) return -1;
- if (loop.conns[fd] == NULL) return 0;
- if (aeApiDelEvent(&loop, fd) == -1)
- return -1;
- loop.conns[fd] = NULL;
- return 0;
-}
-
-static void *worker_main(void *arg) {
- pthread_setcanceltype (PTHREAD_CANCEL_ASYNCHRONOUS, 0);
-
- struct timeval tv = {1, 0};
- while (!daemon_quit) {
- pthread_mutex_lock(&leader);
-
-AGAIN:
- while(loop.nready == 0 && daemon_quit == 0)
- loop.nready = aeApiPoll(&loop, &tv);
- if (daemon_quit) {
- pthread_mutex_unlock(&leader);
- break;
- }
-
- loop.nready --;
- int fd = loop.fired[loop.nready];
- conn *c = loop.conns[fd];
- if (c == NULL){
- fprintf(stderr, "Bug: conn %d should not be NULL\n", fd);
- close(fd);
- goto AGAIN;
- }
- loop.conns[fd] = NULL;
- pthread_mutex_unlock(&leader);
-
- drive_machine(c);
- if (c->ev_flags > 0) {
- update_event(fd, c->ev_flags, c);
- }
- }
- return NULL;
-}
-
-void loop_run(int nthread)
-{
- int i, ret;
- pthread_attr_t attr;
- pthread_attr_init(&attr);
- pthread_t* tids = malloc(sizeof(pthread_t) * nthread);
-
- for (i=0; i<nthread - 1; i++) {
- if ((ret = pthread_create(tids + i, &attr, worker_main, NULL)) != 0) {
- fprintf(stderr, "Can't create thread: %s\n",
- strerror(ret));
- exit(1);
- }
- }
-
- worker_main(NULL);
-
- // wait workers to stop
- for (i=0; i<nthread - 1; i++) {
- (void) pthread_join(tids[i], NULL);
- }
- free(tids);
-}
-
diff --git a/3rdparty/cuttdb/src/test_mt.c b/3rdparty/cuttdb/src/test_mt.c
deleted file mode 100644
index de4d383731a4a66bc2e690f40dfc742ec022073e..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/test_mt.c
+++ /dev/null
@@ -1,149 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include <stdio.h>
-#include <stdlib.h>
-#include <time.h>
-#include <unistd.h>
-#include <pthread.h>
-#include "cuttdb.h"
-
-
-CDB *db;
-
-enum {
- SETOP,
- GETOP,
- DELOP,
-};
-
-#if 1
-static int prob_table1[8] = {SETOP, SETOP, SETOP, SETOP, SETOP, SETOP, DELOP, GETOP};
-static int prob_table2[8] = {SETOP, SETOP, SETOP, SETOP, SETOP, DELOP, DELOP, GETOP};
-static int prob_table3[8] = {SETOP, SETOP, SETOP, DELOP, DELOP, DELOP, DELOP, GETOP};
-#else
-static int prob_table1[8] = {SETOP, SETOP, SETOP, SETOP, SETOP, SETOP, SETOP, GETOP};
-static int prob_table2[8] = {SETOP, SETOP, SETOP, SETOP, SETOP, SETOP, SETOP, GETOP};
-static int prob_table3[8] = {SETOP, SETOP, SETOP, SETOP, SETOP, SETOP, SETOP, GETOP};
-#endif
-int *optable = NULL;
-
-
-long get_rand()
-{
- return (long)rand() * RAND_MAX + rand();
-}
-
-
-void *test_thread(void *arg)
-{
- char key[64];
- char value[128];
- void *v;
- int knum = *(int*)arg;
- while(1) {
- int krand = get_rand() % knum;
- int ksize = snprintf(key, 64, "%ld%ld%ld", krand, krand, krand);
- int vsize = snprintf(value, 128, "%ld%ld%ld%ld%d%ld%ld%ld%ld",
- krand, krand, krand, krand, krand, krand, krand, krand);
- int op = optable[rand() & 0x07];
- int expire = 600 + 20 * (rand() % 1000);
- switch(op) {
- case SETOP:
- if (cdb_set2(db, key, ksize, value, vsize, CDB_OVERWRITE | CDB_INSERTCACHE, expire) < 0)
- printf("ERROR! %s:%d\n", __FILE__, __LINE__);
- break;
- case GETOP:
- if (cdb_get(db, key, ksize, &v, &vsize) == -1)
- printf("ERROR! %s:%d\n", __FILE__, __LINE__);
- if (v)
- cdb_free_val(&v);
- break;
- case DELOP:
- if (cdb_del(db, key, ksize) == -1)
- printf("ERROR! %s:%d\n", __FILE__, __LINE__);
- break;
- default:
- break;
- }
- }
-}
-
-
-
-int main(int argc, char *argv[])
-{
- int thread_num = 2;
- int record_num = 10000000;
- char *db_path = NULL;
- printf("Usage: %s db_path [record_num] [thread_num]\n", argv[0]);
- if (argc >= 2)
- db_path = argv[1];
- else
- return -1;
-
- if (argc >= 3)
- record_num = atoi(argv[2]);
- if (argc >= 4)
- thread_num = atoi(argv[3]);
-
- record_num = record_num < 100? 100: record_num;
- thread_num = thread_num < 1? 1: thread_num;
- srand(time(NULL));
-
- db = cdb_new();
- cdb_option(db, record_num / 100, 0, 1024000);
- if (cdb_open(db, db_path, CDB_CREAT | CDB_TRUNC) < 0) {
- printf("DB Open err\n");
- return -1;
- }
-
-
- optable = prob_table1;
- pthread_t threads[thread_num];
- for(int i = 0; i < thread_num; i++) {
- pthread_create(&threads[i], NULL, test_thread, &record_num);
- }
-
- int clear_interval = 0;
- while(1) {
- CDBSTAT st;
- cdb_stat(db, &st);
- printf("rnum: %lu, rcnum: %lu, pnum: %lu, pcnum %lu, rlatcy: %u wlatcy: %u"
- " rh/m: %lu/%lu ph/m: %lu/%lu\n",
- st.rnum, st.rcnum, st.pnum, st.pcnum, st.rlatcy, st.wlatcy,
- st.rchit, st.rcmiss, st.pchit, st.pcmiss);
- if (++clear_interval % 20 == 0)
- cdb_stat(db, NULL);
-
- if (st.rnum > 0.7 * record_num)
- optable = prob_table2;
- if (st.rnum > 0.9 * record_num)
- optable = prob_table3;
-
- if (st.rnum < 0.8 * record_num)
- optable = prob_table2;
-
- if (st.rnum < 0.6 * record_num)
- optable = prob_table1;
- fflush(stdout);
- sleep(1);
- }
-
- return 0;
-}
-
-
-
diff --git a/3rdparty/cuttdb/src/vio_apnd2.c b/3rdparty/cuttdb/src/vio_apnd2.c
deleted file mode 100644
index 68dc326df18ab8409520da20a7285989bdd09377..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/vio_apnd2.c
+++ /dev/null
@@ -1,2644 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#include "vio_apnd2.h"
-#include "cdb_hashtable.h"
-#include "cdb_bgtask.h"
-#include "cdb_lock.h"
-#include "cuttdb.h"
-#include "cdb_core.h"
-#include "cdb_errno.h"
-#include "cdb_types.h"
-#include "cdb_crc64.h"
-#include <stdio.h>
-#include <stdlib.h>
-#include <string.h>
-#include <errno.h>
-#include <time.h>
-#include <sys/stat.h>
-#ifdef _WIN32
-#include "mman.h"
-#else
-#include <sys/mman.h>
-#endif
-#include <fcntl.h>
-#include <unistd.h>
-#include <dirent.h>
-
-/* record magic bytes */
-#define RECMAGIC 0x19871022
-/* obsoleted, but appeared in some code */
-#define DELRECMAGIC 0x19871023
-#define PAGEMAGIC 0x19890604
-
-/* data buffered before pwrite to disk */
-#define IOBUFSIZE (2 * MB)
-/* structure of deletion buffer differs from the others, buffered DELBUFMAX records at most */
-#define DELBUFMAX 10000
-
-/* index(page) file size limit */
-#define FIDXMAXSIZE (16 * MB)
-/* data file size limit */
-#define FDATMAXSIZE (128 * MB)
-/* all meta information are regulated to fix size */
-#define FILEMETASIZE 64
-/* the file opened simultaneously limit, managed by LRU */
-#define MAXFD 16384
-#define MAX_PATH_LEN 255
-
-#define FILEMAGICHEADER "CuTtDbFiLePaRtIaL"
-#define FILEMAGICLEN (strlen(FILEMAGICHEADER))
-/* page or data records are stored at aligned offset */
-#define ALIGNBYTES 16
-
-/* virtual offset(48bits) transform into real offset(fid,offset) */
-#define VOFF2ROFF(off, fid, roff) do{fid = (off).i4 >> 8; \
- roff = ((off).i4 & 0xff) << 16; roff = (roff | (off).i2) * ALIGNBYTES;}while(0)
-
-/* real offset transform into virtual offset */
-#define ROFF2VOFF(fid, roff, off) do{(off).i4 = fid << 8; \
- (off).i4 |= (roff / ALIGNBYTES) >> 16; (off).i2 = (roff / ALIGNBYTES) & 0xffff;} while(0)
-
-/* align to a integer offset */
-#define OFFALIGNED(off) ((((off)-1) | (ALIGNBYTES - 1)) + 1)
-
-/* used in fd LRU-cached, distinguish index or data files' fd */
-#define VFIDIDX(fid) (fid * 2)
-#define VFIDDAT(fid) (fid * 2 + 1)
-
-/* how often write out buffered data */
-#define FLUSHTIMEOUT 5
-/* how often to check if index file needs space recycle */
-#define RCYLEPAGEINTERVAL 60
-/* how often to check if data file needs space recycle */
-#define RCYLEDATAINTERVAL 120
-/* data file space recycle check interval factor (seconds per data file/128MB)*/
-#define DATARCYLECHECKFACTOR 1800
-
-
-/* three type of file */
-enum {
- /* random value */
- VIOAPND2_INDEX = 0x97,
- VIOAPND2_DATA = 0x98,
- VIOAPND2_DELLOG = 0x99,
-};
-
-
-/* where the record comes from when calling writerec */
-enum {
- VIOAPND2_RECEXTERNAL = 0,
- VIOAPND2_RECINTERNAL = 1,
-};
-
-
-/* a file is writing or full? */
-enum {
- VIOAPND2_WRITING = 0,
- VIOAPND2_FULL = 1,
-};
-
-/* signature in the header file, indicates it's open or be safety closed */
-enum {
- /* any number doens't matter */
- VIOAPND2_SIGOPEN = 2,
- VIOAPND2_SIGCLOSED = 3,
-};
-
-
-/* buffer for IO */
-typedef struct {
- uint32_t limit;
- uint32_t off;
- uint32_t pos;
- uint32_t fid;
- uint64_t oid;
- int fd;
- char buf[IOBUFSIZE];
-} VIOAPND2IOBUF;
-
-
-/* file information for every file */
-typedef struct VIOAPND2FINFO {
- /* fid */
- uint32_t fid;
- /* first oid */
- uint64_t oidf;
- /* last oid */
- uint64_t oidl;
-
- /* next file */
- struct VIOAPND2FINFO *fnext;
- /* prev file */
- struct VIOAPND2FINFO *fprev;
-
- uint32_t fsize;
- /* junk space */
- uint32_t rcyled;
- /* nearest expire time */
- uint32_t nexpire;
- /* last time for recycle check */
- uint32_t lcktime;
- /* index page file or data file? */
- uint8_t ftype;
- /* writing or full? */
- uint8_t fstatus;
- /* ref count, avoid unlink failure */
- uint32_t ref;
- /* whether unlink the file after dereference */
- bool unlink;
-} VIOAPND2FINFO;
-
-
-typedef struct {
- /* a new db? */
- bool create;
- /* fd number limit */
- int maxfds;
- /* opened files' fds cache */
- CDBHASHTABLE *fdcache;
-
- /* number of data file */
- uint32_t dfnum;
- /* number of index file */
- uint32_t ifnum;
-
- /* Buffers */
- VIOAPND2IOBUF dbuf;
- VIOAPND2IOBUF ibuf;
- FOFF delbuf[DELBUFMAX];
- int delbufpos;
-
- /* db path */
- char *filepath;
-
-
- /* file information of index files */
- CDBHASHTABLE *idxmeta;
- VIOAPND2FINFO *idxfhead;
- VIOAPND2FINFO *idxftail;
- /* file information of data files */
- CDBHASHTABLE *datmeta;
- VIOAPND2FINFO *datfhead;
- VIOAPND2FINFO *datftail;
-
- /* fd for db header */
- int hfd;
- /* fd for files meta header */
- int mfd;
- /* fd for deletion log */
- int dfd;
-
- /* lock for all I/O operation */
- CDBLOCK *lock;
-
- int idxitfid;
- uint32_t idxitoff;
- char *idxmmap;
-
-} VIOAPND2;
-
-
-/* iterator for index/data */
-typedef struct {
- /* current open fd */
- int fd;
- /* current offset in file*/
- uint32_t off;
- /* current operation id */
- uint64_t oid;
- /* current file size*/
- uint64_t fsize;
- /* mapped of file */
- char *mmap;
- /* reference of filemeta struct */
- VIOAPND2FINFO *finfo;
-} VIOAPND2ITOR;
-
-
-static int _vio_apnd2_open(CDBVIO *vio, const char *filepath, int flags);
-static int _vio_apnd2_checkpid(CDBVIO *vio);
-static int _vio_apnd2_write(CDBVIO *vio, int fd, void *buf, uint32_t size, bool aligned);
-static int _vio_apnd2_read(CDBVIO *vio, int fd, void *buf, uint32_t size, uint64_t off);
-static int _vio_apnd2_readmeta(CDBVIO *vio, bool overwrite);
-static int _vio_apnd2_writemeta(CDBVIO *vio);
-static int _vio_apnd2_close(CDBVIO *vio);
-static int _vio_apnd2_writerec(CDBVIO *vio, CDBREC *rec, FOFF *off, int ptrtype);
-static int _vio_apnd2_writerecexternal(CDBVIO *vio, CDBREC *rec, FOFF *off);
-static int _vio_apnd2_writerecinternal(CDBVIO *vio, CDBREC *rec, FOFF *off);
-static int _vio_apnd2_deleterec(CDBVIO *vio, CDBREC *rec, FOFF off);
-static int _vio_apnd2_readrec(CDBVIO *vio, CDBREC** rec, FOFF off, bool readval);
-static int _vio_apnd2_writepage(CDBVIO *vio, CDBPAGE *page, FOFF *off);
-static int _vio_apnd2_readpage(CDBVIO *vio, CDBPAGE **page, FOFF off);
-static int _vio_apnd2_sync(CDBVIO *vio);
-static int _vio_apnd2_writehead2(CDBVIO *vio);
-static int _vio_apnd2_writehead(CDBVIO *vio, bool wtable);
-static int _vio_apnd2_readhead2(CDBVIO *vio);
-static int _vio_apnd2_readhead(CDBVIO *vio, bool rtable);
-static int _vio_apnd2_writefmeta(CDBVIO *vio, int fd, VIOAPND2FINFO *finfo);
-static int _vio_apnd2_readfmeta(CDBVIO *vio, int fd, VIOAPND2FINFO *finfo);
-static int _vio_apnd2_flushbuf(CDBVIO *vio, int dtype);
-static void _vio_apnd2_flushtask(void *arg);
-static void _vio_apnd2_rcyledataspacetask(void *arg);
-static void _vio_apnd2_fixcachepageooff(CDB *db, uint32_t bit, FOFF off);
-static void _vio_apnd2_rcylepagespacetask(void *arg);
-static int _vio_apnd2_shiftnew(CDBVIO *vio, int dtype);
-static int _vio_apnd2_recovery(CDBVIO *vio, bool force);
-static void _vio_apnd2_unlink(CDBVIO *vio, VIOAPND2FINFO *finfo, int dtype);
-static VIOAPND2FINFO* _vio_apnd2_fileiternext(CDBVIO *vio, int dtype, uint64_t oid);
-static int _vio_apnd2_iterfirst(CDBVIO *vio, VIOAPND2ITOR *it, int dtype, int64_t oid);
-static int _vio_apnd2_iterfree(CDBVIO *vio, int dtype, VIOAPND2ITOR *it);
-static int _vio_apnd2_pageiternext(CDBVIO *vio, CDBPAGE **page, void *iter);
-static int _vio_apnd2_reciternext(CDBVIO *vio, CDBREC **rec, void *iter);
-static void* _vio_apnd2_reciterfirst(CDBVIO *vio, uint64_t oid);
-static void* _vio_apnd2_pageiterfirst(CDBVIO *vio, uint64_t oid);
-static void _vio_apnd2_reciterdestory(CDBVIO *vio, void *iter);
-static void _vio_apnd2_pageiterdestory(CDBVIO *vio, void *iter);
-static void _vio_apnd2_cleanpoint(CDBVIO *vio);
-static int _vio_apnd2_cmpfuncsreorder(const void *p1, const void *p2);
-static int _vio_apnd2_checkopensig(CDBVIO *vio);
-static int _vio_apnd2_setopensig(CDBVIO *vio, int sig);
-static int _vio_apnd2_rcyledatafile(CDBVIO *vio, VIOAPND2FINFO *finfo, bool rcyle);
-
-
-/* hook the io methods */
-void vio_apnd2_init(CDBVIO *vio)
-{
- vio->close = _vio_apnd2_close;
- vio->open = _vio_apnd2_open;
- vio->rpage = _vio_apnd2_readpage;
- vio->wpage = _vio_apnd2_writepage;
- vio->rrec = _vio_apnd2_readrec;
- vio->drec = _vio_apnd2_deleterec;
- vio->wrec = _vio_apnd2_writerecexternal;
- vio->sync = _vio_apnd2_sync;
- vio->rhead = _vio_apnd2_readhead2;
- vio->whead = _vio_apnd2_writehead2;
- vio->cleanpoint = _vio_apnd2_cleanpoint;
- vio->pageitfirst = _vio_apnd2_pageiterfirst;
- vio->pageitnext = _vio_apnd2_pageiternext;
- vio->pageitdestroy = _vio_apnd2_pageiterdestory;
- vio->recitfirst = _vio_apnd2_reciterfirst;
- vio->recitnext = _vio_apnd2_reciternext;
- vio->recitdestroy = _vio_apnd2_reciterdestory;
-}
-
-/* the hash table used in VIOAPND2 need not rehash, just use the key id is OK */
-static uint32_t _directhash(const void *key, int size)
-{
- (void) size;
- return *(uint32_t*)key;
-}
-
-
-/* allocate a new VIOAPND2 object, called when open db */
-static void _vio_apnd2_new(CDBVIO *vio)
-{
- VIOAPND2 *myio = (VIOAPND2 *)malloc(sizeof(VIOAPND2));
-
- myio->dfnum = myio->ifnum = 0;
-
- myio->dbuf.fid = 0;
- myio->dbuf.pos = 0;
- myio->dbuf.off = 0;
- myio->dbuf.oid = 0;
- memset(myio->dbuf.buf, 0, IOBUFSIZE);
- myio->idxfhead = NULL;
- myio->idxftail = NULL;
-
- myio->ibuf.fid = 0;
- myio->ibuf.pos = 0;
- myio->ibuf.off = 0;
- myio->ibuf.oid = 0;
- memset(myio->ibuf.buf, 0, IOBUFSIZE);
- myio->datfhead = NULL;
- myio->datftail = NULL;
-
- myio->delbufpos = 0;
-
- myio->ifnum = 0;
- myio->dfnum = 0;
-
- myio->mfd = -1;
- myio->hfd = -1;
- myio->dfd = -1;
-
- myio->fdcache = cdb_ht_new(true, _directhash);
- /* the following two are look-up table, need not LRU */
- myio->idxmeta = cdb_ht_new(false, _directhash);
- myio->datmeta = cdb_ht_new(false, _directhash);
-
- myio->lock = cdb_lock_new(CDB_LOCKMUTEX);
-
- myio->create = true;
- myio->maxfds = MAXFD;
- myio->filepath = NULL;
-
- vio->iometa = myio;
-}
-
-
-/* free a VIOAPND2 object, called when close db */
-static void _vio_apnd2_destroy(CDBVIO *vio)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- cdb_ht_destroy(myio->fdcache);
- cdb_ht_destroy(myio->idxmeta);
- cdb_ht_destroy(myio->datmeta);
- cdb_lock_destory(myio->lock);
- if (myio->filepath)
- free(myio->filepath);
- free(myio);
- vio->iometa = NULL;
-}
-
-/* check if another process has already open the current db */
-static int _vio_apnd2_checkpid(CDBVIO *vio)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- struct stat st;
- char filename[MAX_PATH_LEN] = {0};
- char syspidpath[MAX_PATH_LEN] = {0};
- snprintf(filename, MAX_PATH_LEN, "%s/pid.cdb", myio->filepath);
-
- if (stat(filename, &st) == 0) {
- /* pid file exist */
- FILE *f = fopen(filename, "rt");
- int pid = -1;
- if (f == NULL) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- return -1;
- }
-
- int ret = fscanf(f, "%d", &pid);
- fclose(f);
- if (ret == 1 ) {
- /* check if the process still alive */
- snprintf(syspidpath, MAX_PATH_LEN, "/proc/%d", pid);
- if (stat(syspidpath, &st) == 0) {
- cdb_seterrno(vio->db, CDB_PIDEXIST, __FILE__, __LINE__);
- return -1;
- }
- }
- }
-
- /* pid file non-exist or obsoleted */
- FILE *f = fopen(filename, "wt");
- if (f == NULL) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- return -1;
- }
- fprintf(f, "%d\n", getpid());
- fclose(f);
- return 0;
-}
-
-/* open an db by path and mode */
-static int _vio_apnd2_open(CDBVIO *vio, const char *filepath, int flags)
-{
- int rflags = O_RDWR;
- char filename[MAX_PATH_LEN] = {0};
- int fsize;
- int sigstatus;
- VIOAPND2 *myio;
-
- _vio_apnd2_new(vio);
- myio = (VIOAPND2 *)vio->iometa;
- myio->filepath = strdup(filepath);
-
- if (flags & CDB_CREAT)
- rflags |= O_CREAT;
- if (flags & CDB_TRUNC)
- rflags |= O_TRUNC;
-
- if (/*_vio_apnd2_checkpid(vio) <*/ 0) {
- goto ERRRET;
- }
-
- snprintf(filename, MAX_PATH_LEN, "%s/mainindex.cdb", myio->filepath);
- myio->hfd = open(filename, rflags, 0644);
- if (myio->hfd < 0 && errno == ENOENT && (rflags & O_CREAT)) {
- /* try to create, but path not exists */
- cdb_seterrno(vio->db, CDB_DIRNOEXIST, __FILE__, __LINE__);
- goto ERRRET;
- } else if (myio->hfd < 0) {
- /* other open error */
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- goto ERRRET;
- }
-
- fsize = lseek(myio->hfd, 0, SEEK_END);
- if (fsize) {
- myio->create = false;
- sigstatus = _vio_apnd2_checkopensig(vio);
- if (sigstatus < 0) {
- /* main table read error */
- cdb_seterrno(vio->db, CDB_READERR, __FILE__, __LINE__);
- goto ERRRET;
- }
- } else {
- sigstatus = VIOAPND2_SIGCLOSED;
- }
-
- /* */
- struct stat st;
- snprintf(filename, MAX_PATH_LEN, "%s/force_recovery", myio->filepath);
- if (stat(filename, &st) == 0) {
- /* special file exist, force recovery to fix the database */
- _vio_apnd2_recovery(vio, true);
- unlink(filename);
- } else if (sigstatus == VIOAPND2_SIGOPEN) {
- /* didn't properly closed last time */
- _vio_apnd2_recovery(vio, false);
- } else if (sigstatus != VIOAPND2_SIGCLOSED) {
- cdb_seterrno(vio->db, CDB_DATAERRMETA, __FILE__, __LINE__);
- goto ERRRET;
- }
-
- if (_vio_apnd2_setopensig(vio, VIOAPND2_SIGOPEN) < 0) {
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- goto ERRRET;
- }
-
- snprintf(filename, MAX_PATH_LEN, "%s/mainmeta.cdb", myio->filepath);
- myio->mfd = open(filename, rflags, 0644);
- if (myio->mfd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- goto ERRRET;
- }
-
- fsize = lseek(myio->mfd, 0, SEEK_END);
- if (fsize) {
- /* exist database */
- _vio_apnd2_readmeta(vio, false);
-
- /* open current data file and index file for buffer */
- snprintf(filename, MAX_PATH_LEN, "%s/idx%08d.cdb", myio->filepath, myio->ibuf.fid);
- myio->ibuf.fd = open(filename, rflags, 0644);
- myio->ibuf.limit = CDBMIN(IOBUFSIZE, FIDXMAXSIZE - myio->ibuf.off);
- myio->ibuf.pos = 0;
-
- snprintf(filename, MAX_PATH_LEN, "%s/dat%08d.cdb", myio->filepath, myio->dbuf.fid);
- myio->dbuf.fd = open(filename, rflags, 0644);
- myio->dbuf.limit = CDBMIN(IOBUFSIZE, FDATMAXSIZE - myio->dbuf.off);
- myio->dbuf.pos = 0;
- } else {
- /* new database */
- myio->create = true;
- /* remember the bnum */
- _vio_apnd2_writehead(vio, false);
- _vio_apnd2_shiftnew(vio, VIOAPND2_INDEX);
- _vio_apnd2_shiftnew(vio, VIOAPND2_DATA);
- }
-
- snprintf(filename, MAX_PATH_LEN, "%s/dellog.cdb", myio->filepath);
- myio->dfd = open(filename, O_RDWR | O_CREAT | O_TRUNC, 0644);
- if (myio->dfd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- goto ERRRET;
- }
-
- /* set background tasks, flush buffer and recycle space */
- cdb_bgtask_add(vio->db->bgtask, _vio_apnd2_flushtask, vio, FLUSHTIMEOUT);
- cdb_bgtask_add(vio->db->bgtask, _vio_apnd2_rcylepagespacetask, vio, RCYLEPAGEINTERVAL);
- cdb_bgtask_add(vio->db->bgtask, _vio_apnd2_rcyledataspacetask, vio, RCYLEDATAINTERVAL);
- return 0;
-
-ERRRET:
- if (myio->mfd > 0)
- close(myio->mfd);
- if (myio->hfd > 0)
- close(myio->hfd);
- if (myio->dfd > 0)
- close(myio->dfd);
- _vio_apnd2_destroy(vio);
- return -1;
-}
-
-
-/* task for flush buffer */
-static void _vio_apnd2_flushtask(void *arg)
-{
- CDBVIO *vio = (CDBVIO *)arg;
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- cdb_lock_lock(myio->lock);
- _vio_apnd2_flushbuf(vio, VIOAPND2_DATA);
- _vio_apnd2_flushbuf(vio, VIOAPND2_INDEX);
- _vio_apnd2_flushbuf(vio, VIOAPND2_DELLOG);
- cdb_lock_unlock(myio->lock);
-}
-
-
-/* read information for db files, 'overwrite' indicates recovery */
-static int _vio_apnd2_readmeta(CDBVIO *vio, bool overwrite)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- char buf[FILEMETASIZE];
- char *hbuf;
- int hbufsize;
- int pos = 0;
-
- if (pread(myio->mfd, buf, FILEMETASIZE, 0) != FILEMETASIZE) {
- if (overwrite)
- return 0;
- cdb_seterrno(vio->db, CDB_READERR, __FILE__, __LINE__);
- return -1;
- }
-
- if (memcmp(buf, FILEMAGICHEADER, FILEMAGICLEN) != 0) {
- cdb_seterrno(vio->db, CDB_DATAERRMETA, __FILE__, __LINE__);
- return -1;
- }
-
- pos += FILEMAGICLEN;
- cdb_lock_lock(myio->lock);
- if (!overwrite)
- myio->ibuf.off = *(uint32_t*)(buf + pos);
- pos += SI4;
- myio->ibuf.limit = *(uint32_t*)(buf + pos);
- pos += SI4;
- if (!overwrite)
- myio->dbuf.off = *(uint32_t*)(buf + pos);
- pos += SI4;
- myio->dbuf.limit = *(uint32_t*)(buf + pos);
- pos += SI4;
- if (!overwrite)
- myio->ifnum = *(uint32_t*)(buf + pos);
- pos += SI4;
- if (!overwrite)
- myio->dfnum = *(uint32_t*)(buf + pos);
- pos += SI4;
- if (!overwrite)
- myio->ibuf.fid = *(uint32_t*)(buf + pos);
- pos += SI4;
- if (!overwrite)
- myio->dbuf.fid = *(uint32_t*)(buf + pos);
- pos += SI4;
-
- hbufsize = (SI4 + SI4 + SI4 + SI8 + SI8 + 1 + 1) * myio->ifnum;
- hbufsize += (SI4 + SI4 + SI4 + SI4 + SI8 + SI8 + 1 + 1) * myio->dfnum;
- hbuf = (char*)malloc(hbufsize);
- pos = 0;
-
- if (pread(myio->mfd, hbuf, hbufsize, FILEMETASIZE) != hbufsize) {
- cdb_lock_unlock(myio->lock);
- free(hbuf);
- if (overwrite)
- return 0;
- cdb_seterrno(vio->db, CDB_READERR, __FILE__, __LINE__);
- return -1;
- }
-
- for(uint32_t i = 0; i < myio->ifnum; i++) {
- VIOAPND2FINFO finfo, *finfo2;
- finfo.fid = *(uint32_t*)(hbuf + pos);
- pos += SI4;
- finfo.fsize = *(uint32_t*)(hbuf + pos);
- pos += SI4;
- finfo.rcyled = *(uint32_t*)(hbuf + pos);
- pos += SI4;;
- finfo.oidf = *(uint64_t*)(hbuf + pos);
- pos += SI8;
- finfo.oidl = *(uint64_t*)(hbuf + pos);
- pos += SI8;
- finfo.fstatus = *(uint8_t*)(hbuf + pos);
- pos += 1;
- finfo.ftype = *(uint8_t*)(hbuf + pos);
- pos += 1;
- finfo.ref = 0;
- finfo.unlink = false;
- if (overwrite) {
- /* in recovery mode only fix 'recycled size' */
- /* But do nothing with index files */
- continue;
- }
- finfo2 = (VIOAPND2FINFO *)cdb_ht_insert2(myio->idxmeta, &finfo.fid, SI4, &finfo, sizeof(finfo));
- if (myio->idxfhead) {
- finfo2->fprev = myio->idxftail;
- myio->idxftail->fnext = finfo2;
- finfo2->fnext = NULL;
- myio->idxftail = finfo2;
- } else {
- myio->idxfhead = myio->idxftail = finfo2;
- finfo2->fprev = finfo2->fnext = NULL;
- }
- }
-
- for(uint32_t i = 0; i < myio->dfnum; i++) {
- VIOAPND2FINFO finfo, *finfo2;
- finfo.fid = *(uint32_t*)(hbuf + pos);
- pos += SI4;
- finfo.fsize = *(uint32_t*)(hbuf + pos);
- pos += SI4;
- finfo.rcyled = *(uint32_t*)(hbuf + pos);
- pos += SI4;
- finfo.nexpire = *(uint32_t*)(hbuf + pos);
- pos += SI4;
- finfo.oidf = *(uint64_t*)(hbuf + pos);
- pos += SI8;
- finfo.oidl = *(uint64_t*)(hbuf + pos);
- pos += SI8;
- finfo.fstatus = *(uint8_t*)(hbuf + pos);
- pos += 1;
- finfo.ftype = *(uint8_t*)(hbuf + pos);
- pos += 1;
- finfo.ref = 0;
- finfo.unlink = false;
- finfo.lcktime = time(NULL);
- if (overwrite) {
- /* in recovery mode only fix 'recycled size' */
- finfo2 = (VIOAPND2FINFO *)cdb_ht_get2(myio->datmeta, &finfo.fid, SI4, false);
- if (finfo2) {
- finfo2->rcyled = finfo.rcyled;
- finfo2->nexpire = finfo.nexpire;
- }
- continue;
- }
- finfo2 = (VIOAPND2FINFO *)cdb_ht_insert2(myio->datmeta, &finfo.fid, SI4, &finfo, sizeof(finfo));
- if (myio->datfhead) {
- finfo2->fprev = myio->datftail;
- myio->datftail->fnext = finfo2;
- finfo2->fnext = NULL;
- myio->datftail = finfo2;
- } else {
- myio->datfhead = myio->datftail = finfo2;
- finfo2->fprev = finfo2->fnext = NULL;
- }
- }
- cdb_lock_unlock(myio->lock);
- free(hbuf);
-
- return 0;
-}
-
-
-/* flush i/o buffer */
-static int _vio_apnd2_flushbuf(CDBVIO *vio, int dtype)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- VIOAPND2FINFO *finfo;
- VIOAPND2IOBUF *iobuf;
- CDBHASHTABLE *ht;
- uint32_t *fid;
- uint32_t fsizemax;
-
- /* link to the proper operation object */
- if (dtype == VIOAPND2_INDEX) {
- iobuf = &myio->ibuf;
- ht = myio->idxmeta;
- fsizemax = FIDXMAXSIZE;
- } else if (dtype == VIOAPND2_DATA) {
- iobuf = &myio->dbuf;
- ht = myio->datmeta;
- fsizemax = FDATMAXSIZE;
- } else if (dtype == VIOAPND2_DELLOG) {
- /* buffer for deletion is special */
- if (myio->delbufpos == 0)
- return 0;
- if ( (size_t) write(myio->dfd, myio->delbuf, sizeof(FOFF) * (size_t) myio->delbufpos)
- != sizeof(FOFF) * (size_t) myio->delbufpos) {
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- return -1;
- }
- myio->delbufpos = 0;
- return 0;
- } else {
- cdb_seterrno(vio->db, CDB_INTERNALERR, __FILE__, __LINE__);
- return -1;
- }
- fid = &iobuf->fid;
-
- /* get information from table */
- finfo = (VIOAPND2FINFO *)cdb_ht_get2(ht, fid, SI4, false);
- if (finfo == NULL) {
- cdb_seterrno(vio->db, CDB_INTERNALERR, __FILE__, __LINE__);
- return -1;
- }
-
- /* write out if buffered */
- if (iobuf->pos > 0) {
- if (pwrite(iobuf->fd, iobuf->buf, iobuf->pos, iobuf->off) != iobuf->pos) {
- if (ftruncate(iobuf->fd, iobuf->off) < 0)
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- return -1;
- }
- }
-
- /* mark the operation id */
- finfo->oidl = iobuf->oid;
-
- /* reset the buffer information */
- iobuf->pos = 0;
- iobuf->off = lseek(iobuf->fd, 0, SEEK_END);
- /* fix file size info whenever possible */
- finfo->fsize = iobuf->off;
- iobuf->off = OFFALIGNED(iobuf->off);
-
- /* current writing file nearly full? open a new one */
- if (iobuf->off > fsizemax - 16 * KB) {
- finfo->fstatus = VIOAPND2_FULL;
- _vio_apnd2_writefmeta(vio, iobuf->fd, finfo);
- close(iobuf->fd);
- _vio_apnd2_shiftnew(vio, dtype);
- } else
- iobuf->limit = CDBMIN(IOBUFSIZE, fsizemax - iobuf->off) ;
-
- return 0;
-}
-
-/* create a new file for buffer and writing */
-static int _vio_apnd2_shiftnew(CDBVIO *vio, int dtype)
-{
- VIOAPND2 *myio = (VIOAPND2*)vio->iometa;
- VIOAPND2IOBUF *iobuf;
- CDBHASHTABLE *ht;
- uint32_t *fnum;
- uint32_t tryiter, curfid;
- char filename[MAX_PATH_LEN];
- char ipfx[] = "idx";
- char dpfx[] = "dat";
- char *pfx;
-
- /* link to proper object by dtype */
- if (dtype == VIOAPND2_INDEX) {
- iobuf = &myio->ibuf;
- ht = myio->idxmeta;
- fnum = &myio->ifnum;
- pfx = ipfx;
- } else if (dtype == VIOAPND2_DATA) {
- iobuf = &myio->dbuf;
- ht = myio->datmeta;
- fnum = &myio->dfnum;
- pfx = dpfx;
- } else {
- cdb_seterrno(vio->db, CDB_INTERNALERR, __FILE__, __LINE__);
- return -1;
- }
-
- curfid = iobuf->fid;
-
- /* reset invalid buffer, prevent for misuse */
- iobuf->fd = -1;
- iobuf->fid = 0xffffff;
- iobuf->limit = iobuf->pos = iobuf->off = 0xffffffff;
-
- /* find a valid fid, try 16M times at most */
- tryiter = 0;
- while(cdb_ht_exist(ht, &curfid, SI4)) {
- curfid++;
- tryiter++;
- if (tryiter == 0xffffff) {
- cdb_seterrno(vio->db, CDB_NOFID, __FILE__, __LINE__);
- return -1;
- }
- if (curfid == 0xffffff)
- curfid = 0;
- }
-
- /* open new file */
- snprintf(filename, MAX_PATH_LEN, "%s/%s%08d.cdb", myio->filepath, pfx, curfid);
- iobuf->fd = open(filename, O_RDWR | O_CREAT | O_TRUNC, 0644);
- if (iobuf->fd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- return -1;
- }
- iobuf->limit = IOBUFSIZE;
- iobuf->fid = curfid;
- iobuf->off = FILEMETASIZE;
- iobuf->pos = 0;
-
- /* set meta information for new file */
- VIOAPND2FINFO finfo, *finfo2;
- finfo.fsize = lseek(iobuf->fd, 0, SEEK_END);
- finfo.oidf = iobuf->oid;
- finfo.oidl = iobuf->oid;
- finfo.rcyled = 0;
- finfo.lcktime = time(NULL);
- finfo.fstatus = VIOAPND2_WRITING;
- finfo.ftype = dtype;
- finfo.fid = curfid;
- finfo.unlink = false;
- finfo.nexpire = 0xffffffff;
- finfo.ref = 0;
- /* meta information also be written to disk immediately */
- if (_vio_apnd2_writefmeta(vio, iobuf->fd, &finfo) < 0) {
- close(iobuf->fd);
- iobuf->fd = -1;
- iobuf->fid = 0xffffff;
- iobuf->limit = iobuf->pos = iobuf->off = 0xffffffff;
- return -1;
- }
- (*fnum)++;
- finfo2 = cdb_ht_insert2(ht, &curfid, SI4, &finfo, sizeof(VIOAPND2FINFO));
- if (dtype == VIOAPND2_INDEX) {
- if (myio->idxfhead) {
- finfo2->fprev = myio->idxftail;
- myio->idxftail->fnext = finfo2;
- finfo2->fnext = NULL;
- myio->idxftail = finfo2;
- } else {
- myio->idxfhead = myio->idxftail = finfo2;
- finfo2->fprev = finfo2->fnext = NULL;
- }
- } else if (dtype == VIOAPND2_DATA) {
- if (myio->datfhead) {
- finfo2->fprev = myio->datftail;
- myio->datftail->fnext = finfo2;
- finfo2->fnext = NULL;
- myio->datftail = finfo2;
- } else {
- myio->datfhead = myio->datftail = finfo2;
- finfo2->fprev = finfo2->fnext = NULL;
- }
- }
-
- return 0;
-}
-
-
-/* write a single file's meta information */
-static int _vio_apnd2_writefmeta(CDBVIO *vio, int fd, VIOAPND2FINFO *finfo)
-{
- char buf[FILEMETASIZE];
- int pos = 0;
-
- memset(buf, 'X', FILEMETASIZE);
- memcpy(buf, FILEMAGICHEADER, FILEMAGICLEN);
- pos += FILEMAGICLEN;
- *(uint64_t*)(buf + pos) = finfo->oidf;
- pos += SI8;
- *(uint64_t*)(buf + pos) = finfo->oidl;
- pos += SI8;
- *(uint32_t*)(buf + pos) = finfo->fsize;
- pos += SI4;
- *(uint32_t*)(buf + pos) = finfo->fid;
- pos += SI4;
- *(uint8_t*)(buf + pos) = finfo->fstatus;
- pos++;
- *(uint8_t*)(buf + pos) = finfo->ftype;
- pos++;
-
- if (pwrite(fd, buf, FILEMETASIZE, 0) != FILEMETASIZE) {
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- return -1;
- }
- return 0;
-}
-
-/* read a single file's meta information */
-static int _vio_apnd2_readfmeta(CDBVIO *vio, int fd, VIOAPND2FINFO *finfo)
-{
- char buf[FILEMETASIZE];
- int pos = 0;
-
- memset(buf, 'X', FILEMETASIZE);
- if (pread(fd, buf, FILEMETASIZE, 0) != FILEMETASIZE) {
- cdb_seterrno(vio->db, CDB_READERR, __FILE__, __LINE__);
- return -1;
- }
-
- if (memcmp(buf, FILEMAGICHEADER, FILEMAGICLEN)) {
- cdb_seterrno(vio->db, CDB_DATAERRMETA, __FILE__, __LINE__);
- return -1;
- }
-
- pos += FILEMAGICLEN;
- finfo->oidf = *(uint64_t*)(buf + pos);
- pos += SI8;
- finfo->oidl = *(uint64_t*)(buf + pos);
- pos += SI8;
- finfo->fsize = *(uint32_t*)(buf + pos);
- pos += SI4;
- finfo->fid = *(uint32_t*)(buf + pos);
- pos += SI4;
- finfo->fstatus = *(uint8_t*)(buf + pos);
- pos++;
- finfo->ftype = *(uint8_t*)(buf + pos);
- pos++;
- return 0;
-}
-
-
-/* write to disk directly instead of using buffer(Only Appends) */
-static int _vio_apnd2_write(CDBVIO *vio, int fd, void *buf, uint32_t size, bool aligned)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- uint32_t off;
-
- if (size == 0)
- return 0;
-
- off = lseek(fd, 0, SEEK_END);
- if (aligned)
- off = OFFALIGNED(off);
- if (pwrite(fd, buf, size, off) != size) {
- /* to avoid compile warning */
- if (ftruncate(myio->ibuf.fd, off) < 0)
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- return -1;
- }
-
- return size;
-}
-
-
-/* read from disk; if data has not been written, read from buffer */
-static int _vio_apnd2_read(CDBVIO *vio, int fd, void *buf, uint32_t size, uint64_t off)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- int ret;
-
- /* in buffer? */
- if (fd == myio->dbuf.fd && off >= myio->dbuf.off) {
- uint64_t boff = off - myio->dbuf.off;
- ret = CDBMIN(size, myio->dbuf.pos - boff);
- memcpy(buf, myio->dbuf.buf + boff, ret);
- } else if (fd == myio->ibuf.fd && off >= myio->ibuf.off) {
- uint64_t boff = off - myio->ibuf.off;
- ret = CDBMIN(size, myio->ibuf.pos - boff);
- memcpy(buf, myio->ibuf.buf + boff, ret);
- } else {
- /* not in buffer */
- ret = pread(fd, buf, size, off);
- if (ret < 0) {
- cdb_seterrno(vio->db, CDB_READERR, __FILE__, __LINE__);
- return -1;
- }
- }
- return ret;
-}
-
-
-/* write all files meta information into a file */
-static int _vio_apnd2_writemeta(CDBVIO *vio)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- char buf[FILEMETASIZE];
- char *hbuf;
- int hbufsize;
- int pos = 0;
-
- memset(buf, 'X', FILEMETASIZE);
- memcpy(buf, FILEMAGICHEADER, FILEMAGICLEN);
- pos += FILEMAGICLEN;
- cdb_lock_lock(myio->lock);
- *(uint32_t*)(buf + pos) = myio->ibuf.off;
- pos += SI4;
- *(uint32_t*)(buf + pos) = myio->ibuf.limit;
- pos += SI4;
- *(uint32_t*)(buf + pos) = myio->dbuf.off;
- pos += SI4;
- *(uint32_t*)(buf + pos) = myio->dbuf.limit;
- pos += SI4;
- *(uint32_t*)(buf + pos) = myio->ifnum;
- pos += SI4;
- *(uint32_t*)(buf + pos) = myio->dfnum;
- pos += SI4;
- *(uint32_t*)(buf + pos) = myio->ibuf.fid;
- pos += SI4;
- *(uint32_t*)(buf + pos) = myio->dbuf.fid;
- pos += SI4;
-
- hbufsize = (SI4 + SI4 + SI4 + SI8 + SI8 + 1 + 1) * myio->ifnum;
- hbufsize += (SI4 + SI4 + SI4 + SI4 + SI8 + SI8 + 1 + 1) * myio->dfnum;
- hbuf = (char*)malloc(hbufsize);
- memset(hbuf, 'X', hbufsize);
- pos = 0;
- /* iterate all the index files order by oid */
- VIOAPND2FINFO *finfo = myio->idxfhead;
- while(finfo != NULL) {
- *(uint32_t*)(hbuf + pos) = finfo->fid;
- pos += 4;
- *(uint32_t*)(hbuf + pos) = finfo->fsize;
- pos += 4;
- *(uint32_t*)(hbuf + pos) = finfo->rcyled;
- pos += 4;
- *(uint64_t*)(hbuf + pos) = finfo->oidf;
- pos += 8;
- *(uint64_t*)(hbuf + pos) = finfo->oidl;
- pos += 8;
- *(uint8_t*)(hbuf + pos) = finfo->fstatus;
- pos += 1;
- *(uint8_t*)(hbuf + pos) = finfo->ftype;
- pos += 1;
- finfo = finfo->fnext;
- }
-
- /* iterate all the data files order by oid */
- finfo = myio->datfhead;
- while(finfo != NULL) {
- *(uint32_t*)(hbuf + pos) = finfo->fid;
- pos += 4;
- *(uint32_t*)(hbuf + pos) = finfo->fsize;
- pos += 4;
- *(uint32_t*)(hbuf + pos) = finfo->rcyled;
- pos += 4;
- *(uint32_t*)(hbuf + pos) = finfo->nexpire;
- pos += 4;
- *(uint64_t*)(hbuf + pos) = finfo->oidf;
- pos += 8;
- *(uint64_t*)(hbuf + pos) = finfo->oidl;
- pos += 8;
- *(uint8_t*)(hbuf + pos) = finfo->fstatus;
- pos += 1;
- *(uint8_t*)(hbuf + pos) = finfo->ftype;
- pos += 1;
- finfo = finfo->fnext;
- }
- cdb_lock_unlock(myio->lock);
-
- if (pwrite(myio->mfd, buf, FILEMETASIZE, 0) != FILEMETASIZE) {
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- free(hbuf);
- return -1;
- }
-
- if (pwrite(myio->mfd, hbuf, hbufsize, FILEMETASIZE) != hbufsize) {
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- free(hbuf);
- return -1;
- }
- free(hbuf);
-
- return 0;
-}
-
-
-/* close db */
-static int _vio_apnd2_close(CDBVIO *vio)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- CDBHTITEM *item;
- char filename[MAX_PATH_LEN] = {0};
- VIOAPND2FINFO *finfo;
-
- /* flush buffer */
- _vio_apnd2_flushbuf(vio, VIOAPND2_INDEX);
- finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->idxmeta, &myio->ibuf.fid, SI4, false);
- if (finfo)
- _vio_apnd2_writefmeta(vio, myio->ibuf.fd, finfo);
- _vio_apnd2_flushbuf(vio, VIOAPND2_DATA);
- finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->datmeta, &myio->dbuf.fid, SI4, false);
- if (finfo)
- _vio_apnd2_writefmeta(vio, myio->dbuf.fd, finfo);
-
- /* iterate and close the fd cache */
- item = cdb_ht_iterbegin(myio->fdcache);
- while(item != NULL) {
- close(*(int*)cdb_ht_itemval(myio->fdcache, item));
- item = cdb_ht_iternext(myio->fdcache, item);
- }
-
- if (myio->dbuf.fd > 0)
- close(myio->dbuf.fd);
- if (myio->ibuf.fd > 0)
- close(myio->ibuf.fd);
-
- /* rewrite the metafile */
- _vio_apnd2_writemeta(vio);
- /* close all open files */
- snprintf(filename, MAX_PATH_LEN, "%s/pid.cdb", myio->filepath);
- unlink(filename);
- /* dellog only be useful for recovery of database unsafety close */
- snprintf(filename, MAX_PATH_LEN, "%s/dellog.cdb", myio->filepath);
- unlink(filename);
- _vio_apnd2_setopensig(vio, VIOAPND2_SIGCLOSED);
- if (myio->hfd > 0)
- close(myio->hfd);
- if (myio->mfd > 0)
- close(myio->mfd);
- if (myio->dfd > 0)
- close(myio->dfd);
- _vio_apnd2_destroy(vio);
- return 0;
-}
-
-
-/* open a file, and remember its fd. The function runs under lock protection */
-static int _vio_apnd2_loadfd(CDBVIO *vio, uint32_t fid, int dtype)
-{
- VIOAPND2 *myio = (VIOAPND2*)vio->iometa;
- int fd;
- char filename[MAX_PATH_LEN];
- char ipfx[] = "idx";
- char dpfx[] = "dat";
- char *pfx;
- uint32_t vfid;
-
- if (dtype == VIOAPND2_INDEX) {
- pfx = ipfx;
- vfid = VFIDIDX(fid);
- } else if (dtype == VIOAPND2_DATA) {
- pfx = dpfx;
- vfid = VFIDDAT(fid);
- } else {
- cdb_seterrno(vio->db, CDB_INTERNALERR, __FILE__, __LINE__);
- return -1;
- }
-
- snprintf(filename, MAX_PATH_LEN, "%s/%s%08d.cdb", myio->filepath, pfx, fid);
- fd = open(filename, O_RDONLY, 0644);
- if (fd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- return -1;
- }
-
- /* cache the fd, close the oldest file not touched */
- cdb_ht_insert2(myio->fdcache, &vfid, SI4, &fd, sizeof(int));
- while(myio->fdcache->num > (uint32_t) myio->maxfds) {
- CDBHTITEM *item = cdb_ht_poptail(myio->fdcache);
- close(*(int*)cdb_ht_itemval(myio->fdcache, item));
- free(item);
- }
-
- return fd;
-}
-
-/* read a index page */
-static int _vio_apnd2_readpage(CDBVIO *vio, CDBPAGE **page, FOFF off)
-{
- VIOAPND2 *myio = (VIOAPND2*)vio->iometa;
- int ret, fd;
- uint32_t psize;
- uint32_t fid, roff;
- uint32_t fixbufsize = SBUFSIZE - (sizeof(CDBPAGE) - PAGEHSIZE);
- uint32_t areadsize = PAGEAREADSIZE; //vio->db->areadsize;
-
- VOFF2ROFF(off, fid, roff);
- /* avoid dirty memory */
- (*page)->magic = 0;
-
- cdb_lock_lock(myio->lock);
- if (fid == myio->ibuf.fid)
- /* read from current writing file? */
- fd = myio->ibuf.fd;
- else {
- /* old index file */
- int vfid, *fdret;
- vfid = VFIDIDX(fid);
- /* in cache? */
- fdret = cdb_ht_get2(myio->fdcache, &vfid, sizeof(vfid), true);
- if (fdret == NULL) {
- fd = _vio_apnd2_loadfd(vio, fid, VIOAPND2_INDEX);
- if (fd < 0) {
- cdb_lock_unlock(myio->lock);
- return -1;
- }
- } else
- fd = *fdret;
- }
-
- /* NOTICE: the data on disk actually starts at 'magic' field in structure */
- ret = _vio_apnd2_read(vio, fd, &(*page)->magic, areadsize, roff);
- if (ret <= 0) {
- cdb_lock_unlock(myio->lock);
- return -1;
- }
-
- if ((*page)->magic != PAGEMAGIC) {
- cdb_lock_unlock(myio->lock);
- cdb_seterrno(vio->db, CDB_DATAERRIDX, __FILE__, __LINE__);
- return -1;
- }
-
- psize = PAGESIZE(*page);
- if ((uint32_t) ret < areadsize && (uint32_t) ret < psize) {
- cdb_lock_unlock(myio->lock);
- cdb_seterrno(vio->db, CDB_DATAERRIDX, __FILE__, __LINE__);
- return ret;
- } else if (psize > areadsize) {
- /* need another read operation since the page is a large than default read size */
- if (psize > fixbufsize) {
- /* record is larger the stack size */
- CDBPAGE *npage = (CDBPAGE *)malloc(sizeof(CDBPAGE) + (*page)->num * sizeof(PITEM));
- memcpy(&npage->magic, &(*page)->magic, areadsize);
- *page = npage;
- }
-
- ret = _vio_apnd2_read(vio, fd, (char*)&(*page)->magic + areadsize,
- psize - areadsize, roff + areadsize);
- if ((uint32_t) ret < psize - areadsize) {
- cdb_lock_unlock(myio->lock);
- cdb_seterrno(vio->db, CDB_DATAERRIDX, __FILE__, __LINE__);
- return -1;
- }
- }
-
- cdb_lock_unlock(myio->lock);
-
- /* remember where i got the page, calculate into junk space if page is discarded */
- (*page)->osize = OFFALIGNED(psize);
- (*page)->ooff = off;
- (*page)->cap = (*page)->num;
- return 0;
-}
-
-/* read a data record */
-static int _vio_apnd2_readrec(CDBVIO *vio, CDBREC** rec, FOFF off, bool readval)
-{
- VIOAPND2 *myio = (VIOAPND2*)vio->iometa;
- int ret, fd;
- uint32_t rsize;
- uint32_t fid, roff;
- /* the 'rec' is hoped to be fit in stack, the actually size is a little smaller */
- /* because some fields in CDBREC structure are not on disk */
- uint32_t fixbufsize = SBUFSIZE - (sizeof(CDBREC) - RECHSIZE);
- uint32_t areadsize = vio->db->areadsize;
-
- VOFF2ROFF(off, fid, roff);
- /* avoid dirty memory */
- (*rec)->magic = 0;
-
- cdb_lock_lock(myio->lock);
- if (fid == myio->dbuf.fid)
- /* read from current writing file? */
- fd = myio->dbuf.fd;
- else {
- /* read from old data file */
- int vfid, *fdret;
- vfid = VFIDDAT(fid);
- fdret = cdb_ht_get2(myio->fdcache, &vfid, sizeof(vfid), true);
- if (fdret == NULL) {
- fd = _vio_apnd2_loadfd(vio, fid, VIOAPND2_DATA);
- if (fd < 0) {
- cdb_lock_unlock(myio->lock);
- return -1;
- }
- } else
- fd = *fdret;
- }
-
- /* NOTICE: the data on disk actually starts at 'magic' field in structure */
- ret = _vio_apnd2_read(vio, fd, &(*rec)->magic, areadsize, roff);
- if (ret <= 0) {
- cdb_lock_unlock(myio->lock);
- return -1;
- }
-
- if ((*rec)->magic != RECMAGIC) {
- cdb_lock_unlock(myio->lock);
- cdb_seterrno(vio->db, CDB_DATAERRDAT, __FILE__, __LINE__);
- return -1;
- }
-
- uint32_t ovsize = (*rec)->vsize;
- if (!readval)
- /* read key only */
- (*rec)->vsize = 0;
- rsize = RECSIZE(*rec);
-
- if ((uint32_t) ret < areadsize && (uint32_t) ret < rsize) {
- cdb_lock_unlock(myio->lock);
- cdb_seterrno(vio->db, CDB_DATAERRDAT, __FILE__, __LINE__);
- return -1;
- } else if (rsize > areadsize) {
- /* need another read */
- if (rsize > fixbufsize) {
- /* record is larger the stack size */
- CDBREC *nrec = (CDBREC *)malloc(sizeof(CDBREC)+(*rec)->ksize+(*rec)->vsize);
- memcpy(&nrec->magic, &(*rec)->magic, areadsize);
- *rec = nrec;
- }
- ret = _vio_apnd2_read(vio, fd, (char*)&(*rec)->magic + areadsize,
- rsize - areadsize, roff + areadsize);
- if ((uint32_t) ret != rsize - areadsize) {
- cdb_lock_unlock(myio->lock);
- cdb_seterrno(vio->db, CDB_DATAERRDAT, __FILE__, __LINE__);
- return -1;
- }
- }
- cdb_lock_unlock(myio->lock);
-
- /* fix pointer */
- (*rec)->key = (*rec)->buf;
- (*rec)->val = (*rec)->buf + (*rec)->ksize;
-
- /* even if didn't read the value, still keep the complete (old) size */
- if (!readval)
- (*rec)->osize = OFFALIGNED(rsize + ovsize);
- else
- (*rec)->osize = OFFALIGNED(rsize);
-
- (*rec)->ooff = off;
- return 0;
-}
-
-
-/* write a index page, return the written virtual offset */
-static int _vio_apnd2_writepage(CDBVIO *vio, CDBPAGE *page, FOFF *off)
-{
- VIOAPND2 *myio = (VIOAPND2*)vio->iometa;
- VIOAPND2FINFO *finfo;
- uint32_t psize = PAGESIZE(page);
- uint32_t fid, roff;
- uint32_t ofid;
-
- page->magic = PAGEMAGIC;
- page->oid = cdb_genoid(vio->db);
-
- cdb_lock_lock(myio->lock);
- /* buffer ready? */
- if (myio->ibuf.fd < 0) {
- if (_vio_apnd2_shiftnew(vio, VIOAPND2_INDEX) < 0) {
- cdb_lock_unlock(myio->lock);
- return -1;
- }
- }
-
- /* if it was modified from existing page, remember the wasted space */
- if (OFFNOTNULL(page->ooff)) {
- VOFF2ROFF(page->ooff, ofid, roff);
- finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->idxmeta, &ofid, SI4, false);
- if (finfo)
- finfo->rcyled += page->osize;
- }
-
- if (psize > myio->ibuf.limit) {
- /* page too large */
- _vio_apnd2_flushbuf(vio, VIOAPND2_INDEX);
- fid = myio->ibuf.fid;
- roff = myio->ibuf.off;
- _vio_apnd2_write(vio, myio->ibuf.fd, &page->magic, psize, true);
- myio->ibuf.oid = page->oid;
- _vio_apnd2_flushbuf(vio, VIOAPND2_INDEX);
- cdb_lock_unlock(myio->lock);
-
- /* remember last wrote offset */
- ROFF2VOFF(fid, roff, *off);
- page->ooff = *off;
- page->osize = OFFALIGNED(psize);
- return 0;
- } else if (psize + myio->ibuf.pos > myio->ibuf.limit)
- /* buffer is full */
- _vio_apnd2_flushbuf(vio, VIOAPND2_INDEX);
-
- /* copy to buffer */
- fid = myio->ibuf.fid;
- roff = myio->ibuf.off + myio->ibuf.pos;
- memcpy(myio->ibuf.buf + myio->ibuf.pos, &page->magic, psize);
- myio->ibuf.pos += psize;
- myio->ibuf.pos = OFFALIGNED(myio->ibuf.pos);
- myio->ibuf.oid = page->oid;
- cdb_lock_unlock(myio->lock);
- ROFF2VOFF(fid, roff, *off);
-
- /* remember last wrote offset */
- page->ooff = *off;
- page->osize = OFFALIGNED(psize);
- return 0;
-}
-
-
-/* delete a record */
-static int _vio_apnd2_deleterec(CDBVIO *vio, CDBREC *rec, FOFF off)
-{
- VIOAPND2 *myio = (VIOAPND2*)vio->iometa;
- uint32_t ofid, roff;
-
- cdb_lock_lock(myio->lock);
- myio->delbuf[myio->delbufpos] = off;
- if (++myio->delbufpos == DELBUFMAX) {
- if (_vio_apnd2_flushbuf(vio, VIOAPND2_DELLOG) < 0)
- return -1;
- }
-
- /* it is an deleted record, remember the space to be recycled */
- VOFF2ROFF(off, ofid, roff);
- if (OFFNOTNULL(rec->ooff)) {
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->datmeta, &ofid, SI4, false);
- if (finfo) {
- finfo->rcyled += rec->osize;
- }
- }
- cdb_lock_unlock(myio->lock);
- return 0;
-}
-
-
-
-/* write a data record, return the written virtual offset */
-static int _vio_apnd2_writerec(CDBVIO *vio, CDBREC *rec, FOFF *off, int ptrtype) {
- VIOAPND2 *myio = (VIOAPND2*)vio->iometa;
- uint32_t rsize = RECSIZE(rec);
- uint32_t fid, roff, ofid;
- if (ptrtype == VIOAPND2_RECEXTERNAL)
- rec->magic = RECMAGIC;
-
- /* oid always are increment, even if it is a record moved from an old data file */
- rec->oid = cdb_genoid(vio->db);
- cdb_lock_lock(myio->lock);
- /* buffer ready? */
- if (myio->dbuf.fd < 0) {
- if (_vio_apnd2_shiftnew(vio, VIOAPND2_DATA) < 0) {
- cdb_lock_unlock(myio->lock);
- return -1;
- }
- }
- /* it is an overwritten record, remember the space to be recycled */
- if (OFFNOTNULL(rec->ooff)) {
- VOFF2ROFF(rec->ooff, ofid, roff);
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->datmeta, &ofid, SI4, false);
- if (finfo)
- finfo->rcyled += rec->osize;
- }
- if (rsize > myio->dbuf.limit) {
- /* record too large */
- _vio_apnd2_flushbuf(vio, VIOAPND2_DATA);
- fid = myio->dbuf.fid;
- roff = myio->dbuf.off;
- _vio_apnd2_write(vio, myio->dbuf.fd, &rec->magic, RECHSIZE, true);
- if (ptrtype == VIOAPND2_RECINTERNAL)
- _vio_apnd2_write(vio, myio->dbuf.fd, rec->buf, rec->ksize + rec->vsize, false);
- else {
- _vio_apnd2_write(vio, myio->dbuf.fd, rec->key, rec->ksize, false);
- _vio_apnd2_write(vio, myio->dbuf.fd, rec->val, rec->vsize, false);
- }
- /* reset the buffer */
- myio->dbuf.oid = rec->oid;
- _vio_apnd2_flushbuf(vio, VIOAPND2_DATA);
- if (rec->expire) {
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->datmeta, &fid, SI4, false);
- if (finfo) {
- if (finfo->nexpire == 0) {
- finfo->lcktime = time(NULL);
- finfo->nexpire = rec->expire;
- } else if (finfo->nexpire > rec->expire) {
- finfo->nexpire = rec->expire;
- }
- }
- }
- cdb_lock_unlock(myio->lock);
- ROFF2VOFF(fid, roff, *off);
- return 0;
- } else if (rsize + myio->dbuf.pos > myio->dbuf.limit)
- /* buffer is full */
- _vio_apnd2_flushbuf(vio, VIOAPND2_DATA);
- /* copy to buffer */
- fid = myio->dbuf.fid;
- roff = myio->dbuf.off + myio->dbuf.pos;
- memcpy(myio->dbuf.buf + myio->dbuf.pos, &rec->magic, RECHSIZE);
- myio->dbuf.pos += RECHSIZE;
- if (ptrtype == VIOAPND2_RECINTERNAL) {
- memcpy(myio->dbuf.buf + myio->dbuf.pos, rec->buf, rec->ksize + rec->vsize);
- myio->dbuf.pos += rec->ksize + rec->vsize;
- } else {
- memcpy(myio->dbuf.buf + myio->dbuf.pos, rec->key, rec->ksize);
- myio->dbuf.pos += rec->ksize;
- memcpy(myio->dbuf.buf + myio->dbuf.pos, rec->val, rec->vsize);
- myio->dbuf.pos += rec->vsize;
- }
- myio->dbuf.pos = OFFALIGNED(myio->dbuf.pos);
- myio->dbuf.oid = rec->oid;
- if (rec->expire) {
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->datmeta, &fid, SI4, false);
- if (finfo) {
- if (finfo->nexpire == 0) {
- finfo->lcktime = time(NULL);
- finfo->nexpire = rec->expire;
- } else if (finfo->nexpire > rec->expire) {
- finfo->nexpire = rec->expire;
- }
- }
- }
- ROFF2VOFF(fid, roff, *off);
- cdb_lock_unlock(myio->lock);
- rec->osize = rsize;
- rec->ooff = *off;
- return 0;
-}
-
-static int _vio_apnd2_writerecexternal(CDBVIO *vio, CDBREC *rec, FOFF *off)
-{
- return _vio_apnd2_writerec(vio, rec, off, VIOAPND2_RECEXTERNAL);
-}
-
-static int _vio_apnd2_writerecinternal(CDBVIO *vio, CDBREC *rec, FOFF *off)
-{
- return _vio_apnd2_writerec(vio, rec, off, VIOAPND2_RECINTERNAL);
-}
-
-
-/* flush buffers, and sync data to disk from OS cache */
-static int _vio_apnd2_sync(CDBVIO *vio)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- cdb_lock_lock(myio->lock);
- _vio_apnd2_flushbuf(vio, VIOAPND2_DATA);
- _vio_apnd2_flushbuf(vio, VIOAPND2_INDEX);
- if (myio->dbuf.fd > 0)
- fdatasync(myio->dbuf.fd);
- if (myio->ibuf.fd > 0)
- fdatasync(myio->ibuf.fd);
-
- _vio_apnd2_writehead(vio, false);
- cdb_lock_unlock(myio->lock);
- return 0;
-}
-
-
-/* write db information and main index table into a single file */
-static int _vio_apnd2_writehead(CDBVIO *vio, bool wtable)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- CDB *db = vio->db;
- char buf[FILEMETASIZE];
- int pos = 0;
-
- memset(buf, 'X', FILEMETASIZE);
- memcpy(buf, FILEMAGICHEADER, FILEMAGICLEN);
- pos += FILEMAGICLEN;
- *(uint32_t*)(buf + pos) = db->hsize;
- pos += SI4;
- *(uint64_t*)(buf + pos) = db->oid;
- pos += SI8;
- *(uint64_t*)(buf + pos) = db->roid;
- pos += SI8;
- *(uint64_t*)(buf + pos) = db->rnum;
- pos += SI8;
- *(uint32_t*)(buf + pos) = VIOAPND2_SIGOPEN;
- pos += SI4;
-
- if (pwrite(myio->hfd, buf, FILEMETASIZE, 0) != FILEMETASIZE) {
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- return -1;
- }
-
- if (wtable && (size_t) pwrite(myio->hfd, db->mtable, sizeof(FOFF) * db->hsize, FILEMETASIZE)
- != sizeof(FOFF) * db->hsize) {
- cdb_seterrno(vio->db, CDB_WRITEERR, __FILE__, __LINE__);
- return -1;
- }
- return 0;
-}
-
-
-/* wrapped for upper layer */
-static int _vio_apnd2_writehead2(CDBVIO *vio)
-{
- return _vio_apnd2_writehead(vio, true);
-}
-
-
-/* read db information and main index table from a single file */
-static int _vio_apnd2_readhead(CDBVIO *vio, bool rtable)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- CDB *db = vio->db;
- char buf[FILEMETASIZE];
- int pos = 0;
-
- if (myio->create) {
- /* the db is just created, allocate a empty main index table for db */
- db->mtable = (FOFF *)malloc(sizeof(FOFF) * db->hsize);
- memset(db->mtable, 0, sizeof(FOFF) * db->hsize);
- _vio_apnd2_writehead(vio, false);
- return 0;
- }
-
- if (pread(myio->hfd, buf, FILEMETASIZE, 0) != FILEMETASIZE) {
- cdb_seterrno(db, CDB_READERR, __FILE__, __LINE__);
- return -1;
- }
-
- if (memcmp(buf, FILEMAGICHEADER, FILEMAGICLEN)) {
- cdb_seterrno(db, CDB_DATAERRMETA, __FILE__, __LINE__);
- return -1;
- }
-
- pos += FILEMAGICLEN;
- db->hsize = *(uint32_t*)(buf + pos);
- pos += SI4;
- db->oid = *(uint64_t*)(buf + pos);
- pos += SI8;
- db->roid = *(uint64_t*)(buf + pos);
- pos += SI8;
- db->rnum = *(uint64_t*)(buf + pos);
- pos += SI8;
- /* 4 bytes reserved for open status */
- pos += SI4;
-
- if (!rtable)
- return 0;
-
- if (db->mtable)
- free(db->mtable);
- db->mtable = (FOFF *)malloc(sizeof(FOFF) * db->hsize);
- if ( (size_t) pread(myio->hfd, db->mtable, sizeof(FOFF) * db->hsize, FILEMETASIZE) !=
- sizeof(FOFF) * db->hsize) {
- free(db->mtable);
- cdb_seterrno(db, CDB_READERR, __FILE__, __LINE__);
- return -1;
- }
- return 0;
-}
-
-
-/* wrapped for upper layer */
-static int _vio_apnd2_readhead2(CDBVIO *vio)
-{
- return _vio_apnd2_readhead(vio, true);
-}
-
-
-/* check if some dat file has too large junk space */
-static void _vio_apnd2_rcyledataspacetask(void *arg)
-{
- CDBVIO *vio = (CDBVIO *)arg;
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- CDBHTITEM *item;
- uint32_t now = time(NULL);
- uint32_t posblexpnum = 0;
- cdb_lock_lock(myio->lock);
- item = cdb_ht_iterbegin(myio->datmeta);
- while(item != NULL) {
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO*)cdb_ht_itemval(myio->datmeta, item);
- if (finfo->nexpire && finfo->nexpire <= now)
- posblexpnum++;
- item = cdb_ht_iternext(myio->datmeta, item);
- }
-
- item = cdb_ht_iterbegin(myio->datmeta);
- while(item != NULL) {
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO*)cdb_ht_itemval(myio->datmeta, item);
- uint32_t fid = finfo->fid;
- /* rcyled space size is inaccurate */
- if (finfo->rcyled * 2 < finfo->fsize
- /* no data file possibly has expire record */
- && (posblexpnum == 0
- /* long enough time passed since last check on this file */
- || finfo->lcktime + posblexpnum * DATARCYLECHECKFACTOR > now
- /* check the data file most recent expire record */
- || finfo->nexpire > now
- /* no expire record */
- || finfo->nexpire == 0)) {
- item = cdb_ht_iternext(myio->datmeta, item);
- continue;
- }
-
- /* do not work on the writing file or file to be deleted */
- if (finfo->fstatus != VIOAPND2_FULL || finfo->unlink) {
- item = cdb_ht_iternext(myio->datmeta, item);
- continue;
- }
-
- /* have to iterate and calculate recycle space */
- finfo->ref++;
- /* operation on this file should not in lock protection */
- cdb_lock_unlock(myio->lock);
-
- if (finfo->rcyled * 2 < finfo->fsize) {
- _vio_apnd2_rcyledatafile(vio, finfo, false);
- finfo->lcktime = now;
- }
-
- if (finfo->rcyled * 2 >= finfo->fsize) {
- _vio_apnd2_rcyledatafile(vio, finfo, true);
- }
-
- cdb_lock_lock(myio->lock);
- finfo->ref--;
- if (finfo->ref == 0 && finfo->unlink) {
- /* unlink the file */
- _vio_apnd2_unlink(vio, finfo, VIOAPND2_DATA);
- cdb_ht_del2(myio->datmeta, &fid, SI4);
- }
- item = cdb_ht_iterbegin(myio->datmeta);
- }
- cdb_lock_unlock(myio->lock);
-}
-
-/* only be called in _vio_apnd2_rcylepagespacetask; when a page is moved into a new
- index file, its ooff should be changed, also its copy in cache should be updated */
-static void _vio_apnd2_fixcachepageooff(CDB *db, uint32_t bid, FOFF off)
-{
- CDBPAGE *page = NULL;
-
- if (db->pcache) {
- cdb_lock_lock(db->pclock);
- page = cdb_ht_get2(db->pcache, &bid, SI4, true);
- cdb_lock_unlock(db->pclock);
- }
-
- /* not in pcache, exists in dirty page cache? */
- if (page == NULL && db->dpcache) {
- cdb_lock_lock(db->dpclock);
- page = cdb_ht_get2(db->dpcache, &bid, SI4, true);
- cdb_lock_unlock(db->dpclock);
- }
-
- if (page)
- page->ooff = off;
-}
-
-/* check if some index file has too large junk space */
-static void _vio_apnd2_rcylepagespacetask(void *arg)
-{
- CDBVIO *vio = (CDBVIO *)arg;
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- CDBHTITEM *item;
-
- cdb_lock_lock(myio->lock);
- item = cdb_ht_iterbegin(myio->idxmeta);
- while(item != NULL) {
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO*)cdb_ht_itemval(myio->idxmeta, item);
- uint32_t fid = finfo->fid;
-
- /* do not work on the writing file or file to be deleted */
- if (finfo->fstatus != VIOAPND2_FULL || finfo->unlink) {
- item = cdb_ht_iternext(myio->idxmeta, item);
- continue;
- }
-
- /* junk space too large? */
- if (finfo->rcyled * 2 > finfo->fsize) {
- int fd;
- char filename[MAX_PATH_LEN];
- snprintf(filename, MAX_PATH_LEN, "%s/idx%08d.cdb", myio->filepath, fid);
- fd = open(filename, O_RDONLY, 0644);
- if (fd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- item = cdb_ht_iternext(myio->idxmeta, item);
- continue;
- }
- finfo->ref++;
- /* I/O should not block the lock */
- cdb_lock_unlock(myio->lock);
-
- uint32_t fsize = lseek(fd, 0, SEEK_END);
- uint32_t pos = FILEMETASIZE;
- char *map = mmap(NULL, fsize, PROT_READ, MAP_PRIVATE, fd, 0);
- while(pos < fsize) {
- CDBPAGE *page = (CDBPAGE *)&map[pos-(sizeof(CDBPAGE) - PAGEHSIZE)];
- FOFF off;
-
- if (page->magic != PAGEMAGIC) {
- pos += ALIGNBYTES;
- continue;
- }
-
- ROFF2VOFF(fid, pos, off);
- page->ooff = off;
- page->osize = OFFALIGNED(PAGESIZE(page));
- if (OFFEQ(vio->db->mtable[page->bid], off)) {
- FOFF noff;
- _vio_apnd2_writepage(vio, page, &noff);
- /* lock and double check */
- cdb_lock_lock(vio->db->mlock[page->bid % MLOCKNUM]);
- if (OFFEQ(vio->db->mtable[page->bid], off)) {
- vio->db->mtable[page->bid] = noff;
- _vio_apnd2_fixcachepageooff(vio->db, page->bid, noff);
- }
- cdb_lock_unlock(vio->db->mlock[page->bid % MLOCKNUM]);
- }
- pos += OFFALIGNED(PAGESIZE(page));
- }
- munmap(map, fsize);
- close(fd);
-
- cdb_lock_lock(myio->lock);
- /* drop information for the file */
- finfo->ref--;
- finfo->unlink = true;
- if (finfo->ref == 0) {
- /* unlink the file */
- _vio_apnd2_unlink(vio, finfo, VIOAPND2_INDEX);
- cdb_ht_del2(myio->idxmeta, &fid, SI4);
- }
- /* reset the iterator */
- item = cdb_ht_iterbegin(myio->idxmeta);
- continue;
- }
- item = cdb_ht_iternext(myio->idxmeta, item);
- }
- cdb_lock_unlock(myio->lock);
-}
-
-
-/* unlink a file and remove fd from fdcache. The function runs under lock protection */
-static void _vio_apnd2_unlink(CDBVIO *vio, VIOAPND2FINFO *finfo, int dtype)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- char filename[MAX_PATH_LEN];
- char ipfx[] = "idx";
- char dpfx[] = "dat";
- char *pfx;
- uint32_t *fnum;
- uint32_t vfid, fid = finfo->fid;
- VIOAPND2FINFO **fhead, **ftail;
- CDBHTITEM *fditem = NULL;
-
- if (dtype == VIOAPND2_INDEX) {
- pfx = ipfx;
- vfid = VFIDIDX(fid);
- fnum = &myio->ifnum;
- fhead = &myio->idxfhead;
- ftail = &myio->idxftail;
- } else if (dtype == VIOAPND2_DATA) {
- pfx = dpfx;
- vfid = VFIDDAT(fid);
- fnum = &myio->dfnum;
- fhead = &myio->datfhead;
- ftail = &myio->datftail;
- } else
- return;
-
- snprintf(filename, MAX_PATH_LEN, "%s/%s%08d.cdb", myio->filepath, pfx, fid);
- fditem = cdb_ht_del(myio->fdcache, &vfid, SI4);
- if (fditem != NULL) {
- close(*(int*)cdb_ht_itemval(myio->fdcache, fditem));
- free(fditem);
- }
- (*fnum)--;
- unlink(filename);
-
- /* fix linked list of data/index files after remove a finfo from meta table */
- if (finfo->fprev)
- finfo->fprev->fnext = finfo->fnext;
- if (finfo->fnext)
- finfo->fnext->fprev = finfo->fprev;
- if (*fhead == finfo)
- *fhead = finfo->fnext;
- if (*ftail == finfo)
- *ftail = finfo->fprev;
-}
-
-
-/* only be used for sorting files at recovery */
-typedef struct {
- uint32_t fid;
- uint64_t oidf;
-} VIOAPND2SREORDER;
-
-
-static int _vio_apnd2_cmpfuncsreorder(const void *p1, const void *p2)
-{
- VIOAPND2SREORDER *s1, *s2;
- s1 = (VIOAPND2SREORDER *)p1;
- s2 = (VIOAPND2SREORDER *)p2;
- return s1->oidf - s2->oidf;
-}
-
-
-/* recovery the database if it was not close properly
- * or force recovery from roid = 0
- * the procedure runs with no lock protection */
-static int _vio_apnd2_recovery(CDBVIO *vio, bool force)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- CDB *db = vio->db;
- char filename[MAX_PATH_LEN];
- struct dirent *filelist;
- VIOAPND2SREORDER *idxorders;
- int idxpos, idxlimit;
- VIOAPND2SREORDER *datorders;
- int datpos, datlimit;
- uint32_t imaxfid = 0, dmaxfid = 0;
- bool gotmindex = false;
-
-
- idxpos = datpos = 0;
- idxlimit = datlimit = 256;
- idxorders = (VIOAPND2SREORDER *)malloc(idxlimit * sizeof(VIOAPND2SREORDER));
- datorders = (VIOAPND2SREORDER *)malloc(datlimit * sizeof(VIOAPND2SREORDER));
- DIR *dir = opendir(myio->filepath);
- myio->dfnum = myio->ifnum = 0;
- myio->datfhead = myio->datftail = myio->idxfhead = myio->idxftail = NULL;
- /* special value to mark if found current writing file */
- myio->ibuf.fid = myio->dbuf.fid = -1;
- for (filelist = readdir(dir); filelist; filelist = readdir(dir)) {
- // Check file name/type
- const char *cstr = filelist->d_name;
- if (strncmp(cstr + strlen(cstr) - 4, ".cdb", 4) != 0)
- /* not a cuttdb file*/
- continue;
- if (strcmp(cstr, "dellog.cdb") == 0) {
- snprintf(filename, MAX_PATH_LEN, "%s/%s", myio->filepath, cstr);
- myio->dfd = open(filename, O_RDONLY, 0644);
- } else if (strcmp(cstr, "mainindex.cdb") == 0) {
- gotmindex = true;
-// snprintf(filename, MAX_PATH_LEN, "%s/%s", myio->filepath, cstr);
-// myio->hfd = open(filename, O_RDONLY, 0644);
-// if (_vio_apnd2_readhead(vio, false) < 0 || db->hsize == 0) {
-// goto ERRRET;
-// }
-// db->mtable = (FOFF *)malloc(sizeof(FOFF) * db->hsize);
-// gotmindex = true;
-// memset(db->mtable, 0, sizeof(FOFF) * db->hsize);
- } else if (strcmp(cstr, "mainmeta.cdb") == 0) {
- snprintf(filename, MAX_PATH_LEN, "%s/%s", myio->filepath, cstr);
- myio->mfd = open(filename, O_RDWR, 0644);
- if (myio->mfd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- continue;
- }
- } else if (strlen(cstr) == 15
- && (strncmp(cstr, "dat", 3) == 0 || strncmp(cstr, "idx", 3) == 0)) {
- VIOAPND2FINFO finfo;
- uint64_t fsize = 0;
-
- snprintf(filename, MAX_PATH_LEN, "%s/%s", myio->filepath, cstr);
- int fd = open(filename, O_RDWR, 0644);
- if (fd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- continue;
- }
- if (_vio_apnd2_readfmeta(vio, fd, &finfo) < 0) {
- close(fd);
- continue;
- }
- fsize = lseek(fd, 0, SEEK_END);
- finfo.rcyled = 0;
- finfo.ref = 0;
- finfo.unlink = false;
- finfo.fprev = finfo.fnext = NULL;
- if (finfo.ftype == VIOAPND2_INDEX) {
- if (force) {
- /* delete all index file and rebuild them if force to recovery */
- close(fd);
- unlink(filename);
- } else {
- cdb_ht_insert2(myio->idxmeta, &finfo.fid, SI4, &finfo, sizeof(VIOAPND2FINFO));
- idxorders[idxpos].fid = finfo.fid;
- idxorders[idxpos].oidf = finfo.oidf;
- if (++idxpos == idxlimit) {
- VIOAPND2SREORDER *tmp = (VIOAPND2SREORDER *)malloc(idxlimit * 2 * sizeof(VIOAPND2SREORDER));
- memcpy(tmp, idxorders, idxlimit * sizeof(VIOAPND2SREORDER));
- idxlimit *= 2;
- free(idxorders);
- idxorders = tmp;
- }
- if(finfo.fstatus == VIOAPND2_WRITING) {
- myio->ibuf.fid = finfo.fid;
- myio->ibuf.off = OFFALIGNED(fsize);
- myio->ibuf.pos = 0;
- myio->ibuf.fd = fd;
- } else
- close(fd);
- if (finfo.fid > imaxfid)
- imaxfid = finfo.fid;
- myio->ifnum++;
- }
- } else if (finfo.ftype == VIOAPND2_DATA) {
- /* no information about nearest expire record time, make a fake one(non zero) */
- finfo.nexpire = finfo.lcktime = time(NULL);
- cdb_ht_insert2(myio->datmeta, &finfo.fid, SI4, &finfo, sizeof(VIOAPND2FINFO));
- datorders[datpos].fid = finfo.fid;
- datorders[datpos].oidf = finfo.oidf;
- if (++datpos == datlimit) {
- VIOAPND2SREORDER *tmp = (VIOAPND2SREORDER *)malloc(datlimit * 2 * sizeof(VIOAPND2SREORDER));
- memcpy(tmp, datorders, datlimit * sizeof(VIOAPND2SREORDER));
- datlimit *= 2;
- free(datorders);
- datorders = tmp;
- }
- if (finfo.fstatus == VIOAPND2_WRITING) {
- myio->dbuf.fid = finfo.fid;
- myio->dbuf.off = OFFALIGNED(fsize);
- myio->dbuf.pos = 0;
- myio->dbuf.fd = fd;
- } else
- close(fd);
- if (finfo.fid > dmaxfid)
- dmaxfid = finfo.fid;
- myio->dfnum++;
- } else
- close(fd);
- } /* end of else */
- } /* end of for */
-
-
- /* fix recycled size */
- _vio_apnd2_readmeta(vio, true);
- closedir(dir);
-
- if (!gotmindex) {
- /* recovery failed */
- /* return */
- goto ERRRET;
- } else {
- if (_vio_apnd2_readhead(vio, false) < 0)
- goto ERRRET;
- }
-
- if (myio->mfd < 0) {
- snprintf(filename, MAX_PATH_LEN, "%s/mainmeta.cdb", myio->filepath);
- myio->mfd = open(filename, O_RDWR | O_CREAT, 0644);
- if (myio->mfd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- goto ERRRET;
- }
- }
-
- /* index file complele broken, replay all records to build the index */
- if (myio->ifnum == 0 || force)
- db->roid = 0;
- /* re-count records num */
- db->rnum = 0;
-
- /* fix index/data file meta relation */
- qsort(datorders, datpos, sizeof(VIOAPND2SREORDER), _vio_apnd2_cmpfuncsreorder);
- qsort(idxorders, idxpos, sizeof(VIOAPND2SREORDER), _vio_apnd2_cmpfuncsreorder);
-
- VIOAPND2FINFO *lfinfo = NULL;
- for(int i = 0; i < datpos; i++) {
- VIOAPND2FINFO *cfinfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->datmeta, &datorders[i].fid, SI4, false);
- if (cfinfo == NULL)
- continue;
- if (lfinfo)
- lfinfo->fnext = cfinfo;
- else {
- myio->datfhead = cfinfo;
- }
- cfinfo->fprev = lfinfo;
- lfinfo = cfinfo;
- }
- myio->datftail = lfinfo;
- if (lfinfo)
- lfinfo->fnext = NULL;
- lfinfo = NULL;
- for(int i = 0; i < idxpos; i++) {
- VIOAPND2FINFO *cfinfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->idxmeta, &idxorders[i].fid, SI4, false);
- if (cfinfo == NULL)
- continue;
- if (lfinfo)
- lfinfo->fnext = cfinfo;
- else {
- myio->idxfhead = cfinfo;
- }
- cfinfo->fprev = lfinfo;
- lfinfo = cfinfo;
- }
- myio->idxftail = lfinfo;
- if (lfinfo)
- lfinfo->fnext = NULL;
- lfinfo = NULL;
-
- if (myio->ibuf.fid == (uint32_t) -1) {
- myio->ibuf.fid = 0;
- _vio_apnd2_shiftnew(vio, VIOAPND2_INDEX);
- }
- if (myio->dbuf.fid == (uint32_t) -1) {
- myio->dbuf.fid = 0;
- _vio_apnd2_shiftnew(vio, VIOAPND2_DATA);
- }
-
- /* fix offsets in main index table */
- db->mtable = (FOFF *)malloc(db->hsize * sizeof(FOFF));
- memset(db->mtable, 0, db->hsize * sizeof(FOFF));
- void *it = _vio_apnd2_pageiterfirst(vio, 0);
- if (it) {
- char sbuf[SBUFSIZE];
- CDBPAGE *page = (CDBPAGE *)sbuf;
- /* need not use iterator since don't care about contents in page */
- /* I'm just lazy, cpu time is cheap */
- while(_vio_apnd2_pageiternext(vio, &page, it) == 0) {
- if (OFFNOTNULL(db->mtable[page->bid])) {
- /* recalculate the space to be recycled */
- uint32_t ofid, roff;
- char sbuf[SBUFSIZE];
- CDBPAGE *opage = (CDBPAGE *)sbuf;
- _vio_apnd2_readpage(vio, &opage, db->mtable[page->bid]);
- if (OFFNOTNULL(opage->ooff)) {
- VOFF2ROFF(opage->ooff, ofid, roff);
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->idxmeta, &ofid, SI4, false);
- if (finfo)
- finfo->rcyled += opage->osize;
- }
- /* fix impaction of old page */
- db->rnum -= opage->num;
- if (opage != (CDBPAGE *)sbuf)
- free(opage);
- }
- db->mtable[page->bid] = page->ooff;
- db->rnum += page->num;
- if (page != (CDBPAGE *)sbuf) {
- free(page);
- page = (CDBPAGE *)sbuf;
- }
- }
- _vio_apnd2_pageiterdestory(vio, it);
- }
-
- /* like what was did just now */
- it = _vio_apnd2_reciterfirst(vio, db->roid);
- if (it) {
- char sbuf[SBUFSIZE];
- CDBREC *rec = (CDBREC *)sbuf;
- while(_vio_apnd2_reciternext(vio, &rec, it) == 0) {
- FOFF soffs[SFOFFNUM];
- FOFF *soff = soffs, ooff;
- char sbuf2[SBUFSIZE];
- OFFZERO(ooff);
- CDBREC *rrec = (CDBREC*)sbuf2;
- uint64_t hash = CDBHASH64(rec->buf, rec->ksize);
-
- /* check record with duplicate key(old version/overwritten maybe */
- int retnum = cdb_getoff(db, hash, &soff, CDB_NOTLOCKED);
- for(int i = 0; i < retnum; i++) {
- if (rrec != (CDBREC*)sbuf2) {
- free(rrec);
- rrec = (CDBREC*)sbuf2;
- }
-
- int cret = _vio_apnd2_readrec(db->vio, &rrec, soff[i], false);
- if (cret < 0)
- continue;
-
- if (rec->ksize == rrec->ksize && memcmp(rrec->key, rec->key, rec->ksize) == 0) {
- ooff = rrec->ooff;
- break;
- }
- }
- if (soff != soffs)
- free(soff);
- if (rrec != (CDBREC*)sbuf2)
- free(rrec);
-
- if (OFFNOTNULL(ooff))
- /* replace offset in index */
- cdb_replaceoff(db, hash, ooff, rec->ooff, CDB_NOTLOCKED);
- else
- cdb_updatepage(vio->db, hash, rec->ooff, CDB_PAGEINSERTOFF, CDB_NOTLOCKED);
-
- if (rec->oid > db->oid)
- db->oid = rec->oid;
- if (rec != (CDBREC *)sbuf) {
- free(rec);
- rec = (CDBREC *)sbuf;
- }
- }
- _vio_apnd2_reciterdestory(vio, it);
- }
-
- /* replay deletion logs */
- FOFF delitems[1024];
- for(; myio->dfd > 0;) {
- int ret = read(myio->dfd, delitems, 1024 * sizeof(FOFF));
- if (ret > 0) {
- for(uint32_t j = 0; j * sizeof(FOFF) < (uint32_t) ret; j++) {
- char sbuf[SBUFSIZE];
- uint32_t ofid, roff;
- CDBREC *rec = (CDBREC *)sbuf;
- if (_vio_apnd2_readrec(vio, &rec, delitems[j], false) < 0)
- continue;
- if (cdb_updatepage(db, CDBHASH64(rec->key, rec->ksize),
- delitems[j], CDB_PAGEDELETEOFF, CDB_NOTLOCKED) == 0)
- VOFF2ROFF(delitems[j], ofid, roff);
- VIOAPND2FINFO *finfo = (VIOAPND2FINFO *)cdb_ht_get2(myio->datmeta, &ofid, SI4, false);
- if (finfo)
- finfo->rcyled += rec->osize;
- if (rec != (CDBREC *)sbuf)
- free(rec);
- }
- } else {
- close(myio->dfd);
- myio->dfd = -1;
- }
- }
-
- cdb_flushalldpage(db);
- _vio_apnd2_writemeta(vio);
- _vio_apnd2_writehead(vio, true);
- cdb_ht_clean(myio->idxmeta);
- cdb_ht_clean(myio->datmeta);
- free(idxorders);
- free(datorders);
- /* mfd / dfd will be opened again after this function, but hfd won't be */
- myio->datfhead = myio->datftail = myio->idxfhead = myio->idxftail = NULL;
- if (myio->ibuf.fd > 0)
- close(myio->ibuf.fd);
- if (myio->dbuf.fd > 0)
- close(myio->dbuf.fd);
- if (myio->mfd > 0)
- close(myio->mfd);
- if (myio->dfd > 0)
- close(myio->dfd);
- return 0;
-
-ERRRET:
- closedir(dir);
- if (myio->hfd > 0)
- close(myio->hfd);
- if (myio->mfd > 0)
- close(myio->mfd);
- if (myio->dfd > 0)
- close(myio->dfd);
- free(datorders);
- free(idxorders);
- return -1;
-}
-
-
-static VIOAPND2FINFO* _vio_apnd2_fileiternext(CDBVIO *vio, int dtype, uint64_t oid)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- uint64_t foid = (uint64_t)-1;
- CDBHTITEM *item;
- CDBHASHTABLE *ht;
- VIOAPND2FINFO *finfo = NULL;
-
- if (dtype == VIOAPND2_INDEX)
- ht = myio->idxmeta;
- else if (dtype == VIOAPND2_DATA)
- ht = myio->datmeta;
- else
- return NULL;
-
- cdb_lock_lock(myio->lock);
- item = cdb_ht_iterbegin(ht);
- while(item) {
- VIOAPND2FINFO *tfinfo = (VIOAPND2FINFO *)cdb_ht_itemval(ht, item);
- if (tfinfo->oidf < foid && tfinfo->oidf >= oid) {
- foid = tfinfo->oidf;
- finfo = tfinfo;
- }
- item = cdb_ht_iternext(ht, item);
- }
- if (finfo)
- finfo->ref++;
- cdb_lock_unlock(myio->lock);
- return finfo;
-}
-
-static int _vio_apnd2_iterfirst(CDBVIO *vio, VIOAPND2ITOR *it, int dtype, int64_t oid)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- CDBHASHTABLE *tmpcache;
- char filename[MAX_PATH_LEN];
- char ipfx[] = "idx";
- char dpfx[] = "dat";
- char *pfx;
-
- if (dtype == VIOAPND2_INDEX) {
- pfx = ipfx;
- tmpcache = myio->idxmeta;
- } else if (dtype == VIOAPND2_DATA) {
- pfx = dpfx;
- tmpcache = myio->datmeta;
- } else
- return -1;
-
- if (it->finfo == NULL)
- it->finfo = _vio_apnd2_fileiternext(vio, dtype, oid);
- if (it->finfo == NULL) {
- return -1;
- }
-
- snprintf(filename, MAX_PATH_LEN, "%s/%s%08d.cdb", myio->filepath, pfx, it->finfo->fid);
- it->fd = open(filename, O_RDONLY, 0644);
- if (it->fd < 0) {
- cdb_lock_lock(myio->lock);
- it->finfo->ref--;
- if (it->finfo->ref == 0 && it->finfo->unlink) {
- /* unlink the file */
- _vio_apnd2_unlink(vio, it->finfo, dtype);
- cdb_ht_del2(tmpcache, &it->finfo->fid, SI4);
- }
- cdb_lock_unlock(myio->lock);
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- return -1;
- }
-
- it->fsize = lseek(it->fd, 0, SEEK_END);
- it->mmap = mmap(NULL, it->fsize, PROT_READ, MAP_PRIVATE, it->fd, 0);
- it->off = FILEMETASIZE;
- it->oid = oid;
-
- while(it->off < it->fsize) {
- if (dtype == VIOAPND2_INDEX) {
- CDBPAGE *page = (CDBPAGE *)(it->mmap + it->off -(sizeof(CDBPAGE) - PAGEHSIZE));
- if (page->magic != PAGEMAGIC) {
- it->off += ALIGNBYTES;
- continue;
- }
- if (page->oid >= (uint64_t) oid)
- break;
- it->off += OFFALIGNED(PAGESIZE(page));
- } else if (dtype == VIOAPND2_DATA) {
- CDBREC *rec = (CDBREC *)(it->mmap + it->off -(sizeof(CDBREC) - RECHSIZE));
- if (rec->magic != RECMAGIC && rec->magic != DELRECMAGIC) {
- it->off += ALIGNBYTES;
- continue;
- }
- if (rec->oid >= (uint64_t) oid)
- break;
- it->off += OFFALIGNED(RECSIZE(rec));
- }
- }
-
- if (it->off >= it->fsize) {
- munmap(it->mmap, it->fsize);
- close(it->fd);
- cdb_lock_lock(myio->lock);
- it->finfo->ref--;
- if (it->finfo->ref == 0 && it->finfo->unlink) {
- /* unlink the file */
- _vio_apnd2_unlink(vio, it->finfo, dtype);
- cdb_ht_del2(tmpcache, &it->finfo->fid, SI4);
- }
- cdb_lock_unlock(myio->lock);
- return -1;
- }
- return 0;
-}
-
-
-static int _vio_apnd2_pageiternext(CDBVIO *vio, CDBPAGE **page, void *iter)
-{
- VIOAPND2ITOR *it = (VIOAPND2ITOR *)iter;
- CDBPAGE *cpage;
- uint32_t fixbufsize = SBUFSIZE - (sizeof(CDBPAGE) - PAGEHSIZE);
-
- for(;;) {
- if (it->off >= it->fsize) {
- it->oid = CDBMAX(it->oid, it->finfo->oidl);
- _vio_apnd2_iterfree(vio, VIOAPND2_INDEX, it);
- if (_vio_apnd2_iterfirst(vio, it, VIOAPND2_INDEX, it->oid) < 0)
- return -1;
- }
- cpage = (CDBPAGE *)(it->mmap + it->off -(sizeof(CDBPAGE) - PAGEHSIZE));
- if (cpage->magic != PAGEMAGIC) {
- it->off += ALIGNBYTES;
- continue;
- }
- if (PAGESIZE(cpage) <= fixbufsize)
- memcpy(&(*page)->magic, &cpage->magic, PAGESIZE(cpage));
- else {
- *page = (CDBPAGE *)malloc(sizeof(CDBPAGE) + (*page)->num * sizeof(PITEM));
- memcpy(&(*page)->magic, &cpage->magic, PAGESIZE(cpage));
- }
- (*page)->osize = PAGESIZE(cpage);
- (*page)->cap = (*page)->num;
- ROFF2VOFF(it->finfo->fid, it->off, (*page)->ooff);
- /* set iterator to next one */
- it->oid = (*page)->oid + 1;
- it->off += OFFALIGNED(PAGESIZE(cpage));
- return 0;
- }
- return -1;
-}
-
-static int _vio_apnd2_reciternext(CDBVIO *vio, CDBREC **rec, void *iter)
-{
- VIOAPND2ITOR *it = (VIOAPND2ITOR *)iter;
- CDBREC *crec;
- uint32_t fixbufsize = SBUFSIZE - (sizeof(CDBREC) - RECHSIZE);
-
- for(;;) {
- if (it->off >= it->fsize) {
- it->oid = CDBMAX(it->oid, it->finfo->oidl);
- _vio_apnd2_iterfree(vio, VIOAPND2_DATA, it);
- if (_vio_apnd2_iterfirst(vio, it, VIOAPND2_DATA, it->oid) < 0)
- return -1;
- }
- crec = (CDBREC *)(it->mmap + it->off -(sizeof(CDBREC) - RECHSIZE));
- if (crec->magic != RECMAGIC && crec->magic != DELRECMAGIC) {
- it->off += ALIGNBYTES;
- continue;
- }
- if (RECSIZE(crec) <= fixbufsize)
- memcpy(&(*rec)->magic, &crec->magic, RECSIZE(crec));
- else {
- *rec = (CDBREC *)malloc(sizeof(CDBREC) + crec->ksize + crec->vsize);
- memcpy(&(*rec)->magic, &crec->magic, RECSIZE(crec));
- }
-
- (*rec)->osize = RECSIZE(crec);
- (*rec)->expire = crec->expire;
- ROFF2VOFF(it->finfo->fid, it->off, (*rec)->ooff);
- (*rec)->key = (*rec)->buf;
- (*rec)->val = (*rec)->buf + (*rec)->ksize;
-
- /* set iterator to next one */
- it->oid = (*rec)->oid + 1;
- it->off += OFFALIGNED(RECSIZE(crec));
- return 0;
- }
- return -1;
-}
-
-
-static int _vio_apnd2_iterfree(CDBVIO *vio, int dtype, VIOAPND2ITOR *it)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- if (it->mmap) {
- munmap(it->mmap, it->fsize);
- close(it->fd);
- cdb_lock_lock(myio->lock);
- it->finfo->ref--;
- if (it->finfo->ref == 0 && it->finfo->unlink) {
- /* unlink the file */
- VIOAPND2FINFO *tfinfo;
- it->finfo->fnext->fprev = it->finfo->fprev;
- it->finfo->fprev->fnext = it->finfo->fnext;
- tfinfo = it->finfo;
- it->finfo = it->finfo->fnext;
- _vio_apnd2_unlink(vio, tfinfo, dtype);
- if (dtype == VIOAPND2_INDEX)
- cdb_ht_del2(myio->idxmeta, &tfinfo->fid, SI4);
- else if (dtype == VIOAPND2_DATA)
- cdb_ht_del2(myio->datmeta, &tfinfo->fid, SI4);
- } else
- it->finfo = it->finfo->fnext;
- if (it->finfo)
- it->finfo->ref++;
- cdb_lock_unlock(myio->lock);
- it->mmap = NULL;
- }
- return 0;
-}
-
-
-static void* _vio_apnd2_reciterfirst(CDBVIO *vio, uint64_t oid)
-{
- VIOAPND2ITOR *it = (VIOAPND2ITOR *)malloc(sizeof(VIOAPND2ITOR));
-
- /* iterator won't get to buffered data */
- _vio_apnd2_flushbuf(vio, VIOAPND2_DATA);
- it->mmap = NULL;
- it->finfo = NULL;
- if (_vio_apnd2_iterfirst(vio, it, VIOAPND2_DATA, oid) < 0) {
- free(it);
- return NULL;
- }
- return (void*)it;
-}
-
-
-static void _vio_apnd2_reciterdestory(CDBVIO *vio, void *iter)
-{
- if (iter) {
- _vio_apnd2_iterfree(vio, VIOAPND2_DATA, (VIOAPND2ITOR *)iter);
- free(iter);
- }
-}
-
-static void* _vio_apnd2_pageiterfirst(CDBVIO *vio, uint64_t oid)
-{
- VIOAPND2ITOR *it = (VIOAPND2ITOR *)malloc(sizeof(VIOAPND2ITOR));
-
- /* iterator won't get to buffered data */
- _vio_apnd2_flushbuf(vio, VIOAPND2_INDEX);
- it->mmap = NULL;
- it->finfo = NULL;
- if (_vio_apnd2_iterfirst(vio, it, VIOAPND2_INDEX, oid) < 0) {
- free(it);
- return NULL;
- }
- return (void*)it;
-}
-
-
-static void _vio_apnd2_pageiterdestory(CDBVIO *vio, void *iter)
-{
- if (iter) {
- _vio_apnd2_iterfree(vio, VIOAPND2_INDEX, (VIOAPND2ITOR *)iter);
- free(iter);
- }
-}
-
-static int _vio_apnd2_rcyledatafile(CDBVIO *vio, VIOAPND2FINFO *finfo, bool rcyle)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- int fd;
- char filename[MAX_PATH_LEN];
- uint32_t nexpire = 0xffffffff;
-
- snprintf(filename, MAX_PATH_LEN, "%s/dat%08d.cdb", myio->filepath, finfo->fid);
- fd = open(filename, O_RDONLY, 0644);
- if (fd < 0) {
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- return -1;
- }
-
- uint32_t frsize = 0, fsize = lseek(fd, 0, SEEK_END);
- uint32_t pos = FILEMETASIZE;
- char *map = mmap(NULL, fsize, PROT_READ, MAP_PRIVATE, fd, 0);
- uint32_t now = time(NULL);
- while(pos < fsize) {
- CDBREC *rec = (CDBREC *)&map[pos-(sizeof(CDBREC) - RECHSIZE)];
- FOFF off;
- uint64_t hash;
-
- if (rec->magic != RECMAGIC && rec->magic != DELRECMAGIC) {
- pos += ALIGNBYTES;
- continue;
- }
-
- ROFF2VOFF(finfo->fid, pos, off);
- hash = CDBHASH64(rec->buf, rec->ksize);
- if (cdb_checkoff(vio->db, hash, off, CDB_NOTLOCKED)
- /* not expired */
- && (rec->expire > now || rec->expire == 0)) {
- /* nearest expire record in current file */
- if (rec->expire && rec->expire < nexpire)
- nexpire = rec->expire;
-
- /* record exist in index, skip */
- if (rcyle) {
- FOFF noff;
- rec->ooff = off;
- rec->osize = OFFALIGNED(RECSIZE(rec));
- _vio_apnd2_writerecinternal(vio, rec, &noff);
- cdb_replaceoff(vio->db, hash, off, noff, CDB_NOTLOCKED);
- }
- } else {
- if (rcyle && rec->expire && rec->expire < now) {
- /* expired record, delete from index page */
- cdb_updatepage(vio->db, hash, off, CDB_PAGEDELETEOFF, CDB_NOTLOCKED);
- }
- frsize += OFFALIGNED(RECSIZE(rec));
- }
- pos += OFFALIGNED(RECSIZE(rec));
- }
- munmap(map, fsize);
- close(fd);
- cdb_lock_lock(myio->lock);
- /* fix metainfo about nearest expire time in current data file */
- if (nexpire == 0xffffffff)
- finfo->nexpire = 0;
- else
- finfo->nexpire = nexpire;
- finfo->rcyled = frsize;
- if (rcyle) {
- /* unlink */
- finfo->unlink = true;
- }
- cdb_lock_unlock(myio->lock);
- return 0;
-}
-
-
-static void _vio_apnd2_cleanpoint(CDBVIO *vio)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- char filename[MAX_PATH_LEN];
-
- cdb_lock_lock(myio->lock);
- _vio_apnd2_flushbuf(vio, VIOAPND2_DATA);
- _vio_apnd2_flushbuf(vio, VIOAPND2_INDEX);
- _vio_apnd2_writehead(vio, false);
- if (myio->dfd > 0)
- close(myio->dfd);
- snprintf(filename, MAX_PATH_LEN, "%s/dellog.cdb", myio->filepath);
- /* clean the previous deletion log */
- myio->dfd = open(filename, O_WRONLY | O_CREAT | O_TRUNC, 0644);
- /* open failed, whom to tell? */
- if (myio->dfd < 0)
- cdb_seterrno(vio->db, CDB_OPENERR, __FILE__, __LINE__);
- cdb_lock_unlock(myio->lock);
-}
-
-
-static int _vio_apnd2_checkopensig(CDBVIO *vio)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- int pos = 0;
- uint32_t ret;
-
- if (myio->hfd < 0)
- return -1;
-
- pos += FILEMAGICLEN;
- pos += SI4;
- pos += SI8;
- pos += SI8;
- pos += SI8;
- if (pread(myio->hfd, &ret, SI4, pos) != SI4)
- return -1;
-
- return ret;
-}
-
-
-static int _vio_apnd2_setopensig(CDBVIO *vio, int sig)
-{
- VIOAPND2 *myio = (VIOAPND2 *)vio->iometa;
- int pos = 0;
- uint32_t val = sig;
- if (myio->hfd < 0)
- return -1;
-
- pos += FILEMAGICLEN;
- pos += SI4;
- pos += SI8;
- pos += SI8;
- pos += SI8;
- if (pwrite(myio->hfd, &val, SI4, pos) != SI4)
- return -1;
- return 0;
-}
-
-
diff --git a/3rdparty/cuttdb/src/vio_apnd2.h b/3rdparty/cuttdb/src/vio_apnd2.h
deleted file mode 100644
index cb47a7dbd18a0f40ad1f9a571b35b94feaeff6d5..0000000000000000000000000000000000000000
--- a/3rdparty/cuttdb/src/vio_apnd2.h
+++ /dev/null
@@ -1,23 +0,0 @@
-/*
- * CuttDB - a fast key-value storage engine
- *
- *
- * http://code.google.com/p/cuttdb/
- *
- * Copyright (c) 2012, Siyuan Fu. All rights reserved.
- * Use and distribution licensed under the BSD license.
- * See the LICENSE file for full text
- *
- * Author: Siyuan Fu <fusiyuan2010@gmail.com>
- *
- */
-
-
-#ifndef _VIO_APND2_H_
-#define _VIO_APND2_H_
-#include "cdb_vio.h"
-
-
-void vio_apnd2_init(CDBVIO *vio);
-
-#endif
diff --git a/3rdparty/libmdbx/CMakeLists.txt b/3rdparty/libmdbx/CMakeLists.txt
deleted file mode 100644
index 2ddc192beefc7e7f982fee08096f389ddcbc3fbf..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/CMakeLists.txt
+++ /dev/null
@@ -1,943 +0,0 @@
-##
-## Copyright 2020-2022 Leonid Yuriev <leo@yuriev.ru>
-## and other libmdbx authors: please see AUTHORS file.
-## All rights reserved.
-##
-## Redistribution and use in source and binary forms, with or without
-## modification, are permitted only as authorized by the OpenLDAP
-## Public License.
-##
-## A copy of this license is available in the file LICENSE in the
-## top-level directory of the distribution or, alternatively, at
-## <http://www.OpenLDAP.org/license.html>.
-##
-
-##
-## libmdbx = { Revised and extended descendant of Symas LMDB. }
-## Please see README.md at https://gitflic.ru/project/erthink/libmdbx
-##
-## Libmdbx is superior to LMDB in terms of features and reliability,
-## not inferior in performance. libmdbx works on Linux, FreeBSD, MacOS X
-## and other systems compliant with POSIX.1-2008, but also support Windows
-## as a complementary platform.
-##
-## The next version is under active non-public development and will be
-## released as MithrilDB and libmithrildb for libraries & packages.
-## Admittedly mythical Mithril is resembling silver but being stronger and
-## lighter than steel. Therefore MithrilDB is rightly relevant name.
-##
-## MithrilDB will be radically different from libmdbx by the new database
-## format and API based on C++17, as well as the Apache 2.0 License.
-## The goal of this revolution is to provide a clearer and robust API,
-## add more features and new valuable properties of database.
-##
-## The Future will (be) Positive. Ð’ÑÑ‘ будет хорошо.
-##
-
-if(CMAKE_VERSION VERSION_LESS 3.12)
- cmake_minimum_required(VERSION 3.8.2)
-else()
- cmake_minimum_required(VERSION 3.12)
-endif()
-
-cmake_policy(PUSH)
-cmake_policy(VERSION ${CMAKE_MINIMUM_REQUIRED_VERSION})
-if(NOT CMAKE_VERSION VERSION_LESS 3.15)
- cmake_policy(SET CMP0091 NEW)
-endif()
-if(NOT CMAKE_VERSION VERSION_LESS 3.13)
- cmake_policy(SET CMP0077 NEW)
-endif()
-if(NOT CMAKE_VERSION VERSION_LESS 3.12)
- cmake_policy(SET CMP0075 NEW)
-endif()
-if(NOT CMAKE_VERSION VERSION_LESS 3.9)
- cmake_policy(SET CMP0069 NEW)
- include(CheckIPOSupported)
- check_ipo_supported(RESULT CMAKE_INTERPROCEDURAL_OPTIMIZATION_AVAILABLE)
-else()
- set(CMAKE_INTERPROCEDURAL_OPTIMIZATION_AVAILABLE FALSE)
-endif()
-
-if(EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/.git" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/test/CMakeLists.txt" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/src/core.c" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/src/alloy.c" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/src/config.h.in" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/src/version.c.in" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/src/man1" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/src/mdbx_chk.c" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/src/mdbx.c++")
- set(MDBX_AMALGAMATED_SOURCE FALSE)
- find_program(GIT git)
- if(NOT GIT)
- message(SEND_ERROR "Git command-line tool not found")
- endif()
- set(MDBX_SOURCE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/src")
-elseif(EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/VERSION.txt" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/mdbx.c" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/mdbx.c++" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/config.h.in" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/man1" AND
- EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/mdbx_chk.c")
- set(MDBX_AMALGAMATED_SOURCE TRUE)
- set(MDBX_SOURCE_DIR "${CMAKE_CURRENT_SOURCE_DIR}")
-else()
- message(FATAL_ERROR "\n"
- "Please don't use tarballs nor zips which are automatically provided by Github! "
- "These archives do not contain version information and thus are unfit to build libmdbx. "
- "You can vote for ability of disabling auto-creation such unsuitable archives at https://github.community/t/disable-tarball\n"
- "Instead of above, just clone the git repository, either download a tarball or zip with the properly amalgamated source core. "
- "For embedding libmdbx use a git-submodule or the amalgamated source code.\n"
- "Please, avoid using any other techniques.")
-endif()
-
-if(DEFINED PROJECT_NAME)
- option(MDBX_FORCE_BUILD_AS_MAIN_PROJECT "Force libmdbx to full control build options even it added as a subdirectory to your project." OFF)
-endif()
-
-if(DEFINED PROJECT_NAME AND NOT MDBX_FORCE_BUILD_AS_MAIN_PROJECT)
- set(SUBPROJECT ON)
- set(NOT_SUBPROJECT OFF)
- if(NOT MDBX_AMALGAMATED_SOURCE AND NOT DEFINED BUILD_TESTING)
- set(BUILD_TESTING OFF)
- endif()
- enable_language(C)
-else()
- set(SUBPROJECT OFF)
- set(NOT_SUBPROJECT ON)
- project(libmdbx C)
- if(NOT MDBX_AMALGAMATED_SOURCE AND NOT DEFINED BUILD_TESTING)
- set(BUILD_TESTING ON)
- endif()
-endif()
-
-if(NOT MDBX_AMALGAMATED_SOURCE)
- include(CTest)
- option(MDBX_ENABLE_TESTS "Build libmdbx tests." ${BUILD_TESTING})
-elseif(DEFINED MDBX_ENABLE_TESTS AND MDBX_ENABLE_TESTS)
- message(WARNING "MDBX_ENABLE_TESTS=${MDBX_ENABLE_TESTS}: But amalgamated source code don't includes tests.")
- set(MDBX_ENABLE_TESTS OFF)
-endif()
-
-# Try to find a C++ compiler unless sure that this is unnecessary.
-if (NOT CMAKE_CXX_COMPILER_LOADED)
- include(CheckLanguage)
- if(NOT DEFINED MDBX_BUILD_CXX OR MDBX_BUILD_CXX
- OR (NOT MDBX_AMALGAMATED_SOURCE AND (NOT DEFINED MDBX_ENABLE_TESTS OR MDBX_ENABLE_TESTS)))
- check_language(CXX)
- if(CMAKE_CXX_COMPILER)
- enable_language(CXX)
- endif()
- endif()
-else()
- enable_language(CXX)
-endif()
-
-# Set default build type to Release. This is to ease a User's life.
-if(NOT CMAKE_BUILD_TYPE)
- set(CMAKE_BUILD_TYPE Release CACHE STRING
- "Choose the type of build, options are: Debug Release RelWithDebInfo MinSizeRel."
- FORCE)
-endif()
-string(TOUPPER ${CMAKE_BUILD_TYPE} CMAKE_BUILD_TYPE_UPPERCASE)
-
-if(NOT_SUBPROJECT AND (CMAKE_CROSSCOMPILING OR IOS))
- set(CMAKE_TRY_COMPILE_TARGET_TYPE STATIC_LIBRARY)
-endif()
-
-if(NOT "$ENV{TEAMCITY_PROCESS_FLOW_ID}" STREQUAL "")
- set(CI TEAMCITY)
- message(STATUS "TeamCity CI")
-elseif(NOT "$ENV{TRAVIS}" STREQUAL "")
- set(CI TRAVIS)
- message(STATUS "Travis CI")
-elseif(NOT "$ENV{CIRCLECI}" STREQUAL "")
- set(CI CIRCLE)
- message(STATUS "Circle CI")
-elseif(NOT "$ENV{APPVEYOR}" STREQUAL "")
- set(CI APPVEYOR)
- message(STATUS "AppVeyor CI")
-elseif(NOT "$ENV{CI}" STREQUAL "")
- set(CI "$ENV{CI}")
- message(STATUS "Other CI (${CI})")
-else()
- message(STATUS "Assume No any CI environment")
- unset(CI)
-endif()
-
-# output all mdbx-related targets in single directory
-if(NOT DEFINED MDBX_OUTPUT_DIR)
- set(MDBX_OUTPUT_DIR ${CMAKE_CURRENT_BINARY_DIR})
-endif()
-set(CMAKE_ARCHIVE_OUTPUT_DIRECTORY ${MDBX_OUTPUT_DIR})
-set(CMAKE_LIBRARY_OUTPUT_DIRECTORY ${MDBX_OUTPUT_DIR})
-set(CMAKE_PDB_OUTPUT_DIRECTORY ${MDBX_OUTPUT_DIR})
-set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${MDBX_OUTPUT_DIR})
-
-include(CheckFunctionExists)
-include(FindPackageMessage)
-include(GNUInstallDirs)
-
-if(CMAKE_C_COMPILER_ID STREQUAL "MSVC" AND MSVC_VERSION LESS 1900)
- message(SEND_ERROR "MSVC compiler ${MSVC_VERSION} is too old for building MDBX."
- " At least 'Microsoft Visual Studio 2015' is required.")
-endif()
-
-set(CMAKE_THREAD_PREFER_PTHREAD TRUE)
-set(THREADS_PREFER_PTHREAD_FLAG TRUE)
-find_package(Threads REQUIRED)
-
-include(cmake/utils.cmake)
-include(cmake/compiler.cmake)
-include(cmake/profile.cmake)
-
-# Workaround for `-pthread` toolchain/cmake bug
-if(NOT APPLE AND NOT MSVC
- AND CMAKE_USE_PTHREADS_INIT AND NOT CMAKE_THREAD_LIBS_INIT
- AND (CMAKE_COMPILER_IS_GNUCC OR CMAKE_COMPILER_IS_CLANG))
- check_compiler_flag("-pthread" CC_HAS_PTHREAD)
- if(CC_HAS_PTHREAD AND NOT CMAKE_EXE_LINKER_FLAGS MATCHES "-pthread")
- message(STATUS "Force add -pthread for linker flags to avoid troubles")
- set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -pthread")
- set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} -pthread")
- set(CMAKE_MODULE_LINKER_FLAGS "${CMAKE_MODULE_LINKER_FLAGS} -pthread")
- endif()
-endif()
-
-CHECK_FUNCTION_EXISTS(pow NOT_NEED_LIBM)
-if(NOT_NEED_LIBM)
- set(LIB_MATH "")
-else()
- set(CMAKE_REQUIRED_LIBRARIES m)
- CHECK_FUNCTION_EXISTS(pow HAVE_LIBM)
- if(HAVE_LIBM)
- set(LIB_MATH m)
- else()
- message(FATAL_ERROR "No libm found for math support")
- endif()
-endif()
-
-if(SUBPROJECT)
- if(NOT DEFINED BUILD_SHARED_LIBS)
- option(BUILD_SHARED_LIBS "Build shared libraries (DLLs)" OFF)
- endif()
- if(NOT DEFINED CMAKE_POSITION_INDEPENDENT_CODE)
- option(CMAKE_POSITION_INDEPENDENT_CODE "Generate position independent (PIC)" ON)
- endif()
-else()
- option(BUILD_SHARED_LIBS "Build shared libraries (DLLs)" ON)
- option(CMAKE_POSITION_INDEPENDENT_CODE "Generate position independent (PIC)" ON)
- if (CC_HAS_ARCH_NATIVE)
- option(BUILD_FOR_NATIVE_CPU "Generate code for the compiling machine CPU" OFF)
- endif()
-
- if(CMAKE_CONFIGURATION_TYPES OR NOT CMAKE_BUILD_TYPE_UPPERCASE STREQUAL "DEBUG")
- set(INTERPROCEDURAL_OPTIMIZATION_DEFAULT ON)
- else()
- set(INTERPROCEDURAL_OPTIMIZATION_DEFAULT OFF)
- endif()
-
- if(CMAKE_INTERPROCEDURAL_OPTIMIZATION_AVAILABLE
- OR GCC_LTO_AVAILABLE OR MSVC_LTO_AVAILABLE OR
- (CLANG_LTO_AVAILABLE AND
- ((DEFINED MDBX_ENABLE_TESTS AND NOT MDBX_ENABLE_TESTS)
- OR NOT CMAKE_CXX_COMPILER_VERSION VERSION_LESS 4.0)))
- option(INTERPROCEDURAL_OPTIMIZATION "Enable interprocedural/LTO optimization" ${INTERPROCEDURAL_OPTIMIZATION_DEFAULT})
- else()
- set(INTERPROCEDURAL_OPTIMIZATION OFF)
- endif()
-
- if(INTERPROCEDURAL_OPTIMIZATION)
- if(GCC_LTO_AVAILABLE)
- set(LTO_ENABLED TRUE)
- set(CMAKE_AR ${CMAKE_GCC_AR} CACHE PATH "Path to ar program with LTO-plugin" FORCE)
- set(CMAKE_NM ${CMAKE_GCC_NM} CACHE PATH "Path to nm program with LTO-plugin" FORCE)
- set(CMAKE_RANLIB ${CMAKE_GCC_RANLIB} CACHE PATH "Path to ranlib program with LTO-plugin" FORCE)
- message(STATUS "MDBX indulge Link-Time Optimization by GCC")
- elseif(CLANG_LTO_AVAILABLE)
- set(LTO_ENABLED TRUE)
- set(CMAKE_AR ${CMAKE_CLANG_AR} CACHE PATH "Path to ar program with LTO-plugin" FORCE)
- set(CMAKE_NM ${CMAKE_CLANG_NM} CACHE PATH "Path to nm program with LTO-plugin" FORCE)
- set(CMAKE_RANLIB ${CMAKE_CLANG_RANLIB} CACHE PATH "Path to ranlib program with LTO-plugin" FORCE)
- message(STATUS "MDBX indulge Link-Time Optimization by CLANG")
- elseif(MSVC_LTO_AVAILABLE)
- set(LTO_ENABLED TRUE)
- message(STATUS "MDBX indulge Link-Time Optimization by MSVC")
- elseif(CMAKE_INTERPROCEDURAL_OPTIMIZATION_AVAILABLE)
- message(STATUS "MDBX indulge Interprocedural Optimization by CMake")
- set(CMAKE_INTERPROCEDURAL_OPTIMIZATION TRUE)
- set(LTO_ENABLED TRUE)
- else()
- message(WARNING "Unable to engage interprocedural/LTO optimization.")
- endif()
- else()
- set(CMAKE_INTERPROCEDURAL_OPTIMIZATION FALSE)
- set(LTO_ENABLED FALSE)
- endif()
-
- if(NOT MDBX_AMALGAMATED_SOURCE)
- find_program(VALGRIND valgrind)
- if(VALGRIND)
- # LY: cmake is ugly and nasty.
- # - therefore memcheck-options should be defined before including ctest;
- # - otherwise ctest may ignore it.
- set(MEMORYCHECK_SUPPRESSIONS_FILE
- "${CMAKE_CURRENT_SOURCE_DIR}/test/valgrind_suppress.txt"
- CACHE FILEPATH "Suppressions file for Valgrind" FORCE)
- set(MEMORYCHECK_COMMAND_OPTIONS
- "--trace-children=yes --leak-check=full --track-origins=yes --error-exitcode=42 --error-markers=@ --errors-for-leak-kinds=definite --fair-sched=yes --suppressions=${MEMORYCHECK_SUPPRESSIONS_FILE}"
- CACHE STRING "Valgrind options" FORCE)
- set(VALGRIND_COMMAND_OPTIONS "${MEMORYCHECK_COMMAND_OPTIONS}" CACHE STRING "Valgrind options" FORCE)
- endif()
-
- # Enable 'make tags' target.
- find_program(CTAGS ctags)
- if(CTAGS)
- add_custom_target(tags COMMAND ${CTAGS} -R -f tags
- WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR})
- add_custom_target(ctags DEPENDS tags)
- endif(CTAGS)
-
- if(UNIX)
- find_program(CLANG_FORMAT
- NAMES clang-format-13 clang-format)
- if(CLANG_FORMAT)
- execute_process(COMMAND ${CLANG_FORMAT} "--version" OUTPUT_VARIABLE clang_format_version_info)
- string(REGEX MATCH "version ([0-9]+)\\.([0-9]+)\\.([0-9]+)(.*)?" clang_format_version_info CLANG_FORMAT_VERSION)
- if(clang_format_version_info AND NOT CLANG_FORMAT_VERSION VERSION_LESS 13.0)
- # Enable 'make reformat' target.
- add_custom_target(reformat
- VERBATIM
- COMMAND
- git ls-files |
- grep -E \\.\(c|cxx|cc|cpp|h|hxx|hpp\)\(\\.in\)?\$ |
- xargs ${CLANG_FORMAT} -i --style=file
- WORKING_DIRECTORY ${CMAKE_SOURCE_DIR})
- endif()
- endif()
- endif()
-
- if(NOT "${PROJECT_BINARY_DIR}" STREQUAL "${PROJECT_SOURCE_DIR}")
- add_custom_target(distclean)
- add_custom_command(TARGET distclean
- COMMAND ${CMAKE_COMMAND} -E remove_directory "${PROJECT_BINARY_DIR}"
- COMMENT "Removing the build directory and its content")
- elseif(IS_DIRECTORY .git AND GIT)
- add_custom_target(distclean)
- add_custom_command(TARGET distclean
- WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
- COMMAND ${GIT} submodule foreach --recursive git clean -f -X -d
- COMMAND ${GIT} clean -f -X -d
- COMMENT "Removing all build files from the source directory")
- endif()
- endif(NOT MDBX_AMALGAMATED_SOURCE)
-
- setup_compile_flags()
-endif(SUBPROJECT)
-
-list(FIND CMAKE_C_COMPILE_FEATURES c_std_11 HAS_C11)
-list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_11 HAS_CXX11)
-list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_14 HAS_CXX14)
-list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_17 HAS_CXX17)
-list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_20 HAS_CXX20)
-list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_23 HAS_CXX23)
-if(NOT DEFINED MDBX_CXX_STANDARD)
- if(DEFINED ENV{CMAKE_CXX_STANDARD})
- set(CMAKE_CXX_STANDARD $ENV{CMAKE_CXX_STANDARD})
- endif()
- if(DEFINED CMAKE_CXX_STANDARD)
- set(MDBX_CXX_STANDARD ${CMAKE_CXX_STANDARD})
- elseif(NOT HAS_CXX23 LESS 0
- AND NOT (CMAKE_COMPILER_IS_CLANG AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 12))
- set(MDBX_CXX_STANDARD 23)
- elseif(NOT HAS_CXX20 LESS 0
- AND NOT (CMAKE_COMPILER_IS_CLANG AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 10))
- set(MDBX_CXX_STANDARD 20)
- elseif(NOT HAS_CXX17 LESS 0
- AND NOT (CMAKE_COMPILER_IS_CLANG AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 5))
- set(MDBX_CXX_STANDARD 17)
- elseif(NOT HAS_CXX14 LESS 0)
- set(MDBX_CXX_STANDARD 14)
- elseif(NOT HAS_CXX11 LESS 0)
- set(MDBX_CXX_STANDARD 11)
- else()
- set(MDBX_CXX_STANDARD 98)
- endif()
-endif()
-if(NOT DEFINED MDBX_C_STANDARD)
- # MSVC >= 19.28 (Microsoft Visual Studio 16.8) is mad!
- # It unable process Windows SDK headers in the C11 mode!
- if(MSVC AND MSVC_VERSION GREATER 1927 AND NOT MSVC_VERSION GREATER 1929)
- set(MDBX_C_STANDARD 99)
- set(C_FALLBACK_11 OFF)
- set(C_FALLBACK_GNU11 OFF)
- elseif(HAS_C11 LESS 0 AND NOT C_FALLBACK_GNU11 AND NOT C_FALLBACK_11)
- set(MDBX_C_STANDARD 99)
- else()
- set(MDBX_C_STANDARD 11)
- endif()
-endif()
-
-if(${CMAKE_SYSTEM_NAME} STREQUAL "Windows" AND EXISTS "${MDBX_SOURCE_DIR}/ntdll.def")
- if(MSVC)
- if(NOT MSVC_LIB_EXE)
- # Find lib.exe
- get_filename_component(CL_NAME ${CMAKE_C_COMPILER} NAME)
- string(REPLACE cl.exe lib.exe MSVC_LIB_EXE ${CL_NAME})
- find_program(MSVC_LIB_EXE ${MSVC_LIB_EXE})
- endif()
- if(MSVC_LIB_EXE)
- message(STATUS "Found MSVC's lib tool: ${MSVC_LIB_EXE}")
- set(MDBX_NTDLL_EXTRA_IMPLIB "${CMAKE_CURRENT_BINARY_DIR}/mdbx_ntdll_extra.lib")
- add_custom_command(OUTPUT "${MDBX_NTDLL_EXTRA_IMPLIB}"
- COMMENT "Create extra-import-library for ntdll.dll"
- MAIN_DEPENDENCY "${MDBX_SOURCE_DIR}/ntdll.def"
- COMMAND ${MSVC_LIB_EXE} /def:"${MDBX_SOURCE_DIR}/ntdll.def" /out:"${MDBX_NTDLL_EXTRA_IMPLIB}" ${INITIAL_CMAKE_STATIC_LINKER_FLAGS})
- else()
- message(WARNING "MSVC's lib tool not found")
- endif()
- elseif(MINGW OR MINGW64)
- if(NOT DLLTOOL)
- # Find dlltool
- get_filename_component(GCC_NAME ${CMAKE_C_COMPILER} NAME)
- string(REPLACE gcc dlltool DLLTOOL_NAME ${GCC_NAME})
- find_program(DLLTOOL NAMES ${DLLTOOL_NAME})
- endif()
- if(DLLTOOL)
- message(STATUS "Found dlltool: ${DLLTOOL}")
- set(MDBX_NTDLL_EXTRA_IMPLIB "${CMAKE_CURRENT_BINARY_DIR}/mdbx_ntdll_extra.a")
- add_custom_command(OUTPUT "${MDBX_NTDLL_EXTRA_IMPLIB}"
- COMMENT "Create extra-import-library for ntdll.dll"
- MAIN_DEPENDENCY "${MDBX_SOURCE_DIR}/ntdll.def"
- COMMAND ${DLLTOOL} -d "${MDBX_SOURCE_DIR}/ntdll.def" -l "${MDBX_NTDLL_EXTRA_IMPLIB}")
- else()
- message(WARNING "dlltool not found")
- endif()
- endif()
-
- if(MDBX_NTDLL_EXTRA_IMPLIB)
- # LY: Sometimes CMake requires a nightmarish magic for simple things.
- # 1) create a target out of the library compilation result
- add_custom_target(ntdll_extra_target DEPENDS "${MDBX_NTDLL_EXTRA_IMPLIB}")
- # 2) create an library target out of the library compilation result
- add_library(ntdll_extra STATIC IMPORTED GLOBAL)
- add_dependencies(ntdll_extra ntdll_extra_target)
- # 3) specify where the library is (and where to find the headers)
- set_target_properties(ntdll_extra
- PROPERTIES
- IMPORTED_LOCATION "${MDBX_NTDLL_EXTRA_IMPLIB}")
- endif()
-endif()
-
-################################################################################
-################################################################################
-#
-# #### ##### ##### # #### # # ####
-# # # # # # # # # ## # #
-# # # # # # # # # # # # ####
-# # # ##### # # # # # # # #
-# # # # # # # # # ## # #
-# #### # # # #### # # ####
-#
-
-set(MDBX_BUILD_OPTIONS ENABLE_UBSAN ENABLE_ASAN MDBX_USE_VALGRIND ENABLE_GPROF ENABLE_GCOV)
-macro(add_mdbx_option NAME DESCRIPTION DEFAULT)
- list(APPEND MDBX_BUILD_OPTIONS ${NAME})
- if(NOT ${DEFAULT} STREQUAL "AUTO")
- option(${NAME} "${DESCRIPTION}" ${DEFAULT})
- elseif(NOT DEFINED ${NAME})
- set(${NAME}_AUTO ON)
- endif()
-endmacro()
-
-if(IOS)
- set(MDBX_BUILD_TOOLS_DEFAULT OFF)
- if(NOT_SUBPROJECT)
- cmake_policy(SET CMP0006 OLD)
- set(CMAKE_XCODE_ATTRIBUTE_CODE_SIGNING_ALLOWED "NO")
- endif()
-else()
- set(MDBX_BUILD_TOOLS_DEFAULT ON)
-endif()
-
-add_mdbx_option(MDBX_INSTALL_STATIC "Build and install libmdbx for static linking" OFF)
-add_mdbx_option(MDBX_BUILD_SHARED_LIBRARY "Build libmdbx as shared library (DLL)" ${BUILD_SHARED_LIBS})
-add_mdbx_option(MDBX_BUILD_TOOLS "Build MDBX tools (mdbx_chk/stat/dump/load/copy)" ${MDBX_BUILD_TOOLS_DEFAULT})
-CMAKE_DEPENDENT_OPTION(MDBX_INSTALL_MANPAGES "Install man-pages for MDBX tools (mdbx_chk/stat/dump/load/copy)" ON MDBX_BUILD_TOOLS OFF)
-add_mdbx_option(MDBX_TXN_CHECKOWNER "Checking transaction matches the calling thread inside libmdbx's API" ON)
-add_mdbx_option(MDBX_ENV_CHECKPID "Paranoid checking PID inside libmdbx's API" AUTO)
-mark_as_advanced(MDBX_ENV_CHECKPID)
-if(${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
- add_mdbx_option(MDBX_DISABLE_GNU_SOURCE "Don't use GNU/Linux libc extensions" OFF)
- mark_as_advanced(MDBX_DISABLE_GNU_SOURCE)
-endif()
-if(${CMAKE_SYSTEM_NAME} STREQUAL "Darwin" OR IOS)
- add_mdbx_option(MDBX_OSX_SPEED_INSTEADOF_DURABILITY "Disable use fcntl(F_FULLFSYNC) in favor of speed" OFF)
- mark_as_advanced(MDBX_OSX_SPEED_INSTEADOF_DURABILITY)
-endif()
-if(${CMAKE_SYSTEM_NAME} STREQUAL "Windows")
- if(MDBX_NTDLL_EXTRA_IMPLIB)
- add_mdbx_option(MDBX_WITHOUT_MSVC_CRT "Avoid dependence from MSVC CRT and use ntdll.dll instead" OFF)
- endif()
-else()
- add_mdbx_option(MDBX_USE_OFDLOCKS "Use Open file description locks (aka OFD locks, non-POSIX)" AUTO)
- mark_as_advanced(MDBX_USE_OFDLOCKS)
-endif()
-add_mdbx_option(MDBX_LOCKING "Locking method (Win32=-1, SysV=5, POSIX=1988, POSIX=2001, POSIX=2008, Futexes=1995)" AUTO)
-mark_as_advanced(MDBX_LOCKING)
-add_mdbx_option(MDBX_TRUST_RTC "Does a system have battery-backed Real-Time Clock or just a fake" AUTO)
-mark_as_advanced(MDBX_TRUST_RTC)
-option(MDBX_FORCE_ASSERTIONS "Force enable assertion checking" OFF)
-option(MDBX_DISABLE_PAGECHECKS "Disable some checks to reduce an overhead and detection probability of database corruption to a values closer to the LMDB" OFF)
-
-if(NOT MDBX_AMALGAMATED_SOURCE)
- if(CMAKE_CONFIGURATION_TYPES OR CMAKE_BUILD_TYPE_UPPERCASE STREQUAL "DEBUG")
- set(MDBX_ALLOY_BUILD_DEFAULT OFF)
- else()
- set(MDBX_ALLOY_BUILD_DEFAULT ON)
- endif()
- add_mdbx_option(MDBX_ALLOY_BUILD "Build MDBX library through single/alloyed object file" ${MDBX_ALLOY_BUILD_DEFAULT})
-endif()
-
-if((MDBX_BUILD_TOOLS OR MDBX_ENABLE_TESTS) AND MDBX_BUILD_SHARED_LIBRARY)
- add_mdbx_option(MDBX_LINK_TOOLS_NONSTATIC "Link MDBX tools with non-static libmdbx" OFF)
-else()
- unset(MDBX_LINK_TOOLS_NONSTATIC CACHE)
-endif()
-
-if(CMAKE_CXX_COMPILER_LOADED AND MDBX_CXX_STANDARD GREATER_EQUAL 11 AND MDBX_CXX_STANDARD LESS 83)
- if(NOT MDBX_AMALGAMATED_SOURCE)
- option(MDBX_ENABLE_TESTS "Build MDBX tests" ${BUILD_TESTING})
- endif()
- if(NOT MDBX_WITHOUT_MSVC_CRT
- AND NOT (CMAKE_COMPILER_IS_GNUCXX AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 4.8)
- AND NOT (CMAKE_COMPILER_IS_CLANG AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 3.9)
- AND NOT (MSVC AND MSVC_VERSION LESS 1900))
- option(MDBX_BUILD_CXX "Build C++ portion" ON)
- else()
- set(MDBX_BUILD_CXX FALSE)
- endif()
-else()
- set(MDBX_BUILD_CXX FALSE)
- set(MDBX_ENABLE_TESTS FALSE)
-endif()
-
-################################################################################
-################################################################################
-
-if(MDBX_BUILD_CXX AND NOT CMAKE_CXX_COMPILER_LOADED)
- message(FATAL_ERROR "MDBX_BUILD_CXX=${MDBX_BUILD_CXX}: The C++ compiler is required to build the C++API.")
-endif()
-
-if(MDBX_BUILD_CXX)
- # determine library for C++ std::filesystem
- probe_libcxx_filesystem()
-endif()
-
-# Get version
-fetch_version(MDBX "${CMAKE_CURRENT_SOURCE_DIR}" FALSE)
-message(STATUS "libmdbx version is ${MDBX_VERSION}")
-
-# sources list
-set(LIBMDBX_PUBLIC_HEADERS mdbx.h)
-set(LIBMDBX_SOURCES mdbx.h "${CMAKE_CURRENT_BINARY_DIR}/config.h")
-if(MDBX_AMALGAMATED_SOURCE)
- list(APPEND LIBMDBX_SOURCES mdbx.c)
-else()
- # generate version file
- configure_file("${MDBX_SOURCE_DIR}/version.c.in"
- "${CMAKE_CURRENT_BINARY_DIR}/version.c" ESCAPE_QUOTES)
- file(SHA256 "${CMAKE_CURRENT_BINARY_DIR}/version.c" MDBX_SOURCERY_DIGEST)
- string(MAKE_C_IDENTIFIER "${MDBX_GIT_DESCRIBE}" MDBX_SOURCERY_SUFFIX)
- set(MDBX_BUILD_SOURCERY "${MDBX_SOURCERY_DIGEST}_${MDBX_SOURCERY_SUFFIX}")
-
- if(MDBX_ALLOY_BUILD)
- list(APPEND LIBMDBX_SOURCES "${MDBX_SOURCE_DIR}/alloy.c")
- include_directories("${MDBX_SOURCE_DIR}" "${CMAKE_CURRENT_BINARY_DIR}")
- else()
- list(APPEND LIBMDBX_SOURCES
- "${CMAKE_CURRENT_BINARY_DIR}/version.c"
- "${MDBX_SOURCE_DIR}/options.h" "${MDBX_SOURCE_DIR}/defs.h"
- "${MDBX_SOURCE_DIR}/internals.h" "${MDBX_SOURCE_DIR}/osal.h"
- "${MDBX_SOURCE_DIR}/core.c" "${MDBX_SOURCE_DIR}/osal.c"
- "${MDBX_SOURCE_DIR}/lck-posix.c" "${MDBX_SOURCE_DIR}/lck-windows.c")
- include_directories("${MDBX_SOURCE_DIR}")
- endif()
-endif(MDBX_AMALGAMATED_SOURCE)
-if(MDBX_BUILD_CXX)
- message(STATUS "Use C${MDBX_C_STANDARD} and C++${MDBX_CXX_STANDARD} for libmdbx")
- list(APPEND LIBMDBX_PUBLIC_HEADERS mdbx.h++)
- list(APPEND LIBMDBX_SOURCES "${MDBX_SOURCE_DIR}/mdbx.c++" mdbx.h++)
-else()
- message(STATUS "Use C${MDBX_C_STANDARD} for libmdbx but C++ portion is disabled")
-endif()
-
-if(SUBPROJECT AND MSVC)
- if(MSVC_VERSION LESS 1900)
- message(FATAL_ERROR "At least \"Microsoft C/C++ Compiler\" version 19.0.24234.1 (Visual Studio 2015 Update 3) is required.")
- endif()
- add_compile_options("/utf-8")
-endif()
-
-macro(target_setup_options TARGET)
- if(DEFINED INTERPROCEDURAL_OPTIMIZATION)
- set_target_properties(${TARGET} PROPERTIES
- INTERPROCEDURAL_OPTIMIZATION $<BOOL:${INTERPROCEDURAL_OPTIMIZATION}>)
- endif()
- if(NOT C_FALLBACK_GNU11 AND NOT C_FALLBACK_11)
- set_target_properties(${TARGET} PROPERTIES
- C_STANDARD ${MDBX_C_STANDARD} C_STANDARD_REQUIRED ON)
- endif()
- if(MDBX_BUILD_CXX)
- set_target_properties(${TARGET} PROPERTIES
- CXX_STANDARD ${MDBX_CXX_STANDARD} CXX_STANDARD_REQUIRED ON)
- if(MSVC AND NOT MSVC_VERSION LESS 1910)
- target_compile_options(${TARGET} INTERFACE "/Zc:__cplusplus")
- endif()
- endif()
- if(CC_HAS_FASTMATH
- AND NOT (CMAKE_COMPILER_IS_CLANG AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 10))
- target_compile_options(${TARGET} PRIVATE "-ffast-math")
- endif()
- if(CC_HAS_VISIBILITY)
- target_compile_options(${TARGET} PRIVATE "-fvisibility=hidden")
- endif()
- if(BUILD_FOR_NATIVE_CPU AND CC_HAS_ARCH_NATIVE)
- target_compile_options(${TARGET} PRIVATE "-march=native")
- endif()
-endmacro()
-
-macro(libmdbx_setup_libs TARGET MODE)
- target_link_libraries(${TARGET} ${MODE} Threads::Threads)
- if(${CMAKE_SYSTEM_NAME} STREQUAL "Windows")
- target_link_libraries(${TARGET} ${MODE} ntdll advapi32)
- if(MDBX_NTDLL_EXTRA_IMPLIB AND MDBX_WITHOUT_MSVC_CRT)
- target_link_libraries(${TARGET} ${MODE} ntdll_extra)
- endif()
- elseif(${CMAKE_SYSTEM_NAME} STREQUAL "SunOS" OR ${CMAKE_SYSTEM_NAME} STREQUAL "Solaris")
- target_link_libraries(${TARGET} ${MODE} kstat)
- elseif(${CMAKE_SYSTEM_NAME} STREQUAL "Android")
- target_link_libraries(${TARGET} ${MODE} log)
- endif()
- if(LIBCXX_FILESYSTEM AND MDBX_BUILD_CXX)
- if(CMAKE_COMPILER_IS_ELBRUSCXX AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 1.25.23
- AND NOT CMAKE_VERSION VERSION_LESS 3.13)
- target_link_options(${TARGET} PUBLIC "-Wl,--allow-multiple-definition")
- endif()
- target_link_libraries(${TARGET} PUBLIC ${LIBCXX_FILESYSTEM})
- endif()
-endmacro()
-
-# build static library
-if(MDBX_INSTALL_STATIC)
- add_library(mdbx-static STATIC ${LIBMDBX_SOURCES})
-else()
- add_library(mdbx-static STATIC EXCLUDE_FROM_ALL ${LIBMDBX_SOURCES})
-endif()
-set_target_properties(mdbx-static PROPERTIES PUBLIC_HEADER "${LIBMDBX_PUBLIC_HEADERS}")
-target_compile_definitions(mdbx-static PRIVATE MDBX_BUILD_SHARED_LIBRARY=0)
-target_setup_options(mdbx-static)
-libmdbx_setup_libs(mdbx-static INTERFACE)
-if(MDBX_BUILD_SHARED_LIBRARY)
- set_target_properties(mdbx-static PROPERTIES OUTPUT_NAME mdbx-static)
-else()
- set_target_properties(mdbx-static PROPERTIES OUTPUT_NAME mdbx)
-endif()
-target_include_directories(mdbx-static INTERFACE "${CMAKE_CURRENT_SOURCE_DIR}")
-
-################################################################################
-
-# build shared library
-if(MDBX_BUILD_SHARED_LIBRARY)
- add_library(mdbx SHARED ${LIBMDBX_SOURCES})
- set_target_properties(mdbx PROPERTIES PUBLIC_HEADER "${LIBMDBX_PUBLIC_HEADERS}")
- target_compile_definitions(mdbx PRIVATE LIBMDBX_EXPORTS MDBX_BUILD_SHARED_LIBRARY=1 INTERFACE LIBMDBX_IMPORTS)
- target_setup_options(mdbx)
- libmdbx_setup_libs(mdbx PRIVATE)
- if(MSVC)
- if(MDBX_NTDLL_EXTRA_IMPLIB AND MDBX_WITHOUT_MSVC_CRT)
- set_property(TARGET mdbx PROPERTY LINKER_FLAGS "/NODEFAULTLIB")
- else()
- set_property(TARGET mdbx PROPERTY MSVC_RUNTIME_LIBRARY "MultiThreaded$<$<CONFIG:Debug>:Debug>DLL")
- endif()
- endif()
- if(CC_HAS_VISIBILITY AND (LTO_ENABLED OR INTERPROCEDURAL_OPTIMIZATION))
- set_target_properties(mdbx PROPERTIES LINK_FLAGS "-fvisibility=hidden")
- endif()
- list(APPEND MDBX_BUILD_FLAGS ${CMAKE_SHARED_LINKER_FLAGS})
- target_include_directories(mdbx INTERFACE "${CMAKE_CURRENT_SOURCE_DIR}")
-endif()
-
-if(MDBX_BUILD_SHARED_LIBRARY AND MDBX_LINK_TOOLS_NONSTATIC)
- set(TOOL_MDBX_LIB mdbx)
-
- # use, i.e. don't skip the full RPATH for the build tree
- set(CMAKE_SKIP_BUILD_RPATH FALSE)
-
- # when building, don't use the install RPATH already (but later on when installing)
- set(CMAKE_BUILD_WITH_INSTALL_RPATH FALSE)
-
- # add the automatically determined parts of the RPATH
- # which point to directories outside the build tree to the install RPATH
- set(CMAKE_INSTALL_RPATH_USE_LINK_PATH TRUE)
-
- # the RPATH to be used when installing, but only if it's not a system directory
- list(FIND CMAKE_PLATFORM_IMPLICIT_LINK_DIRECTORIES "${CMAKE_INSTALL_PREFIX}/lib" isSystemDir)
- if(isSystemDir EQUAL -1)
- if(${CMAKE_SYSTEM_NAME} STREQUAL "Darwin")
- set(CMAKE_INSTALL_RPATH "@executable_path/../lib")
- else()
- set(CMAKE_INSTALL_RPATH "\$ORIGIN/../lib")
- endif()
- endif()
-else()
- set(TOOL_MDBX_LIB mdbx-static)
-endif()
-
-# build mdbx-tools
-if(MDBX_BUILD_TOOLS)
- if(NOT MDBX_AMALGAMATED_SOURCE AND ${CMAKE_SYSTEM_NAME} STREQUAL "Windows")
- set(WINGETOPT_SRC ${MDBX_SOURCE_DIR}/wingetopt.c ${MDBX_SOURCE_DIR}/wingetopt.h)
- else()
- set(WINGETOPT_SRC "")
- endif()
-
- foreach(TOOL mdbx_chk mdbx_copy mdbx_stat mdbx_dump mdbx_load mdbx_drop)
- add_executable(${TOOL} mdbx.h ${MDBX_SOURCE_DIR}/${TOOL}.c ${WINGETOPT_SRC})
- if(NOT C_FALLBACK_GNU11 AND NOT C_FALLBACK_11)
- set_target_properties(${TOOL} PROPERTIES
- C_STANDARD ${MDBX_C_STANDARD} C_STANDARD_REQUIRED ON)
- endif()
- target_setup_options(${TOOL})
- target_link_libraries(${TOOL} ${TOOL_MDBX_LIB})
- endforeach()
- if(LIB_MATH)
- target_link_libraries(mdbx_chk ${LIB_MATH})
- target_link_libraries(mdbx_stat ${LIB_MATH})
- endif()
-endif()
-
-################################################################################
-
-# mdbx-shared-lib installation
-if(NOT DEFINED MDBX_DLL_INSTALL_DESTINATION)
- if(WIN32)
- set(MDBX_DLL_INSTALL_DESTINATION ${CMAKE_INSTALL_BINDIR})
- else()
- set(MDBX_DLL_INSTALL_DESTINATION ${CMAKE_INSTALL_LIBDIR})
- endif()
-endif()
-if(MDBX_BUILD_SHARED_LIBRARY)
- if(CMAKE_VERSION VERSION_LESS 3.12)
- install(TARGETS mdbx EXPORT libmdbx
- LIBRARY DESTINATION ${MDBX_DLL_INSTALL_DESTINATION} COMPONENT runtime
- OBJECTS DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} COMPONENT devel
- INCLUDES DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} COMPONENT devel)
- else()
- install(TARGETS mdbx EXPORT libmdbx
- LIBRARY DESTINATION ${MDBX_DLL_INSTALL_DESTINATION} COMPONENT runtime
- NAMELINK_COMPONENT devel
- OBJECTS DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} COMPONENT devel
- INCLUDES DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} COMPONENT devel)
- endif()
-endif(MDBX_BUILD_SHARED_LIBRARY)
-
-# mdbx-tools installation
-if(MDBX_BUILD_TOOLS)
- if(NOT DEFINED MDBX_TOOLS_INSTALL_DESTINATION)
- set(MDBX_TOOLS_INSTALL_DESTINATION ${CMAKE_INSTALL_BINDIR})
- endif()
- install(
- TARGETS
- mdbx_chk
- mdbx_stat
- mdbx_copy
- mdbx_dump
- mdbx_load
- mdbx_drop
- RUNTIME
- DESTINATION ${MDBX_TOOLS_INSTALL_DESTINATION}
- COMPONENT runtime)
- if(MDBX_INSTALL_MANPAGES)
- if(NOT DEFINED MDBX_MAN_INSTALL_DESTINATION)
- set(MDBX_MAN_INSTALL_DESTINATION ${CMAKE_INSTALL_MANDIR}/man1)
- endif()
- install(
- FILES
- "${MDBX_SOURCE_DIR}/man1/mdbx_chk.1"
- "${MDBX_SOURCE_DIR}/man1/mdbx_stat.1"
- "${MDBX_SOURCE_DIR}/man1/mdbx_copy.1"
- "${MDBX_SOURCE_DIR}/man1/mdbx_dump.1"
- "${MDBX_SOURCE_DIR}/man1/mdbx_load.1"
- "${MDBX_SOURCE_DIR}/man1/mdbx_drop.1"
- DESTINATION ${MDBX_MAN_INSTALL_DESTINATION}
- COMPONENT doc)
- endif()
-endif(MDBX_BUILD_TOOLS)
-
-# mdbx-static-lib installation
-if(MDBX_INSTALL_STATIC)
- if(CMAKE_VERSION VERSION_LESS 3.12)
- install(TARGETS mdbx-static EXPORT libmdbx
- LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- OBJECTS DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} COMPONENT devel
- INCLUDES DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} COMPONENT devel)
- else()
- install(TARGETS mdbx-static EXPORT libmdbx
- LIBRARY DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- NAMELINK_COMPONENT devel
- OBJECTS DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR} COMPONENT devel
- PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} COMPONENT devel
- INCLUDES DESTINATION ${CMAKE_INSTALL_INCLUDEDIR} COMPONENT devel)
- endif()
-endif(MDBX_INSTALL_STATIC)
-
-################################################################################
-
-# collect options & build info
-if(NOT DEFINED MDBX_BUILD_TIMESTAMP)
- string(TIMESTAMP MDBX_BUILD_TIMESTAMP UTC)
-endif()
-set(MDBX_BUILD_FLAGS ${CMAKE_C_FLAGS})
-if(MDBX_BUILD_CXX)
- set(MDBX_BUILD_FLAGS ${CMAKE_CXX_FLAGS})
-endif()
-
-# append cmake's build-type flags and defines
-if(NOT CMAKE_CONFIGURATION_TYPES)
- list(APPEND MDBX_BUILD_FLAGS ${CMAKE_C_FLAGS_${CMAKE_BUILD_TYPE_UPPERCASE}})
- if(MDBX_BUILD_CXX)
- list(APPEND MDBX_BUILD_FLAGS ${CMAKE_CXX_FLAGS_${CMAKE_BUILD_TYPE_UPPERCASE}})
- endif()
-endif()
-
-# choice target to fetch definitions and options
-if(MDBX_BUILD_SHARED_LIBRARY)
- set(target4fetch mdbx)
-else()
- set(target4fetch mdbx-static)
-endif()
-
-# get definitions
-get_target_property(defs_list ${target4fetch} COMPILE_DEFINITIONS)
-if(defs_list)
- list(APPEND MDBX_BUILD_FLAGS ${defs_list})
-endif()
-
-# get target compile options
-get_target_property(options_list ${target4fetch} COMPILE_OPTIONS)
-if(options_list)
- list(APPEND MDBX_BUILD_FLAGS ${options_list})
-endif()
-
-list(REMOVE_DUPLICATES MDBX_BUILD_FLAGS)
-string(REPLACE ";" " " MDBX_BUILD_FLAGS "${MDBX_BUILD_FLAGS}")
-if(CMAKE_CONFIGURATION_TYPES)
- # add dynamic part via per-configuration define
- message(STATUS "MDBX Compile Flags: ${MDBX_BUILD_FLAGS} <AND CONFIGURATION DEPENDENT>")
- add_definitions(-DMDBX_BUILD_FLAGS_CONFIG="$<$<CONFIG:Debug>:${CMAKE_C_FLAGS_DEBUG} ${CMAKE_C_DEFINES_DEBUG}>$<$<CONFIG:Release>:${CMAKE_C_FLAGS_RELEASE} ${CMAKE_C_DEFINES_RELEASE}>$<$<CONFIG:RelWithDebInfo>:${CMAKE_C_FLAGS_RELWITHDEBINFO} ${CMAKE_C_DEFINES_RELWITHDEBINFO}>$<$<CONFIG:MinSizeRel>:${CMAKE_C_FLAGS_MINSIZEREL} ${CMAKE_C_DEFINES_MINSIZEREL}>")
-else()
- message(STATUS "MDBX Compile Flags: ${MDBX_BUILD_FLAGS}")
-endif()
-
-# get compiler info
-execute_process(COMMAND sh -c "${CMAKE_C_COMPILER} --version | head -1"
- OUTPUT_VARIABLE MDBX_BUILD_COMPILER
- OUTPUT_STRIP_TRAILING_WHITESPACE
- ERROR_QUIET
- RESULT_VARIABLE rc)
-if(rc OR NOT MDBX_BUILD_COMPILER)
- string(STRIP "${CMAKE_C_COMPILER_ID}-${CMAKE_C_COMPILER_VERSION}" MDBX_BUILD_COMPILER)
-endif()
-
-# make a build-target triplet
-if(CMAKE_C_COMPILER_TARGET)
- set(MDBX_BUILD_TARGET "${CMAKE_C_COMPILER_TARGET}")
-else()
- if(CMAKE_C_COMPILER_ARCHITECTURE_ID)
- string(STRIP "${CMAKE_C_COMPILER_ARCHITECTURE_ID}" MDBX_BUILD_TARGET)
- elseif(CMAKE_GENERATOR_PLATFORM AND NOT CMAKE_GENERATOR_PLATFORM STREQUAL CMAKE_SYSTEM_NAME)
- string(STRIP "${CMAKE_GENERATOR_PLATFORM}" MDBX_BUILD_TARGET)
- elseif(CMAKE_SYSTEM_ARCH)
- string(STRIP "${CMAKE_SYSTEM_ARCH}" MDBX_BUILD_TARGET)
- elseif(CMAKE_LIBRARY_ARCHITECTURE)
- string(STRIP "${CMAKE_LIBRARY_ARCHITECTURE}" MDBX_BUILD_TARGET)
- elseif(CMAKE_SYSTEM_PROCESSOR)
- string(STRIP "${CMAKE_SYSTEM_PROCESSOR}" MDBX_BUILD_TARGET)
- else()
- set(MDBX_BUILD_TARGET "unknown")
- endif()
- if(CMAKE_C_COMPILER_ABI
- AND NOT (CMAKE_C_COMPILER_ABI MATCHES ".*${MDBX_BUILD_TARGET}.*" OR MDBX_BUILD_TARGET MATCHES ".*${CMAKE_C_COMPILER_ABI}.*"))
- string(APPEND MDBX_BUILD_TARGET "-${CMAKE_C_COMPILER_ABI}")
- endif()
- if(CMAKE_C_PLATFORM_ID
- AND NOT (CMAKE_SYSTEM_NAME
- AND (CMAKE_C_PLATFORM_ID MATCHES ".*${CMAKE_SYSTEM_NAME}.*" OR CMAKE_SYSTEM_NAME MATCHES ".*${CMAKE_C_PLATFORM_ID}.*"))
- AND NOT (CMAKE_C_PLATFORM_ID MATCHES ".*${CMAKE_C_PLATFORM_ID}.*" OR MDBX_BUILD_TARGET MATCHES ".*${CMAKE_C_PLATFORM_ID}.*"))
- string(APPEND MDBX_BUILD_TARGET "-${CMAKE_C_COMPILER_ABI}")
- endif()
- if(CMAKE_SYSTEM_NAME)
- string(APPEND MDBX_BUILD_TARGET "-${CMAKE_SYSTEM_NAME}")
- endif()
-endif()
-
-# provide build-type
-if(CMAKE_CONFIGURATION_TYPES)
- # via per-configuration define
- add_definitions(-DMDBX_BUILD_TYPE="$<CONFIG>")
- set(MDBX_BUILD_TYPE "<CONFIGURATION DEPENDENT>")
-else()
- set(MDBX_BUILD_TYPE ${CMAKE_BUILD_TYPE})
-endif()
-
-# options
-set(options VERSION C_COMPILER CXX_COMPILER MDBX_BUILD_TARGET MDBX_BUILD_TYPE ${MDBX_BUILD_OPTIONS})
-foreach(item IN LISTS options)
- if(DEFINED ${item})
- set(value "${${item}}")
- elseif(DEFINED MDBX_${item})
- set(item MDBX_${item})
- set(value "${${item}}")
- elseif(DEFINED CMAKE_${item})
- set(item CMAKE_${item})
- set(value "${${item}}")
- else()
- set(value "AUTO (not pre-defined explicitly)")
- endif()
- message(STATUS "${item}: ${value}")
-endforeach(item)
-
-# provide config.h for library build info
-configure_file("${MDBX_SOURCE_DIR}/config.h.in"
- "${CMAKE_CURRENT_BINARY_DIR}/config.h" ESCAPE_QUOTES)
-add_definitions(-DMDBX_CONFIG_H="${CMAKE_CURRENT_BINARY_DIR}/config.h")
-
-################################################################################
-
-if(NOT MDBX_AMALGAMATED_SOURCE AND MDBX_ENABLE_TESTS)
- if(NOT CMAKE_CXX_COMPILER_LOADED)
- message(FATAL_ERROR "MDBX_ENABLE_TESTS=${MDBX_ENABLE_TESTS}: The C++ compiler is required to build the tests.")
- endif()
- add_subdirectory(test)
-endif()
-
-################################################################################
-
-set(PACKAGE "libmdbx")
-set(CPACK_PACKAGE_VERSION_MAJOR ${MDBX_VERSION_MAJOR})
-set(CPACK_PACKAGE_VERSION_MINOR ${MDBX_VERSION_MINOR})
-set(CPACK_PACKAGE_VERSION_PATCH ${MDBX_VERSION_RELEASE})
-set(CPACK_PACKAGE_VERSION_COMMIT ${MDBX_VERSION_REVISION})
-set(PACKAGE_VERSION "${CPACK_PACKAGE_VERSION_MAJOR}.${CPACK_PACKAGE_VERSION_MINOR}.${CPACK_PACKAGE_VERSION_PATCH}.${CPACK_PACKAGE_VERSION_COMMIT}")
-message(STATUS "libmdbx package version is ${PACKAGE_VERSION}")
-
-cmake_policy(POP)
diff --git a/3rdparty/libmdbx/ChangeLog.md b/3rdparty/libmdbx/ChangeLog.md
deleted file mode 100644
index f677fed8d9822d262f63df2d8f2fe49b40df40c3..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/ChangeLog.md
+++ /dev/null
@@ -1,777 +0,0 @@
-ChangeLog
----------
-
-### TODO
-
- - [Engage an "overlapped I/O" on Windows](todo4recovery://erased_by_github/libmdbx/issues/224).
- - [Simple careful mode for working with corrupted DB](todo4recovery://erased_by_github/libmdbx/issues/223).
- - [Move most of `mdbx_chk` functional to the library API](todo4recovery://erased_by_github/libmdbx/issues/204).
- - [Replace SRW-lock on Windows to allow shrink DB with `MDBX_NOTLS` option](todo4recovery://erased_by_github/libmdbx/issues/210).
- - [More flexible support of asynchronous runtime/framework(s)](todo4recovery://erased_by_github/libmdbx/issues/200).
- - [Migration guide from LMDB to MDBX](todo4recovery://erased_by_github/libmdbx/issues/199).
- - [Get rid of dirty-pages list in MDBX_WRITEMAP mode](todo4recovery://erased_by_github/libmdbx/issues/193).
- - [Large/Overflow pages accounting for dirty-room](todo4recovery://erased_by_github/libmdbx/issues/192).
- - [Support for RAW devices](todo4recovery://erased_by_github/libmdbx/issues/124).
- - [Support MessagePack for Keys & Values](todo4recovery://erased_by_github/libmdbx/issues/115).
- - [Engage new terminology](todo4recovery://erased_by_github/libmdbx/issues/137).
- - Packages for [Astra Linux](https://astralinux.ru/), [ALT Linux](https://www.altlinux.org/), [ROSA Linux](https://www.rosalinux.ru/), etc.
-
-
-## v0.11.7 at 2022-04-22
-
-The stable risen release after the Github's intentional malicious disaster.
-
-#### We have migrated to a reliable trusted infrastructure
-The origin for now is at [GitFlic](https://gitflic.ru/project/erthink/libmdbx)
-since on 2022-04-15 the Github administration, without any warning nor
-explanation, deleted _libmdbx_ along with a lot of other projects,
-simultaneously blocking access for many developers.
-For the same reason ~~Github~~ is blacklisted forever.
-
-GitFlic already support Russian and English languages, plan to support more,
-including å’Œ ä¸æ–‡. You are welcome!
-
-New:
-
- - Added the `tools-static` make target to build statically linked MDBX tools.
- - Support for Microsoft Visual Studio 2022.
- - Support build by MinGW' make from command line without CMake.
- - Added `mdbx::filesystem` C++ API namespace that corresponds to `std::filesystem` or `std::experimental::filesystem`.
- - Created [website](https://libmdbx.website.yandexcloud.net/) for online auto-generated documentation.
- - Used `todo4recovery://erased_by_github/` for dead (or temporarily lost) resources deleted by ~~Github~~.
- - Added `--loglevel=` command-line option to the `mdbx_test` tool.
- - Added few fast smoke-like tests into CMake builds.
-
-Fixes:
-
- - Fixed a race between starting a transaction and creating a DBI descriptor that could lead to `SIGSEGV` in the cursor tracking code.
- - Clarified description of `MDBX_EPERM` error returned from `mdbx_env_set_geometry()`.
- - Fixed non-promoting the parent transaction to be dirty in case the undo of the geometry update failed during abortion of a nested transaction.
- - Resolved linking issues with `libstdc++fs`/`libc++fs`/`libc++experimental` for C++ `std::filesystem` or `std::experimental::filesystem` for legacy compilers.
- - Added workaround for GNU Make 3.81 and earlier.
- - Added workaround for Elbrus/LCC 1.25 compiler bug of class inline `static constexpr` member field.
- - [Fixed](https://github.com/ledgerwatch/erigon/issues/3874) minor assertion regression (only debug builds were affected).
- - Fixed detection of `C++20` concepts accessibility.
- - Fixed detection of Clang's LTO availability for Android.
- - Fixed build for ARM/ARM64 by MSVC.
- - Fixed non-x86 Windows builds with `MDBX_WITHOUT_MSVC_CRT=ON` and `MDBX_BUILD_SHARED_LIBRARY=ON`.
-
-Minors:
-
- - Resolve minor MSVC warnings: avoid `/INCREMENTAL[:YES]` with `/LTCG`, `/W4` with `/W3`, the `C5105` warning.
- - Switched to using `MDBX_EPERM` instead of `MDBX_RESULT_TRUE' to indicate that the geometry cannot be updated.
- - Added `NULL` checking during memory allocation inside `mdbx_chk`.
- - Resolved all warnings from MinGW while used without CMake.
- - Added inheretable `target_include_directories()` to `CMakeLists.txt` for easy integration.
- - Added build-time checks and paranoid runtime assertions for the `off_t` arguments of `fcntl()` which are used for locking.
- - Added `-Wno-lto-type-mismatch` to avoid false-positive warnings from old GCC during LTO-enabled builds.
- - Added checking for TID (system thread id) to avoid hang on 32-bit Bionic/Android within `pthread_mutex_lock()`.
- - Reworked `MDBX_BUILD_TARGET` of CMake builds.
- - Added `CMAKE_HOST_ARCH` and `CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET`.
-
-
-## v0.11.6 at 2022-03-24
-
-The stable release with the complete workaround for an incoherence flaw of Linux unified page/buffer cache.
-Nonetheless the cause for this trouble may be an issue of Intel CPU cache/MESI.
-See [issue#269](todo4recovery://erased_by_github/libmdbx/issues/269) for more information.
-
-Acknowledgements:
-
- - [David Bouyssié](https://github.com/david-bouyssie) for [Scala bindings](https://github.com/david-bouyssie/mdbx4s).
- - [Michelangelo Riccobene](https://github.com/mriccobene) for reporting and testing.
-
-Fixes:
-
- - [Added complete workaround](todo4recovery://erased_by_github/libmdbx/issues/269) for an incoherence flaw of Linux unified page/buffer cache.
- - [Fixed](todo4recovery://erased_by_github/libmdbx/issues/272) cursor reusing for read-only transactions.
- - Fixed copy&paste typo inside `mdbx::cursor::find_multivalue()`.
-
-Minors:
-
- - Minor refine C++ API for convenience.
- - Minor internals refines.
- - Added `lib-static` and `lib-shared` targets for make.
- - Added minor workaround for AppleClang 13.3 bug.
- - Clarified error messages of a signature/version mismatch.
-
-
-## v0.11.5 at 2022-02-23
-
-The release with the temporary hotfix for a flaw of Linux unified page/buffer cache.
-See [issue#269](todo4recovery://erased_by_github/libmdbx/issues/269) for more information.
-
-Acknowledgements:
-
- - [Simon Leier](https://github.com/leisim) for reporting and testing.
- - [Kai Wetlesen](https://github.com/kaiwetlesen) for [RPMs](http://copr.fedorainfracloud.org/coprs/kwetlesen/libmdbx/).
- - [Tullio Canepa](https://github.com/canepat) for reporting C++ API issue and contributing.
-
-Fixes:
-
- - [Added hotfix](todo4recovery://erased_by_github/libmdbx/issues/269) for a flaw of Linux unified page/buffer cache.
- - [Fixed/Reworked](todo4recovery://erased_by_github/libmdbx/pull/270) move-assignment operators for "managed" classes of C++ API.
- - Fixed potential `SIGSEGV` while open DB with overrided non-default page size.
- - [Made](todo4recovery://erased_by_github/libmdbx/issues/267) `mdbx_env_open()` idempotence in failure cases.
- - Refined/Fixed pages reservation inside `mdbx_update_gc()` to avoid non-reclamation in a rare cases.
- - Fixed typo in a retained space calculation for the hsr-callback.
-
-Minors:
-
- - Reworked functions for meta-pages, split-off non-volatile.
- - Disentangled C11-atomic fences/barriers and pure-functions (with `__attribute__((__pure__))`) to avoid compiler misoptimization.
- - Fixed hypotetic unaligned access to 64-bit dwords on ARM with `__ARM_FEATURE_UNALIGNED` defined.
- - Reasonable paranoia that makes clarity for code readers.
- - Minor fixes Doxygen references, comments, descriptions, etc.
-
-
-## v0.11.4 at 2022-02-02
-
-The stable release with fixes for large and huge databases sized of 4..128 TiB.
-
-Acknowledgements:
-
- - [Ledgerwatch](https://github.com/ledgerwatch), [Binance](https://github.com/binance-chain) and [Positive Technologies](https://www.ptsecurity.com/) teams for reporting, assistance in investigation and testing.
- - [Alex Sharov](https://github.com/AskAlexSharov) for reporting, testing and provide resources for remote debugging/investigation.
- - [Kris Zyp](https://github.com/kriszyp) for [Deno](https://deno.land/) support.
-
-New features, extensions and improvements:
-
- - Added treating the `UINT64_MAX` value as maximum for given option inside `mdbx_env_set_option()`.
- - Added `to_hex/to_base58/to_base64::output(std::ostream&)` overloads without using temporary string objects as buffers.
- - Added `--geometry-jitter=YES|no` option to the test framework.
- - Added support for [Deno](https://deno.land/) support by [Kris Zyp](https://github.com/kriszyp).
-
-Fixes:
-
- - Fixed handling `MDBX_opt_rp_augment_limit` for GC's records from huge transactions (Erigon/Akula/Ethereum).
- - [Fixed](todo4recovery://erased_by_github/libmdbx/issues/258) build on Android (avoid including `sys/sem.h`).
- - [Fixed](todo4recovery://erased_by_github/libmdbx/pull/261) missing copy assignment operator for `mdbx::move_result`.
- - Fixed missing `&` for `std::ostream &operator<<()` overloads.
- - Fixed unexpected `EXDEV` (Cross-device link) error from `mdbx_env_copy()`.
- - Fixed base64 encoding/decoding bugs in auxillary C++ API.
- - Fixed overflow of `pgno_t` during checking PNL on 64-bit platforms.
- - [Fixed](todo4recovery://erased_by_github/libmdbx/issues/260) excessive PNL checking after sort for spilling.
- - Reworked checking `MAX_PAGENO` and DB upper-size geometry limit.
- - [Fixed](todo4recovery://erased_by_github/libmdbx/issues/265) build for some combinations of versions of MSVC and Windows SDK.
-
-Minors:
-
- - Added workaround for CLANG bug [D79919/PR42445](https://reviews.llvm.org/D79919).
- - Fixed build test on Android (using `pthread_barrier_t` stub).
- - Disabled C++20 concepts for CLANG < 14 on Android.
- - Fixed minor `unused parameter` warning.
- - Added CI for Android.
- - Refine/cleanup internal logging.
- - Refined line splitting inside hex/base58/base64 encoding to avoid `\n` at the end.
- - Added workaround for modern libstdc++ with CLANG < 4.x
- - Relaxed txn-check rules for auxiliary functions.
- - Clarified a comments and descriptions, etc.
- - Using the `-fno-semantic interposition` option to reduce the overhead to calling self own public functions.
-
-
-## v0.11.3 at 2021-12-31
-
-Acknowledgements:
-
- - [gcxfd <i@rmw.link>](https://github.com/gcxfd) for reporting, contributing and testing.
- - [장세연 (Чан Се Ен)](https://github.com/sasgas) for reporting and testing.
- - [Alex Sharov](https://github.com/AskAlexSharov) for reporting, testing and provide resources for remote debugging/investigation.
-
-New features, extensions and improvements:
-
- - [Added](todo4recovery://erased_by_github/libmdbx/issues/236) `mdbx_cursor_get_batch()`.
- - [Added](todo4recovery://erased_by_github/libmdbx/issues/250) `MDBX_SET_UPPERBOUND`.
- - C++ API is finalized now.
- - The GC update stage has been [significantly speeded](todo4recovery://erased_by_github/libmdbx/issues/254) when fixing huge Erigon's transactions (Ethereum ecosystem).
-
-Fixes:
-
- - Disabled C++20 concepts for stupid AppleClang 13.x
- - Fixed internal collision of `MDBX_SHRINK_ALLOWED` with `MDBX_ACCEDE`.
-
-Minors:
-
- - Fixed returning `MDBX_RESULT_TRUE` (unexpected -1) from `mdbx_env_set_option()`.
- - Added `mdbx_env_get_syncbytes()` and `mdbx_env_get_syncperiod()`.
- - [Clarified](todo4recovery://erased_by_github/libmdbx/pull/249) description of `MDBX_INTEGERKEY`.
- - Reworked/simplified `mdbx_env_sync_internal()`.
- - [Fixed](todo4recovery://erased_by_github/libmdbx/issues/248) extra assertion inside `mdbx_cursor_put()` for `MDBX_DUPFIXED` cases.
- - Avoiding extra looping inside `mdbx_env_info_ex()`.
- - Explicitly enabled core dumps from stochastic tests scripts on Linux.
- - [Fixed](todo4recovery://erased_by_github/libmdbx/issues/253) `mdbx_override_meta()` to avoid false-positive assertions.
- - For compatibility reverted returning `MDBX_ENODATA`for some cases.
-
-
-## v0.11.2 at 2021-12-02
-
-Acknowledgements:
-
- - [장세연 (Чан Се Ен)](https://github.com/sasgas) for contributing to C++ API.
- - [Alain Picard](https://github.com/castortech) for [Java bindings](https://github.com/castortech/mdbxjni).
- - [Alex Sharov](https://github.com/AskAlexSharov) for reporting and testing.
- - [Kris Zyp](https://github.com/kriszyp) for reporting and testing.
- - [Artem Vorotnikov](https://github.com/vorot93) for support [Rust wrapper](https://github.com/vorot93/libmdbx-rs).
-
-Fixes:
-
- - [Fixed compilation](todo4recovery://erased_by_github/libmdbx/pull/239) with `devtoolset-9` on CentOS/RHEL 7.
- - [Fixed unexpected `MDBX_PROBLEM` error](todo4recovery://erased_by_github/libmdbx/issues/242) because of update an obsolete meta-page.
- - [Fixed returning `MDBX_NOTFOUND` error](todo4recovery://erased_by_github/libmdbx/issues/243) in case an inexact value found for `MDBX_GET_BOTH` operation.
- - [Fixed compilation](todo4recovery://erased_by_github/libmdbx/issues/245) without kernel/libc-devel headers.
-
-Minors:
-
- - Fixed `constexpr`-related macros for legacy compilers.
- - Allowed to define 'CMAKE_CXX_STANDARD` using an environment variable.
- - Simplified collection statistics of page operation .
- - Added `MDBX_FORCE_BUILD_AS_MAIN_PROJECT` cmake option.
- - Remove unneeded `#undef P_DIRTY`.
-
-
-## v0.11.1 at 2021-10-23
-
-### Backward compatibility break:
-
-The database format signature has been changed to prevent
-forward-interoperability with an previous releases, which may lead to a
-[false positive diagnosis of database corruption](todo4recovery://erased_by_github/libmdbx/issues/238)
-due to flaws of an old library versions.
-
-This change is mostly invisible:
-
- - previously versions are unable to read/write a new DBs;
- - but the new release is able to handle an old DBs and will silently upgrade ones.
-
-Acknowledgements:
-
- - [Alex Sharov](https://github.com/AskAlexSharov) for reporting and testing.
-
-
-## v0.10.5 at 2021-10-13 (obsolete, please use v0.11.1)
-
-Unfortunately, the `v0.10.5` accidentally comes not full-compatible with previous releases:
-
- - `v0.10.5` can read/processing DBs created by previous releases, i.e. the backward-compatibility is provided;
- - however, previous releases may lead to false-corrupted state with DB that was touched by `v0.10.5`, i.e. the forward-compatibility is broken for `v0.10.4` and earlier.
-
-This cannot be fixed, as it requires fixing past versions, which as a result we will just get a current version.
-Therefore, it is recommended to use `v0.11.1` instead of `v0.10.5`.
-
-Acknowledgements:
-
- - [Noel Kuntze](https://github.com/Thermi) for immediately bug reporting.
-
-Fixes:
-
- - Fixed unaligned access regression after the `#pragma pack` fix for modern compilers.
- - Added UBSAN-test to CI to avoid a regression(s) similar to lately fixed.
- - Fixed possibility of meta-pages clashing after manually turn to a particular meta-page using `mdbx_chk` utility.
-
-Minors:
-
- - Refined handling of weak or invalid meta-pages while a DB opening.
- - Refined providing information for the `@MAIN` and `@GC` sub-databases of a last committed modification transaction's ID.
-
-
-## v0.10.4 at 2021-10-10
-
-Acknowledgements:
-
- - [Artem Vorotnikov](https://github.com/vorot93) for support [Rust wrapper](https://github.com/vorot93/libmdbx-rs).
- - [Andrew Ashikhmin](https://github.com/yperbasis) for contributing to C++ API.
-
-Fixes:
-
- - Fixed possibility of looping update GC during transaction commit (no public issue since the problem was discovered inside [Positive Technologies](https://www.ptsecurity.ru)).
- - Fixed `#pragma pack` to avoid provoking some compilers to generate code with [unaligned access](todo4recovery://erased_by_github/libmdbx/issues/235).
- - Fixed `noexcept` for potentially throwing `txn::put()` of C++ API.
-
-Minors:
-
- - Added stochastic test script for checking small transactions cases.
- - Removed extra transaction commit/restart inside test framework.
- - In debugging builds fixed a too small (single page) by default DB shrink threshold.
-
-
-## v0.10.3 at 2021-08-27
-
-Acknowledgements:
-
- - [Francisco Vallarino](https://github.com/fjvallarino) for [Haskell bindings for libmdbx](https://hackage.haskell.org/package/libmdbx).
- - [Alex Sharov](https://github.com/AskAlexSharov) for reporting and testing.
- - [Andrea Lanfranchi](https://github.com/AndreaLanfranchi) for contributing.
-
-Extensions and improvements:
-
- - Added `cursor::erase()` overloads for `key` and for `key-value`.
- - Resolve minor Coverity Scan issues (no fixes but some hint/comment were added).
- - Resolve minor UndefinedBehaviorSanitizer issues (no fixes but some workaround were added).
-
-Fixes:
-
- - Always setup `madvise` while opening DB (fixes todo4recovery://erased_by_github/libmdbx/issues/231).
- - Fixed checking legacy `P_DIRTY` flag (`0x10`) for nested/sub-pages.
-
-Minors:
-
- - Fixed getting revision number from middle of history during amalgamation (GNU Makefile).
- - Fixed search GCC tools for LTO (CMake scripts).
- - Fixed/reorder dirs list for search CLANG tools for LTO (CMake scripts).
- - Fixed/workarounds for CLANG < 9.x
- - Fixed CMake warning about compatibility with 3.8.2
-
-
-## v0.10.2 at 2021-07-26
-
-Acknowledgements:
-
- - [Alex Sharov](https://github.com/AskAlexSharov) for reporting and testing.
- - [Andrea Lanfranchi](https://github.com/AndreaLanfranchi) for reporting bugs.
- - [Lionel Debroux](https://github.com/debrouxl) for fuzzing tests and reporting bugs.
- - [Sergey Fedotov](https://github.com/SergeyFromHell/) for [`node-mdbx` NodeJS bindings](https://www.npmjs.com/package/node-mdbx).
- - [Kris Zyp](https://github.com/kriszyp) for [`lmdbx-store` NodeJS bindings](https://github.com/kriszyp/lmdbx-store).
- - [Noel Kuntze](https://github.com/Thermi) for [draft Python bindings](todo4recovery://erased_by_github/libmdbx/commits/python-bindings).
-
-New features, extensions and improvements:
-
- - [Allow to predefine/override `MDBX_BUILD_TIMESTAMP` for builds reproducibility](todo4recovery://erased_by_github/libmdbx/issues/201).
- - Added options support for `long-stochastic` script.
- - Avoided `MDBX_TXN_FULL` error for large transactions when possible.
- - The `MDBX_READERS_LIMIT` increased to `32767`.
- - Raise `MDBX_TOO_LARGE` under Valgrind/ASAN if being opened DB is 100 larger than RAM (to avoid hangs and OOM).
- - Minimized the size of poisoned/unpoisoned regions to avoid Valgrind/ASAN stuck.
- - Added more workarounds for QEMU for testing builds for 32-bit platforms, Alpha and Sparc architectures.
- - `mdbx_chk` now skips iteration & checking of DB' records if corresponding page-tree is corrupted (to avoid `SIGSEGV`, ASAN failures, etc).
- - Added more checks for [rare/fuzzing corruption cases](todo4recovery://erased_by_github/libmdbx/issues/217).
-
-Backward compatibility break:
-
- - Use file `VERSION.txt` for version information instead of `VERSION` to avoid collision with `#include <version>`.
- - Rename `slice::from/to_FOO_bytes()` to `slice::envisage_from/to_FOO_length()'.
- - Rename `MDBX_TEST_EXTRA` make's variable to `MDBX_SMOKE_EXTRA`.
- - Some details of the C++ API have been changed for subsequent freezing.
-
-Fixes:
-
- - Fixed excess meta-pages checks in case `mdbx_chk` is called to check the DB for a specific meta page and thus could prevent switching to the selected meta page, even if the check passed without errors.
- - Fixed [recursive use of SRW-lock on Windows cause by `MDBX_NOTLS` option](todo4recovery://erased_by_github/libmdbx/issues/203).
- - Fixed [log a warning during a new DB creation](todo4recovery://erased_by_github/libmdbx/issues/205).
- - Fixed [false-negative `mdbx_cursor_eof()` result](todo4recovery://erased_by_github/libmdbx/issues/207).
- - Fixed [`make install` with non-GNU `install` utility (OSX, BSD)](todo4recovery://erased_by_github/libmdbx/issues/208).
- - Fixed [installation by `CMake` in special cases by complete use `GNUInstallDirs`'s variables](todo4recovery://erased_by_github/libmdbx/issues/209).
- - Fixed [C++ Buffer issue with `std::string` and alignment](todo4recovery://erased_by_github/libmdbx/issues/191).
- - Fixed `safe64_reset()` for platforms without atomic 64-bit compare-and-swap.
- - Fixed hang/shutdown on big-endian platforms without `__cxa_thread_atexit()`.
- - Fixed [using bad meta-pages if DB was partially/recoverable corrupted](todo4recovery://erased_by_github/libmdbx/issues/217).
- - Fixed extra `noexcept` for `buffer::&assign_reference()`.
- - Fixed `bootid` generation on Windows for case of change system' time.
- - Fixed [test framework keygen-related issue](todo4recovery://erased_by_github/libmdbx/issues/127).
-
-
-## v0.10.1 at 2021-06-01
-
-Acknowledgements:
-
- - [Alexey Akhunov](https://github.com/AlexeyAkhunov) and [Alex Sharov](https://github.com/AskAlexSharov) for bug reporting and testing.
- - [Andrea Lanfranchi](https://github.com/AndreaLanfranchi) for bug reporting and testing related to WSL2.
-
-New features:
-
- - Added `-p` option to `mdbx_stat` utility for printing page operations statistic.
- - Added explicit checking for and warning about using unfit github's archives.
- - Added fallback from [OFD locking](https://bit.ly/3yFRtYC) to legacy non-OFD POSIX file locks on an `EINVAL` error.
- - Added [Plan 9](https://en.wikipedia.org/wiki/9P_(protocol)) network file system to the whitelist for an ability to open a DB in exclusive mode.
- - Support for opening from WSL2 environment a DB hosted on Windows drive and mounted via [DrvFs](https://docs.microsoft.com/it-it/archive/blogs/wsl/wsl-file-system-support#drvfs) (i.e by Plan 9 noted above).
-
-Fixes:
-
- - Fixed minor "foo not used" warnings from modern C++ compilers when building the C++ part of the library.
- - Fixed confusing/messy errors when build library from unfit github's archives (todo4recovery://erased_by_github/libmdbx/issues/197).
- - Fixed `#​e​l​s​i​f` typo.
- - Fixed rare unexpected `MDBX_PROBLEM` error during altering data in huge transactions due to wrong spilling/oust of dirty pages (todo4recovery://erased_by_github/libmdbx/issues/195).
- - Re-Fixed WSL1/WSL2 detection with distinguishing (todo4recovery://erased_by_github/libmdbx/issues/97).
-
-
-## v0.10.0 at 2021-05-09
-
-Acknowledgements:
-
- - [Mahlon E. Smith](https://github.com/mahlonsmith) for [Ruby bindings](https://rubygems.org/gems/mdbx/).
- - [Alex Sharov](https://github.com/AskAlexSharov) for [mdbx-go](https://github.com/torquem-ch/mdbx-go), bug reporting and testing.
- - [Artem Vorotnikov](https://github.com/vorot93) for bug reporting and PR.
- - [Paolo Rebuffo](https://www.linkedin.com/in/paolo-rebuffo-8255766/), [Alexey Akhunov](https://github.com/AlexeyAkhunov) and Mark Grosberg for donations.
- - [Noel Kuntze](https://github.com/Thermi) for preliminary [Python bindings](https://github.com/Thermi/libmdbx/tree/python-bindings)
-
-New features:
-
- - Added `mdbx_env_set_option()` and `mdbx_env_get_option()` for controls
- various runtime options for an environment (announce of this feature was missed in a previous news).
- - Added `MDBX_DISABLE_PAGECHECKS` build option to disable some checks to reduce an overhead
- and detection probability of database corruption to a values closer to the LMDB.
- The `MDBX_DISABLE_PAGECHECKS=1` provides a performance boost of about 10% in CRUD scenarios,
- and conjointly with the `MDBX_ENV_CHECKPID=0` and `MDBX_TXN_CHECKOWNER=0` options can yield
- up to 30% more performance compared to LMDB.
- - Using float point (exponential quantized) representation for internal 16-bit values
- of grow step and shrink threshold when huge ones (todo4recovery://erased_by_github/libmdbx/issues/166).
- To minimize the impact on compatibility, only the odd values inside the upper half
- of the range (i.e. 32769..65533) are used for the new representation.
- - Added the `mdbx_drop` similar to LMDB command-line tool to purge or delete (sub)database(s).
- - [Ruby bindings](https://rubygems.org/gems/mdbx/) is available now by [Mahlon E. Smith](https://github.com/mahlonsmith).
- - Added `MDBX_ENABLE_MADVISE` build option which controls the use of POSIX `madvise()` hints and friends.
- - The internal node sizes were refined, resulting in a reduction in large/overflow pages in some use cases
- and a slight increase in limits for a keys size to ≈½ of page size.
- - Added to `mdbx_chk` output number of keys/items on pages.
- - Added explicit `install-strip` and `install-no-strip` targets to the `Makefile` (todo4recovery://erased_by_github/libmdbx/pull/180).
- - Major rework page splitting (af9b7b560505684249b76730997f9e00614b8113) for
- - An "auto-appending" feature upon insertion for both ascending and
- descending key sequences. As a result, the optimality of page filling
- increases significantly (more densely, less slackness) while
- inserting ordered sequences of keys,
- - A "splitting at middle" to make page tree more balanced on average.
- - Added `mdbx_get_sysraminfo()` to the API.
- - Added guessing a reasonable maximum DB size for the default upper limit of geometry (todo4recovery://erased_by_github/libmdbx/issues/183).
- - Major rework internal labeling of a dirty pages (958fd5b9479f52f2124ab7e83c6b18b04b0e7dda) for
- a "transparent spilling" feature with the gist to make a dirty pages
- be ready to spilling (writing to a disk) without further altering ones.
- Thus in the `MDBX_WRITEMAP` mode the OS kernel able to oust dirty pages
- to DB file without further penalty during transaction commit.
- As a result, page swapping and I/O could be significantly reduced during extra large transactions and/or lack of memory.
- - Minimized reading leaf-pages during dropping subDB(s) and nested trees.
- - Major rework a spilling of dirty pages to support [LRU](https://en.wikipedia.org/wiki/Cache_replacement_policies#Least_recently_used_(LRU))
- policy and prioritization for a large/overflow pages.
- - Statistics of page operations (split, merge, copy, spill, etc) now available through `mdbx_env_info_ex()`.
- - Auto-setup limit for length of dirty pages list (`MDBX_opt_txn_dp_limit` option).
- - Support `make options` to list available build options.
- - Support `make help` to list available make targets.
- - Silently `make`'s build by default.
- - Preliminary [Python bindings](https://github.com/Thermi/libmdbx/tree/python-bindings) is available now
- by [Noel Kuntze](https://github.com/Thermi) (todo4recovery://erased_by_github/libmdbx/issues/147).
-
-Backward compatibility break:
-
- - The `MDBX_AVOID_CRT` build option was renamed to `MDBX_WITHOUT_MSVC_CRT`.
- This option is only relevant when building for Windows.
- - The `mdbx_env_stat()` always, and `mdbx_env_stat_ex()` when called with the zeroed transaction parameter,
- now internally start temporary read transaction and thus may returns `MDBX_BAD_RSLOT` error.
- So, just never use deprecated `mdbx_env_stat()' and call `mdbx_env_stat_ex()` with transaction parameter.
- - The build option `MDBX_CONFIG_MANUAL_TLS_CALLBACK` was removed and now just a non-zero value of
- the `MDBX_MANUAL_MODULE_HANDLER` macro indicates the requirement to manually call `mdbx_module_handler()`
- when loading libraries and applications uses statically linked libmdbx on an obsolete Windows versions.
-
-Fixes:
-
- - Fixed performance regression due non-optimal C11 atomics usage (todo4recovery://erased_by_github/libmdbx/issues/160).
- - Fixed "reincarnation" of subDB after it deletion (todo4recovery://erased_by_github/libmdbx/issues/168).
- - Fixed (disallowing) implicit subDB deletion via operations on `@MAIN`'s DBI-handle.
- - Fixed a crash of `mdbx_env_info_ex()` in case of a call for a non-open environment (todo4recovery://erased_by_github/libmdbx/issues/171).
- - Fixed the selecting/adjustment values inside `mdbx_env_set_geometry()` for implicit out-of-range cases (todo4recovery://erased_by_github/libmdbx/issues/170).
- - Fixed `mdbx_env_set_option()` for set initial and limit size of dirty page list ((todo4recovery://erased_by_github/libmdbx/issues/179).
- - Fixed an unreasonably huge default upper limit for DB geometry (todo4recovery://erased_by_github/libmdbx/issues/183).
- - Fixed `constexpr` specifier for the `slice::invalid()`.
- - Fixed (no)readahead auto-handling (todo4recovery://erased_by_github/libmdbx/issues/164).
- - Fixed non-alloy build for Windows.
- - Switched to using Heap-functions instead of LocalAlloc/LocalFree on Windows.
- - Fixed `mdbx_env_stat_ex()` to returning statistics of the whole environment instead of MainDB only (todo4recovery://erased_by_github/libmdbx/issues/190).
- - Fixed building by GCC 4.8.5 (added workaround for a preprocessor's bug).
- - Fixed building C++ part for iOS <= 13.0 (unavailability of `std::filesystem::path`).
- - Fixed building for Windows target versions prior to Windows Vista (`WIN32_WINNT < 0x0600`).
- - Fixed building by MinGW for Windows (todo4recovery://erased_by_github/libmdbx/issues/155).
-
-
-## v0.9.3 at 2021-02-02
-
-Acknowledgements:
-
- - [Mahlon E. Smith](http://www.martini.nu/) for [FreeBSD port of libmdbx](https://svnweb.freebsd.org/ports/head/databases/mdbx/).
- - [장세연](http://www.castis.com) for bug fixing and PR.
- - [Clément Renault](https://github.com/Kerollmops/heed) for [Heed](https://github.com/Kerollmops/heed) fully typed Rust wrapper.
- - [Alex Sharov](https://github.com/AskAlexSharov) for bug reporting.
- - [Noel Kuntze](https://github.com/Thermi) for bug reporting.
-
-Removed options and features:
-
- - Drop `MDBX_HUGE_TRANSACTIONS` build-option (now no longer required).
-
-New features:
-
- - Package for FreeBSD is available now by Mahlon E. Smith.
- - New API functions to get/set various options (todo4recovery://erased_by_github/libmdbx/issues/128):
- - the maximum number of named databases for the environment;
- - the maximum number of threads/reader slots;
- - threshold (since the last unsteady commit) to force flush the data buffers to disk;
- - relative period (since the last unsteady commit) to force flush the data buffers to disk;
- - limit to grow a list of reclaimed/recycled page's numbers for finding a sequence of contiguous pages for large data items;
- - limit to grow a cache of dirty pages for reuse in the current transaction;
- - limit of a pre-allocated memory items for dirty pages;
- - limit of dirty pages for a write transaction;
- - initial allocation size for dirty pages list of a write transaction;
- - maximal part of the dirty pages may be spilled when necessary;
- - minimal part of the dirty pages should be spilled when necessary;
- - how much of the parent transaction dirty pages will be spilled while start each child transaction;
- - Unlimited/Dynamic size of retired and dirty page lists (todo4recovery://erased_by_github/libmdbx/issues/123).
- - Added `-p` option (purge subDB before loading) to `mdbx_load` tool.
- - Reworked spilling of large transaction and committing of nested transactions:
- - page spilling code reworked to avoid the flaws and bugs inherited from LMDB;
- - limit for number of dirty pages now is controllable at runtime;
- - a spilled pages, including overflow/large pages, now can be reused and refunded/compactified in nested transactions;
- - more effective refunding/compactification especially for the loosed page cache.
- - Added `MDBX_ENABLE_REFUND` and `MDBX_PNL_ASCENDING` internal/advanced build options.
- - Added `mdbx_default_pagesize()` function.
- - Better support architectures with a weak/relaxed memory consistency model (ARM, AARCH64, PPC, MIPS, RISC-V, etc) by means [C11 atomics](https://en.cppreference.com/w/c/atomic).
- - Speed up page number lists and dirty page lists (todo4recovery://erased_by_github/libmdbx/issues/132).
- - Added `LIBMDBX_NO_EXPORTS_LEGACY_API` build option.
-
-Fixes:
-
- - Fixed missing cleanup (null assigned) in the C++ commit/abort (todo4recovery://erased_by_github/libmdbx/pull/143).
- - Fixed `mdbx_realloc()` for case of nullptr and `MDBX_WITHOUT_MSVC_CRT=ON` for Windows.
- - Fixed the possibility to use invalid and renewed (closed & re-opened, dropped & re-created) DBI-handles (todo4recovery://erased_by_github/libmdbx/issues/146).
- - Fixed 4-byte aligned access to 64-bit integers, including access to the `bootid` meta-page's field (todo4recovery://erased_by_github/libmdbx/issues/153).
- - Fixed minor/potential memory leak during page flushing and unspilling.
- - Fixed handling states of cursors's and subDBs's for nested transactions.
- - Fixed page leak in extra rare case the list of retired pages changed during update GC on transaction commit.
- - Fixed assertions to avoid false-positive UB detection by CLANG/LLVM (todo4recovery://erased_by_github/libmdbx/issues/153).
- - Fixed `MDBX_TXN_FULL` and regressive `MDBX_KEYEXIST` during large transaction commit with `MDBX_LIFORECLAIM` (todo4recovery://erased_by_github/libmdbx/issues/123).
- - Fixed auto-recovery (`weak->steady` with the same boot-id) when Database size at last weak checkpoint is large than at last steady checkpoint.
- - Fixed operation on systems with unusual small/large page size, including PowerPC (todo4recovery://erased_by_github/libmdbx/issues/157).
-
-
-## v0.9.2 at 2020-11-27
-
-Acknowledgements:
-
- - Jens Alfke (Mobile Architect at [Couchbase](https://www.couchbase.com/)) for [NimDBX](https://github.com/snej/nimdbx).
- - Clément Renault (CTO at [MeiliSearch](https://www.meilisearch.com/)) for [mdbx-rs](https://github.com/Kerollmops/mdbx-rs).
- - Alex Sharov (Go-Lang Teach Lead at [TurboGeth/Ethereum](https://ethereum.org/)) for an extreme test cases and bug reporting.
- - George Hazan (CTO at [Miranda NG](https://www.miranda-ng.org/)) for bug reporting.
- - [Positive Technologies](https://www.ptsecurity.com/) for funding and [The Standoff](https://standoff365.com/).
-
-Added features:
-
- - Provided package for [buildroot](https://buildroot.org/).
- - Binding for Nim is [available](https://github.com/snej/nimdbx) now by Jens Alfke.
- - Added `mdbx_env_delete()` for deletion an environment files in a proper and multiprocess-safe way.
- - Added `mdbx_txn_commit_ex()` with collecting latency information.
- - Fast completion pure nested transactions.
- - Added `LIBMDBX_INLINE_API` macro and inline versions of some API functions.
- - Added `mdbx_cursor_copy()` function.
- - Extended tests for checking cursor tracking.
- - Added `MDBX_SET_LOWERBOUND` operation for `mdbx_cursor_get()`.
-
-Fixes:
-
- - Fixed missing installation of `mdbx.h++`.
- - Fixed use of obsolete `__noreturn`.
- - Fixed use of `yield` instruction on ARM if unsupported.
- - Added pthread workaround for buggy toolchain/cmake/buildroot.
- - Fixed use of `pthread_yield()` for non-GLIBC.
- - Fixed use of `RegGetValueA()` on Windows 2000/XP.
- - Fixed use of `GetTickCount64()` on Windows 2000/XP.
- - Fixed opening DB on a network shares (in the exclusive mode).
- - Fixed copy&paste typos.
- - Fixed minor false-positive GCC warning.
- - Added workaround for broken `DEFINE_ENUM_FLAG_OPERATORS` from Windows SDK.
- - Fixed cursor state after multimap/dupsort repeated deletes (todo4recovery://erased_by_github/libmdbx/issues/121).
- - Added `SIGPIPE` suppression for internal thread during `mdbx_env_copy()`.
- - Fixed extra-rare `MDBX_KEY_EXIST` error during `mdbx_commit()` (todo4recovery://erased_by_github/libmdbx/issues/131).
- - Fixed spilled pages checking (todo4recovery://erased_by_github/libmdbx/issues/126).
- - Fixed `mdbx_load` for 'plain text' and without `-s name` cases (todo4recovery://erased_by_github/libmdbx/issues/136).
- - Fixed save/restore/commit of cursors for nested transactions.
- - Fixed cursors state in rare/special cases (move next beyond end-of-data, after deletion and so on).
- - Added workaround for MSVC 19.28 (Visual Studio 16.8) (but may still hang during compilation).
- - Fixed paranoidal Clang C++ UB for bitwise operations with flags defined by enums.
- - Fixed large pages checking (for compatibility and to avoid false-positive errors from `mdbx_chk`).
- - Added workaround for Wine (https://github.com/miranda-ng/miranda-ng/issues/1209).
- - Fixed `ERROR_NOT_SUPPORTED` while opening DB by UNC pathnames (https://github.com/miranda-ng/miranda-ng/issues/2627).
- - Added handling `EXCEPTION_POSSIBLE_DEADLOCK` condition for Windows.
-
-
-## v0.9.1 2020-09-30
-
-Added features:
-
- - Preliminary C++ API with support for C++17 polymorphic allocators.
- - [Online C++ API reference](https://erased_by_github_and_to_be_restored.todo/libmdbx/) by Doxygen.
- - Quick reference for Insert/Update/Delete operations.
- - Explicit `MDBX_SYNC_DURABLE` to sync modes for API clarity.
- - Explicit `MDBX_ALLDUPS` and `MDBX_UPSERT` for API clarity.
- - Support for read transactions preparation (`MDBX_TXN_RDONLY_PREPARE` flag).
- - Support for cursor preparation/(pre)allocation and reusing (`mdbx_cursor_create()` and `mdbx_cursor_bind()` functions).
- - Support for checking database using specified meta-page (see `mdbx_chk -h`).
- - Support for turn to the specific meta-page after checking (see `mdbx_chk -h`).
- - Support for explicit reader threads (de)registration.
- - The `mdbx_txn_break()` function to explicitly mark a transaction as broken.
- - Improved handling of corrupted databases by `mdbx_chk` utility and `mdbx_walk_tree()` function.
- - Improved DB corruption detection by checking parent-page-txnid.
- - Improved opening large DB (> 4Gb) from 32-bit code.
- - Provided `pure-function` and `const-function` attributes to C API.
- - Support for user-settable context for transactions & cursors.
- - Revised API and documentation related to Handle-Slow-Readers callback feature.
-
-Deprecated functions and flags:
-
- - For clarity and API simplification the `MDBX_MAPASYNC` flag is deprecated.
- Just use `MDBX_SAFE_NOSYNC` or `MDBX_UTTERLY_NOSYNC` instead of it.
- - `MDBX_oom_func`, `mdbx_env_set_oomfunc()` and `mdbx_env_get_oomfunc()`
- replaced with `MDBX_hsr_func`, `mdbx_env_get_hsr` and `mdbx_env_get_hsr()`.
-
-Fixes:
-
- - Fix `mdbx_strerror()` for `MDBX_BUSY` error (no error description is returned).
- - Fix update internal meta-geo information in read-only mode (`EACCESS` or `EBADFD` error).
- - Fix `mdbx_page_get()` null-defer when DB corrupted (crash by `SIGSEGV`).
- - Fix `mdbx_env_open()` for re-opening after non-fatal errors (`mdbx_chk` unexpected failures).
- - Workaround for MSVC 19.27 `static_assert()` bug.
- - Doxygen descriptions and refinement.
- - Update Valgrind's suppressions.
- - Workaround to avoid infinite loop of 'nested' testcase on MIPS under QEMU.
- - Fix a lot of typos & spelling (Thanks to Josh Soref for PR).
- - Fix `getopt()` messages for Windows (Thanks to Andrey Sporaw for reporting).
- - Fix MSVC compiler version requirements (Thanks to Andrey Sporaw for reporting).
- - Workarounds for QEMU's bugs to run tests for cross-builded library under QEMU.
- - Now C++ compiler optional for building by CMake.
-
-
-## v0.9.0 2020-07-31 (not a release, but API changes)
-
-Added features:
-
- - [Online C API reference](https://erased_by_github_and_to_be_restored.todo/libmdbx/) by Doxygen.
- - Separated enums for environment, sub-databases, transactions, copying and data-update flags.
-
-Deprecated functions and flags:
-
- - Usage of custom comparators and the `mdbx_dbi_open_ex()` are deprecated, since such databases couldn't be checked by the `mdbx_chk` utility.
- Please use the value-to-key functions to provide keys that are compatible with the built-in libmdbx comparators.
-
-
-## v0.8.2 2020-07-06
-- Added support multi-opening the same DB in a process with SysV locking (BSD).
-- Fixed warnings & minors for LCC compiler (E2K).
-- Enabled to simultaneously open the same database from processes with and without the `MDBX_WRITEMAP` option.
-- Added key-to-value, `mdbx_get_keycmp()` and `mdbx_get_datacmp()` functions (helpful to avoid using custom comparators).
-- Added `ENABLE_UBSAN` CMake option to enabling the UndefinedBehaviorSanitizer from GCC/CLANG.
-- Workaround for [CLANG bug](https://bugs.llvm.org/show_bug.cgi?id=43275).
-- Returning `MDBX_CORRUPTED` in case all meta-pages are weak and no other error.
-- Refined mode bits while auto-creating LCK-file.
-- Avoids unnecessary database file re-mapping in case geometry changed by another process(es).
- From the user's point of view, the `MDBX_UNABLE_EXTEND_MAPSIZE` error will now be returned less frequently and only when using the DB in the current process really requires it to be reopened.
-- Remapping on-the-fly and of the database file was implemented.
- Now remapping with a change of address is performed automatically if there are no dependent readers in the current process.
-
-
-## v0.8.1 2020-06-12
-- Minor change versioning. The last number in the version now means the number of commits since last release/tag.
-- Provide ChangeLog file.
-- Fix for using libmdbx as a C-only sub-project with CMake.
-- Fix `mdbx_env_set_geometry()` for case it is called from an opened environment outside of a write transaction.
-- Add support for huge transactions and `MDBX_HUGE_TRANSACTIONS` build-option (default `OFF`).
-- Refine LTO (link time optimization) for clang.
-- Force enabling exceptions handling for MSVC (`/EHsc` option).
-
-
-## v0.8.0 2020-06-05
-- Support for Android/Bionic.
-- Support for iOS.
-- Auto-handling `MDBX_NOSUBDIR` while opening for any existing database.
-- Engage github-actions to make release-assets.
-- Clarify API description.
-- Extended keygen-cases in stochastic test.
-- Fix fetching of first/lower key from LEAF2-page during page merge.
-- Fix missing comma in array of error messages.
-- Fix div-by-zero while copy-with-compaction for non-resizable environments.
-- Fixes & enhancements for custom-comparators.
-- Fix `MDBX_WITHOUT_MSVC_CRT` option and missing `ntdll.def`.
-- Fix `mdbx_env_close()` to work correctly called concurrently from several threads.
-- Fix null-deref in an ASAN-enabled builds while opening the environment with error and/or read-only.
-- Fix AddressSanitizer errors after closing the environment.
-- Fix/workaround to avoid GCC 10.x pedantic warnings.
-- Fix using `ENODATA` for FreeBSD.
-- Avoid invalidation of DBI-handle(s) when it just closes.
-- Avoid using `pwritev()` for single-writes (up to 10% speedup for some kernels & scenarios).
-- Avoiding `MDBX_UTTERLY_NOSYNC` as result of flags merge.
-- Add `mdbx_dbi_dupsort_depthmask()` function.
-- Add `MDBX_CP_FORCE_RESIZEABLE` option.
-- Add deprecated `MDBX_MAP_RESIZED` for compatibility.
-- Add `MDBX_BUILD_TOOLS` option (default `ON`).
-- Refine `mdbx_dbi_open_ex()` to safe concurrently opening the same handle from different threads.
-- Truncate clk-file during environment closing. So a zero-length lck-file indicates that the environment was closed properly.
-- Refine `mdbx_update_gc()` for huge transactions with small sizes of database page.
-- Extends dump/load to support all MDBX attributes.
-- Avoid upsertion the same key-value data, fix related assertions.
-- Rework min/max length checking for keys & values.
-- Checking the order of keys on all pages during checking.
-- Support `CFLAGS_EXTRA` make-option for convenience.
-- Preserve the last txnid while copying with compactification.
-- Auto-reset running transaction in mdbx_txn_renew().
-- Automatically abort errored transaction in mdbx_txn_commit().
-- Auto-choose page size for large databases.
-- Rearrange source files, rework build, options-support by CMake.
-- Crutch for WSL1 (Windows subsystem for Linux).
-- Refine install/uninstall targets.
-- Support for Valgrind 3.14 and later.
-- Add check-analyzer check-ubsan check-asan check-leak targets to Makefile.
-- Minor fix/workaround to avoid UBSAN traps for `memcpy(ptr, NULL, 0)`.
-- Avoid some GCC-analyzer false-positive warnings.
-
-
-## v0.7.0 2020-03-18
-- Workarounds for Wine (Windows compatibility layer for Linux).
-- `MDBX_MAP_RESIZED` renamed to `MDBX_UNABLE_EXTEND_MAPSIZE`.
-- Clarify API description, fix typos.
-- Speedup runtime checks in debug/checked builds.
-- Added checking for read/write transactions overlapping for the same thread, added `MDBX_TXN_OVERLAPPING` error and `MDBX_DBG_LEGACY_OVERLAP` option.
-- Added `mdbx_key_from_jsonInteger()`, `mdbx_key_from_double()`, `mdbx_key_from_float()`, `mdbx_key_from_int64()` and `mdbx_key_from_int32()` functions. See `mdbx.h` for description.
-- Fix compatibility (use zero for invalid DBI).
-- Refine/clarify error messages.
-- Avoids extra error messages "bad txn" from mdbx_chk when DB is corrupted.
-
-
-## v0.6.0 2020-01-21
-- Fix `mdbx_load` utility for custom comparators.
-- Fix checks related to `MDBX_APPEND` flag inside `mdbx_cursor_put()`.
-- Refine/fix dbi_bind() internals.
-- Refine/fix handling `STATUS_CONFLICTING_ADDRESSES`.
-- Rework `MDBX_DBG_DUMP` option to avoid disk I/O performance degradation.
-- Add built-in help to test tool.
-- Fix `mdbx_env_set_geometry()` for large page size.
-- Fix env_set_geometry() for large pagesize.
-- Clarify API description & comments, fix typos.
-
-
-## v0.5.0 2019-12-31
-- Fix returning MDBX_RESULT_TRUE from page_alloc().
-- Fix false-positive ASAN issue.
-- Fix assertion for `MDBX_NOTLS` option.
-- Rework `MADV_DONTNEED` threshold.
-- Fix `mdbx_chk` utility for don't checking some numbers if walking on the B-tree was disabled.
-- Use page's mp_txnid for basic integrity checking.
-- Add `MDBX_FORCE_ASSERTIONS` built-time option.
-- Rework `MDBX_DBG_DUMP` to avoid performance degradation.
-- Rename `MDBX_NOSYNC` to `MDBX_SAFE_NOSYNC` for clarity.
-- Interpret `ERROR_ACCESS_DENIED` from `OpenProcess()` as 'process exists'.
-- Avoid using `FILE_FLAG_NO_BUFFERING` for compatibility with small database pages.
-- Added install section for CMake.
-
-
-## v0.4.0 2019-12-02
-- Support for Mac OSX, FreeBSD, NetBSD, OpenBSD, DragonFly BSD, OpenSolaris, OpenIndiana (AIX and HP-UX pending).
-- Use bootid for decisions of rollback.
-- Counting retired pages and extended transaction info.
-- Add `MDBX_ACCEDE` flag for database opening.
-- Using OFD-locks and tracking for in-process multi-opening.
-- Hot backup into pipe.
-- Support for cmake & amalgamated sources.
-- Fastest internal sort implementation.
-- New internal dirty-list implementation with lazy sorting.
-- Support for lazy-sync-to-disk with polling.
-- Extended key length.
-- Last update transaction number for each sub-database.
-- Automatic read ahead enabling/disabling.
-- More auto-compactification.
-- Using -fsanitize=undefined and -Wpedantic options.
-- Rework page merging.
-- Nested transactions.
-- API description.
-- Checking for non-local filesystems to avoid DB corruption.
diff --git a/3rdparty/libmdbx/GNUmakefile b/3rdparty/libmdbx/GNUmakefile
deleted file mode 100644
index d6b9fc7a7cf14a26ce72e8c1f8f54d404c603173..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/GNUmakefile
+++ /dev/null
@@ -1,375 +0,0 @@
-# This makefile is for GNU Make 3.80 or above, and nowadays provided
-# just for compatibility and preservation of traditions.
-#
-# Please use CMake in case of any difficulties or
-# problems with this old-school's magic.
-#
-################################################################################
-#
-# Basic internal definitios. For a customizable variables and options see below.
-#
-$(info // The GNU Make $(MAKE_VERSION))
-SHELL := $(shell env bash -c 'echo $$BASH')
-MAKE_VERx3 := $(shell printf "%3s%3s%3s" $(subst ., ,$(MAKE_VERSION)))
-make_lt_3_81 := $(shell expr "$(MAKE_VERx3)" "<" " 3 81")
-ifneq ($(make_lt_3_81),0)
-$(error Please use GNU Make 3.81 or above)
-endif
-make_ge_4_1 := $(shell expr "$(MAKE_VERx3)" ">=" " 4 1")
-SRC_PROBE_C := $(shell [ -f mdbx.c ] && echo mdbx.c || echo src/osal.c)
-SRC_PROBE_CXX := $(shell [ -f mdbx.c++ ] && echo mdbx.c++ || echo src/mdbx.c++)
-UNAME := $(shell uname -s 2>/dev/null || echo Unknown)
-
-define cxx_filesystem_probe
- int main(int argc, const char*argv[]) {
- mdbx::filesystem::path probe(argv[0]);
- if (argc != 1) throw mdbx::filesystem::filesystem_error(std::string("fake"), std::error_code());
- return mdbx::filesystem::is_directory(probe.relative_path());
- }
-endef
-#
-################################################################################
-#
-# Use `make options` to list the available libmdbx build options.
-#
-# Note that the defaults should already be correct for most platforms;
-# you should not need to change any of these. Read their descriptions
-# in README and source code (see src/options.h) if you do.
-#
-
-# install sandbox
-DESTDIR ?=
-INSTALL ?= install
-# install prefixes (inside sandbox)
-prefix ?= /usr/local
-mandir ?= $(prefix)/man
-# lib/bin suffix for multiarch/biarch, e.g. '.x86_64'
-suffix ?=
-
-# toolchain
-CC ?= gcc
-CXX ?= g++
-CFLAGS_EXTRA ?=
-LD ?= ld
-
-# build options
-MDBX_BUILD_OPTIONS ?=-DNDEBUG=1
-MDBX_BUILD_TIMESTAMP ?=$(shell date +%Y-%m-%dT%H:%M:%S%z)
-
-# probe and compose common compiler flags with variable expansion trick (seems this work two times per session for GNU Make 3.81)
-CFLAGS ?= $(strip $(eval CFLAGS := -std=gnu11 -O2 -g -Wall -Werror -Wextra -Wpedantic -ffunction-sections -fPIC -fvisibility=hidden -pthread -Wno-error=attributes $$(shell for opt in -fno-semantic-interposition -Wno-unused-command-line-argument -Wno-tautological-compare; do [ -z "$$$$($(CC) '-DMDBX_BUILD_FLAGS="probe"' $$$${opt} -c $(SRC_PROBE_C) -o /dev/null >/dev/null 2>&1 || echo failed)" ] && echo "$$$${opt} "; done)$(CFLAGS_EXTRA))$(CFLAGS))
-
-# choosing C++ standard with variable expansion trick (seems this work two times per session for GNU Make 3.81)
-CXXSTD ?= $(eval CXXSTD := $$(shell for std in gnu++23 c++23 gnu++2b c++2b gnu++20 c++20 gnu++2a c++2a gnu++17 c++17 gnu++1z c++1z gnu++14 c++14 gnu++1y c++1y gnu+11 c++11 gnu++0x c++0x; do $(CXX) -std=$$$${std} -c $(SRC_PROBE_CXX) -o /dev/null 2>probe4std-$$$${std}.err >/dev/null && echo "-std=$$$${std}" && exit; done))$(CXXSTD)
-CXXFLAGS ?= $(strip $(CXXSTD) $(filter-out -std=gnu11,$(CFLAGS)))
-
-# libraries and options for linking
-EXE_LDFLAGS ?= -pthread
-ifneq ($(make_ge_4_1),1)
-# don't use variable expansion trick as workaround for bugs of GNU Make before 4.1
-LIBS ?= $(shell $(uname2libs))
-LDFLAGS ?= $(shell $(uname2ldflags))
-LIB_STDCXXFS ?= $(shell echo '$(cxx_filesystem_probe)' | cat mdbx.h++ - | sed $$'1s/\xef\xbb\xbf//' | $(CXX) -x c++ $(CXXFLAGS) -Wno-error - -Wl,--allow-multiple-definition -lstdc++fs $(LIBS) $(LDFLAGS) $(EXE_LDFLAGS) -o /dev/null 2>probe4lstdfs.err >/dev/null && echo '-Wl,--allow-multiple-definition -lstdc++fs')
-else
-# using variable expansion trick to avoid repeaded probes
-LIBS ?= $(eval LIBS := $$(shell $$(uname2libs)))$(LIBS)
-LDFLAGS ?= $(eval LDFLAGS := $$(shell $$(uname2ldflags)))$(LDFLAGS)
-LIB_STDCXXFS ?= $(eval LIB_STDCXXFS := $$(shell echo '$$(cxx_filesystem_probe)' | cat mdbx.h++ - | sed $$$$'1s/\xef\xbb\xbf//' | $(CXX) -x c++ $(CXXFLAGS) -Wno-error - -Wl,--allow-multiple-definition -lstdc++fs $(LIBS) $(LDFLAGS) $(EXE_LDFLAGS) -o /dev/null 2>probe4lstdfs.err >/dev/null && echo '-Wl,--allow-multiple-definition -lstdc++fs'))$(LIB_STDCXXFS)
-endif
-
-################################################################################
-
-define uname2sosuffix
- case "$(UNAME)" in
- Darwin*|Mach*) echo dylib;;
- CYGWIN*|MINGW*|MSYS*|Windows*) echo dll;;
- *) echo so;;
- esac
-endef
-
-define uname2ldflags
- case "$(UNAME)" in
- CYGWIN*|MINGW*|MSYS*|Windows*)
- echo '-Wl,--gc-sections,-O1';
- ;;
- *)
- $(LD) --help 2>/dev/null | grep -q -- --gc-sections && echo '-Wl,--gc-sections,-z,relro,-O1';
- $(LD) --help 2>/dev/null | grep -q -- -dead_strip && echo '-Wl,-dead_strip';
- ;;
- esac
-endef
-
-# TIP: try add the'-Wl, --no-as-needed,-lrt' for ability to built with modern glibc, but then use with the old.
-define uname2libs
- case "$(UNAME)" in
- CYGWIN*|MINGW*|MSYS*|Windows*)
- echo '-lm -lntdll -lwinmm';
- ;;
- *SunOS*|*Solaris*)
- echo '-lm -lkstat -lrt';
- ;;
- *Darwin*|OpenBSD*)
- echo '-lm';
- ;;
- *)
- echo '-lm -lrt';
- ;;
- esac
-endef
-
-SO_SUFFIX := $(shell $(uname2sosuffix))
-HEADERS := mdbx.h mdbx.h++
-LIBRARIES := libmdbx.a libmdbx.$(SO_SUFFIX)
-TOOLS := mdbx_stat mdbx_copy mdbx_dump mdbx_load mdbx_chk mdbx_drop
-MANPAGES := mdbx_stat.1 mdbx_copy.1 mdbx_dump.1 mdbx_load.1 mdbx_chk.1 mdbx_drop.1
-TIP := // TIP:
-
-.PHONY: all help options lib libs tools clean install uninstall check_buildflags_tag tools-static
-.PHONY: install-strip install-no-strip strip libmdbx mdbx show-options lib-static lib-shared
-
-ifeq ("$(origin V)", "command line")
- MDBX_BUILD_VERBOSE := $(V)
-endif
-ifndef MDBX_BUILD_VERBOSE
- MDBX_BUILD_VERBOSE := 0
-endif
-
-ifeq ($(MDBX_BUILD_VERBOSE),1)
- QUIET :=
- HUSH :=
- $(info $(TIP) Use `make V=0` for quiet.)
-else
- QUIET := @
- HUSH := >/dev/null
- $(info $(TIP) Use `make V=1` for verbose.)
-endif
-
-all: show-options $(LIBRARIES) $(TOOLS)
-
-help:
- @echo " make all - build libraries and tools"
- @echo " make help - print this help"
- @echo " make options - list build options"
- @echo " make lib - build libraries, also lib-static and lib-shared"
- @echo " make tools - build the tools"
- @echo " make tools-static - build the tools with statically linking with system libraries and compiler runtime"
- @echo " make clean "
- @echo " make install "
- @echo " make uninstall "
- @echo ""
- @echo " make strip - strip debug symbols from binaries"
- @echo " make install-no-strip - install explicitly without strip"
- @echo " make install-strip - install explicitly with strip"
- @echo ""
- @echo " make bench - run ioarena-benchmark"
- @echo " make bench-couple - run ioarena-benchmark for mdbx and lmdb"
- @echo " make bench-triplet - run ioarena-benchmark for mdbx, lmdb, sqlite3"
- @echo " make bench-quartet - run ioarena-benchmark for mdbx, lmdb, rocksdb, wiredtiger"
- @echo " make bench-clean - remove temp database(s) after benchmark"
-
-show-options:
- @echo " MDBX_BUILD_OPTIONS = $(MDBX_BUILD_OPTIONS)"
- @echo " MDBX_BUILD_TIMESTAMP = $(MDBX_BUILD_TIMESTAMP)"
- @echo '$(TIP) Use `make options` to listing available build options.'
- @echo " CC =`which $(CC)` | `$(CC) --version | head -1`"
- @echo " CFLAGS =$(CFLAGS)"
- @echo " CXXFLAGS =$(CXXFLAGS)"
- @echo " LDFLAGS =$(LDFLAGS) $(LIB_STDCXXFS) $(LIBS) $(EXE_LDFLAGS)"
- @echo '$(TIP) Use `make help` to listing available targets.'
-
-options:
- @echo " INSTALL =$(INSTALL)"
- @echo " DESTDIR =$(DESTDIR)"
- @echo " prefix =$(prefix)"
- @echo " mandir =$(mandir)"
- @echo " suffix =$(suffix)"
- @echo ""
- @echo " CC =$(CC)"
- @echo " CFLAGS_EXTRA =$(CFLAGS_EXTRA)"
- @echo " CFLAGS =$(CFLAGS)"
- @echo " CXX =$(CXX)"
- @echo " CXXSTD =$(CXXSTD)"
- @echo " CXXFLAGS =$(CXXFLAGS)"
- @echo ""
- @echo " LD =$(LD)"
- @echo " LDFLAGS =$(LDFLAGS)"
- @echo " EXE_LDFLAGS =$(EXE_LDFLAGS)"
- @echo " LIBS =$(LIBS)"
- @echo ""
- @echo " MDBX_BUILD_OPTIONS = $(MDBX_BUILD_OPTIONS)"
- @echo " MDBX_BUILD_TIMESTAMP = $(MDBX_BUILD_TIMESTAMP)"
- @echo ""
- @echo "## Assortment items for MDBX_BUILD_OPTIONS:"
- @echo "## Note that the defaults should already be correct for most platforms;"
- @echo "## you should not need to change any of these. Read their descriptions"
- @echo "## in README and source code (see mdbx.c) if you do."
- @grep -h '#ifndef MDBX_' mdbx.c | grep -v BUILD | uniq | sed 's/#ifndef / /'
-
-lib libs libmdbx mdbx: libmdbx.a libmdbx.$(SO_SUFFIX)
-
-tools: $(TOOLS)
-tools-static: $(addsuffix .static,$(TOOLS)) $(addsuffix .static-lto,$(TOOLS))
-
-strip: all
- @echo ' STRIP libmdbx.$(SO_SUFFIX) $(TOOLS)'
- $(TRACE )strip libmdbx.$(SO_SUFFIX) $(TOOLS)
-
-clean:
- @echo ' REMOVE ...'
- $(QUIET)rm -rf $(TOOLS) mdbx_test @* *.[ao] *.[ls]o *.$(SO_SUFFIX) *.dSYM *~ tmp.db/* \
- *.gcov *.log *.err src/*.o test/*.o mdbx_example dist \
- config.h src/config.h src/version.c *.tar* buildflags.tag \
- mdbx_*.static mdbx_*.static-lto
-
-MDBX_BUILD_FLAGS =$(strip $(MDBX_BUILD_OPTIONS) $(CXXSTD) $(CFLAGS) $(LDFLAGS) $(LIBS))
-check_buildflags_tag:
- $(QUIET)if [ "$(MDBX_BUILD_FLAGS)" != "$$(cat buildflags.tag 2>&1)" ]; then \
- echo -n " CLEAN for build with specified flags..." && \
- $(MAKE) IOARENA=false CXXSTD= -s clean >/dev/null && echo " Ok" && \
- echo '$(MDBX_BUILD_FLAGS)' > buildflags.tag; \
- fi
-
-buildflags.tag: check_buildflags_tag
-
-lib-static libmdbx.a: mdbx-static.o mdbx++-static.o
- @echo ' AR $@'
- $(QUIET)$(AR) rcs $@ $? $(HUSH)
-
-lib-shared libmdbx.$(SO_SUFFIX): mdbx-dylib.o mdbx++-dylib.o
- @echo ' LD $@'
- $(QUIET)$(CXX) $(CXXFLAGS) $^ -pthread -shared $(LDFLAGS) $(LIB_STDCXXFS) $(LIBS) -o $@
-
-
-################################################################################
-# Amalgamated source code, i.e. distributed after `make dist`
-MAN_SRCDIR := man1/
-
-config.h: buildflags.tag mdbx.c $(lastword $(MAKEFILE_LIST))
- @echo ' MAKE $@'
- $(QUIET)(echo '#define MDBX_BUILD_TIMESTAMP "$(MDBX_BUILD_TIMESTAMP)"' \
- && echo "#define MDBX_BUILD_FLAGS \"$$(cat buildflags.tag)\"" \
- && echo '#define MDBX_BUILD_COMPILER "$(shell (LC_ALL=C $(CC) --version || echo 'Please use GCC or CLANG compatible compiler') | head -1)"' \
- && echo '#define MDBX_BUILD_TARGET "$(shell set -o pipefail; (LC_ALL=C $(CC) -v 2>&1 | grep -i '^Target:' | cut -d ' ' -f 2- || (LC_ALL=C $(CC) --version | grep -qi e2k && echo E2K) || echo 'Please use GCC or CLANG compatible compiler') | head -1)"' \
- ) >$@
-
-mdbx-dylib.o: config.h mdbx.c mdbx.h $(lastword $(MAKEFILE_LIST))
- @echo ' CC $@'
- $(QUIET)$(CC) $(CFLAGS) $(MDBX_BUILD_OPTIONS) '-DMDBX_CONFIG_H="config.h"' -DLIBMDBX_EXPORTS=1 -c mdbx.c -o $@
-
-mdbx-static.o: config.h mdbx.c mdbx.h $(lastword $(MAKEFILE_LIST))
- @echo ' CC $@'
- $(QUIET)$(CC) $(CFLAGS) $(MDBX_BUILD_OPTIONS) '-DMDBX_CONFIG_H="config.h"' -ULIBMDBX_EXPORTS -c mdbx.c -o $@
-
-mdbx++-dylib.o: config.h mdbx.c++ mdbx.h mdbx.h++ $(lastword $(MAKEFILE_LIST))
- @echo ' CC $@'
- $(QUIET)$(CXX) $(CXXFLAGS) $(MDBX_BUILD_OPTIONS) '-DMDBX_CONFIG_H="config.h"' -DLIBMDBX_EXPORTS=1 -c mdbx.c++ -o $@
-
-mdbx++-static.o: config.h mdbx.c++ mdbx.h mdbx.h++ $(lastword $(MAKEFILE_LIST))
- @echo ' CC $@'
- $(QUIET)$(CXX) $(CXXFLAGS) $(MDBX_BUILD_OPTIONS) '-DMDBX_CONFIG_H="config.h"' -ULIBMDBX_EXPORTS -c mdbx.c++ -o $@
-
-mdbx_%: mdbx_%.c mdbx-static.o
- @echo ' CC+LD $@'
- $(QUIET)$(CC) $(CFLAGS) $(MDBX_BUILD_OPTIONS) '-DMDBX_CONFIG_H="config.h"' $^ $(EXE_LDFLAGS) $(LIBS) -o $@
-
-mdbx_%.static: mdbx_%.c mdbx-static.o
- @echo ' CC+LD $@'
- $(QUIET)$(CC) $(CFLAGS) $(MDBX_BUILD_OPTIONS) '-DMDBX_CONFIG_H="config.h"' $^ $(EXE_LDFLAGS) -static -Wl,--strip-all -o $@
-
-mdbx_%.static-lto: mdbx_%.c config.h mdbx.c mdbx.h
- @echo ' CC+LD $@'
- $(QUIET)$(CC) $(CFLAGS) -Os -flto $(MDBX_BUILD_OPTIONS) '-DLIBMDBX_API=' '-DMDBX_CONFIG_H="config.h"' \
- $< mdbx.c $(EXE_LDFLAGS) $(LIBS) -static -Wl,--strip-all -o $@
-
-
-install: $(LIBRARIES) $(TOOLS) $(HEADERS)
- @echo ' INSTALLING...'
- $(QUIET)mkdir -p $(DESTDIR)$(prefix)/bin$(suffix) && \
- $(INSTALL) -p $(EXE_INSTALL_FLAGS) $(TOOLS) $(DESTDIR)$(prefix)/bin$(suffix)/ && \
- mkdir -p $(DESTDIR)$(prefix)/lib$(suffix)/ && \
- $(INSTALL) -p $(EXE_INSTALL_FLAGS) $(filter-out libmdbx.a,$(LIBRARIES)) $(DESTDIR)$(prefix)/lib$(suffix)/ && \
- mkdir -p $(DESTDIR)$(prefix)/lib$(suffix)/ && \
- $(INSTALL) -p libmdbx.a $(DESTDIR)$(prefix)/lib$(suffix)/ && \
- mkdir -p $(DESTDIR)$(prefix)/include/ && \
- $(INSTALL) -p -m 444 $(HEADERS) $(DESTDIR)$(prefix)/include/ && \
- mkdir -p $(DESTDIR)$(mandir)/man1/ && \
- $(INSTALL) -p -m 444 $(addprefix $(MAN_SRCDIR), $(MANPAGES)) $(DESTDIR)$(mandir)/man1/
-
-install-strip: EXE_INSTALL_FLAGS = -s
-install-strip: install
-
-install-no-strip: EXE_INSTALL_FLAGS =
-install-no-strip: install
-
-uninstall:
- @echo ' UNINSTALLING/REMOVE...'
- $(QUIET)rm -f $(addprefix $(DESTDIR)$(prefix)/bin$(suffix)/,$(TOOLS)) \
- $(addprefix $(DESTDIR)$(prefix)/lib$(suffix)/,$(LIBRARIES)) \
- $(addprefix $(DESTDIR)$(prefix)/include/,$(HEADERS)) \
- $(addprefix $(DESTDIR)$(mandir)/man1/,$(MANPAGES))
-
-################################################################################
-# Benchmarking by ioarena
-
-ifeq ($(origin IOARENA),undefined)
-IOARENA := $(shell \
- (test -x ../ioarena/@BUILD/src/ioarena && echo ../ioarena/@BUILD/src/ioarena) || \
- (test -x ../../@BUILD/src/ioarena && echo ../../@BUILD/src/ioarena) || \
- (test -x ../../src/ioarena && echo ../../src/ioarena) || which ioarena 2>&- || \
- (echo false && echo '$(TIP) Clone and build the https://github.com/pmwkaa/ioarena.git within a neighbouring directory for availability of benchmarking.' >&2))
-endif
-NN ?= 25000000
-BENCH_CRUD_MODE ?= nosync
-
-bench-clean:
- @echo ' REMOVE bench-*.txt _ioarena/*'
- $(QUIET)rm -rf bench-*.txt _ioarena/*
-
-re-bench: bench-clean bench
-
-ifeq ($(or $(IOARENA),false),false)
-bench bench-quartet bench-triplet bench-couple:
- $(QUIET)echo 'The `ioarena` benchmark is required.' >&2 && \
- echo 'Please clone and build the https://github.com/pmwkaa/ioarena.git within a neighbouring `ioarena` directory.' >&2 && \
- false
-
-else
-
-.PHONY: bench bench-clean bench-couple re-bench bench-quartet bench-triplet
-
-define bench-rule
-bench-$(1)_$(2).txt: $(3) $(IOARENA) $(lastword $(MAKEFILE_LIST))
- @echo ' RUNNING ioarena for $1/$2...'
- $(QUIET)LD_LIBRARY_PATH="./:$$$${LD_LIBRARY_PATH}" \
- $(IOARENA) -D $(1) -B crud -m $(BENCH_CRUD_MODE) -n $(2) \
- | tee $$@ | grep throughput && \
- LD_LIBRARY_PATH="./:$$$${LD_LIBRARY_PATH}" \
- $(IOARENA) -D $(1) -B get,iterate -m $(BENCH_CRUD_MODE) -r 4 -n $(2) \
- | tee -a $$@ | grep throughput \
- || mv -f $$@ $$@.error
-
-endef
-
-$(eval $(call bench-rule,mdbx,$(NN),libmdbx.$(SO_SUFFIX)))
-
-$(eval $(call bench-rule,sophia,$(NN)))
-$(eval $(call bench-rule,leveldb,$(NN)))
-$(eval $(call bench-rule,rocksdb,$(NN)))
-$(eval $(call bench-rule,wiredtiger,$(NN)))
-$(eval $(call bench-rule,forestdb,$(NN)))
-$(eval $(call bench-rule,lmdb,$(NN)))
-$(eval $(call bench-rule,nessdb,$(NN)))
-$(eval $(call bench-rule,sqlite3,$(NN)))
-$(eval $(call bench-rule,ejdb,$(NN)))
-$(eval $(call bench-rule,vedisdb,$(NN)))
-$(eval $(call bench-rule,dummy,$(NN)))
-bench: bench-mdbx_$(NN).txt
-bench-quartet: bench-mdbx_$(NN).txt bench-lmdb_$(NN).txt bench-rocksdb_$(NN).txt bench-wiredtiger_$(NN).txt
-bench-triplet: bench-mdbx_$(NN).txt bench-lmdb_$(NN).txt bench-sqlite3_$(NN).txt
-bench-couple: bench-mdbx_$(NN).txt bench-lmdb_$(NN).txt
-
-# $(eval $(call bench-rule,debug,10))
-# .PHONY: bench-debug
-# bench-debug: bench-debug_10.txt
-
-endif
diff --git a/3rdparty/libmdbx/LICENSE b/3rdparty/libmdbx/LICENSE
deleted file mode 100644
index 05ad7571e448b9d83ead5d4691274d9484574714..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/LICENSE
+++ /dev/null
@@ -1,47 +0,0 @@
-The OpenLDAP Public License
- Version 2.8, 17 August 2003
-
-Redistribution and use of this software and associated documentation
-("Software"), with or without modification, are permitted provided
-that the following conditions are met:
-
-1. Redistributions in source form must retain copyright statements
- and notices,
-
-2. Redistributions in binary form must reproduce applicable copyright
- statements and notices, this list of conditions, and the following
- disclaimer in the documentation and/or other materials provided
- with the distribution, and
-
-3. Redistributions must contain a verbatim copy of this document.
-
-The OpenLDAP Foundation may revise this license from time to time.
-Each revision is distinguished by a version number. You may use
-this Software under terms of this license revision or under the
-terms of any subsequent revision of the license.
-
-THIS SOFTWARE IS PROVIDED BY THE OPENLDAP FOUNDATION AND ITS
-CONTRIBUTORS ``AS IS'' AND ANY EXPRESSED OR IMPLIED WARRANTIES,
-INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
-AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT
-SHALL THE OPENLDAP FOUNDATION, ITS CONTRIBUTORS, OR THE AUTHOR(S)
-OR OWNER(S) OF THE SOFTWARE BE LIABLE FOR ANY DIRECT, INDIRECT,
-INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
-BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
-LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
-CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
-LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
-ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
-POSSIBILITY OF SUCH DAMAGE.
-
-The names of the authors and copyright holders must not be used in
-advertising or otherwise to promote the sale, use or other dealing
-in this Software without specific, written prior permission. Title
-to copyright in this Software shall at all times remain with copyright
-holders.
-
-OpenLDAP is a registered trademark of the OpenLDAP Foundation.
-
-Copyright 1999-2003 The OpenLDAP Foundation, Redwood City,
-California, USA. All Rights Reserved. Permission to copy and
-distribute verbatim copies of this document is granted.
diff --git a/3rdparty/libmdbx/README.md b/3rdparty/libmdbx/README.md
deleted file mode 100644
index b27fadb211a231b7f5f9106aac68afa34cafe184..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/README.md
+++ /dev/null
@@ -1,760 +0,0 @@
-<!-- Required extensions: pymdownx.betterem, pymdownx.tilde, pymdownx.emoji, pymdownx.tasklist, pymdownx.superfences -->
-
-### The origin has been migrated to [GitFlic](https://gitflic.ru/project/erthink/libmdbx)
-since on 2022-04-15 the Github administration, without any warning
-nor explanation, deleted _libmdbx_ along with a lot of other projects,
-simultaneously blocking access for many developers.
-For the same reason ~~Github~~ is blacklisted forever.
-
-GitFlic's developers plan to support other languages,
-including English å’Œ ä¸æ–‡, in the near future.
-
-### ОÑновной репозиторий перемещен на [GitFlic](https://gitflic.ru/project/erthink/libmdbx)
-так как 15 Ð°Ð¿Ñ€ÐµÐ»Ñ 2022 админиÑÑ‚Ñ€Ð°Ñ†Ð¸Ñ Github без Ð¿Ñ€ÐµÐ´ÑƒÐ¿Ñ€ÐµÐ¶Ð´ÐµÐ½Ð¸Ñ Ð¸
-объÑÑÐ½ÐµÐ½Ð¸Ñ Ð¿Ñ€Ð¸Ñ‡Ð¸Ð½ удалила _libmdbx_ вмеÑте Ñ Ð¼Ð°ÑÑой других проектов,
-одновременно заблокировав доÑтуп многим разработчикам.
-По Ñтой же причине ~~Github~~ навÑегда занеÑен в черный ÑпиÑок.
-
---------------------------------------------------------------------------------
-
-*The Future will (be) [Positive](https://www.ptsecurity.com). Ð’ÑÑ‘ будет хорошо.*
-
-> Please refer to the online [documentation](https://libmdbx.website.yandexcloud.net)
-> with [`C` API description](https://libmdbx.website.yandexcloud.net/group__c__api.html)
-> and pay attention to the [`C++` API](https://gitflic.ru/project/erthink/libmdbx/blob?file=mdbx.h%2B%2B#line-num-1).
-
-> Questions, feedback and suggestions are welcome to the [Telegram' group](https://t.me/libmdbx).
-
-> For NEWS take a look to the [ChangeLog](https://gitflic.ru/project/erthink/libmdbx/blob?file=ChangeLog.md).
-
-
-libmdbx
-========
-
-<!-- section-begin overview -->
-
-_libmdbx_ is an extremely fast, compact, powerful, embedded, transactional
-[key-value database](https://en.wikipedia.org/wiki/Key-value_database),
-with [permissive license](https://gitflic.ru/project/erthink/libmdbx/blob?file=LICENSE).
-_libmdbx_ has a specific set of properties and capabilities,
-focused on creating unique lightweight solutions.
-
-1. Allows **a swarm of multi-threaded processes to
-[ACID](https://en.wikipedia.org/wiki/ACID)ly read and update** several
-key-value [maps](https://en.wikipedia.org/wiki/Associative_array) and
-[multimaps](https://en.wikipedia.org/wiki/Multimap) in a locally-shared
-database.
-
-2. Provides **extraordinary performance**, minimal overhead through
-[Memory-Mapping](https://en.wikipedia.org/wiki/Memory-mapped_file) and
-`Olog(N)` operations costs by virtue of [B+
-tree](https://en.wikipedia.org/wiki/B%2B_tree).
-
-3. Requires **no maintenance and no crash recovery** since it doesn't use
-[WAL](https://en.wikipedia.org/wiki/Write-ahead_logging), but that might
-be a caveat for write-intensive workloads with durability requirements.
-
-4. **Compact and friendly for fully embedding**. Only ≈25KLOC of `C11`,
-≈64K x86 binary code of core, no internal threads neither server process(es),
-but implements a simplified variant of the [Berkeley
-DB](https://en.wikipedia.org/wiki/Berkeley_DB) and
-[dbm](https://en.wikipedia.org/wiki/DBM_(computing)) API.
-
-5. Enforces [serializability](https://en.wikipedia.org/wiki/Serializability) for
-writers just by single
-[mutex](https://en.wikipedia.org/wiki/Mutual_exclusion) and affords
-[wait-free](https://en.wikipedia.org/wiki/Non-blocking_algorithm#Wait-freedom)
-for parallel readers without atomic/interlocked operations, while
-**writing and reading transactions do not block each other**.
-
-6. **Guarantee data integrity** after crash unless this was explicitly
-neglected in favour of write performance.
-
-7. Supports Linux, Windows, MacOS, Android, iOS, FreeBSD, DragonFly, Solaris,
-OpenSolaris, OpenIndiana, NetBSD, OpenBSD and other systems compliant with
-**POSIX.1-2008**.
-
-<!-- section-end -->
-
-Historically, _libmdbx_ is a deeply revised and extended descendant of the amazing
-[Lightning Memory-Mapped Database](https://en.wikipedia.org/wiki/Lightning_Memory-Mapped_Database).
-_libmdbx_ inherits all benefits from _LMDB_, but resolves some issues and adds [a set of improvements](#improvements-beyond-lmdb).
-
-<!-- section-begin mithril -->
-
-The next version is under active non-public development from scratch and will be
-released as **MithrilDB** and `libmithrildb` for libraries & packages.
-Admittedly mythical [Mithril](https://en.wikipedia.org/wiki/Mithril) is
-resembling silver but being stronger and lighter than steel. Therefore
-_MithrilDB_ is a rightly relevant name.
-
- > _MithrilDB_ will be radically different from _libmdbx_ by the new
- > database format and API based on C++17, as well as the [Apache 2.0
- > License](https://www.apache.org/licenses/LICENSE-2.0). The goal of this
- > revolution is to provide a clearer and robust API, add more features and
- > new valuable properties of the database.
-
-<!-- section-end -->
-
------
-
-## Table of Contents
-- [Characteristics](#characteristics)
- - [Features](#features)
- - [Limitations](#limitations)
- - [Gotchas](#gotchas)
- - [Comparison with other databases](#comparison-with-other-databases)
- - [Improvements beyond LMDB](#improvements-beyond-lmdb)
- - [History & Acknowledgments](#history)
-- [Usage](#usage)
- - [Building and Testing](#building-and-testing)
- - [API description](#api-description)
- - [Bindings](#bindings)
-- [Performance comparison](#performance-comparison)
- - [Integral performance](#integral-performance)
- - [Read scalability](#read-scalability)
- - [Sync-write mode](#sync-write-mode)
- - [Lazy-write mode](#lazy-write-mode)
- - [Async-write mode](#async-write-mode)
- - [Cost comparison](#cost-comparison)
-
-# Characteristics
-
-<!-- section-begin characteristics -->
-
-## Features
-
-- Key-value data model, keys are always sorted.
-
-- Fully [ACID](https://en.wikipedia.org/wiki/ACID)-compliant, through to
-[MVCC](https://en.wikipedia.org/wiki/Multiversion_concurrency_control)
-and [CoW](https://en.wikipedia.org/wiki/Copy-on-write).
-
-- Multiple key-value sub-databases within a single datafile.
-
-- Range lookups, including range query estimation.
-
-- Efficient support for short fixed length keys, including native 32/64-bit integers.
-
-- Ultra-efficient support for [multimaps](https://en.wikipedia.org/wiki/Multimap). Multi-values sorted, searchable and iterable. Keys stored without duplication.
-
-- Data is [memory-mapped](https://en.wikipedia.org/wiki/Memory-mapped_file) and accessible directly/zero-copy. Traversal of database records is extremely-fast.
-
-- Transactions for readers and writers, ones do not block others.
-
-- Writes are strongly serialized. No transaction conflicts nor deadlocks.
-
-- Readers are [non-blocking](https://en.wikipedia.org/wiki/Non-blocking_algorithm), notwithstanding [snapshot isolation](https://en.wikipedia.org/wiki/Snapshot_isolation).
-
-- Nested write transactions.
-
-- Reads scale linearly across CPUs.
-
-- Continuous zero-overhead database compactification.
-
-- Automatic on-the-fly database size adjustment.
-
-- Customizable database page size.
-
-- `Olog(N)` cost of lookup, insert, update, and delete operations by virtue of [B+ tree characteristics](https://en.wikipedia.org/wiki/B%2B_tree#Characteristics).
-
-- Online hot backup.
-
-- Append operation for efficient bulk insertion of pre-sorted data.
-
-- No [WAL](https://en.wikipedia.org/wiki/Write-ahead_logging) nor any
-transaction journal. No crash recovery needed. No maintenance is required.
-
-- No internal cache and/or memory management, all done by basic OS services.
-
-## Limitations
-
-- **Page size**: a power of 2, minimum `256` (mostly for testing), maximum `65536` bytes, default `4096` bytes.
-- **Key size**: minimum `0`, maximum ≈½ pagesize (`2022` bytes for default 4K pagesize, `32742` bytes for 64K pagesize).
-- **Value size**: minimum `0`, maximum `2146435072` (`0x7FF00000`) bytes for maps, ≈½ pagesize for multimaps (`2022` bytes for default 4K pagesize, `32742` bytes for 64K pagesize).
-- **Write transaction size**: up to `1327217884` pages (`4.944272` TiB for default 4K pagesize, `79.108351` TiB for 64K pagesize).
-- **Database size**: up to `2147483648` pages (≈`8.0` TiB for default 4K pagesize, ≈`128.0` TiB for 64K pagesize).
-- **Maximum sub-databases**: `32765`.
-
-## Gotchas
-
-1. There cannot be more than one writer at a time, i.e. no more than one write transaction at a time.
-
-2. _libmdbx_ is based on [B+ tree](https://en.wikipedia.org/wiki/B%2B_tree), so access to database pages is mostly random.
-Thus SSDs provide a significant performance boost over spinning disks for large databases.
-
-3. _libmdbx_ uses [shadow paging](https://en.wikipedia.org/wiki/Shadow_paging) instead of [WAL](https://en.wikipedia.org/wiki/Write-ahead_logging).
-Thus syncing data to disk might be a bottleneck for write intensive workload.
-
-4. _libmdbx_ uses [copy-on-write](https://en.wikipedia.org/wiki/Copy-on-write) for [snapshot isolation](https://en.wikipedia.org/wiki/Snapshot_isolation) during updates,
-but read transactions prevents recycling an old retired/freed pages, since it read ones. Thus altering of data during a parallel
-long-lived read operation will increase the process work set, may exhaust entire free database space,
-the database can grow quickly, and result in performance degradation.
-Try to avoid long running read transactions.
-
-5. _libmdbx_ is extraordinarily fast and provides minimal overhead for data access,
-so you should reconsider using brute force techniques and double check your code.
-On the one hand, in the case of _libmdbx_, a simple linear search may be more profitable than complex indexes.
-On the other hand, if you make something suboptimally, you can notice detrimentally only on sufficiently large data.
-
-## Comparison with other databases
-For now please refer to [chapter of "BoltDB comparison with other
-databases"](https://github.com/coreos/bbolt#comparison-with-other-databases)
-which is also (mostly) applicable to _libmdbx_.
-
-<!-- section-end -->
-
-<!-- section-begin improvements -->
-
-Improvements beyond LMDB
-========================
-
-_libmdbx_ is superior to legendary _[LMDB](https://symas.com/lmdb/)_ in
-terms of features and reliability, not inferior in performance. In
-comparison to _LMDB_, _libmdbx_ make things "just work" perfectly and
-out-of-the-box, not silently and catastrophically break down. The list
-below is pruned down to the improvements most notable and obvious from
-the user's point of view.
-
-## Added Features
-
-1. Keys could be more than 2 times longer than _LMDB_.
-
- > For DB with default page size _libmdbx_ support keys up to 2022 bytes
- > and up to 32742 bytes for 64K page size. _LMDB_ allows key size up to
- > 511 bytes and may silently loses data with large values.
-
-2. Up to 30% faster than _LMDB_ in [CRUD](https://en.wikipedia.org/wiki/Create,_read,_update_and_delete) benchmarks.
-
- > Benchmarks of the in-[tmpfs](https://en.wikipedia.org/wiki/Tmpfs) scenarios,
- > that tests the speed of the engine itself, showned that _libmdbx_ 10-20% faster than _LMDB_,
- > and up to 30% faster when _libmdbx_ compiled with specific build options
- > which downgrades several runtime checks to be match with LMDB behaviour.
- >
- > These and other results could be easily reproduced with [ioArena](https://github.com/pmwkaa/ioarena) just by `make bench-quartet` command,
- > including comparisons with [RockDB](https://en.wikipedia.org/wiki/RocksDB)
- > and [WiredTiger](https://en.wikipedia.org/wiki/WiredTiger).
-
-3. Automatic on-the-fly database size adjustment, both increment and reduction.
-
- > _libmdbx_ manages the database size according to parameters specified
- > by `mdbx_env_set_geometry()` function,
- > ones include the growth step and the truncation threshold.
- >
- > Unfortunately, on-the-fly database size adjustment doesn't work under [Wine](https://en.wikipedia.org/wiki/Wine_(software))
- > due to its internal limitations and unimplemented functions, i.e. the `MDBX_UNABLE_EXTEND_MAPSIZE` error will be returned.
-
-4. Automatic continuous zero-overhead database compactification.
-
- > During each commit _libmdbx_ merges a freeing pages which adjacent with the unallocated area
- > at the end of file, and then truncates unused space when a lot enough of.
-
-5. The same database format for 32- and 64-bit builds.
-
- > _libmdbx_ database format depends only on the [endianness](https://en.wikipedia.org/wiki/Endianness) but not on the [bitness](https://en.wiktionary.org/wiki/bitness).
-
-6. LIFO policy for Garbage Collection recycling. This can significantly increase write performance due write-back disk cache up to several times in a best case scenario.
-
- > LIFO means that for reuse will be taken the latest becomes unused pages.
- > Therefore the loop of database pages circulation becomes as short as possible.
- > In other words, the set of pages, that are (over)written in memory and on disk during a series of write transactions, will be as small as possible.
- > Thus creates ideal conditions for the battery-backed or flash-backed disk cache efficiency.
-
-7. Fast estimation of range query result volume, i.e. how many items can
-be found between a `KEY1` and a `KEY2`. This is a prerequisite for build
-and/or optimize query execution plans.
-
- > _libmdbx_ performs a rough estimate based on common B-tree pages of the paths from root to corresponding keys.
-
-8. `mdbx_chk` utility for database integrity check.
-Since version 0.9.1, the utility supports checking the database using any of the three meta pages and the ability to switch to it.
-
-9. Support for opening databases in the exclusive mode, including on a network share.
-
-10. Zero-length for keys and values.
-
-11. Ability to determine whether the particular data is on a dirty page
-or not, that allows to avoid copy-out before updates.
-
-12. Extended information of whole-database, sub-databases, transactions, readers enumeration.
-
- > _libmdbx_ provides a lot of information, including dirty and leftover pages
- > for a write transaction, reading lag and holdover space for read transactions.
-
-13. Extended update and delete operations.
-
- > _libmdbx_ allows one _at once_ with getting previous value
- > and addressing the particular item from multi-value with the same key.
-
-14. Useful runtime options for tuning engine to application's requirements and use cases specific.
-
-15. Automated steady sync-to-disk upon several thresholds and/or timeout via cheap polling.
-
-16. Sequence generation and three persistent 64-bit markers.
-
-17. Handle-Slow-Readers callback to resolve a database full/overflow issues due to long-lived read transaction(s).
-
-18. Ability to determine whether the cursor is pointed to a key-value
-pair, to the first, to the last, or not set to anything.
-
-
-## Other fixes and specifics
-
-1. Fixed more than 10 significant errors, in particular: page leaks,
-wrong sub-database statistics, segfault in several conditions,
-nonoptimal page merge strategy, updating an existing record with
-a change in data size (including for multimap), etc.
-
-2. All cursors can be reused and should be closed explicitly,
-regardless ones were opened within a write or read transaction.
-
-3. Opening database handles are spared from race conditions and
-pre-opening is not needed.
-
-4. Returning `MDBX_EMULTIVAL` error in case of ambiguous update or delete.
-
-5. Guarantee of database integrity even in asynchronous unordered write-to-disk mode.
-
- > _libmdbx_ propose additional trade-off by `MDBX_SAFE_NOSYNC` with append-like manner for updates,
- > that avoids database corruption after a system crash contrary to LMDB.
- > Nevertheless, the `MDBX_UTTERLY_NOSYNC` mode is available to match LMDB's behaviour for `MDB_NOSYNC`.
-
-6. On **MacOS & iOS** the `fcntl(F_FULLFSYNC)` syscall is used _by
-default_ to synchronize data with the disk, as this is [the only way to
-guarantee data
-durability](https://developer.apple.com/library/archive/documentation/System/Conceptual/ManPages_iPhoneOS/man2/fsync.2.html)
-in case of power failure. Unfortunately, in scenarios with high write
-intensity, the use of `F_FULLFSYNC` significantly degrades performance
-compared to LMDB, where the `fsync()` syscall is used. Therefore,
-_libmdbx_ allows you to override this behavior by defining the
-`MDBX_OSX_SPEED_INSTEADOF_DURABILITY=1` option while build the library.
-
-7. On **Windows** the `LockFileEx()` syscall is used for locking, since
-it allows place the database on network drives, and provides protection
-against incompetent user actions (aka
-[poka-yoke](https://en.wikipedia.org/wiki/Poka-yoke)). Therefore
-_libmdbx_ may be a little lag in performance tests from LMDB where the
-named mutexes are used.
-
-<!-- section-end -->
-<!-- section-begin history -->
-
-# History
-
-Historically, _libmdbx_ is a deeply revised and extended descendant of the
-[Lightning Memory-Mapped Database](https://en.wikipedia.org/wiki/Lightning_Memory-Mapped_Database).
-At first the development was carried out within the
-[ReOpenLDAP](todo4recovery://erased_by_github/ReOpenLDAP) project. About a
-year later _libmdbx_ was separated into a standalone project, which was
-[presented at Highload++ 2015
-conference](http://www.highload.ru/2015/abstracts/1831.html).
-
-Since 2017 _libmdbx_ is used in [Fast Positive Tables](https://gitflic.ru/project/erthink/libfpta),
-and development is funded by [Positive Technologies](https://www.ptsecurity.com).
-
-On 2022-04-15 the Github administration, without any warning nor
-explanation, deleted _libmdbx_ along with a lot of other projects,
-simultaneously blocking access for many developers. Therefore on
-2022-04-21 we have migrated to a reliable trusted infrastructure.
-The origin for now is at [GitFlic](https://gitflic.ru/project/erthink/libmdbx)
-with backup at [ABF by ROSA Лаб](https://abf.io/erthink/libmdbx).
-For the same reason ~~Github~~ is blacklisted forever.
-
-## Acknowledgments
-Howard Chu <hyc@openldap.org> is the author of LMDB, from which
-originated the _libmdbx_ in 2015.
-
-Martin Hedenfalk <martin@bzero.se> is the author of `btree.c` code, which
-was used to begin development of LMDB.
-
-<!-- section-end -->
-
---------------------------------------------------------------------------------
-
-Usage
-=====
-
-<!-- section-begin usage -->
-
-Currently, libmdbx is only available in a
-[source code](https://en.wikipedia.org/wiki/Source_code) form.
-Packages support for common Linux distributions is planned in the future,
-since release the version 1.0.
-
-## Source code embedding
-
-_libmdbx_ provides two official ways for integration in source code form:
-
-1. Using the amalgamated source code.
-
- > The amalgamated source code includes all files required to build and
- > use _libmdbx_, but not for testing _libmdbx_ itself.
-
-2. Adding the complete original source code as a `git submodule`.
-
- > This allows you to build as _libmdbx_ and testing tool.
- > On the other hand, this way requires you to pull git tags, and use C++11 compiler for test tool.
-
-_**Please, avoid using any other techniques.**_ Otherwise, at least
-don't ask for support and don't name such chimeras `libmdbx`.
-
-The amalgamated source code could be created from the original clone of git
-repository on Linux by executing `make dist`. As a result, the desired
-set of files will be formed in the `dist` subdirectory.
-
-
-## Building and Testing
-
-Both amalgamated and original source code provides build through the use
-[CMake](https://cmake.org/) or [GNU
-Make](https://www.gnu.org/software/make/) with
-[bash](https://en.wikipedia.org/wiki/Bash_(Unix_shell)). All build ways
-are completely traditional and have minimal prerequirements like
-`build-essential`, i.e. the non-obsolete C/C++ compiler and a
-[SDK](https://en.wikipedia.org/wiki/Software_development_kit) for the
-target platform. Obviously you need building tools itself, i.e. `git`,
-`cmake` or GNU `make` with `bash`. For your convenience, `make help`
-and `make options` are also available for listing existing targets
-and build options respectively.
-
-The only significant specificity is that git' tags are required
-to build from complete (not amalgamated) source codes.
-Executing **`git fetch --tags --force --prune`** is enough to get ones,
-and `--unshallow` or `--update-shallow` is required for shallow cloned case.
-
-So just using CMake or GNU Make in your habitual manner and feel free to
-fill an issue or make pull request in the case something will be
-unexpected or broken down.
-
-### Testing
-The amalgamated source code does not contain any tests for or several reasons.
-Please read [the explanation](todo4recovery://erased_by_github/libmdbx/issues/214#issuecomment-870717981) and don't ask to alter this.
-So for testing _libmdbx_ itself you need a full source code, i.e. the clone of a git repository, there is no option.
-
-The full source code of _libmdbx_ has a [`test` subdirectory](https://gitflic.ru/project/erthink/libmdbx/tree/master/test) with minimalistic test "framework".
-Actually yonder is a source code of the `mdbx_test` – console utility which has a set of command-line options that allow construct and run a reasonable enough test scenarios.
-This test utility is intended for _libmdbx_'s developers for testing library itself, but not for use by users.
-Therefore, only basic information is provided:
-
- - There are few CRUD-based test cases (hill, TTL, nested, append, jitter, etc),
- which can be combined to test the concurrent operations within shared database in a multi-processes environment.
- This is the `basic` test scenario.
- - The `Makefile` provide several self-described targets for testing: `smoke`, `test`, `check`, `memcheck`, `test-valgrind`,
- `test-asan`, `test-leak`, `test-ubsan`, `cross-gcc`, `cross-qemu`, `gcc-analyzer`, `smoke-fault`, `smoke-singleprocess`,
- `test-singleprocess`, 'long-test'. Please run `make --help` if doubt.
- - In addition to the `mdbx_test` utility, there is the script [`long_stochastic.sh`](https://gitflic.ru/project/erthink/libmdbx/blob/master/test/long_stochastic.sh),
- which calls `mdbx_test` by going through set of modes and options, with gradually increasing the number of operations and the size of transactions.
- This script is used for mostly of all automatic testing, including `Makefile` targets and Continuous Integration.
- - Brief information of available command-line options is available by `--help`.
- However, you should dive into source code to get all, there is no option.
-
-Anyway, no matter how thoroughly the _libmdbx_ is tested, you should rely only on your own tests for a few reasons:
-
-1. Mostly of all use cases are unique.
- So it is no warranty that your use case was properly tested, even the _libmdbx_'s tests engages stochastic approach.
-2. If there are problems, then your test on the one hand will help to verify whether you are using _libmdbx_ correctly,
- on the other hand it will allow to reproduce the problem and insure against regression in a future.
-3. Actually you should rely on than you checked by yourself or take a risk.
-
-
-### Common important details
-
-#### Build reproducibility
-By default _libmdbx_ track build time via `MDBX_BUILD_TIMESTAMP` build option and macro.
-So for a [reproducible builds](https://en.wikipedia.org/wiki/Reproducible_builds) you should predefine/override it to known fixed string value.
-For instance:
-
- - for reproducible build with make: `make MDBX_BUILD_TIMESTAMP=unknown ` ...
- - or during configure by CMake: `cmake -DMDBX_BUILD_TIMESTAMP:STRING=unknown ` ...
-
-Of course, in addition to this, your toolchain must ensure the reproducibility of builds.
-For more information please refer to [reproducible-builds.org](https://reproducible-builds.org/).
-
-#### Containers
-There are no special traits nor quirks if you use libmdbx ONLY inside the single container.
-But in a cross-container cases or with a host-container(s) mix the two major things MUST be
-guaranteed:
-
-1. Coherence of memory mapping content and unified page cache inside OS kernel for host and all container(s) operated with a DB.
-Basically this means must be only a single physical copy of each memory mapped DB' page in the system memory.
-
-2. Uniqueness of [PID](https://en.wikipedia.org/wiki/Process_identifier) values and/or a common space for ones:
- - for POSIX systems: PID uniqueness for all processes operated with a DB.
- I.e. the `--pid=host` is required for run DB-aware processes inside Docker,
- either without host interaction a `--pid=container:<name|id>` with the same name/id.
- - for non-POSIX (i.e. Windows) systems: inter-visibility of processes handles.
- I.e. the `OpenProcess(SYNCHRONIZE, ..., PID)` must return reasonable error,
- including `ERROR_ACCESS_DENIED`,
- but not the `ERROR_INVALID_PARAMETER` as for an invalid/non-existent PID.
-
-#### DSO/DLL unloading and destructors of Thread-Local-Storage objects
-When building _libmdbx_ as a shared library or use static _libmdbx_ as a
-part of another dynamic library, it is advisable to make sure that your
-system ensures the correctness of the call destructors of
-Thread-Local-Storage objects when unloading dynamic libraries.
-
-If this is not the case, then unloading a dynamic-link library with
-_libmdbx_ code inside, can result in either a resource leak or a crash
-due to calling destructors from an already unloaded DSO/DLL object. The
-problem can only manifest in a multithreaded application, which makes
-the unloading of shared dynamic libraries with _libmdbx_ code inside,
-after using _libmdbx_. It is known that TLS-destructors are properly
-maintained in the following cases:
-
-- On all modern versions of Windows (Windows 7 and later).
-
-- On systems with the
-[`__cxa_thread_atexit_impl()`](https://sourceware.org/glibc/wiki/Destructor%20support%20for%20thread_local%20variables)
-function in the standard C library, including systems with GNU libc
-version 2.18 and later.
-
-- On systems with libpthread/ntpl from GNU libc with bug fixes
-[#21031](https://sourceware.org/bugzilla/show_bug.cgi?id=21031) and
-[#21032](https://sourceware.org/bugzilla/show_bug.cgi?id=21032), or
-where there are no similar bugs in the pthreads implementation.
-
-### Linux and other platforms with GNU Make
-To build the library it is enough to execute `make all` in the directory
-of source code, and `make check` to execute the basic tests.
-
-If the `make` installed on the system is not GNU Make, there will be a
-lot of errors from make when trying to build. In this case, perhaps you
-should use `gmake` instead of `make`, or even `gnu-make`, etc.
-
-### FreeBSD and related platforms
-As a rule on BSD and it derivatives the default is to use Berkeley Make and
-[Bash](https://en.wikipedia.org/wiki/Bash_(Unix_shell)) is not installed.
-
-So you need to install the required components: GNU Make, Bash, C and C++
-compilers compatible with GCC or CLANG. After that, to build the
-library, it is enough to execute `gmake all` (or `make all`) in the
-directory with source code, and `gmake check` (or `make check`) to run
-the basic tests.
-
-### Windows
-For build _libmdbx_ on Windows the _original_ CMake and [Microsoft Visual
-Studio 2019](https://en.wikipedia.org/wiki/Microsoft_Visual_Studio) are
-recommended. Please use the recent versions of CMake, Visual Studio and Windows
-SDK to avoid troubles with C11 support and `alignas()` feature.
-
-For build by MinGW the 10.2 or recent version coupled with a modern CMake are required.
-So it is recommended to use [chocolatey](https://chocolatey.org/) to install and/or update the ones.
-
-Another ways to build is potentially possible but not supported and will not.
-The `CMakeLists.txt` or `GNUMakefile` scripts will probably need to be modified accordingly.
-Using other methods do not forget to add the `ntdll.lib` to linking.
-
-It should be noted that in _libmdbx_ was efforts to avoid
-runtime dependencies from CRT and other MSVC libraries.
-For this is enough to pass the `-DMDBX_WITHOUT_MSVC_CRT:BOOL=ON` option
-during configure by CMake.
-
-An example of running a basic test script can be found in the
-[CI-script](appveyor.yml) for [AppVeyor](https://www.appveyor.com/). To
-run the [long stochastic test scenario](test/long_stochastic.sh),
-[bash](https://en.wikipedia.org/wiki/Bash_(Unix_shell)) is required, and
-such testing is recommended with placing the test data on the
-[RAM-disk](https://en.wikipedia.org/wiki/RAM_drive).
-
-### Windows Subsystem for Linux
-_libmdbx_ could be used in [WSL2](https://en.wikipedia.org/wiki/Windows_Subsystem_for_Linux#WSL_2)
-but NOT in [WSL1](https://en.wikipedia.org/wiki/Windows_Subsystem_for_Linux#WSL_1) environment.
-This is a consequence of the fundamental shortcomings of _WSL1_ and cannot be fixed.
-To avoid data loss, _libmdbx_ returns the `ENOLCK` (37, "No record locks available")
-error when opening the database in a _WSL1_ environment.
-
-### MacOS
-Current [native build tools](https://en.wikipedia.org/wiki/Xcode) for
-MacOS include GNU Make, CLANG and an outdated version of Bash.
-Therefore, to build the library, it is enough to run `make all` in the
-directory with source code, and run `make check` to execute the base
-tests. If something goes wrong, it is recommended to install
-[Homebrew](https://brew.sh/) and try again.
-
-To run the [long stochastic test scenario](test/long_stochastic.sh), you
-will need to install the current (not outdated) version of
-[Bash](https://en.wikipedia.org/wiki/Bash_(Unix_shell)). To do this, we
-recommend that you install [Homebrew](https://brew.sh/) and then execute
-`brew install bash`.
-
-### Android
-We recommend using CMake to build _libmdbx_ for Android.
-Please refer to the [official guide](https://developer.android.com/studio/projects/add-native-code).
-
-### iOS
-To build _libmdbx_ for iOS, we recommend using CMake with the
-["toolchain file"](https://cmake.org/cmake/help/latest/variable/CMAKE_TOOLCHAIN_FILE.html)
-from the [ios-cmake](https://github.com/leetal/ios-cmake) project.
-
-<!-- section-end -->
-
-## API description
-
-Please refer to the online [_libmdbx_ API reference](https://libmdbx.website.yandexcloud.net/docs)
-and/or see the [mdbx.h++](mdbx.h%2B%2B) and [mdbx.h](mdbx.h) headers.
-
-<!-- section-begin bindings -->
-
-Bindings
-========
-
-| Runtime | Repo | Author |
-| ------- | ------ | ------ |
-| Scala | [mdbx4s](https://github.com/david-bouyssie/mdbx4s) | [David Bouyssié](https://github.com/david-bouyssie) |
-| Haskell | [libmdbx-hs](https://hackage.haskell.org/package/libmdbx) | [Francisco Vallarino](https://github.com/fjvallarino) |
-| NodeJS, [Deno](https://deno.land/) | [lmdbx-js](https://github.com/kriszyp/lmdbx-js) | [Kris Zyp](https://github.com/kriszyp/)
-| NodeJS | [node-mdbx](https://www.npmjs.com/package/node-mdbx/) | [Сергей Федотов](mailto:sergey.fedotov@corp.mail.ru) |
-| Ruby | [ruby-mdbx](https://rubygems.org/gems/mdbx/) | [Mahlon E. Smith](https://github.com/mahlonsmith) |
-| Go | [mdbx-go](https://github.com/torquem-ch/mdbx-go) | [Alex Sharov](https://github.com/AskAlexSharov) |
-| [Nim](https://en.wikipedia.org/wiki/Nim_(programming_language)) | [NimDBX](https://github.com/snej/nimdbx) | [Jens Alfke](https://github.com/snej)
-| Rust | [libmdbx-rs](https://github.com/vorot93/libmdbx-rs) | [Artem Vorotnikov](https://github.com/vorot93) |
-| Rust | [mdbx](https://crates.io/crates/mdbx) | [gcxfd](https://github.com/gcxfd) |
-| Java | [mdbxjni](https://github.com/castortech/mdbxjni) | [Castor Technologies](https://castortech.com/) |
-| Python (draft) | [python-bindings](todo4recovery://erased_by_github/libmdbx/commits/python-bindings) branch | [Noel Kuntze](https://github.com/Thermi)
-| .NET (obsolete) | [mdbx.NET](https://github.com/wangjia184/mdbx.NET) | [Jerry Wang](https://github.com/wangjia184) |
-
-<!-- section-end -->
-
---------------------------------------------------------------------------------
-
-<!-- section-begin performance -->
-
-Performance comparison
-======================
-
-All benchmarks were done in 2015 by [IOArena](https://github.com/pmwkaa/ioarena)
-and multiple [scripts](https://github.com/pmwkaa/ioarena/tree/HL%2B%2B2015)
-runs on Lenovo Carbon-2 laptop, i7-4600U 2.1 GHz (2 physical cores, 4 HyperThreading cores), 8 Gb RAM,
-SSD SAMSUNG MZNTD512HAGL-000L1 (DXT23L0Q) 512 Gb.
-
-## Integral performance
-
-Here showed sum of performance metrics in 3 benchmarks:
-
- - Read/Search on the machine with 4 logical CPUs in HyperThreading mode (i.e. actually 2 physical CPU cores);
-
- - Transactions with [CRUD](https://en.wikipedia.org/wiki/CRUD)
- operations in sync-write mode (fdatasync is called after each
- transaction);
-
- - Transactions with [CRUD](https://en.wikipedia.org/wiki/CRUD)
- operations in lazy-write mode (moment to sync data to persistent storage
- is decided by OS).
-
-*Reasons why asynchronous mode isn't benchmarked here:*
-
- 1. It doesn't make sense as it has to be done with DB engines, oriented
- for keeping data in memory e.g. [Tarantool](https://tarantool.io/),
- [Redis](https://redis.io/)), etc.
-
- 2. Performance gap is too high to compare in any meaningful way.
-
-
-
---------------------------------------------------------------------------------
-
-## Read Scalability
-
-Summary performance with concurrent read/search queries in 1-2-4-8
-threads on the machine with 4 logical CPUs in HyperThreading mode (i.e. actually 2 physical CPU cores).
-
-
-
---------------------------------------------------------------------------------
-
-## Sync-write mode
-
- - Linear scale on left and dark rectangles mean arithmetic mean
- transactions per second;
-
- - Logarithmic scale on right is in seconds and yellow intervals mean
- execution time of transactions. Each interval shows minimal and maximum
- execution time, cross marks standard deviation.
-
-**10,000 transactions in sync-write mode**. In case of a crash all data
-is consistent and conforms to the last successful transaction. The
-[fdatasync](https://linux.die.net/man/2/fdatasync) syscall is used after
-each write transaction in this mode.
-
-In the benchmark each transaction contains combined CRUD operations (2
-inserts, 1 read, 1 update, 1 delete). Benchmark starts on an empty database
-and after full run the database contains 10,000 small key-value records.
-
-
-
---------------------------------------------------------------------------------
-
-## Lazy-write mode
-
- - Linear scale on left and dark rectangles mean arithmetic mean of
- thousands transactions per second;
-
- - Logarithmic scale on right in seconds and yellow intervals mean
- execution time of transactions. Each interval shows minimal and maximum
- execution time, cross marks standard deviation.
-
-**100,000 transactions in lazy-write mode**. In case of a crash all data
-is consistent and conforms to the one of last successful transactions, but
-transactions after it will be lost. Other DB engines use
-[WAL](https://en.wikipedia.org/wiki/Write-ahead_logging) or transaction
-journal for that, which in turn depends on order of operations in the
-journaled filesystem. _libmdbx_ doesn't use WAL and hands I/O operations
-to filesystem and OS kernel (mmap).
-
-In the benchmark each transaction contains combined CRUD operations (2
-inserts, 1 read, 1 update, 1 delete). Benchmark starts on an empty database
-and after full run the database contains 100,000 small key-value
-records.
-
-
-
-
---------------------------------------------------------------------------------
-
-## Async-write mode
-
- - Linear scale on left and dark rectangles mean arithmetic mean of
- thousands transactions per second;
-
- - Logarithmic scale on right in seconds and yellow intervals mean
- execution time of transactions. Each interval shows minimal and maximum
- execution time, cross marks standard deviation.
-
-**1,000,000 transactions in async-write mode**.
-In case of a crash all data is consistent and conforms to the one of last successful transactions,
-but lost transaction count is much higher than in
-lazy-write mode. All DB engines in this mode do as little writes as
-possible on persistent storage. _libmdbx_ uses
-[msync(MS_ASYNC)](https://linux.die.net/man/2/msync) in this mode.
-
-In the benchmark each transaction contains combined CRUD operations (2
-inserts, 1 read, 1 update, 1 delete). Benchmark starts on an empty database
-and after full run the database contains 10,000 small key-value records.
-
-
-
---------------------------------------------------------------------------------
-
-## Cost comparison
-
-Summary of used resources during lazy-write mode benchmarks:
-
- - Read and write IOPs;
-
- - Sum of user CPU time and sys CPU time;
-
- - Used space on persistent storage after the test and closed DB, but not
- waiting for the end of all internal housekeeping operations (LSM
- compactification, etc).
-
-_ForestDB_ is excluded because benchmark showed it's resource
-consumption for each resource (CPU, IOPs) much higher than other engines
-which prevents to meaningfully compare it with them.
-
-All benchmark data is gathered by
-[getrusage()](http://man7.org/linux/man-pages/man2/getrusage.2.html)
-syscall and by scanning the data directory.
-
-
-
-<!-- section-end -->
diff --git a/3rdparty/libmdbx/VERSION.txt b/3rdparty/libmdbx/VERSION.txt
deleted file mode 100644
index a9da9632c3d3a984f84670e0e4d1458e18896c49..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/VERSION.txt
+++ /dev/null
@@ -1 +0,0 @@
-0.11.7.0
diff --git a/3rdparty/libmdbx/cmake/compiler.cmake b/3rdparty/libmdbx/cmake/compiler.cmake
deleted file mode 100644
index 17c468b7395273f43a5bfd8e413ad1329fb3aa80..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/cmake/compiler.cmake
+++ /dev/null
@@ -1,1023 +0,0 @@
-## Copyright (c) 2012-2022 Leonid Yuriev <leo@yuriev.ru>.
-##
-## Licensed under the Apache License, Version 2.0 (the "License");
-## you may not use this file except in compliance with the License.
-## You may obtain a copy of the License at
-##
-## http://www.apache.org/licenses/LICENSE-2.0
-##
-## Unless required by applicable law or agreed to in writing, software
-## distributed under the License is distributed on an "AS IS" BASIS,
-## WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-## See the License for the specific language governing permissions and
-## limitations under the License.
-##
-
-if(CMAKE_VERSION VERSION_LESS 3.12)
- cmake_minimum_required(VERSION 3.8.2)
-else()
- cmake_minimum_required(VERSION 3.12)
-endif()
-
-cmake_policy(PUSH)
-cmake_policy(VERSION ${CMAKE_MINIMUM_REQUIRED_VERSION})
-
-if(CMAKE_VERSION MATCHES ".*MSVC.*" AND CMAKE_VERSION VERSION_LESS 3.16)
- message(FATAL_ERROR "CMake from MSVC kit is unfit! "
- "Please use MSVC2019 with modern CMake the original CMake from https://cmake.org/download/")
-endif()
-
-if(NOT (CMAKE_C_COMPILER_LOADED OR CMAKE_CXX_COMPILER_LOADED))
- message(FATAL_ERROR "This module required C or C++ to be enabled")
-endif()
-
-if(CMAKE_CXX_COMPILER_LOADED)
- include(CheckCXXSourceRuns)
- include(CheckCXXSourceCompiles)
- include(CheckCXXCompilerFlag)
-endif()
-if(CMAKE_C_COMPILER_LOADED)
- include(CheckCSourceRuns)
- include(CheckCSourceCompiles)
- include(CheckCCompilerFlag)
-endif()
-include(CMakeDependentOption)
-include(CheckLibraryExists)
-include(CheckIncludeFiles)
-
-# Check if the same compile family is used for both C and CXX
-if(CMAKE_C_COMPILER_LOADED AND CMAKE_CXX_COMPILER_LOADED AND
- NOT (CMAKE_C_COMPILER_ID STREQUAL CMAKE_CXX_COMPILER_ID))
- message(WARNING "CMAKE_C_COMPILER_ID (${CMAKE_C_COMPILER_ID}) is different "
- "from CMAKE_CXX_COMPILER_ID (${CMAKE_CXX_COMPILER_ID}). "
- "The final binary may be unusable.")
-endif()
-
-if(CMAKE_CXX_COMPILER_LOADED)
- set(CMAKE_PRIMARY_LANG "CXX")
-else()
- set(CMAKE_PRIMARY_LANG "C")
-endif()
-
-macro(check_compiler_flag flag variable)
- if(CMAKE_CXX_COMPILER_LOADED)
- check_cxx_compiler_flag(${flag} ${variable})
- else()
- check_c_compiler_flag(${flag} ${variable})
- endif()
-endmacro(check_compiler_flag)
-
-# We support building with Clang and gcc. First check
-# what we're using for build.
-if(CMAKE_C_COMPILER_LOADED AND CMAKE_C_COMPILER_ID MATCHES ".*[Cc][Ll][Aa][Nn][Gg].*")
- set(CMAKE_COMPILER_IS_CLANG ON)
- set(CMAKE_COMPILER_IS_GNUCC OFF)
-endif()
-if(CMAKE_CXX_COMPILER_LOADED AND CMAKE_CXX_COMPILER_ID MATCHES ".*[Cc][Ll][Aa][Nn][Gg].*")
- set(CMAKE_COMPILER_IS_CLANG ON)
- set(CMAKE_COMPILER_IS_GNUCXX OFF)
-endif()
-
-if(CMAKE_C_COMPILER_LOADED)
- # Check for Elbrus lcc
- execute_process(COMMAND ${CMAKE_C_COMPILER} --version
- OUTPUT_VARIABLE tmp_lcc_probe_version
- RESULT_VARIABLE tmp_lcc_probe_result ERROR_QUIET)
- if(tmp_lcc_probe_result EQUAL 0)
- string(FIND "${tmp_lcc_probe_version}" "lcc:" tmp_lcc_marker)
- string(FIND "${tmp_lcc_probe_version}" ":e2k-" tmp_e2k_marker)
- if(tmp_lcc_marker GREATER -1 AND tmp_e2k_marker GREATER tmp_lcc_marker)
- execute_process(COMMAND ${CMAKE_C_COMPILER} -print-version
- OUTPUT_VARIABLE CMAKE_C_COMPILER_VERSION
- RESULT_VARIABLE tmp_lcc_probe_result
- OUTPUT_STRIP_TRAILING_WHITESPACE)
- set(CMAKE_COMPILER_IS_ELBRUSC ON)
- set(CMAKE_C_COMPILER_ID "Elbrus")
- message(STATUS "Detected Elbrus C compiler ${CMAKE_C_COMPILER_VERSION}")
- else()
- set(CMAKE_COMPILER_IS_ELBRUSC OFF)
- endif()
- unset(tmp_lcc_marker)
- unset(tmp_e2k_marker)
- endif()
- unset(tmp_lcc_probe_version)
- unset(tmp_lcc_probe_result)
-endif()
-
-if(CMAKE_CXX_COMPILER_LOADED)
- # Check for Elbrus l++
- execute_process(COMMAND ${CMAKE_CXX_COMPILER} --version
- OUTPUT_VARIABLE tmp_lxx_probe_version
- RESULT_VARIABLE tmp_lxx_probe_result ERROR_QUIET)
- if(tmp_lxx_probe_result EQUAL 0)
- string(FIND "${tmp_lxx_probe_version}" "lcc:" tmp_lcc_marker)
- string(FIND "${tmp_lxx_probe_version}" ":e2k-" tmp_e2k_marker)
- if(tmp_lcc_marker GREATER -1 AND tmp_e2k_marker GREATER tmp_lcc_marker)
- execute_process(COMMAND ${CMAKE_CXX_COMPILER} -print-version
- OUTPUT_VARIABLE CMAKE_CXX_COMPILER_VERSION
- RESULT_VARIABLE tmp_lxx_probe_result
- OUTPUT_STRIP_TRAILING_WHITESPACE)
- set(CMAKE_COMPILER_IS_ELBRUSCXX ON)
- set(CMAKE_CXX_COMPILER_ID "Elbrus")
- message(STATUS "Detected Elbrus C++ compiler ${CMAKE_CXX_COMPILER_VERSION}")
- else()
- set(CMAKE_COMPILER_IS_ELBRUSCXX OFF)
- endif()
- unset(tmp_lcc_marker)
- unset(tmp_e2k_marker)
- endif()
- unset(tmp_lxx_probe_version)
- unset(tmp_lxx_probe_result)
-endif()
-
-# Hard coding the compiler version is ugly from cmake POV, but
-# at least gives user a friendly error message. The most critical
-# demand for C++ compiler is support of C++11 lambdas, added
-# only in version 4.5 https://gcc.gnu.org/projects/cxx0x.html
-if(CMAKE_COMPILER_IS_GNUCC)
- if(CMAKE_C_COMPILER_VERSION VERSION_LESS 4.5
- AND NOT CMAKE_COMPILER_IS_ELBRUSC)
- message(FATAL_ERROR "
- Your GCC version is ${CMAKE_C_COMPILER_VERSION}, please update")
- endif()
-endif()
-if(CMAKE_COMPILER_IS_GNUCXX)
- if(CMAKE_CXX_COMPILER_VERSION VERSION_LESS 4.5
- AND NOT CMAKE_COMPILER_IS_ELBRUSCXX)
- message(FATAL_ERROR "
- Your G++ version is ${CMAKE_CXX_COMPILER_VERSION}, please update")
- endif()
-endif()
-
-if(CMAKE_CL_64)
- set(MSVC64 1)
-endif()
-if(WIN32 AND CMAKE_COMPILER_IS_GNU${CMAKE_PRIMARY_LANG})
- execute_process(COMMAND ${CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER} -dumpmachine
- OUTPUT_VARIABLE __GCC_TARGET_MACHINE
- OUTPUT_STRIP_TRAILING_WHITESPACE)
- if(__GCC_TARGET_MACHINE MATCHES "amd64|x86_64|AMD64")
- set(MINGW64 1)
- endif()
- unset(__GCC_TARGET_MACHINE)
-endif()
-
-if(NOT DEFINED IOS)
- if(APPLE AND (CMAKE_SYSTEM_NAME STREQUAL "iOS"
- OR DEFINED CMAKE_IOS_DEVELOPER_ROOT
- OR DEFINED IOS_PLATFORM OR DEFINED IOS_ARCH))
- set(IOS TRUE)
- else()
- set(IOS FALSE)
- endif()
-endif()
-
-if(NOT DEFINED CMAKE_TARGET_BITNESS)
- if (CMAKE_SIZEOF_VOID_P LESS 4)
- set(CMAKE_TARGET_BITNESS 16)
- elseif (CMAKE_SIZEOF_VOID_P LESS 8)
- set(CMAKE_TARGET_BITNESS 32)
- else()
- set(CMAKE_TARGET_BITNESS 64)
- endif()
-endif()
-
-if(NOT CMAKE_SYSTEM_ARCH)
- if(CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_ARCHITECTURE_ID)
- set(CMAKE_SYSTEM_ARCH "${CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_ARCHITECTURE_ID}")
- elseif(CMAKE_ANDROID_ARCH_ABI)
- set(CMAKE_SYSTEM_ARCH "${CMAKE_ANDROID_ARCH_ABI}")
- elseif(CMAKE_COMPILER_IS_ELBRUSC OR CMAKE_COMPILER_IS_ELBRUSCXX
- OR CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_ID STREQUAL "LCC"
- OR CMAKE_SYSTEM_PROCESSOR MATCHES "e2k.*|E2K.*|elbrus.*|ELBRUS.*")
- set(E2K TRUE)
- set(CMAKE_SYSTEM_ARCH "Elbrus")
- elseif(MSVC64 OR MINGW64 OR MINGW OR (MSVC AND NOT CMAKE_CROSSCOMPILING))
- if(CMAKE_TARGET_BITNESS EQUAL 64)
- set(X86_64 TRUE)
- set(CMAKE_SYSTEM_ARCH "x86_64")
- else()
- set(X86_32 TRUE)
- set(CMAKE_SYSTEM_ARCH "x86")
- endif()
- elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "amd64.*|[xXiI]86_64.*|AMD64.*|[iI][3-6]86.*|[xXiI]86.*")
- if(CMAKE_TARGET_BITNESS EQUAL 64)
- set(X86_64 TRUE)
- set(CMAKE_SYSTEM_ARCH "x86_64")
- else()
- set(X86_32 TRUE)
- set(CMAKE_SYSTEM_ARCH "x86")
- endif()
- elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(aarch.*|AARCH.*|arm.*|ARM.*)")
- if(CMAKE_TARGET_BITNESS EQUAL 64)
- set(AARCH64 TRUE)
- set(CMAKE_SYSTEM_ARCH "ARM64")
- else()
- set(ARM32 TRUE)
- set(CMAKE_SYSTEM_ARCH "ARM")
- endif()
- elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc).*")
- if(CMAKE_TARGET_BITNESS EQUAL 64)
- set(PPC64 TRUE)
- set(CMAKE_SYSTEM_ARCH "PPC64")
- else()
- set(PPC32 TRUE)
- set(CMAKE_SYSTEM_ARCH "PPC")
- endif()
- elseif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(mips|MIPS).*")
- if(CMAKE_TARGET_BITNESS EQUAL 64)
- set(MIPS64 TRUE)
- set(CMAKE_SYSTEM_ARCH "MIPS64")
- else()
- set(MIPS32 TRUE)
- set(CMAKE_SYSTEM_ARCH "MIPS")
- endif()
- endif()
-endif()
-if(CMAKE_C_COMPILER_LOADED AND NOT CMAKE_C_COMPILER_ARCHITECTURE_ID)
- set(CMAKE_C_COMPILER_ARCHITECTURE_ID "${CMAKE_SYSTEM_ARCH}")
-endif()
-if(CMAKE_CXX_COMPILER_LOADED AND NOT CMAKE_CXX_COMPILER_ARCHITECTURE_ID)
- set(CMAKE_CXX_COMPILER_ARCHITECTURE_ID "${CMAKE_SYSTEM_ARCH}")
-endif()
-
-if(NOT CMAKE_HOST_ARCH)
- if(CMAKE_VS_PLATFORM_TOOLSET_HOST_ARCHITECTURE)
- set(CMAKE_HOST_ARCH "${CMAKE_VS_PLATFORM_TOOLSET_HOST_ARCHITECTURE}")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "e2k.*|E2K.*|elbrus.*|ELBRUS.*")
- set(CMAKE_HOST_ARCH "Elbrus")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "amd64.*|[xXiI]86_64.*|AMD64.*")
- set(CMAKE_HOST_ARCH "x86_64")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "[iI][3-6]86.*|[xXiI]86.*")
- set(CMAKE_HOST_ARCH "x86")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "^(aarch64.*|AARCH64.*|ARM64.*)")
- set(CMAKE_HOST_ARCH "ARM64")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "^(arm.*|ARM.*)")
- set(CMAKE_HOST_ARCH "ARM")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "^(powerpc|ppc)64.*")
- set(CMAKE_HOST_ARCH "PPC64")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "^(powerpc|ppc).*")
- set(CMAKE_HOST_ARCH "PPC")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "^(mips|MIPS)64.*")
- set(CMAKE_HOST_ARCH "MIPS64")
- elseif(CMAKE_HOST_PROCESSOR MATCHES "^(mips|MIPS).*")
- set(CMAKE_HOST_ARCH "MIPS")
- else()
- set(CMAKE_HOST_ARCH "${CMAKE_HOST_SYSTEM_PROCESSOR}")
- endif()
-endif()
-
-if(NOT DEFINED CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET)
- if(CMAKE_CROSSCOMPILING AND CMAKE_CROSSCOMPILING_EMULATOR)
- set(CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET TRUE)
- elseif(CMAKE_CROSSCOMPILING AND NOT CMAKE_CROSSCOMPILING_EMULATOR)
- set(CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET FALSE)
- elseif(CMAKE_SYSTEM_NAME STREQUAL CMAKE_HOST_SYSTEM_NAME
- AND ((CMAKE_SYSTEM_PROCESSOR STREQUAL CMAKE_HOST_PROCESSOR)
- OR (CMAKE_SYSTEM_ARCH STREQUAL CMAKE_HOST_ARCH)
- OR (WIN32 AND CMAKE_HOST_WIN32 AND X86_32 AND CMAKE_HOST_ARCH STREQUAL "x86_64")))
- set(CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET TRUE)
- message(STATUS
- "Assume СAN RUN A BUILT EXECUTABLES,"
- " since host (${CMAKE_HOST_SYSTEM_NAME}-${CMAKE_HOST_ARCH})"
- " match target (${CMAKE_SYSTEM_NAME}-${CMAKE_SYSTEM_ARCH})")
- else()
- if(CMAKE_C_COMPILER_LOADED)
- include(CheckCSourceRuns)
- check_c_source_runs("int main(void) { return 0; }" CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET)
- elseif(CMAKE_CXX_COMPILER_LOADED)
- include(CheckCXXSourceRuns)
- check_cxx_source_runs("int main(void) { return 0; }" CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET)
- endif()
- if(NOT CMAKE_HOST_CAN_RUN_EXECUTABLES_BUILT_FOR_TARGET)
- message(STATUS "Force CMAKE_CROSSCOMPILING to TRUE")
- set(CMAKE_CROSSCOMPILING TRUE)
- endif()
- endif()
-endif()
-
-if(MSVC)
- check_compiler_flag("/WX" CC_HAS_WERROR)
-else()
- #
- # GCC started to warn for unused result starting from 4.2, and
- # this is when it introduced -Wno-unused-result
- # GCC can also be built on top of llvm runtime (on mac).
- check_compiler_flag("-Wno-unknown-pragmas" CC_HAS_WNO_UNKNOWN_PRAGMAS)
- check_compiler_flag("-Wextra" CC_HAS_WEXTRA)
- check_compiler_flag("-Werror" CC_HAS_WERROR)
- check_compiler_flag("-fexceptions" CC_HAS_FEXCEPTIONS)
- check_compiler_flag("-fno-semantic-interposition" CC_HAS_FNO_SEMANTIC_INTERPOSITION)
- if(CMAKE_CXX_COMPILER_LOADED)
- check_cxx_compiler_flag("-fcxx-exceptions" CC_HAS_FCXX_EXCEPTIONS)
- endif()
- check_compiler_flag("-funwind-tables" CC_HAS_FUNWIND_TABLES)
- check_compiler_flag("-fno-omit-frame-pointer" CC_HAS_FNO_OMIT_FRAME_POINTER)
- check_compiler_flag("-fno-common" CC_HAS_FNO_COMMON)
- check_compiler_flag("-ggdb" CC_HAS_GGDB)
- check_compiler_flag("-fvisibility=hidden" CC_HAS_VISIBILITY)
- check_compiler_flag("-march=native" CC_HAS_ARCH_NATIVE)
- check_compiler_flag("-Og" CC_HAS_DEBUG_FRIENDLY_OPTIMIZATION)
- check_compiler_flag("-Wall" CC_HAS_WALL)
- check_compiler_flag("-Ominimal" CC_HAS_OMINIMAL)
- check_compiler_flag("-ffunction-sections -fdata-sections" CC_HAS_SECTIONS)
- check_compiler_flag("-ffast-math" CC_HAS_FASTMATH)
- check_compiler_flag("-Wno-attributes" CC_HAS_WNO_ATTRIBUTES)
-
- # Check for an omp support
- set(CMAKE_REQUIRED_FLAGS "-fopenmp -Werror")
- if(CMAKE_CXX_COMPILER_LOADED)
- check_cxx_source_compiles("int main(void) {
- #pragma omp parallel
- return 0;
- }" HAVE_OPENMP)
- else()
- check_c_source_compiles("int main(void) {
- #pragma omp parallel
- return 0;
- }" HAVE_OPENMP)
- endif()
- set(CMAKE_REQUIRED_FLAGS "")
-endif()
-
-# Crutch for old C++ compilers and/or CMake to enabling C++11
-if(CMAKE_CXX_COMPILER_LOADED)
- list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_11 HAS_CXX11)
- if(HAS_CXX11 LESS 0)
- check_cxx_compiler_flag("-std=gnu++11" CXX_FALLBACK_GNU11)
- if(NOT CXX_FALLBACK_GNU11)
- check_cxx_compiler_flag("-std=c++11" CXX_FALLBACK_11)
- endif()
- endif()
-endif()
-
-# Crutch for mad C compilers and/or CMake to enabling C11
-if(CMAKE_C_COMPILER_LOADED)
- list(FIND CMAKE_C_COMPILE_FEATURES c_std_11 HAS_C11)
- if(HAS_C11 LESS 0)
- if (MSVC)
- check_c_compiler_flag("/std:c11" C_FALLBACK_11)
- else()
- check_c_compiler_flag("-std=gnu11" C_FALLBACK_GNU11)
- if(NOT C_FALLBACK_GNU11)
- check_c_compiler_flag("-std=c11" C_FALLBACK_11)
- endif()
- endif()
- endif()
-endif()
-
-# Check for LTO support by GCC
-if(CMAKE_COMPILER_IS_GNU${CMAKE_PRIMARY_LANG} AND NOT CMAKE_COMPILER_IS_ELBRUSC AND NOT CMAKE_COMPILER_IS_ELBRUSCXX)
- unset(gcc_collect)
- unset(gcc_lto_wrapper)
-
- if(NOT CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 4.7)
- execute_process(COMMAND ${CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER} -v
- OUTPUT_VARIABLE gcc_info_v ERROR_VARIABLE gcc_info_v)
-
- string(REGEX MATCH "^(.+\nCOLLECT_GCC=)([^ \n]+)(\n.+)$" gcc_collect_valid ${gcc_info_v})
- if(gcc_collect_valid)
- string(REGEX REPLACE "^(.+\nCOLLECT_GCC=)([^ \n]+)(\n.+)$" "\\2" gcc_collect ${gcc_info_v})
- endif()
-
- string(REGEX MATCH "^(.+\nCOLLECT_LTO_WRAPPER=)([^ \n]+/lto-wrapper)(\n.+)$" gcc_lto_wrapper_valid ${gcc_info_v})
- if(gcc_lto_wrapper_valid)
- string(REGEX REPLACE "^(.+\nCOLLECT_LTO_WRAPPER=)([^ \n]+/lto-wrapper)(\n.+)$" "\\2" gcc_lto_wrapper ${gcc_info_v})
- endif()
-
- set(gcc_suffix "")
- if(gcc_collect_valid AND gcc_collect)
- string(REGEX MATCH "^(.*(cc|\\+\\+))(-.+)$" gcc_suffix_valid ${gcc_collect})
- if(gcc_suffix_valid)
- string(REGEX REPLACE "^(.*(cc|\\+\\+))(-.+)$" "\\3" gcc_suffix ${gcc_collect})
- endif()
- endif()
-
- get_filename_component(gcc_dir ${CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER} DIRECTORY)
- if(NOT CMAKE_GCC_AR)
- find_program(CMAKE_GCC_AR NAMES "gcc${gcc_suffix}-ar" "gcc-ar${gcc_suffix}" PATHS "${gcc_dir}" NO_DEFAULT_PATH)
- endif()
- if(NOT CMAKE_GCC_NM)
- find_program(CMAKE_GCC_NM NAMES "gcc${gcc_suffix}-nm" "gcc-nm${gcc_suffix}" PATHS "${gcc_dir}" NO_DEFAULT_PATH)
- endif()
- if(NOT CMAKE_GCC_RANLIB)
- find_program(CMAKE_GCC_RANLIB NAMES "gcc${gcc_suffix}-ranlib" "gcc-ranlib${gcc_suffix}" PATHS "${gcc_dir}" NO_DEFAULT_PATH)
- endif()
-
- unset(gcc_dir)
- unset(gcc_suffix_valid)
- unset(gcc_suffix)
- unset(gcc_lto_wrapper_valid)
- unset(gcc_collect_valid)
- unset(gcc_collect)
- unset(gcc_info_v)
- endif()
-
- if(CMAKE_GCC_AR AND CMAKE_GCC_NM AND CMAKE_GCC_RANLIB AND gcc_lto_wrapper)
- message(STATUS "Found GCC's LTO toolset: ${gcc_lto_wrapper}, ${CMAKE_GCC_AR}, ${CMAKE_GCC_RANLIB}")
- set(GCC_LTO_CFLAGS "-flto -fno-fat-lto-objects -fuse-linker-plugin")
- set(GCC_LTO_AVAILABLE TRUE)
- message(STATUS "Link-Time Optimization by GCC is available")
- else()
- set(GCC_LTO_AVAILABLE FALSE)
- message(STATUS "Link-Time Optimization by GCC is NOT available")
- endif()
- unset(gcc_lto_wrapper)
-endif()
-
-# check for LTO by MSVC
-if(MSVC AND NOT CMAKE_COMPILER_IS_CLANG)
- if(NOT MSVC_VERSION LESS 1600)
- set(MSVC_LTO_AVAILABLE TRUE)
- message(STATUS "Link-Time Optimization by MSVC is available")
- else()
- set(MSVC_LTO_AVAILABLE FALSE)
- message(STATUS "Link-Time Optimization by MSVC is NOT available")
- endif()
-endif()
-
-# Check for LTO support by CLANG
-if(CMAKE_COMPILER_IS_CLANG)
- if(NOT CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 3.5)
- execute_process(COMMAND ${CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER} -print-search-dirs
- OUTPUT_VARIABLE clang_search_dirs RESULT_VARIABLE clang_probe_result ERROR_QUIET)
-
- unset(clang_bindirs)
- unset(clang_bindirs_x)
- unset(clang_libdirs)
- unset(clang_libdirs_x)
- if(clang_probe_result EQUAL 0)
- string(REGEX MATCH "(^|\n.*)(.*programs: =)([^\n]+)((\n.*)|$)" regexp_valid ${clang_search_dirs})
- if(regexp_valid)
- string(REGEX REPLACE "(^|\n.*)(.*programs: =)([^\n]+)((\n.*)|$)" "\\3" list ${clang_search_dirs})
- string(REPLACE ":" ";" list "${list}")
- foreach(dir IN LISTS list)
- get_filename_component(dir "${dir}" REALPATH)
- if(dir MATCHES ".*llvm.*" OR dir MATCHES ".*clang.*")
- list(APPEND clang_bindirs "${dir}")
- else()
- list(APPEND clang_bindirs_x "${dir}")
- endif()
- endforeach()
- list(APPEND clang_bindirs "${clang_bindirs_x}")
- list(REMOVE_DUPLICATES clang_bindirs)
- endif()
- string(REGEX MATCH "(^|\n.*)(.*libraries: =)([^\n]+)((\n.*)|$)" regexp_valid ${clang_search_dirs})
- if(regexp_valid)
- string(REGEX REPLACE "(^|\n.*)(.*libraries: =)([^\n]+)((\n.*)|$)" "\\3" list ${clang_search_dirs})
- string(REPLACE ":" ";" list "${list}")
- foreach(dir IN LISTS list)
- get_filename_component(dir "${dir}" REALPATH)
- if(dir MATCHES ".*llvm.*" OR dir MATCHES ".*clang.*")
- list(APPEND clang_libdirs "${dir}")
- else()
- list(APPEND clang_libdirs_x "${dir}")
- endif()
- endforeach()
- list(APPEND clang_libdirs "${clang_libdirs_x}")
- list(REMOVE_DUPLICATES clang_libdirs)
- endif()
- else()
- get_filename_component(clang_bindirs ${CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER} DIRECTORY)
- if(CMAKE_HOST_SYSTEM_NAME STREQUAL "Windows")
- set(clang_libdirs ${clang_bindirs})
- else()
- get_filename_component(clang_libdirs "${CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER}/../lib" REALPATH)
- endif()
- endif()
-
- if(clang_bindirs AND clang_libdirs)
- message(STATUS "Found CLANG/LLVM directories: ${clang_bindirs}, ${clang_libdirs}")
- else()
- message(STATUS "Could NOT find CLANG/LLVM directories (bin and/or lib).")
- endif()
-
- if(NOT CMAKE_CLANG_LD AND clang_bindirs)
- find_program(CMAKE_CLANG_LD NAMES lld-link ld.lld "ld${CMAKE_TARGET_BITNESS}.lld" lld llvm-link llvm-ld PATHS ${clang_bindirs} NO_DEFAULT_PATH)
- endif()
- if(NOT CMAKE_CLANG_AR AND clang_bindirs)
- find_program(CMAKE_CLANG_AR NAMES llvm-ar ar PATHS ${clang_bindirs} NO_DEFAULT_PATH)
- endif()
- if(NOT CMAKE_CLANG_NM AND clang_bindirs)
- find_program(CMAKE_CLANG_NM NAMES llvm-nm nm PATHS ${clang_bindirs} NO_DEFAULT_PATH)
- endif()
- if(NOT CMAKE_CLANG_RANLIB AND clang_bindirs)
- find_program(CMAKE_CLANG_RANLIB NAMES llvm-ranlib ranlib PATHS ${clang_bindirs} NO_DEFAULT_PATH)
- endif()
-
- set(clang_lto_plugin_name "LLVMgold${CMAKE_SHARED_LIBRARY_SUFFIX}")
- if(NOT CMAKE_LD_GOLD AND clang_bindirs)
- find_program(CMAKE_LD_GOLD NAMES ld.gold PATHS ${clang_bindirs})
- endif()
- if(NOT CLANG_LTO_PLUGIN AND clang_libdirs)
- find_file(CLANG_LTO_PLUGIN ${clang_lto_plugin_name} PATHS ${clang_libdirs} NO_DEFAULT_PATH)
- endif()
-
- if(CLANG_LTO_PLUGIN)
- message(STATUS "Found CLANG/LLVM's plugin for LTO: ${CLANG_LTO_PLUGIN}")
- else()
- message(STATUS "Could NOT find CLANG/LLVM's plugin (${clang_lto_plugin_name}) for LTO.")
- endif()
-
- if(CMAKE_CLANG_LD)
- message(STATUS "Found CLANG/LLVM's linker for LTO: ${CMAKE_CLANG_LD}")
- else()
- message(STATUS "Could NOT find CLANG/LLVM's linker (lld, llvm-ld, llvm-link) for LTO.")
- endif()
-
- if(CMAKE_CLANG_AR AND CMAKE_CLANG_RANLIB AND CMAKE_CLANG_NM)
- message(STATUS "Found CLANG/LLVM's binutils for LTO: ${CMAKE_CLANG_AR}, ${CMAKE_CLANG_RANLIB}, ${CMAKE_CLANG_NM}")
- else()
- message(STATUS "Could NOT find CLANG/LLVM's binutils (ar, ranlib, nm) for LTO.")
- endif()
-
- unset(clang_lto_plugin_name)
- unset(clang_libdir)
- unset(clang_bindir_valid)
- unset(clang_bindir)
- unset(clang_search_dirs)
- endif()
-
- if(CMAKE_CLANG_AR AND CMAKE_CLANG_NM AND CMAKE_CLANG_RANLIB
- AND ((CLANG_LTO_PLUGIN AND CMAKE_LD_GOLD)
- OR (CMAKE_CLANG_LD
- AND NOT (CMAKE_HOST_SYSTEM_NAME STREQUAL "Linux"
- AND CMAKE_SYSTEM_NAME STREQUAL "Linux"))
- OR APPLE))
- if(ANDROID AND CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 12)
- set(CLANG_LTO_AVAILABLE FALSE)
- message(STATUS "Link-Time Optimization by CLANG/LLVM is available but unusable due https://reviews.llvm.org/D79919")
- else()
- set(CLANG_LTO_AVAILABLE TRUE)
- message(STATUS "Link-Time Optimization by CLANG/LLVM is available")
- endif()
- elseif(CMAKE_TOOLCHAIN_FILE AND NOT CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 7.0)
- set(CLANG_LTO_AVAILABLE TRUE)
- if(NOT CMAKE_CLANG_LD)
- set(CMAKE_CLANG_LD ${CMAKE_LINKER})
- endif()
- if(NOT CMAKE_CLANG_AR)
- set(CMAKE_CLANG_AR ${CMAKE_AR})
- endif()
- if(NOT CMAKE_CLANG_NM)
- set(CMAKE_CLANG_NM ${CMAKE_NM})
- endif()
- if(NOT CMAKE_CLANG_RANLIB)
- set(CMAKE_CLANG_RANLIB ${CMAKE_RANLIB})
- endif()
- message(STATUS "Assume Link-Time Optimization by CLANG/LLVM is available via ${CMAKE_TOOLCHAIN_FILE}")
- else()
- set(CLANG_LTO_AVAILABLE FALSE)
- message(STATUS "Link-Time Optimization by CLANG/LLVM is NOT available")
- endif()
-endif()
-
-# Perform build type specific configuration.
-option(ENABLE_BACKTRACE "Enable output of fiber backtrace information in 'show
- fiber' administrative command. Only works on x86 architectures, if compiled
- with gcc. If GNU binutils and binutils-dev libraries are installed, backtrace
- is output with resolved function (symbol) names. Otherwise only frame
- addresses are printed." OFF)
-
-set(HAVE_BFD FALSE)
-if(ENABLE_BACKTRACE)
- if(NOT (X86_32 OR X86_64) OR NOT CMAKE_COMPILER_IS_GNU${CMAKE_PRIMARY_LANG})
- # We only know this option to work with gcc
- message(FATAL_ERROR "ENABLE_BACKTRACE option is set but the system
- is not x86 based (${CMAKE_SYSTEM_PROCESSOR}) or the compiler
- is not GNU GCC (${CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER}).")
- endif()
- # Use GNU bfd if present.
- find_library(BFD_LIBRARY NAMES libbfd.a)
- if(BFD_LIBRARY)
- check_library_exists(${BFD_LIBRARY} bfd_init "" HAVE_BFD_LIB)
- endif()
- find_library(IBERTY_LIBRARY NAMES libiberty.a)
- if(IBERTY_LIBRARY)
- check_library_exists(${IBERTY_LIBRARY} cplus_demangle "" HAVE_IBERTY_LIB)
- endif()
- set(CMAKE_REQUIRED_DEFINITIONS -DPACKAGE=${PACKAGE} -DPACKAGE_VERSION=${PACKAGE_VERSION})
- check_include_files(bfd.h HAVE_BFD_H)
- set(CMAKE_REQUIRED_DEFINITIONS)
- find_package(ZLIB)
- if(HAVE_BFD_LIB AND HAVE_BFD_H AND HAVE_IBERTY_LIB AND ZLIB_FOUND)
- set(HAVE_BFD ON)
- set(BFD_LIBRARIES ${BFD_LIBRARY} ${IBERTY_LIBRARY} ${ZLIB_LIBRARIES})
- find_package_message(BFD_LIBRARIES "Found libbfd and dependencies"
- ${BFD_LIBRARIES})
- if(TARGET_OS_FREEBSD AND NOT TARGET_OS_DEBIAN_FREEBSD)
- set(BFD_LIBRARIES ${BFD_LIBRARIES} iconv)
- endif()
- endif()
-endif()
-
-macro(setup_compile_flags)
- # save initial C/CXX flags
- if(NOT INITIAL_CMAKE_FLAGS_SAVED)
- if(CMAKE_CXX_COMPILER_LOADED)
- set(INITIAL_CMAKE_CXX_FLAGS ${CMAKE_CXX_FLAGS} CACHE STRING "Initial CMake's flags" FORCE)
- endif()
- if(CMAKE_C_COMPILER_LOADED)
- set(INITIAL_CMAKE_C_FLAGS ${CMAKE_C_FLAGS} CACHE STRING "Initial CMake's flags" FORCE)
- endif()
- set(INITIAL_CMAKE_EXE_LINKER_FLAGS ${CMAKE_EXE_LINKER_FLAGS} CACHE STRING "Initial CMake's flags" FORCE)
- set(INITIAL_CMAKE_SHARED_LINKER_FLAGS ${CMAKE_SHARED_LINKER_FLAGS} CACHE STRING "Initial CMake's flags" FORCE)
- set(INITIAL_CMAKE_STATIC_LINKER_FLAGS ${CMAKE_STATIC_LINKER_FLAGS} CACHE STRING "Initial CMake's flags" FORCE)
- set(INITIAL_CMAKE_MODULE_LINKER_FLAGS ${CMAKE_MODULE_LINKER_FLAGS} CACHE STRING "Initial CMake's flags" FORCE)
- set(INITIAL_CMAKE_FLAGS_SAVED TRUE CACHE INTERNAL "State of initial CMake's flags" FORCE)
- endif()
-
- # reset C/CXX flags
- if(CMAKE_CXX_COMPILER_LOADED)
- set(CXX_FLAGS ${INITIAL_CMAKE_CXX_FLAGS})
- # Crutch for old C++ compilers and/or CMake to enabling C++11
- if(CXX_FALLBACK_GNU11)
- add_compile_flags("CXX" "-std=gnu++11")
- elseif(CXX_FALLBACK_11)
- add_compile_flags("CXX" "-std=c++11")
- endif()
- endif()
- if(CMAKE_C_COMPILER_LOADED)
- set(C_FLAGS ${INITIAL_CMAKE_C_FLAGS})
- # Crutch for mad C compilers and/or CMake to enabling C11
- if(C_FALLBACK_GNU11)
- add_compile_flags("C" "-std=gnu11")
- elseif(C_FALLBACK_11)
- if(MSVC)
- add_compile_flags("C" "/std:c11")
- else()
- add_compile_flags("C" "-std=c11")
- endif()
- endif()
- endif()
- set(EXE_LINKER_FLAGS ${INITIAL_CMAKE_EXE_LINKER_FLAGS})
- set(SHARED_LINKER_FLAGS ${INITIAL_CMAKE_SHARED_LINKER_FLAGS})
- set(STATIC_LINKER_FLAGS ${INITIAL_CMAKE_STATIC_LINKER_FLAGS})
- set(MODULE_LINKER_FLAGS ${INITIAL_CMAKE_MODULE_LINKER_FLAGS})
-
- if(CC_HAS_FEXCEPTIONS)
- add_compile_flags("C;CXX" "-fexceptions")
- endif()
- if(CC_HAS_FCXX_EXCEPTIONS)
- add_compile_flags("CXX" "-fcxx-exceptions" "-frtti")
- endif()
- if(CC_HAS_FNO_SEMANTIC_INTERPOSITION AND NOT CMAKE_COMPILER_IS_CLANG)
- add_compile_flags("C;CXX" "-fno-semantic-interposition")
- endif()
- if(MSVC)
- # checks for /EHa or /clr options exists,
- # i.e. is enabled structured async WinNT exceptions
- string(REGEX MATCH "^(.* )*[-/]EHc*a( .*)*$" msvc_async_eh_enabled "${CXX_FLAGS}" "${C_FLAGS}")
- string(REGEX MATCH "^(.* )*[-/]clr( .*)*$" msvc_clr_enabled "${CXX_FLAGS}" "${C_FLAGS}")
- # remote any /EH? options
- string(REGEX REPLACE "( *[-/]-*EH[csa]+ *)+" "" CXX_FLAGS "${CXX_FLAGS}")
- string(REGEX REPLACE "( *[-/]-*EH[csa]+ *)+" "" C_FLAGS "${C_FLAGS}")
- if (msvc_clr_enabled STREQUAL "")
- if(NOT msvc_async_eh_enabled STREQUAL "")
- add_compile_flags("C;CXX" "/EHa")
- else()
- add_compile_flags("C;CXX" "/EHsc")
- endif()
- endif()
- endif(MSVC)
-
- if(CC_HAS_WNO_ATTRIBUTES AND CMAKE_COMPILER_IS_GNU${CMAKE_PRIMARY_LANG}
- AND CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 9)
- # GCC < 9.x generates false-positive warnings for optimization attributes
- add_compile_flags("C;CXX" "-Wno-attributes")
- if(LTO_ENABLED)
- add_compile_flags("C;CXX" "-Wno-lto-type-mismatch")
- endif()
- endif()
-
- # In C a global variable without a storage specifier (static/extern) and
- # without an initialiser is called a ’tentative definition’. The
- # language permits multiple tentative definitions in the single
- # translation unit; i.e. int foo; int foo; is perfectly ok. GNU
- # toolchain goes even further, allowing multiple tentative definitions
- # in *different* translation units. Internally, variables introduced via
- # tentative definitions are implemented as ‘common’ symbols. Linker
- # permits multiple definitions if they are common symbols, and it picks
- # one arbitrarily for inclusion in the binary being linked.
- #
- # -fno-common forces GNU toolchain to behave in a more
- # standard-conformant way in respect to tentative definitions and it
- # prevents common symbols generation. Since we are a cross-platform
- # project it really makes sense. There are toolchains that don’t
- # implement GNU style handling of the tentative definitions and there
- # are platforms lacking proper support for common symbols (osx).
- if(CC_HAS_FNO_COMMON)
- add_compile_flags("C;CXX" "-fno-common")
- endif()
-
- if(CC_HAS_GGDB)
- add_compile_flags("C;CXX" "-ggdb")
- endif()
-
- if(CC_HAS_WNO_UNKNOWN_PRAGMAS AND NOT HAVE_OPENMP)
- add_compile_flags("C;CXX" "-Wno-unknown-pragmas")
- endif()
-
- if(CC_HAS_SECTIONS)
- add_compile_flags("C;CXX" "-ffunction-sections" "-fdata-sections")
- elseif(MSVC)
- add_compile_flags("C;CXX" "/Gy")
- endif()
-
- # We must set -fno-omit-frame-pointer here, since we rely
- # on frame pointer when getting a backtrace, and it must
- # be used consistently across all object files.
- # The same reasoning applies to -fno-stack-protector switch.
- if(ENABLE_BACKTRACE)
- if(CC_HAS_FNO_OMIT_FRAME_POINTER)
- add_compile_flags("C;CXX" "-fno-omit-frame-pointer")
- endif()
- endif()
-
- if(MSVC)
- if(MSVC_VERSION LESS 1900)
- message(FATAL_ERROR "At least \"Microsoft C/C++ Compiler\" version 19.0.24234.1 (Visual Studio 2015 Update 3) is required.")
- endif()
- if(NOT MSVC_VERSION LESS 1910)
- add_compile_flags("CXX" "/Zc:__cplusplus")
- endif()
- remove_compile_flag("C;CXX" "/W3")
- add_compile_flags("C;CXX" "/W4 /utf-8")
- else()
- if(CC_HAS_WALL)
- add_compile_flags("C;CXX" "-Wall")
- endif()
- if(CC_HAS_WEXTRA)
- add_compile_flags("C;CXX" "-Wextra")
- endif()
- endif()
-
- add_definitions("-D__STDC_FORMAT_MACROS=1")
- add_definitions("-D__STDC_LIMIT_MACROS=1")
- add_definitions("-D__STDC_CONSTANT_MACROS=1")
- add_definitions("-D_HAS_EXCEPTIONS=1")
-
- # Only add -Werror if it's a debug build, done by developers.
- # Release builds should not cause extra trouble.
- if(CC_HAS_WERROR AND (CI OR CMAKE_CONFIGURATION_TYPES OR CMAKE_BUILD_TYPE STREQUAL "Debug"))
- if(MSVC)
- add_compile_flags("C;CXX" "/WX")
- elseif(CMAKE_COMPILER_IS_CLANG)
- if(NOT CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 6)
- add_compile_flags("C;CXX" "-Werror")
- endif()
- elseif(CMAKE_COMPILER_IS_GNUCC)
- if(NOT CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 6)
- add_compile_flags("C;CXX" "-Werror")
- endif()
- else()
- add_compile_flags("C;CXX" "-Werror")
- endif()
- endif()
-
-
- if(CMAKE_COMPILER_IS_GNU${CMAKE_PRIMARY_LANG}
- AND CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 5)
- # G++ bug. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31488
- add_compile_flags("CXX" "-Wno-invalid-offsetof")
- endif()
- if(MINGW)
- # Disable junk MINGW's warnings that issued due to incompatibilities
- # and shortcomings of MINGW,
- # since the code is checked by builds with GCC, CLANG and MSVC.
- add_compile_flags("C;CXX" "-Wno-format-extra-args" "-Wno-format" "-Wno-cast-function-type" "-Wno-implicit-fallthrough")
- endif()
-
- if(ENABLE_ASAN)
- add_compile_flags("C;CXX" "-fsanitize=address")
- endif()
-
- if(ENABLE_UBSAN)
- add_compile_flags("C;CXX" "-fsanitize=undefined" "-fsanitize-undefined-trap-on-error")
- endif()
-
- if(ENABLE_GCOV)
- if(NOT HAVE_GCOV)
- message(FATAL_ERROR
- "ENABLE_GCOV option requested but gcov library is not found")
- endif()
-
- add_compile_flags("C;CXX" "-fprofile-arcs" "-ftest-coverage")
- set(EXE_LINKER_FLAGS "${EXE_LINKER_FLAGS} -fprofile-arcs -ftest-coverage")
- set(SHARED_LINKER_FLAGS "${SHARED_LINKER_FLAGS} -fprofile-arcs -ftest-coverage")
- set(MODULE_LINKER_FLAGS "${MODULE_LINKER_FLAGS} -fprofile-arcs -ftest-coverage")
- # add_library(gcov SHARED IMPORTED)
- endif()
-
- if(ENABLE_GPROF)
- add_compile_flags("C;CXX" "-pg")
- endif()
-
- if(CMAKE_COMPILER_IS_GNUCC AND LTO_ENABLED)
- add_compile_flags("C;CXX" ${GCC_LTO_CFLAGS})
- set(EXE_LINKER_FLAGS "${EXE_LINKER_FLAGS} ${GCC_LTO_CFLAGS} -fverbose-asm -fwhole-program")
- set(SHARED_LINKER_FLAGS "${SHARED_LINKER_FLAGS} ${GCC_LTO_CFLAGS} -fverbose-asm")
- set(MODULE_LINKER_FLAGS "${MODULE_LINKER_FLAGS} ${GCC_LTO_CFLAGS} -fverbose-asm")
- if(CMAKE_CXX_COMPILER_VERSION VERSION_LESS 5)
- # Pass the same optimization flags to the linker
- set(compile_flags "${CMAKE_C_FLAGS} ${CMAKE_C_FLAGS_${CMAKE_BUILD_TYPE_UPPERCASE}}")
- set(EXE_LINKER_FLAGS "${EXE_LINKER_FLAGS} ${compile_flags}")
- set(SHARED_LINKER_FLAGS "${SHARED_LINKER_FLAGS} ${compile_flags}")
- set(MODULE_LINKER_FLAGS "${MODULE_LINKER_FLAGS} ${compile_flags}")
- unset(compile_flags)
- else()
- add_compile_flags("CXX" "-flto-odr-type-merging")
- endif()
- endif()
-
- if(MSVC AND NOT CMAKE_COMPILER_IS_CLANG AND LTO_ENABLED)
- add_compile_flags("C;CXX" "/GL")
- foreach(linkmode IN ITEMS EXE SHARED STATIC MODULE)
- set(${linkmode}_LINKER_FLAGS "${${linkmode}_LINKER_FLAGS} /LTCG")
- string(REGEX REPLACE "^(.*)(/INCREMENTAL)(:YES)?(:NO)?( ?.*)$" "\\1\\2:NO\\5" ${linkmode}_LINKER_FLAGS "${${linkmode}_LINKER_FLAGS}")
- string(STRIP "${${linkmode}_LINKER_FLAGS}" ${linkmode}_LINKER_FLAGS)
- foreach(config IN LISTS CMAKE_CONFIGURATION_TYPES ITEMS Release MinSizeRel RelWithDebInfo Debug)
- string(TOUPPER "${config}" config_uppercase)
- if(DEFINED "CMAKE_${linkmode}_LINKER_FLAGS_${config_uppercase}")
- string(REGEX REPLACE "^(.*)(/INCREMENTAL)(:YES)?(:NO)?( ?.*)$" "\\1\\2:NO\\5" altered_flags "${CMAKE_${linkmode}_LINKER_FLAGS_${config_uppercase}}")
- string(STRIP "${altered_flags}" altered_flags)
- if(NOT "${altered_flags}" STREQUAL "${CMAKE_${linkmode}_LINKER_FLAGS_${config_uppercase}}")
- set(CMAKE_${linkmode}_LINKER_FLAGS_${config_uppercase} "${altered_flags}" CACHE STRING "Altered: '/INCREMENTAL' removed for LTO" FORCE)
- endif()
- endif()
- endforeach(config)
- endforeach(linkmode)
- unset(linkmode)
-
- foreach(config IN LISTS CMAKE_CONFIGURATION_TYPES ITEMS Release MinSizeRel RelWithDebInfo)
- foreach(lang IN ITEMS C CXX)
- string(TOUPPER "${config}" config_uppercase)
- if(DEFINED "CMAKE_${lang}_FLAGS_${config_uppercase}")
- string(REPLACE "/O2" "/Ox" altered_flags "${CMAKE_${lang}_FLAGS_${config_uppercase}}")
- if(NOT "${altered_flags}" STREQUAL "${CMAKE_${lang}_FLAGS_${config_uppercase}}")
- set(CMAKE_${lang}_FLAGS_${config_uppercase} "${altered_flags}" CACHE STRING "Altered: '/O2' replaced by '/Ox' for LTO" FORCE)
- endif()
- endif()
- unset(config_uppercase)
- endforeach(lang)
- endforeach(config)
- unset(altered_flags)
- unset(lang)
- unset(config)
- endif()
-
- if(CMAKE_COMPILER_IS_CLANG AND OSX_ARCHITECTURES)
- set(EXE_LINKER_FLAGS "${EXE_LINKER_FLAGS} -Wl,-keep_dwarf_unwind")
- set(SHARED_LINKER_FLAGS "${SHARED_LINKER_FLAGS} -Wl,-keep_dwarf_unwind")
- set(MODULE_LINKER_FLAGS "${MODULE_LINKER_FLAGS} -Wl,-keep_dwarf_unwind")
- endif()
-
- if(CMAKE_COMPILER_IS_CLANG AND LTO_ENABLED)
- if(CMAKE_${CMAKE_PRIMARY_LANG}_COMPILER_VERSION VERSION_LESS 3.9)
- set(CLANG_LTO_FLAG "-flto")
- else()
- set(CLANG_LTO_FLAG "-flto=thin")
- endif()
- add_compile_flags("C;CXX" ${CLANG_LTO_FLAG})
- if(NOT MSVC)
- set(EXE_LINKER_FLAGS "${EXE_LINKER_FLAGS} ${CLANG_LTO_FLAG} -fverbose-asm -fwhole-program")
- set(SHARED_LINKER_FLAGS "${SHARED_LINKER_FLAGS} ${CLANG_LTO_FLAG} -fverbose-asm")
- set(MODULE_LINKER_FLAGS "${MODULE_LINKER_FLAGS} ${CLANG_LTO_FLAG} -fverbose-asm")
- endif()
- endif()
-
- # push C/CXX flags into the cache
- if(CMAKE_CXX_COMPILER_LOADED)
- set(CMAKE_CXX_FLAGS ${CXX_FLAGS} CACHE STRING "Flags used by the C++ compiler during all build types" FORCE)
- unset(CXX_FLAGS)
- endif()
- if(CMAKE_C_COMPILER_LOADED)
- set(CMAKE_C_FLAGS ${C_FLAGS} CACHE STRING "Flags used by the C compiler during all build types" FORCE)
- unset(C_FLAGS)
- endif()
- set(CMAKE_EXE_LINKER_FLAGS ${EXE_LINKER_FLAGS} CACHE STRING "Flags used by the linker" FORCE)
- set(CMAKE_SHARED_LINKER_FLAGS ${SHARED_LINKER_FLAGS} CACHE STRING "Flags used by the linker during the creation of dll's" FORCE)
- set(CMAKE_STATIC_LINKER_FLAGS ${STATIC_LINKER_FLAGS} CACHE STRING "Flags used by the linker during the creation of static libraries" FORCE)
- set(CMAKE_MODULE_LINKER_FLAGS ${MODULE_LINKER_FLAGS} CACHE STRING "Flags used by the linker during the creation of modules" FORCE)
- unset(EXE_LINKER_FLAGS)
- unset(SHARED_LINKER_FLAGS)
- unset(STATIC_LINKER_FLAGS)
- unset(MODULE_LINKER_FLAGS)
-endmacro(setup_compile_flags)
-
-macro(probe_libcxx_filesystem)
- if(CMAKE_CXX_COMPILER_LOADED AND NOT DEFINED LIBCXX_FILESYSTEM)
- list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_11 HAS_CXX11)
- if(NOT HAS_CXX11 LESS 0)
- include(CMakePushCheckState)
- include(CheckCXXSourceCompiles)
- cmake_push_check_state()
- set(stdfs_probe_save_libraries ${CMAKE_REQUIRED_LIBRARIES})
- set(stdfs_probe_save_flags ${CMAKE_REQUIRED_FLAGS})
- set(stdfs_probe_save_link_options ${CMAKE_REQUIRED_LINK_OPTIONS})
- unset(stdfs_probe_clear_cxx_standard)
- if(NOT DEFINED CMAKE_CXX_STANDARD)
- list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_14 HAS_CXX14)
- list(FIND CMAKE_CXX_COMPILE_FEATURES cxx_std_17 HAS_CXX17)
- if(NOT HAS_CXX17 LESS 0
- AND NOT (CMAKE_COMPILER_IS_CLANG AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 5))
- set(CMAKE_CXX_STANDARD 17)
- elseif(NOT HAS_CXX14 LESS 0)
- set(CMAKE_CXX_STANDARD 14)
- else()
- set(CMAKE_CXX_STANDARD 11)
- endif()
- set(stdfs_probe_clear_cxx_standard ON)
- endif()
- if(CMAKE_COMPILER_IS_ELBRUSCXX AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 1.25.23)
- if(CMAKE_VERSION VERSION_LESS 3.14)
- set(CMAKE_REQUIRED_FLAGS ${stdfs_probe_save_flags} "-Wl,--allow-multiple-definition")
- else()
- set(CMAKE_REQUIRED_LINK_OPTIONS ${stdfs_probe_save_link_options} "-Wl,--allow-multiple-definition")
- endif()
- endif()
-
- set(stdfs_probe_code [[
- #if defined(__SIZEOF_INT128__) && !defined(__GLIBCXX_TYPE_INT_N_0) && defined(__clang__) && __clang_major__ < 4
- #define __GLIBCXX_BITSIZE_INT_N_0 128
- #define __GLIBCXX_TYPE_INT_N_0 __int128
- #endif
-
- #ifndef __has_include
- #define __has_include(header) (0)
- #endif
- #if __has_include(<version>)
- #include <version>
- #endif
- #include <cstdlib>
- #include <string>
- #if defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L
- #include <string_view>
- #endif
-
- #if defined(__cpp_lib_filesystem) && __cpp_lib_filesystem >= 201703L
- #include <filesystem>
- #else
- #include <experimental/filesystem>
- #endif
-
- #if (defined(__cpp_lib_filesystem) && __cpp_lib_filesystem >= 201703L && (!defined(__MAC_OS_X_VERSION_MIN_REQUIRED) || __MAC_OS_X_VERSION_MIN_REQUIRED >= 101500) && (!defined(__IPHONE_OS_VERSION_MIN_REQUIRED) || __IPHONE_OS_VERSION_MIN_REQUIRED >= 130100))
- namespace fs = ::std::filesystem;
- #elif defined(__cpp_lib_experimental_filesystem) && __cpp_lib_experimental_filesystem >= 201406L
- namespace fs = ::std::experimental::filesystem;
- #endif
-
- int main(int argc, const char*argv[]) {
- fs::path probe(argv[0]);
- if (argc != 1) throw fs::filesystem_error(std::string("fake"), std::error_code());
- return fs::is_directory(probe.relative_path());
- }
- ]])
- set(LIBCXX_FILESYSTEM "")
-
- check_cxx_source_compiles("${stdfs_probe_code}" LIBCXX_FILESYSTEM_none)
- if(LIBCXX_FILESYSTEM_none)
- message(STATUS "No linking with additional library needed for std::filesystem")
- else()
- set(CMAKE_REQUIRED_LIBRARIES ${stdfs_probe_save_libraries} "stdc++fs")
- check_cxx_source_compiles("${stdfs_probe_code}" LIBCXX_FILESYSTEM_stdcxxfs)
- if(LIBCXX_FILESYSTEM_stdcxxfs)
- set(LIBCXX_FILESYSTEM "stdc++fs")
- message(STATUS "Linking with ${LIBCXX_FILESYSTEM} is required for std::filesystem")
- else()
- set(CMAKE_REQUIRED_LIBRARIES ${stdfs_probe_save_libraries} "c++fs")
- check_cxx_source_compiles("${stdfs_probe_code}" LIBCXX_FILESYSTEM_cxxfs)
- if(LIBCXX_FILESYSTEM_cxxfs)
- set(LIBCXX_FILESYSTEM "c++fs")
- message(STATUS "Linking with ${LIBCXX_FILESYSTEM} is required for std::filesystem")
- else()
- set(CMAKE_REQUIRED_LIBRARIES ${stdfs_probe_save_libraries} "c++experimental")
- check_cxx_source_compiles("${stdfs_probe_code}" LIBCXX_FILESYSTEM_cxxexperimental)
- if(LIBCXX_FILESYSTEM_cxxexperimental)
- set(LIBCXX_FILESYSTEM "c++experimental")
- message(STATUS "Linking with ${LIBCXX_FILESYSTEM} is required for std::filesystem")
- else()
- message(STATUS "No support for std::filesystem")
- endif()
- endif()
- endif()
- endif()
-
- set(CMAKE_REQUIRED_LINK_OPTIONS ${stdfs_probe_save_link_options})
- set(CMAKE_REQUIRED_FLAGS ${stdfs_probe_save_flags})
- set(CMAKE_REQUIRED_LIBRARIES ${stdfs_probe_save_libraries})
- if(stdfs_probe_clear_cxx_standard)
- unset(CMAKE_CXX_STANDARD)
- endif()
- unset(stdfs_probe_clear_cxx_standard)
- unset(stdfs_probe_save_link_options)
- unset(stdfs_probe_save_flags)
- unset(stdfs_probe_save_libraries)
- unset(stdfs_probe_code)
- unset(stdfs_probe_rc)
- cmake_pop_check_state()
- endif()
- endif()
-endmacro(probe_libcxx_filesystem)
-
-cmake_policy(POP)
diff --git a/3rdparty/libmdbx/cmake/profile.cmake b/3rdparty/libmdbx/cmake/profile.cmake
deleted file mode 100644
index 6fe9c8217fa5b054cc402a70b136ebc1f1f24e2b..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/cmake/profile.cmake
+++ /dev/null
@@ -1,53 +0,0 @@
-## Copyright (c) 2012-2022 Leonid Yuriev <leo@yuriev.ru>.
-##
-## Licensed under the Apache License, Version 2.0 (the "License");
-## you may not use this file except in compliance with the License.
-## You may obtain a copy of the License at
-##
-## http://www.apache.org/licenses/LICENSE-2.0
-##
-## Unless required by applicable law or agreed to in writing, software
-## distributed under the License is distributed on an "AS IS" BASIS,
-## WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-## See the License for the specific language governing permissions and
-## limitations under the License.
-##
-
-if(CMAKE_VERSION VERSION_LESS 3.12)
- cmake_minimum_required(VERSION 3.8.2)
-else()
- cmake_minimum_required(VERSION 3.12)
-endif()
-
-cmake_policy(PUSH)
-cmake_policy(VERSION ${CMAKE_MINIMUM_REQUIRED_VERSION})
-
-include(CheckLibraryExists)
-check_library_exists(gcov __gcov_flush "" HAVE_GCOV)
-
-option(ENABLE_GCOV
- "Enable integration with gcov, a code coverage program" OFF)
-
-option(ENABLE_GPROF
- "Enable integration with gprof, a performance analyzing tool" OFF)
-
-if(CMAKE_CXX_COMPILER_LOADED)
- include(CheckIncludeFileCXX)
- check_include_file_cxx(valgrind/memcheck.h HAVE_VALGRIND_MEMCHECK_H)
-else()
- include(CheckIncludeFile)
- check_include_file(valgrind/memcheck.h HAVE_VALGRIND_MEMCHECK_H)
-endif()
-
-option(MDBX_USE_VALGRIND "Enable integration with valgrind, a memory analyzing tool" OFF)
-if(MDBX_USE_VALGRIND AND NOT HAVE_VALGRIND_MEMCHECK_H)
- message(FATAL_ERROR "MDBX_USE_VALGRIND option is set but valgrind/memcheck.h is not found")
-endif()
-
-option(ENABLE_ASAN
- "Enable AddressSanitizer, a fast memory error detector based on compiler instrumentation" OFF)
-
-option(ENABLE_UBSAN
- "Enable UndefinedBehaviorSanitizer, a fast undefined behavior detector based on compiler instrumentation" OFF)
-
-cmake_policy(POP)
diff --git a/3rdparty/libmdbx/cmake/utils.cmake b/3rdparty/libmdbx/cmake/utils.cmake
deleted file mode 100644
index a461cc27d1ba0b57bb9564c107f790fad886abfe..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/cmake/utils.cmake
+++ /dev/null
@@ -1,284 +0,0 @@
-## Copyright (c) 2012-2022 Leonid Yuriev <leo@yuriev.ru>.
-##
-## Licensed under the Apache License, Version 2.0 (the "License");
-## you may not use this file except in compliance with the License.
-## You may obtain a copy of the License at
-##
-## http://www.apache.org/licenses/LICENSE-2.0
-##
-## Unless required by applicable law or agreed to in writing, software
-## distributed under the License is distributed on an "AS IS" BASIS,
-## WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-## See the License for the specific language governing permissions and
-## limitations under the License.
-##
-
-if(CMAKE_VERSION VERSION_LESS 3.12)
- cmake_minimum_required(VERSION 3.8.2)
-else()
- cmake_minimum_required(VERSION 3.12)
-endif()
-
-cmake_policy(PUSH)
-cmake_policy(VERSION ${CMAKE_MINIMUM_REQUIRED_VERSION})
-
-macro(add_compile_flags languages)
- foreach(_lang ${languages})
- string(REPLACE ";" " " _flags "${ARGN}")
- if(CMAKE_CXX_COMPILER_LOADED AND _lang STREQUAL "CXX")
- set("${_lang}_FLAGS" "${${_lang}_FLAGS} ${_flags}")
- endif()
- if(CMAKE_C_COMPILER_LOADED AND _lang STREQUAL "C")
- set("${_lang}_FLAGS" "${${_lang}_FLAGS} ${_flags}")
- endif()
- endforeach()
- unset(_lang)
- unset(_flags)
-endmacro(add_compile_flags)
-
-macro(remove_flag varname flag)
- string(REGEX REPLACE "^(.*)( ${flag} )(.*)$" "\\1 \\3" ${varname} ${${varname}})
- string(REGEX REPLACE "^((.+ )*)(${flag})(( .+)*)$" "\\1\\4" ${varname} ${${varname}})
-endmacro(remove_flag)
-
-macro(remove_compile_flag languages flag)
- foreach(_lang ${languages})
- if(CMAKE_CXX_COMPILER_LOADED AND _lang STREQUAL "CXX")
- remove_flag(${_lang}_FLAGS ${flag})
- endif()
- if(CMAKE_C_COMPILER_LOADED AND _lang STREQUAL "C")
- remove_flag(${_lang}_FLAGS ${flag})
- endif()
- endforeach()
- unset(_lang)
-endmacro(remove_compile_flag)
-
-macro(set_source_files_compile_flags)
- foreach(file ${ARGN})
- get_filename_component(_file_ext ${file} EXT)
- set(_lang "")
- if("${_file_ext}" STREQUAL ".m")
- set(_lang OBJC)
- # CMake believes that Objective C is a flavor of C++, not C,
- # and uses g++ compiler for .m files.
- # LANGUAGE property forces CMake to use CC for ${file}
- set_source_files_properties(${file} PROPERTIES LANGUAGE C)
- elseif("${_file_ext}" STREQUAL ".mm")
- set(_lang OBJCXX)
- endif()
-
- if(_lang)
- get_source_file_property(_flags ${file} COMPILE_FLAGS)
- if("${_flags}" STREQUAL "NOTFOUND")
- set(_flags "${CMAKE_${_lang}_FLAGS}")
- else()
- set(_flags "${_flags} ${CMAKE_${_lang}_FLAGS}")
- endif()
- # message(STATUS "Set (${file} ${_flags}")
- set_source_files_properties(${file} PROPERTIES COMPILE_FLAGS
- "${_flags}")
- endif()
- endforeach()
- unset(_file_ext)
- unset(_lang)
-endmacro(set_source_files_compile_flags)
-
-macro(fetch_version name source_root_directory parent_scope)
- set(${name}_VERSION "")
- set(${name}_GIT_DESCRIBE "")
- set(${name}_GIT_TIMESTAMP "")
- set(${name}_GIT_TREE "")
- set(${name}_GIT_COMMIT "")
- set(${name}_GIT_REVISION 0)
- set(${name}_GIT_VERSION "")
- if(GIT AND EXISTS "${source_root_directory}/.git")
- execute_process(COMMAND ${GIT} show --no-patch --format=%cI HEAD
- OUTPUT_VARIABLE ${name}_GIT_TIMESTAMP
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc OR "${name}_GIT_TIMESTAMP" STREQUAL "%cI")
- execute_process(COMMAND ${GIT} show --no-patch --format=%ci HEAD
- OUTPUT_VARIABLE ${name}_GIT_TIMESTAMP
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc OR "${name}_GIT_TIMESTAMP" STREQUAL "%ci")
- message(FATAL_ERROR "Please install latest version of git (`show --no-patch --format=%cI HEAD` failed)")
- endif()
- endif()
-
- execute_process(COMMAND ${GIT} show --no-patch --format=%T HEAD
- OUTPUT_VARIABLE ${name}_GIT_TREE
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc OR "${name}_GIT_TREE" STREQUAL "")
- message(FATAL_ERROR "Please install latest version of git (`show --no-patch --format=%T HEAD` failed)")
- endif()
-
- execute_process(COMMAND ${GIT} show --no-patch --format=%H HEAD
- OUTPUT_VARIABLE ${name}_GIT_COMMIT
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc OR "${name}_GIT_COMMIT" STREQUAL "")
- message(FATAL_ERROR "Please install latest version of git (`show --no-patch --format=%H HEAD` failed)")
- endif()
-
- execute_process(COMMAND ${GIT} rev-list --tags --count
- OUTPUT_VARIABLE tag_count
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc)
- message(FATAL_ERROR "Please install latest version of git (`git rev-list --tags --count` failed)")
- endif()
-
- if(tag_count EQUAL 0)
- execute_process(COMMAND ${GIT} rev-list --all --count
- OUTPUT_VARIABLE whole_count
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc)
- message(FATAL_ERROR "Please install latest version of git (`git rev-list --all --count` failed)")
- endif()
- if(whole_count GREATER 42)
- message(FATAL_ERROR "Please fetch tags (no any tags for ${whole_count} commits)")
- endif()
- set(${name}_GIT_VERSION "0;0;0")
- execute_process(COMMAND ${GIT} rev-list --count --all --no-merges
- OUTPUT_VARIABLE ${name}_GIT_REVISION
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc OR "${name}_GIT_REVISION" STREQUAL "")
- message(FATAL_ERROR "Please install latest version of git (`rev-list --count --all --no-merges` failed)")
- endif()
- else(tag_count EQUAL 0)
- execute_process(COMMAND ${GIT} describe --tags --long --dirty=-dirty "--match=v[0-9]*"
- OUTPUT_VARIABLE ${name}_GIT_DESCRIBE
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc OR "${name}_GIT_DESCRIBE" STREQUAL "")
- if(_whole_count GREATER 42)
- message(FATAL_ERROR "Please fetch tags (`describe --tags --long --dirty --match=v[0-9]*` failed)")
- else()
- execute_process(COMMAND ${GIT} describe --all --long --dirty=-dirty
- OUTPUT_VARIABLE ${name}_GIT_DESCRIBE
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc OR "${name}_GIT_DESCRIBE" STREQUAL "")
- message(FATAL_ERROR "Please install latest version of git (`git rev-list --tags --count` and/or `git rev-list --all --count` failed)")
- endif()
- endif()
- endif()
-
- execute_process(COMMAND ${GIT} describe --tags --abbrev=0 "--match=v[0-9]*"
- OUTPUT_VARIABLE last_release_tag
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc)
- message(FATAL_ERROR "Please install latest version of git (`describe --tags --abbrev=0 --match=v[0-9]*` failed)")
- endif()
- if (last_release_tag)
- set(git_revlist_arg "${last_release_tag}..HEAD")
- else()
- execute_process(COMMAND ${GIT} tag --sort=-version:refname
- OUTPUT_VARIABLE tag_list
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc)
- message(FATAL_ERROR "Please install latest version of git (`tag --sort=-version:refname` failed)")
- endif()
- string(REGEX REPLACE "\n" ";" tag_list "${tag_list}")
- set(git_revlist_arg "HEAD")
- foreach(tag IN LISTS tag_list)
- if(NOT last_release_tag)
- string(REGEX MATCH "^v[0-9]+(\.[0-9]+)+" last_release_tag "${tag}")
- set(git_revlist_arg "${tag}..HEAD")
- endif()
- endforeach(tag)
- endif()
- execute_process(COMMAND ${GIT} rev-list --count "${git_revlist_arg}"
- OUTPUT_VARIABLE ${name}_GIT_REVISION
- OUTPUT_STRIP_TRAILING_WHITESPACE
- WORKING_DIRECTORY ${source_root_directory}
- RESULT_VARIABLE rc)
- if(rc OR "${name}_GIT_REVISION" STREQUAL "")
- message(FATAL_ERROR "Please install latest version of git (`rev-list --count ${git_revlist_arg}` failed)")
- endif()
-
- string(REGEX MATCH "^(v)?([0-9]+)\\.([0-9]+)\\.([0-9]+)(.*)?" git_version_valid "${${name}_GIT_DESCRIBE}")
- if(git_version_valid)
- string(REGEX REPLACE "^(v)?([0-9]+)\\.([0-9]+)\\.([0-9]+)(.*)?" "\\2;\\3;\\4" ${name}_GIT_VERSION ${${name}_GIT_DESCRIBE})
- else()
- string(REGEX MATCH "^(v)?([0-9]+)\\.([0-9]+)(.*)?" git_version_valid "${${name}_GIT_DESCRIBE}")
- if(git_version_valid)
- string(REGEX REPLACE "^(v)?([0-9]+)\\.([0-9]+)(.*)?" "\\2;\\3;0" ${name}_GIT_VERSION ${${name}_GIT_DESCRIBE})
- else()
- message(AUTHOR_WARNING "Bad ${name} version \"${${name}_GIT_DESCRIBE}\"; falling back to 0.0.0 (have you made an initial release?)")
- set(${name}_GIT_VERSION "0;0;0")
- endif()
- endif()
- endif(tag_count EQUAL 0)
- endif()
-
- if(NOT ${name}_GIT_VERSION OR NOT ${name}_GIT_TIMESTAMP OR ${name}_GIT_REVISION STREQUAL "")
- if(GIT AND EXISTS "${source_root_directory}/.git")
- message(WARNING "Unable to retrieve ${name} version from git.")
- endif()
- set(${name}_GIT_VERSION "0;0;0;0")
- set(${name}_GIT_TIMESTAMP "")
- set(${name}_GIT_REVISION 0)
-
- # Try to get version from VERSION file
- set(version_file "${source_root_directory}/VERSION.txt")
- if(NOT EXISTS "${version_file}")
- set(version_file "${source_root_directory}/VERSION")
- endif()
- if(EXISTS "${version_file}")
- file(STRINGS "${version_file}" ${name}_VERSION LIMIT_COUNT 1 LIMIT_INPUT 42)
- endif()
-
- if(NOT ${name}_VERSION)
- message(WARNING "Unable to retrieve ${name} version from \"${version_file}\" file.")
- set(${name}_VERSION_LIST ${${name}_GIT_VERSION})
- string(REPLACE ";" "." ${name}_VERSION "${${name}_GIT_VERSION}")
- else()
- string(REPLACE "." ";" ${name}_VERSION_LIST ${${name}_VERSION})
- endif()
-
- else()
- list(APPEND ${name}_GIT_VERSION ${${name}_GIT_REVISION})
- set(${name}_VERSION_LIST ${${name}_GIT_VERSION})
- string(REPLACE ";" "." ${name}_VERSION "${${name}_GIT_VERSION}")
- endif()
-
- list(GET ${name}_VERSION_LIST 0 "${name}_VERSION_MAJOR")
- list(GET ${name}_VERSION_LIST 1 "${name}_VERSION_MINOR")
- list(GET ${name}_VERSION_LIST 2 "${name}_VERSION_RELEASE")
- list(GET ${name}_VERSION_LIST 3 "${name}_VERSION_REVISION")
-
- if(${parent_scope})
- set(${name}_VERSION_MAJOR "${${name}_VERSION_MAJOR}" PARENT_SCOPE)
- set(${name}_VERSION_MINOR "${${name}_VERSION_MINOR}" PARENT_SCOPE)
- set(${name}_VERSION_RELEASE "${${name}_VERSION_RELEASE}" PARENT_SCOPE)
- set(${name}_VERSION_REVISION "${${name}_VERSION_REVISION}" PARENT_SCOPE)
- set(${name}_VERSION "${${name}_VERSION}" PARENT_SCOPE)
-
- set(${name}_GIT_DESCRIBE "${${name}_GIT_DESCRIBE}" PARENT_SCOPE)
- set(${name}_GIT_TIMESTAMP "${${name}_GIT_TIMESTAMP}" PARENT_SCOPE)
- set(${name}_GIT_TREE "${${name}_GIT_TREE}" PARENT_SCOPE)
- set(${name}_GIT_COMMIT "${${name}_GIT_COMMIT}" PARENT_SCOPE)
- set(${name}_GIT_REVISION "${${name}_GIT_REVISION}" PARENT_SCOPE)
- set(${name}_GIT_VERSION "${${name}_GIT_VERSION}" PARENT_SCOPE)
- endif()
-endmacro(fetch_version)
-
-cmake_policy(POP)
diff --git a/3rdparty/libmdbx/config.h b/3rdparty/libmdbx/config.h
deleted file mode 100644
index b8983250d7e7eccd96e832d50ebb0aca5ad225d2..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/config.h
+++ /dev/null
@@ -1,63 +0,0 @@
-/* This is CMake-template for libmdbx's config.h
- ******************************************************************************/
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-/* #undef LTO_ENABLED */
-/* #undef MDBX_USE_VALGRIND */
-/* #undef ENABLE_GPROF */
-/* #undef ENABLE_GCOV */
-/* #undef ENABLE_ASAN */
-/* #undef ENABLE_UBSAN */
-#define MDBX_FORCE_ASSERTIONS 0
-
-/* Common */
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_ENV_CHECKPID_AUTO
-#ifndef MDBX_ENV_CHECKPID_AUTO
-#define MDBX_ENV_CHECKPID 0
-#endif
-#define MDBX_LOCKING_AUTO
-#ifndef MDBX_LOCKING_AUTO
-/* #undef MDBX_LOCKING */
-#endif
-#define MDBX_TRUST_RTC_AUTO
-#ifndef MDBX_TRUST_RTC_AUTO
-#define MDBX_TRUST_RTC 0
-#endif
-#define MDBX_DISABLE_PAGECHECKS 0
-
-/* Windows */
-#define MDBX_WITHOUT_MSVC_CRT 0
-
-/* MacOS & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY 0
-
-/* POSIX */
-#define MDBX_DISABLE_GNU_SOURCE 0
-#define MDBX_USE_OFDLOCKS_AUTO
-#ifndef MDBX_USE_OFDLOCKS_AUTO
-#define MDBX_USE_OFDLOCKS 0
-#endif
-
-/* Build Info */
-#ifndef MDBX_BUILD_TIMESTAMP
-#define MDBX_BUILD_TIMESTAMP "2022-11-07T16:21:48Z"
-#endif
-#ifndef MDBX_BUILD_TARGET
-#define MDBX_BUILD_TARGET "x86_64-ELF-Linux"
-#endif
-#ifndef MDBX_BUILD_TYPE
-#define MDBX_BUILD_TYPE "Debug"
-#endif
-#ifndef MDBX_BUILD_COMPILER
-#define MDBX_BUILD_COMPILER "gcc (Debian 10.2.1-6) 10.2.1 20210110"
-#endif
-#ifndef MDBX_BUILD_FLAGS
-#define MDBX_BUILD_FLAGS "-g MDBX_BUILD_SHARED_LIBRARY=0 -ffast-math -fvisibility=hidden"
-#endif
-/* #undef MDBX_BUILD_SOURCERY */
-
-/* *INDENT-ON* */
-/* clang-format on */
diff --git a/3rdparty/libmdbx/config.h.in b/3rdparty/libmdbx/config.h.in
deleted file mode 100644
index 7959699a302dbcf5422ead3a7c19d9faa49a8cff..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/config.h.in
+++ /dev/null
@@ -1,63 +0,0 @@
-/* This is CMake-template for libmdbx's config.h
- ******************************************************************************/
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#cmakedefine LTO_ENABLED
-#cmakedefine MDBX_USE_VALGRIND
-#cmakedefine ENABLE_GPROF
-#cmakedefine ENABLE_GCOV
-#cmakedefine ENABLE_ASAN
-#cmakedefine ENABLE_UBSAN
-#cmakedefine01 MDBX_FORCE_ASSERTIONS
-
-/* Common */
-#cmakedefine01 MDBX_TXN_CHECKOWNER
-#cmakedefine MDBX_ENV_CHECKPID_AUTO
-#ifndef MDBX_ENV_CHECKPID_AUTO
-#cmakedefine01 MDBX_ENV_CHECKPID
-#endif
-#cmakedefine MDBX_LOCKING_AUTO
-#ifndef MDBX_LOCKING_AUTO
-#cmakedefine MDBX_LOCKING @MDBX_LOCKING@
-#endif
-#cmakedefine MDBX_TRUST_RTC_AUTO
-#ifndef MDBX_TRUST_RTC_AUTO
-#cmakedefine01 MDBX_TRUST_RTC
-#endif
-#cmakedefine01 MDBX_DISABLE_PAGECHECKS
-
-/* Windows */
-#cmakedefine01 MDBX_WITHOUT_MSVC_CRT
-
-/* MacOS & iOS */
-#cmakedefine01 MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-
-/* POSIX */
-#cmakedefine01 MDBX_DISABLE_GNU_SOURCE
-#cmakedefine MDBX_USE_OFDLOCKS_AUTO
-#ifndef MDBX_USE_OFDLOCKS_AUTO
-#cmakedefine01 MDBX_USE_OFDLOCKS
-#endif
-
-/* Build Info */
-#ifndef MDBX_BUILD_TIMESTAMP
-#cmakedefine MDBX_BUILD_TIMESTAMP "@MDBX_BUILD_TIMESTAMP@"
-#endif
-#ifndef MDBX_BUILD_TARGET
-#cmakedefine MDBX_BUILD_TARGET "@MDBX_BUILD_TARGET@"
-#endif
-#ifndef MDBX_BUILD_TYPE
-#cmakedefine MDBX_BUILD_TYPE "@MDBX_BUILD_TYPE@"
-#endif
-#ifndef MDBX_BUILD_COMPILER
-#cmakedefine MDBX_BUILD_COMPILER "@MDBX_BUILD_COMPILER@"
-#endif
-#ifndef MDBX_BUILD_FLAGS
-#cmakedefine MDBX_BUILD_FLAGS "@MDBX_BUILD_FLAGS@"
-#endif
-#cmakedefine MDBX_BUILD_SOURCERY @MDBX_BUILD_SOURCERY@
-
-/* *INDENT-ON* */
-/* clang-format on */
diff --git a/3rdparty/libmdbx/man1/mdbx_chk.1 b/3rdparty/libmdbx/man1/mdbx_chk.1
deleted file mode 100644
index a4e335d930534958bc29786c850d5086d6bcbfb4..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/man1/mdbx_chk.1
+++ /dev/null
@@ -1,99 +0,0 @@
-.\" Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>.
-.\" Copying restrictions apply. See COPYRIGHT/LICENSE.
-.TH MDBX_CHK 1 "2022-04-22" "MDBX 0.11.7"
-.SH NAME
-mdbx_chk \- MDBX checking tool
-.SH SYNOPSIS
-.B mdbx_chk
-[\c
-.BR \-V ]
-[\c
-.BR \-v [ v [ v ]]]
-[\c
-.BR \-n ]
-[\c
-.BR \-q ]
-[\c
-.BR \-c ]
-[\c
-.BR \-w ]
-[\c
-.BR \-d ]
-[\c
-.BR \-i ]
-[\c
-.BI \-s \ subdb\fR]
-.BR \ dbpath
-.SH DESCRIPTION
-The
-.B mdbx_chk
-utility intended to check an MDBX database file.
-.SH OPTIONS
-.TP
-.BR \-V
-Write the library version number to the standard output, and exit.
-.TP
-.BR \-v
-Produce verbose output, including summarize space and page usage statistics.
-If \fB\-vv\fP is given, be more verbose, show summarized B-tree info
-and space allocation.
-If \fB\-vvv\fP is given, be more verbose, include summarized statistics
-of leaf B-tree pages.
-If \fB\-vvvv\fP is given, be even more verbose, show info of each page
-during B-tree traversal and basic info of each GC record.
-If \fB\-vvvvv\fP is given, turn maximal verbosity, display the full list
-of page IDs in the GC records and size of each key-value pair of database(s).
-.TP
-.BR \-q
-Be quiet; do not output anything even if an error was detected.
-.TP
-.BR \-c
-Force using cooperative mode while opening environment, i.e. don't try to open
-in exclusive/monopolistic mode. Only exclusive/monopolistic mode allow complete
-check, including full check of all meta-pages and actual size of database file.
-.TP
-.BR \-w
-Open environment in read-write mode and lock for writing while checking.
-This could be impossible if environment already used by another process(s)
-in an incompatible read-write mode. This allow rollback to last steady commit
-(in case environment was not closed properly) and then check transaction IDs
-of meta-pages. Otherwise, without \fB\-w\fP option environment will be
-opened in read-only mode.
-.TP
-.BR \-d
-Disable page-by-page traversal of B-tree. In this case, without B-tree
-traversal, it is unable to check for lost-unused pages nor for double-used
-pages.
-.TP
-.BR \-i
-Ignore wrong order errors, which will likely false-positive if custom
-comparator(s) was used.
-.TP
-.BR \-s \ subdb
-Verify and show info only for a specific subdatabase.
-.TP
-.BR \-0 | \-1 | \-2
-Using specific meta-page 0, or 2 for checking.
-.TP
-.BR \-t
-Turn to a specified meta-page on successful check.
-.TP
-.BR \-T
-Turn to a specified meta-page EVEN ON UNSUCCESSFUL CHECK!
-.TP
-.BR \-n
-Open MDBX environment(s) which do not use subdirectories.
-This is legacy option. For now MDBX handles this automatically.
-
-.SH DIAGNOSTICS
-Exit status is zero if no errors occur. Errors result in a non-zero exit status
-and a diagnostic message being written to standard error
-if no quiet mode was requested.
-.SH "SEE ALSO"
-.BR mdbx_stat (1),
-.BR mdbx_copy (1),
-.BR mdbx_dump (1),
-.BR mdbx_load (1)
-.BR mdbx_drop (1)
-.SH AUTHOR
-Leonid Yuriev <https://gitflic.ru/user/erthink>
diff --git a/3rdparty/libmdbx/man1/mdbx_copy.1 b/3rdparty/libmdbx/man1/mdbx_copy.1
deleted file mode 100644
index e06347f6fb0972bf270215789139b087f5ee7d42..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/man1/mdbx_copy.1
+++ /dev/null
@@ -1,68 +0,0 @@
-.\" Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>.
-.\" Copyright 2012-2015 Howard Chu, Symas Corp. All Rights Reserved.
-.\" Copyright 2015,2016 Peter-Service R&D LLC <http://billing.ru/>.
-.\" Copying restrictions apply. See COPYRIGHT/LICENSE.
-.TH MDBX_COPY 1 "2022-04-22" "MDBX 0.11.7"
-.SH NAME
-mdbx_copy \- MDBX environment copy tool
-.SH SYNOPSIS
-.B mdbx_copy
-[\c
-.BR \-V ]
-[\c
-.BR \-q ]
-[\c
-.BR \-c ]
-[\c
-.BR \-n ]
-.B src_path
-[\c
-.BR dest_path ]
-.SH DESCRIPTION
-The
-.B mdbx_copy
-utility copies an MDBX environment. The environment can
-be copied regardless of whether it is currently in use.
-No lockfile is created, since it gets recreated at need.
-
-If
-.I dest_path
-is specified it must be the path of an empty directory
-for storing the backup. Otherwise, the backup will be
-written to stdout.
-
-.SH OPTIONS
-.TP
-.BR \-V
-Write the library version number to the standard output, and exit.
-.TP
-.BR \-q
-Be quiet.
-.TP
-.BR \-c
-Compact while copying. Only current data pages will be copied; freed
-or unused pages will be omitted from the copy. This option will
-slow down the backup process as it is more CPU-intensive.
-Currently it fails if the environment has suffered a page leak.
-.TP
-.BR \-n
-Open MDBX environment(s) which do not use subdirectories.
-This is legacy option. For now MDBX handles this automatically.
-
-.SH DIAGNOSTICS
-Exit status is zero if no errors occur.
-Errors result in a non-zero exit status and
-a diagnostic message being written to standard error.
-.SH CAVEATS
-This utility can trigger significant file size growth if run
-in parallel with write transactions, because pages which they
-free during copying cannot be reused until the copy is done.
-.SH "SEE ALSO"
-.BR mdbx_dump (1),
-.BR mdbx_chk (1),
-.BR mdbx_stat (1),
-.BR mdbx_load (1)
-.BR mdbx_drop (1)
-.SH AUTHOR
-Howard Chu of Symas Corporation <http://www.symas.com>,
-Leonid Yuriev <https://gitflic.ru/user/erthink>
diff --git a/3rdparty/libmdbx/man1/mdbx_drop.1 b/3rdparty/libmdbx/man1/mdbx_drop.1
deleted file mode 100644
index b9577fcafa8f3b090436424886804efc6d9ffad0..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/man1/mdbx_drop.1
+++ /dev/null
@@ -1,48 +0,0 @@
-.\" Copyright 2021-2022 Leonid Yuriev <leo@yuriev.ru>.
-.\" Copyright 2014-2021 Howard Chu, Symas Corp. All Rights Reserved.
-.\" Copying restrictions apply. See COPYRIGHT/LICENSE.
-.TH MDBX_DROP 1 "2022-04-22" "MDBX 0.11.7"
-.SH NAME
-mdbx_drop \- MDBX database delete tool
-.SH SYNOPSIS
-.B mdbx_drop
-[\c
-.BR \-V ]
-[\c
-.BR \-d ]
-[\c
-.BI \-s \ subdb\fR]
-[\c
-.BR \-n ]
-.BR \ dbpath
-.SH DESCRIPTION
-The
-.B mdbx_drop
-utility empties or deletes a database in the specified
-environment.
-.SH OPTIONS
-.TP
-.BR \-V
-Write the library version number to the standard output, and exit.
-.TP
-.BR \-d
-Delete the specified database, don't just empty it.
-.TP
-.BR \-s \ subdb
-Operate on a specific subdatabase. If no database is specified, only the main database is dropped.
-.TP
-.BR \-n
-Dump an MDBX database which does not use subdirectories.
-This is legacy option. For now MDBX handles this automatically.
-
-.SH DIAGNOSTICS
-Exit status is zero if no errors occur.
-Errors result in a non-zero exit status and
-a diagnostic message being written to standard error.
-.SH "SEE ALSO"
-.BR mdbx_load (1),
-.BR mdbx_copy (1),
-.BR mdbx_chk (1),
-.BR mdbx_stat (1)
-.SH AUTHOR
-Howard Chu of Symas Corporation <http://www.symas.com>
diff --git a/3rdparty/libmdbx/man1/mdbx_dump.1 b/3rdparty/libmdbx/man1/mdbx_dump.1
deleted file mode 100644
index 469f13b65fb0ba83ecadb060388e5f3afa9dbb3d..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/man1/mdbx_dump.1
+++ /dev/null
@@ -1,94 +0,0 @@
-.\" Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>.
-.\" Copyright 2014-2015 Howard Chu, Symas Corp. All Rights Reserved.
-.\" Copyright 2015,2016 Peter-Service R&D LLC <http://billing.ru/>.
-.\" Copying restrictions apply. See COPYRIGHT/LICENSE.
-.TH MDBX_DUMP 1 "2022-04-22" "MDBX 0.11.7"
-.SH NAME
-mdbx_dump \- MDBX environment export tool
-.SH SYNOPSIS
-.B mdbx_dump
-[\c
-.BR \-V ]
-[\c
-.BR \-q ]
-[\c
-.BI \-f \ file\fR]
-[\c
-.BR \-l ]
-[\c
-.BR \-p ]
-[\c
-.BR \-a \ |
-.BI \-s \ subdb\fR]
-[\c
-.BR \-r ]
-[\c
-.BR \-n ]
-.BR \ dbpath
-.SH DESCRIPTION
-The
-.B mdbx_dump
-utility reads a database and writes its contents to the
-standard output using a portable flat-text format
-understood by the
-.BR mdbx_load (1)
-utility.
-.SH OPTIONS
-.TP
-.BR \-V
-Write the library version number to the standard output, and exit.
-.TP
-.BR \-q
-Be quiet.
-.TP
-.BR \-f \ file
-Write to the specified file instead of to the standard output.
-.TP
-.BR \-l
-List the databases stored in the environment. Just the
-names will be listed, no data will be output.
-.TP
-.BR \-p
-If characters in either the key or data items are printing characters (as
-defined by isprint(3)), output them directly. This option permits users to
-use standard text editors and tools to modify the contents of databases.
-
-Note: different systems may have different notions about what characters
-are considered printing characters, and databases dumped in this manner may
-be less portable to external systems.
-.TP
-.BR \-a
-Dump all of the subdatabases in the environment.
-.TP
-.BR \-s \ subdb
-Dump a specific subdatabase. If no database is specified, only the main database is dumped.
-.TP
-.BR \-r
-Rescure mode. Ignore some errors to dump corrupted DB.
-.TP
-.BR \-n
-Dump an MDBX database which does not use subdirectories.
-This is legacy option. For now MDBX handles this automatically.
-
-.SH DIAGNOSTICS
-Exit status is zero if no errors occur.
-Errors result in a non-zero exit status and
-a diagnostic message being written to standard error.
-
-Dumping and reloading databases that use user-defined comparison functions
-will result in new databases that use the default comparison functions.
-\fBIn this case it is quite likely that the reloaded database will be
-damaged beyond repair permitting neither record storage nor retrieval.\fP
-
-The only available workaround is to modify the source for the
-.BR mdbx_load (1)
-utility to load the database using the correct comparison functions.
-.SH "SEE ALSO"
-.BR mdbx_load (1),
-.BR mdbx_copy (1),
-.BR mdbx_chk (1),
-.BR mdbx_stat (1)
-.BR mdbx_drop (1)
-.SH AUTHOR
-Howard Chu of Symas Corporation <http://www.symas.com>,
-Leonid Yuriev <https://gitflic.ru/user/erthink>
diff --git a/3rdparty/libmdbx/man1/mdbx_load.1 b/3rdparty/libmdbx/man1/mdbx_load.1
deleted file mode 100644
index 850d4d1c0ee48cbdd4771f2a027fd165b0cb1ef6..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/man1/mdbx_load.1
+++ /dev/null
@@ -1,105 +0,0 @@
-.\" Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>.
-.\" Copyright 2014-2015 Howard Chu, Symas Corp. All Rights Reserved.
-.\" Copyright 2015,2016 Peter-Service R&D LLC <http://billing.ru/>.
-.\" Copying restrictions apply. See COPYRIGHT/LICENSE.
-.TH MDBX_LOAD 1 "2022-04-22" "MDBX 0.11.7"
-.SH NAME
-mdbx_load \- MDBX environment import tool
-.SH SYNOPSIS
-.B mdbx_load
-[\c
-.BR \-V ]
-[\c
-.BR \-q ]
-[\c
-.BR \-a ]
-[\c
-.BI \-f \ file\fR]
-[\c
-.BI \-s \ subdb\fR]
-[\c
-.BR \-N ]
-[\c
-.BR \-T ]
-[\c
-.BR \-r ]
-[\c
-.BR \-n ]
-.BR \ dbpath
-.SH DESCRIPTION
-The
-.B mdbx_load
-utility reads from the standard input and loads it into the
-MDBX environment
-.BR dbpath .
-
-The input to
-.B mdbx_load
-must be in the output format specified by the
-.BR mdbx_dump (1)
-utility or as specified by the
-.B -T
-option below.
-
-A simple escape mechanism, where newline and backslash (\\) characters are special, is
-applied to the text input. Newline characters are interpreted as record separators.
-Backslash characters in the text will be interpreted in one of two ways: If the backslash
-character precedes another backslash character, the pair will be interpreted as a literal
-backslash. If the backslash character precedes any other character, the two characters
-following the backslash will be interpreted as a hexadecimal specification of a single
-character; for example, \\0a is a newline character in the ASCII character set.
-
-For this reason, any backslash or newline characters that naturally occur in the text
-input must be escaped to avoid misinterpretation by
-.BR mdbx_load .
-
-.SH OPTIONS
-.TP
-.BR \-V
-Write the library version number to the standard output, and exit.
-.TP
-.BR \-q
-Be quiet.
-.TP
-.BR \-a
-Append all records in the order they appear in the input. The input is assumed to already be
-in correctly sorted order and no sorting or checking for redundant values will be performed.
-This option must be used to reload data that was produced by running
-.B mdbx_dump
-on a database that uses custom compare functions.
-.TP
-.BR \-f \ file
-Read from the specified file instead of from the standard input.
-.TP
-.BR \-s \ subdb
-Load a specific subdatabase. If no database is specified, data is loaded into the main database.
-.TP
-.BR \-N
-Don't overwrite existing records when loading into an already existing database; just skip them.
-.TP
-.BR \-T
-Load data from simple text files. The input must be paired lines of text, where the first
-line of the pair is the key item, and the second line of the pair is its corresponding
-data item.
-.TP
-.BR \-r
-Rescure mode. Ignore errors to load corrupted DB dump.
-.TP
-.BR \-n
-Load an MDBX database which does not use subdirectories.
-This is legacy option. For now MDBX handles this automatically.
-
-.SH DIAGNOSTICS
-Exit status is zero if no errors occur.
-Errors result in a non-zero exit status and
-a diagnostic message being written to standard error.
-
-.SH "SEE ALSO"
-.BR mdbx_dump (1),
-.BR mdbx_chk (1),
-.BR mdbx_stat (1),
-.BR mdbx_copy (1)
-.BR mdbx_drop (1)
-.SH AUTHOR
-Howard Chu of Symas Corporation <http://www.symas.com>,
-Leonid Yuriev <https://gitflic.ru/user/erthink>
diff --git a/3rdparty/libmdbx/man1/mdbx_stat.1 b/3rdparty/libmdbx/man1/mdbx_stat.1
deleted file mode 100644
index 8055d6e1dd7c08b39a8e6f1cab52089bf22d9cec..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/man1/mdbx_stat.1
+++ /dev/null
@@ -1,86 +0,0 @@
-.\" Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>.
-.\" Copyright 2012-2015 Howard Chu, Symas Corp. All Rights Reserved.
-.\" Copyright 2015,2016 Peter-Service R&D LLC <http://billing.ru/>.
-.\" Copying restrictions apply. See COPYRIGHT/LICENSE.
-.TH MDBX_STAT 1 "2022-04-22" "MDBX 0.11.7"
-.SH NAME
-mdbx_stat \- MDBX environment status tool
-.SH SYNOPSIS
-.B mdbx_stat
-[\c
-.BR \-V ]
-[\c
-.BR \-q ]
-[\c
-.BR \-p ]
-[\c
-.BR \-e ]
-[\c
-.BR \-f [ f [ f ]]]
-[\c
-.BR \-r [ r ]]
-[\c
-.BR \-a \ |
-.BI \-s \ subdb\fR]
-.BR \ dbpath
-[\c
-.BR \-n ]
-.SH DESCRIPTION
-The
-.B mdbx_stat
-utility displays the status of an MDBX environment.
-.SH OPTIONS
-.TP
-.BR \-V
-Write the library version number to the standard output, and exit.
-.TP
-.BR \-q
-Be quiet.
-.TP
-.BR \-p
-Display overall statistics of page operations of all (running, completed
-and aborted) transactions in the current multi-process session (since the
-first process opened the database after everyone had previously closed it).
-.TP
-.BR \-e
-Display information about the database environment.
-.TP
-.BR \-f
-Display information about the environment GC.
-If \fB\-ff\fP is given, summarize each GC/freelist entry.
-If \fB\-fff\fP is given, display the full list of page IDs in the GC/freelist.
-.TP
-.BR \-r
-Display information about the environment reader table.
-Shows the process ID, thread ID, and transaction ID for each active
-reader slot. The process ID and transaction ID are in decimal, the
-thread ID is in hexadecimal. The transaction ID is displayed as "-"
-if the reader does not currently have a read transaction open.
-If \fB\-rr\fP is given, check for stale entries in the reader
-table and clear them. The reader table will be printed again
-after the check is performed.
-.TP
-.BR \-a
-Display the status of all of the subdatabases in the environment.
-.TP
-.BR \-s \ subdb
-Display the status of a specific subdatabase.
-.TP
-.BR \-n
-Display the status of an MDBX database which does not use subdirectories.
-This is legacy option. For now MDBX handles this automatically
-for existing databases, but may be required while creating new.
-
-.SH DIAGNOSTICS
-Exit status is zero if no errors occur.
-Errors result in a non-zero exit status and
-a diagnostic message being written to standard error.
-.SH "SEE ALSO"
-.BR mdbx_chk (1),
-.BR mdbx_copy (1),
-.BR mdbx_dump (1),
-.BR mdbx_load (1)
-.BR mdbx_drop (1)
-.SH AUTHOR
-Howard Chu of Symas Corporation <http://www.symas.com>,
-Leonid Yuriev <https://gitflic.ru/user/erthink>
diff --git a/3rdparty/libmdbx/mdbx.c b/3rdparty/libmdbx/mdbx.c
deleted file mode 100644
index a6db608ee4d17cc4a67e0cc44342f5658c34f42a..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx.c
+++ /dev/null
@@ -1,30924 +0,0 @@
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#define xMDBX_ALLOY 1
-#define MDBX_BUILD_SOURCERY c510b93d78b5848764f976ea67f06d278a7b99f141b4c98ceb095285f834f31c_v0_11_7_0_g40ec559c
-#ifdef MDBX_CONFIG_H
-#include MDBX_CONFIG_H
-#endif
-
-#define LIBMDBX_INTERNALS
-#ifdef xMDBX_TOOLS
-#define MDBX_DEPRECATED
-#endif /* xMDBX_TOOLS */
-
-#ifdef xMDBX_ALLOY
-/* Amalgamated build */
-#define MDBX_INTERNAL_FUNC static
-#define MDBX_INTERNAL_VAR static
-#else
-/* Non-amalgamated build */
-#define MDBX_INTERNAL_FUNC
-#define MDBX_INTERNAL_VAR extern
-#endif /* xMDBX_ALLOY */
-
-/** Disables using GNU/Linux libc extensions.
- * \ingroup build_option
- * \note This option couldn't be moved to the options.h since dependant
- * control macros/defined should be prepared before include the options.h */
-#ifndef MDBX_DISABLE_GNU_SOURCE
-#define MDBX_DISABLE_GNU_SOURCE 0
-#endif
-#if MDBX_DISABLE_GNU_SOURCE
-#undef _GNU_SOURCE
-#elif (defined(__linux__) || defined(__gnu_linux__)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif /* MDBX_DISABLE_GNU_SOURCE */
-
-/*----------------------------------------------------------------------------*/
-
-/* Should be defined before any includes */
-#if !defined(_FILE_OFFSET_BITS) && !defined(__ANDROID_API__) && \
- !defined(ANDROID)
-#define _FILE_OFFSET_BITS 64
-#endif
-
-#ifdef __APPLE__
-#define _DARWIN_C_SOURCE
-#endif
-
-#ifdef _MSC_VER
-#if _MSC_FULL_VER < 190024234
-/* Actually libmdbx was not tested with compilers older than 19.00.24234 (Visual
- * Studio 2015 Update 3). But you could remove this #error and try to continue
- * at your own risk. In such case please don't rise up an issues related ONLY to
- * old compilers.
- *
- * NOTE:
- * Unfortunately, there are several different builds of "Visual Studio" that
- * are called "Visual Studio 2015 Update 3".
- *
- * The 190024234 is used here because it is minimal version of Visual Studio
- * that was used for build and testing libmdbx in recent years. Soon this
- * value will be increased to 19.0.24241.7, since build and testing using
- * "Visual Studio 2015" will be performed only at https://ci.appveyor.com.
- *
- * Please ask Microsoft (but not us) for information about version differences
- * and how to and where you can obtain the latest "Visual Studio 2015" build
- * with all fixes.
- */
-#error \
- "At least \"Microsoft C/C++ Compiler\" version 19.00.24234 (Visual Studio 2015 Update 3) is required."
-#endif
-#ifndef _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#if _MSC_VER > 1913
-#pragma warning(disable : 5045) /* Compiler will insert Spectre mitigation... \
- */
-#endif
-#if _MSC_VER > 1914
-#pragma warning( \
- disable : 5105) /* winbase.h(9531): warning C5105: macro expansion \
- producing 'defined' has undefined behavior */
-#endif
-#pragma warning(disable : 4710) /* 'xyz': function not inlined */
-#pragma warning(disable : 4711) /* function 'xyz' selected for automatic \
- inline expansion */
-#pragma warning( \
- disable : 4201) /* nonstandard extension used : nameless struct / union */
-#pragma warning(disable : 4702) /* unreachable code */
-#pragma warning(disable : 4706) /* assignment within conditional expression */
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4324) /* 'xyz': structure was padded due to \
- alignment specifier */
-#pragma warning(disable : 4310) /* cast truncates constant value */
-#pragma warning( \
- disable : 4820) /* bytes padding added after data member for alignment */
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4366) /* the result of the unary '&' operator may be \
- unaligned */
-#pragma warning(disable : 4200) /* nonstandard extension used: zero-sized \
- array in struct/union */
-#pragma warning(disable : 4204) /* nonstandard extension used: non-constant \
- aggregate initializer */
-#pragma warning( \
- disable : 4505) /* unreferenced local function has been removed */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(__GNUC__) && __GNUC__ < 9
-#pragma GCC diagnostic ignored "-Wattributes"
-#endif /* GCC < 9 */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-#include "mdbx.h"
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#ifndef __GNUC_PREREQ
-# if defined(__GNUC__) && defined(__GNUC_MINOR__)
-# define __GNUC_PREREQ(maj, min) \
- ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GNUC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GNUC_PREREQ */
-
-#ifndef __CLANG_PREREQ
-# ifdef __clang__
-# define __CLANG_PREREQ(maj,min) \
- ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min))
-# else
-# define __CLANG_PREREQ(maj,min) (0)
-# endif
-#endif /* __CLANG_PREREQ */
-
-#ifndef __GLIBC_PREREQ
-# if defined(__GLIBC__) && defined(__GLIBC_MINOR__)
-# define __GLIBC_PREREQ(maj, min) \
- ((__GLIBC__ << 16) + __GLIBC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GLIBC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GLIBC_PREREQ */
-
-#ifndef __has_warning
-# define __has_warning(x) (0)
-#endif
-
-#ifndef __has_include
-# define __has_include(x) (0)
-#endif
-
-#if __has_feature(thread_sanitizer)
-# define __SANITIZE_THREAD__ 1
-#endif
-
-#if __has_feature(address_sanitizer)
-# define __SANITIZE_ADDRESS__ 1
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __extern_C
-# ifdef __cplusplus
-# define __extern_C extern "C"
-# else
-# define __extern_C
-# endif
-#endif /* __extern_C */
-
-#if !defined(nullptr) && !defined(__cplusplus) || (__cplusplus < 201103L && !defined(_MSC_VER))
-# define nullptr NULL
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __always_inline
-# if defined(__GNUC__) || __has_attribute(__always_inline__)
-# define __always_inline __inline __attribute__((__always_inline__))
-# elif defined(_MSC_VER)
-# define __always_inline __forceinline
-# else
-# define __always_inline
-# endif
-#endif /* __always_inline */
-
-#ifndef __noinline
-# if defined(__GNUC__) || __has_attribute(__noinline__)
-# define __noinline __attribute__((__noinline__))
-# elif defined(_MSC_VER)
-# define __noinline __declspec(noinline)
-# else
-# define __noinline
-# endif
-#endif /* __noinline */
-
-#ifndef __must_check_result
-# if defined(__GNUC__) || __has_attribute(__warn_unused_result__)
-# define __must_check_result __attribute__((__warn_unused_result__))
-# else
-# define __must_check_result
-# endif
-#endif /* __must_check_result */
-
-#if !defined(__noop) && !defined(_MSC_VER)
-# define __noop(...) do {} while(0)
-#endif /* __noop */
-
-#if defined(__fallthrough) && \
- (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__))
-#undef __fallthrough
-#endif /* __fallthrough workaround for MinGW */
-
-#ifndef __fallthrough
-# if defined(__cplusplus) && (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || __cplusplus >= 201703L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(8, 0) && defined(__cplusplus) && __cplusplus >= 201103L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(7, 0) && \
- (!defined(__LCC__) || (__LCC__ == 124 && __LCC_MINOR__ >= 12) || \
- (__LCC__ == 125 && __LCC_MINOR__ >= 5) || (__LCC__ >= 126))
-# define __fallthrough __attribute__((__fallthrough__))
-# elif defined(__clang__) && defined(__cplusplus) && __cplusplus >= 201103L &&\
- __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
-# define __fallthrough [[clang::fallthrough]]
-# else
-# define __fallthrough
-# endif
-#endif /* __fallthrough */
-
-#ifndef __unreachable
-# if __GNUC_PREREQ(4,5) || __has_builtin(__builtin_unreachable)
-# define __unreachable() __builtin_unreachable()
-# elif defined(_MSC_VER)
-# define __unreachable() __assume(0)
-# else
-# define __unreachable() __noop()
-# endif
-#endif /* __unreachable */
-
-#ifndef __prefetch
-# if defined(__GNUC__) || defined(__clang__) || __has_builtin(__builtin_prefetch)
-# define __prefetch(ptr) __builtin_prefetch(ptr)
-# else
-# define __prefetch(ptr) __noop(ptr)
-# endif
-#endif /* __prefetch */
-
-#ifndef __nothrow
-# if defined(__cplusplus)
-# if __cplusplus < 201703L
-# define __nothrow throw()
-# else
-# define __nothrow noexcept(true)
-# endif /* __cplusplus */
-# elif defined(__GNUC__) || __has_attribute(__nothrow__)
-# define __nothrow __attribute__((__nothrow__))
-# elif defined(_MSC_VER) && defined(__cplusplus)
-# define __nothrow __declspec(nothrow)
-# else
-# define __nothrow
-# endif
-#endif /* __nothrow */
-
-#ifndef __hidden
-# if defined(__GNUC__) || __has_attribute(__visibility__)
-# define __hidden __attribute__((__visibility__("hidden")))
-# else
-# define __hidden
-# endif
-#endif /* __hidden */
-
-#ifndef __optimize
-# if defined(__OPTIMIZE__)
-# if (defined(__GNUC__) && !defined(__clang__)) || __has_attribute(__optimize__)
-# define __optimize(ops) __attribute__((__optimize__(ops)))
-# else
-# define __optimize(ops)
-# endif
-# else
-# define __optimize(ops)
-# endif
-#endif /* __optimize */
-
-#ifndef __hot
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __hot __attribute__((__hot__)) __optimize(3)
-# elif defined(__clang__) && !__has_attribute(__hot_) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put frequently used functions in separate section */
-# define __hot __attribute__((__section__("text.hot"))) __optimize("O3")
-# elif defined(__GNUC__) || __has_attribute(__hot__)
-# define __hot __attribute__((__hot__)) __optimize("O3")
-# else
-# define __hot __optimize("O3")
-# endif
-# else
-# define __hot
-# endif
-#endif /* __hot */
-
-#ifndef __cold
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __cold __attribute__((__cold__)) __optimize(1)
-# elif defined(__clang__) && !__has_attribute(cold) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put infrequently used functions in separate section */
-# define __cold __attribute__((__section__("text.unlikely"))) __optimize("Os")
-# elif defined(__GNUC__) || __has_attribute(cold)
-# define __cold __attribute__((__cold__)) __optimize("Os")
-# else
-# define __cold __optimize("Os")
-# endif
-# else
-# define __cold
-# endif
-#endif /* __cold */
-
-#ifndef __flatten
-# if defined(__OPTIMIZE__) && (defined(__GNUC__) || __has_attribute(__flatten__))
-# define __flatten __attribute__((__flatten__))
-# else
-# define __flatten
-# endif
-#endif /* __flatten */
-
-#ifndef likely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define likely(cond) __builtin_expect(!!(cond), 1)
-# else
-# define likely(x) (!!(x))
-# endif
-#endif /* likely */
-
-#ifndef unlikely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define unlikely(cond) __builtin_expect(!!(cond), 0)
-# else
-# define unlikely(x) (!!(x))
-# endif
-#endif /* unlikely */
-
-#ifndef __anonymous_struct_extension__
-# if defined(__GNUC__)
-# define __anonymous_struct_extension__ __extension__
-# else
-# define __anonymous_struct_extension__
-# endif
-#endif /* __anonymous_struct_extension__ */
-
-#ifndef __Wpedantic_format_voidptr
- MDBX_MAYBE_UNUSED MDBX_PURE_FUNCTION static __inline const void*
- __Wpedantic_format_voidptr(const void* ptr) {return ptr;}
-# define __Wpedantic_format_voidptr(ARG) __Wpedantic_format_voidptr(ARG)
-#endif /* __Wpedantic_format_voidptr */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(MDBX_USE_VALGRIND)
-# include <valgrind/memcheck.h>
-# ifndef VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE
- /* LY: available since Valgrind 3.10 */
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# endif
-#elif !defined(RUNNING_ON_VALGRIND)
-# define VALGRIND_CREATE_MEMPOOL(h,r,z)
-# define VALGRIND_DESTROY_MEMPOOL(h)
-# define VALGRIND_MEMPOOL_TRIM(h,a,s)
-# define VALGRIND_MEMPOOL_ALLOC(h,a,s)
-# define VALGRIND_MEMPOOL_FREE(h,a)
-# define VALGRIND_MEMPOOL_CHANGE(h,a,b,s)
-# define VALGRIND_MAKE_MEM_NOACCESS(a,s)
-# define VALGRIND_MAKE_MEM_DEFINED(a,s)
-# define VALGRIND_MAKE_MEM_UNDEFINED(a,s)
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_CHECK_MEM_IS_ADDRESSABLE(a,s) (0)
-# define VALGRIND_CHECK_MEM_IS_DEFINED(a,s) (0)
-# define RUNNING_ON_VALGRIND (0)
-#endif /* MDBX_USE_VALGRIND */
-
-#ifdef __SANITIZE_ADDRESS__
-# include <sanitizer/asan_interface.h>
-#elif !defined(ASAN_POISON_MEMORY_REGION)
-# define ASAN_POISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-# define ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-#endif /* __SANITIZE_ADDRESS__ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef ARRAY_LENGTH
-# ifdef __cplusplus
- template <typename T, size_t N>
- char (&__ArraySizeHelper(T (&array)[N]))[N];
-# define ARRAY_LENGTH(array) (sizeof(::__ArraySizeHelper(array)))
-# else
-# define ARRAY_LENGTH(array) (sizeof(array) / sizeof(array[0]))
-# endif
-#endif /* ARRAY_LENGTH */
-
-#ifndef ARRAY_END
-# define ARRAY_END(array) (&array[ARRAY_LENGTH(array)])
-#endif /* ARRAY_END */
-
-#define CONCAT(a,b) a##b
-#define XCONCAT(a,b) CONCAT(a,b)
-
-#ifndef offsetof
-# define offsetof(type, member) __builtin_offsetof(type, member)
-#endif /* offsetof */
-
-#ifndef container_of
-# define container_of(ptr, type, member) \
- ((type *)((char *)(ptr) - offsetof(type, member)))
-#endif /* container_of */
-
-#define MDBX_TETRAD(a, b, c, d) \
- ((uint32_t)(a) << 24 | (uint32_t)(b) << 16 | (uint32_t)(c) << 8 | (d))
-
-#define MDBX_STRING_TETRAD(str) MDBX_TETRAD(str[0], str[1], str[2], str[3])
-
-#define FIXME "FIXME: " __FILE__ ", " MDBX_STRINGIFY(__LINE__)
-
-#ifndef STATIC_ASSERT_MSG
-# if defined(static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) static_assert(expr, msg)
-# elif defined(_STATIC_ASSERT)
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif defined(_MSC_VER)
-# include <crtdbg.h>
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) \
- || __has_feature(c_static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) _Static_assert(expr, msg)
-# else
-# define STATIC_ASSERT_MSG(expr, msg) switch (0) {case 0:case (expr):;}
-# endif
-#endif /* STATIC_ASSERT */
-
-#ifndef STATIC_ASSERT
-# define STATIC_ASSERT(expr) STATIC_ASSERT_MSG(expr, #expr)
-#endif
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__GNUC__) && !__GNUC_PREREQ(4, 2)
-/* Actually libmdbx was not tested with compilers older than GCC 4.2.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required GCC >= 4.2"
-#endif
-
-#if defined(__clang__) && !__CLANG_PREREQ(3, 8)
-/* Actually libmdbx was not tested with CLANG older than 3.8.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required CLANG >= 3.8"
-#endif
-
-#if defined(__GLIBC__) && !__GLIBC_PREREQ(2, 12)
-/* Actually libmdbx was not tested with something older than glibc 2.12.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old systems.
- */
-#warning "libmdbx was only tested with GLIBC >= 2.12."
-#endif
-
-#ifdef __SANITIZE_THREAD__
-#warning \
- "libmdbx don't compatible with ThreadSanitizer, you will get a lot of false-positive issues."
-#endif /* __SANITIZE_THREAD__ */
-
-#if __has_warning("-Wnested-anon-types")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wnested-anon-types"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wnested-anon-types"
-#else
-#pragma warning disable "nested-anon-types"
-#endif
-#endif /* -Wnested-anon-types */
-
-#if __has_warning("-Wconstant-logical-operand")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wconstant-logical-operand"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wconstant-logical-operand"
-#else
-#pragma warning disable "constant-logical-operand"
-#endif
-#endif /* -Wconstant-logical-operand */
-
-#if defined(__LCC__) && (__LCC__ <= 121)
-/* bug #2798 */
-#pragma diag_suppress alignment_reduction_ignored
-#elif defined(__ICC)
-#pragma warning(disable : 3453 1366)
-#elif __has_warning("-Walignment-reduction-ignored")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Walignment-reduction-ignored"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Walignment-reduction-ignored"
-#else
-#pragma warning disable "alignment-reduction-ignored"
-#endif
-#endif /* -Walignment-reduction-ignored */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(_WIN32) || defined(_WIN64)
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if !defined(_NO_CRT_STDIO_INLINE) && MDBX_BUILD_SHARED_LIBRARY && \
- !defined(xMDBX_TOOLS) && MDBX_WITHOUT_MSVC_CRT
-#define _NO_CRT_STDIO_INLINE
-#endif
-#elif !defined(_POSIX_C_SOURCE)
-#define _POSIX_C_SOURCE 200809L
-#endif /* Windows */
-
-/*----------------------------------------------------------------------------*/
-/* C99 includes */
-#include <inttypes.h>
-#include <stddef.h>
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <assert.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-/* C11' alignas() */
-#if __has_include(<stdalign.h>)
-#include <stdalign.h>
-#endif
-#if defined(alignas) || defined(__cplusplus)
-#define MDBX_ALIGNAS(N) alignas(N)
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
-#define MDBX_ALIGNAS(N) _Alignas(N)
-#elif defined(_MSC_VER)
-#define MDBX_ALIGNAS(N) __declspec(align(N))
-#elif __has_attribute(__aligned__) || defined(__GNUC__)
-#define MDBX_ALIGNAS(N) __attribute__((__aligned__(N)))
-#else
-#error "FIXME: Required alignas() or equivalent."
-#endif /* MDBX_ALIGNAS */
-
-/*----------------------------------------------------------------------------*/
-/* Systems includes */
-
-#ifdef __APPLE__
-#include <TargetConditionals.h>
-#endif /* Apple OSX & iOS */
-
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__)
-#include <sys/cdefs.h>
-#include <sys/mount.h>
-#include <sys/sysctl.h>
-#include <sys/types.h>
-#if defined(__FreeBSD__) || defined(__DragonFly__)
-#include <vm/vm_param.h>
-#elif defined(__OpenBSD__) || defined(__NetBSD__)
-#include <uvm/uvm_param.h>
-#else
-#define SYSCTL_LEGACY_NONCONST_MIB
-#endif
-#ifndef __MACH__
-#include <sys/vmmeter.h>
-#endif
-#else
-#include <malloc.h>
-#if !(defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(_WIN32) || defined(_WIN64))
-#include <mntent.h>
-#endif /* !Solaris */
-#endif /* !xBSD */
-
-#if defined(__FreeBSD__) || __has_include(<malloc_np.h>)
-#include <malloc_np.h>
-#endif
-
-#if defined(__APPLE__) || defined(__MACH__) || __has_include(<malloc/malloc.h>)
-#include <malloc/malloc.h>
-#endif /* MacOS */
-
-#if defined(__MACH__)
-#include <mach/host_info.h>
-#include <mach/mach_host.h>
-#include <mach/mach_port.h>
-#include <uuid/uuid.h>
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-#include <sched.h>
-#include <sys/sendfile.h>
-#include <sys/statfs.h>
-#endif /* Linux */
-
-#ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 0
-#endif
-
-#ifndef _XOPEN_SOURCE_EXTENDED
-#define _XOPEN_SOURCE_EXTENDED 0
-#else
-#include <utmpx.h>
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#include <kstat.h>
-#include <sys/mnttab.h>
-/* On Solaris, it's easier to add a missing prototype rather than find a
- * combination of #defines that break nothing. */
-__extern_C key_t ftok(const char *, int);
-#endif /* SunOS/Solaris */
-
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef _WIN32_WINNT
-#define _WIN32_WINNT 0x0601 /* Windows 7 */
-#elif _WIN32_WINNT < 0x0500
-#error At least 'Windows 2000' API is required for libmdbx.
-#endif /* _WIN32_WINNT */
-#if (defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* MinGW */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif /* WIN32_LEAN_AND_MEAN */
-#include <excpt.h>
-#include <tlhelp32.h>
-#include <windows.h>
-#include <winnt.h>
-#include <winternl.h>
-#define HAVE_SYS_STAT_H
-#define HAVE_SYS_TYPES_H
-typedef HANDLE mdbx_thread_t;
-typedef unsigned mdbx_thread_key_t;
-#define MAP_FAILED NULL
-#define HIGH_DWORD(v) ((DWORD)((sizeof(v) > 4) ? ((uint64_t)(v) >> 32) : 0))
-#define THREAD_CALL WINAPI
-#define THREAD_RESULT DWORD
-typedef struct {
- HANDLE mutex;
- HANDLE event[2];
-} mdbx_condpair_t;
-typedef CRITICAL_SECTION mdbx_fastmutex_t;
-
-#if !defined(_MSC_VER) && !defined(__try)
-/* *INDENT-OFF* */
-/* clang-format off */
-#define __try
-#define __except(COND) if(false)
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* stub for MSVC's __try/__except */
-
-#if MDBX_WITHOUT_MSVC_CRT
-
-#ifndef mdbx_malloc
-static inline void *mdbx_malloc(size_t bytes) {
- return HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_malloc */
-
-#ifndef mdbx_calloc
-static inline void *mdbx_calloc(size_t nelem, size_t size) {
- return HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, nelem * size);
-}
-#endif /* mdbx_calloc */
-
-#ifndef mdbx_realloc
-static inline void *mdbx_realloc(void *ptr, size_t bytes) {
- return ptr ? HeapReAlloc(GetProcessHeap(), 0, ptr, bytes)
- : HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_realloc */
-
-#ifndef mdbx_free
-static inline void mdbx_free(void *ptr) { HeapFree(GetProcessHeap(), 0, ptr); }
-#endif /* mdbx_free */
-
-#else /* MDBX_WITHOUT_MSVC_CRT */
-
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup _strdup
-
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-#ifndef snprintf
-#define snprintf _snprintf /* ntdll */
-#endif
-
-#ifndef vsnprintf
-#define vsnprintf _vsnprintf /* ntdll */
-#endif
-
-#else /*----------------------------------------------------------------------*/
-
-#include <unistd.h>
-#if !defined(_POSIX_MAPPED_FILES) || _POSIX_MAPPED_FILES < 1
-#error "libmdbx requires the _POSIX_MAPPED_FILES feature"
-#endif /* _POSIX_MAPPED_FILES */
-
-#include <pthread.h>
-#include <semaphore.h>
-#include <signal.h>
-#include <sys/file.h>
-#include <sys/ipc.h>
-#include <sys/mman.h>
-#include <sys/param.h>
-#include <sys/stat.h>
-#include <sys/statvfs.h>
-#include <sys/uio.h>
-typedef pthread_t mdbx_thread_t;
-typedef pthread_key_t mdbx_thread_key_t;
-#define INVALID_HANDLE_VALUE (-1)
-#define THREAD_CALL
-#define THREAD_RESULT void *
-typedef struct {
- pthread_mutex_t mutex;
- pthread_cond_t cond[2];
-} mdbx_condpair_t;
-typedef pthread_mutex_t mdbx_fastmutex_t;
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup strdup
-#endif /* Platform */
-
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
-/* malloc_usable_size() already provided */
-#elif defined(__APPLE__)
-#define malloc_usable_size(ptr) malloc_size(ptr)
-#elif defined(_MSC_VER) && !MDBX_WITHOUT_MSVC_CRT
-#define malloc_usable_size(ptr) _msize(ptr)
-#endif /* malloc_usable_size */
-
-#ifdef __ANDROID_API__
-#include <android/log.h>
-#if __ANDROID_API__ >= 21
-#include <sys/sendfile.h>
-#endif
-#endif /* Android */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(HAVE_SYS_STAT_H) || __has_include(<sys/stat.h>)
-#include <sys/stat.h>
-#endif
-#if defined(HAVE_SYS_TYPES_H) || __has_include(<sys/types.h>)
-#include <sys/types.h>
-#endif
-#if defined(HAVE_SYS_FILE_H) || __has_include(<sys/file.h>)
-#include <sys/file.h>
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-
-#ifndef SSIZE_MAX
-#define SSIZE_MAX INTPTR_MAX
-#endif
-
-#if !defined(MADV_DODUMP) && defined(MADV_CORE)
-#define MADV_DODUMP MADV_CORE
-#endif /* MADV_CORE -> MADV_DODUMP */
-
-#if !defined(MADV_DONTDUMP) && defined(MADV_NOCORE)
-#define MADV_DONTDUMP MADV_NOCORE
-#endif /* MADV_NOCORE -> MADV_DONTDUMP */
-
-#if defined(i386) || defined(__386) || defined(__i386) || defined(__i386__) || \
- defined(i486) || defined(__i486) || defined(__i486__) || \
- defined(i586) | defined(__i586) || defined(__i586__) || defined(i686) || \
- defined(__i686) || defined(__i686__) || defined(_M_IX86) || \
- defined(_X86_) || defined(__THW_INTEL__) || defined(__I86__) || \
- defined(__INTEL__) || defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64__) || defined(__amd64) || defined(_M_X64) || \
- defined(_M_AMD64) || defined(__IA32__) || defined(__INTEL__)
-#ifndef __ia32__
-/* LY: define neutral __ia32__ for x86 and x86-64 */
-#define __ia32__ 1
-#endif /* __ia32__ */
-#if !defined(__amd64__) && (defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64) || defined(_M_X64))
-/* LY: define trusty __amd64__ for all AMD64/x86-64 arch */
-#define __amd64__ 1
-#endif /* __amd64__ */
-#endif /* all x86 */
-
-#if (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
-#error \
- "Sanity checking failed: Two's complement, reasonably sized integer types"
-#endif
-
-#if UINTPTR_MAX > 0xffffFFFFul || ULONG_MAX > 0xffffFFFFul
-#define MDBX_WORDBITS 64
-#else
-#define MDBX_WORDBITS 32
-#endif /* MDBX_WORDBITS */
-
-#if defined(__ANDROID_API__) || defined(ANDROID)
-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "_FILE_OFFSET_BITS != MDBX_WORDBITS" (_FILE_OFFSET_BITS != MDBX_WORDBITS)
-#elif defined(__FILE_OFFSET_BITS) && __FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "__FILE_OFFSET_BITS != MDBX_WORDBITS" (__FILE_OFFSET_BITS != MDBX_WORDBITS)
-#endif
-#endif /* Android */
-
-/*----------------------------------------------------------------------------*/
-/* Compiler's includes for builtins/intrinsics */
-
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
-#include <intrin.h>
-#elif __GNUC_PREREQ(4, 4) || defined(__clang__)
-#if defined(__ia32__) || defined(__e2k__)
-#include <x86intrin.h>
-#endif /* __ia32__ */
-#if defined(__ia32__)
-#include <cpuid.h>
-#endif /* __ia32__ */
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
-#include <mbarrier.h>
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
-#include <machine/sys/inline.h>
-#elif defined(__IBMC__) && defined(__powerpc)
-#include <atomic.h>
-#elif defined(_AIX)
-#include <builtins.h>
-#include <sys/atomic_op.h>
-#elif (defined(__osf__) && defined(__DECC)) || defined(__alpha)
-#include <c_asm.h>
-#include <machine/builtins.h>
-#elif defined(__MWERKS__)
-/* CodeWarrior - troubles ? */
-#pragma gcc_extensions
-#elif defined(__SNC__)
-/* Sony PS3 - troubles ? */
-#elif defined(__hppa__) || defined(__hppa)
-#include <machine/inline.h>
-#else
-#error Unsupported C compiler, please use GNU C 4.4 or newer
-#endif /* Compiler */
-
-/*----------------------------------------------------------------------------*/
-/* Byteorder */
-
-#if !defined(__BYTE_ORDER__) || !defined(__ORDER_LITTLE_ENDIAN__) || \
- !defined(__ORDER_BIG_ENDIAN__)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(__GLIBC__) || defined(__GNU_LIBRARY__) || defined(__ANDROID_API__) || \
- defined(HAVE_ENDIAN_H) || __has_include(<endian.h>)
-#include <endian.h>
-#elif defined(__APPLE__) || defined(__MACH__) || defined(__OpenBSD__) || \
- defined(HAVE_MACHINE_ENDIAN_H) || __has_include(<machine/endian.h>)
-#include <machine/endian.h>
-#elif defined(HAVE_SYS_ISA_DEFS_H) || __has_include(<sys/isa_defs.h>)
-#include <sys/isa_defs.h>
-#elif (defined(HAVE_SYS_TYPES_H) && defined(HAVE_SYS_ENDIAN_H)) || \
- (__has_include(<sys/types.h>) && __has_include(<sys/endian.h>))
-#include <sys/endian.h>
-#include <sys/types.h>
-#elif defined(__bsdi__) || defined(__DragonFly__) || defined(__FreeBSD__) || \
- defined(__NetBSD__) || \
- defined(HAVE_SYS_PARAM_H) || __has_include(<sys/param.h>)
-#include <sys/param.h>
-#endif /* OS */
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-
-#else
-#error __BYTE_ORDER__ should be defined.
-#endif /* Arch */
-
-#endif
-#endif /* __BYTE_ORDER__ || __ORDER_LITTLE_ENDIAN__ || __ORDER_BIG_ENDIAN__ */
-
-/* Get the size of a memory page for the system.
- * This is the basic size that the platform's memory manager uses, and is
- * fundamental to the use of memory-mapped files. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline size_t
-mdbx_syspagesize(void) {
-#if defined(_WIN32) || defined(_WIN64)
- SYSTEM_INFO si;
- GetSystemInfo(&si);
- return si.dwPageSize;
-#else
- return sysconf(_SC_PAGE_SIZE);
-#endif
-}
-
-typedef struct mdbx_mmap_param {
- union {
- void *address;
- uint8_t *dxb;
- struct MDBX_lockinfo *lck;
- };
- mdbx_filehandle_t fd;
- size_t limit; /* mapping length, but NOT a size of file nor DB */
- size_t current; /* mapped region size, i.e. the size of file and DB */
- uint64_t filesize /* in-process cache of a file size */;
-#if defined(_WIN32) || defined(_WIN64)
- HANDLE section; /* memory-mapped section handle */
-#endif
-} mdbx_mmap_t;
-
-typedef union bin128 {
- __anonymous_struct_extension__ struct { uint64_t x, y; };
- __anonymous_struct_extension__ struct { uint32_t a, b, c, d; };
-} bin128_t;
-
-#if defined(_WIN32) || defined(_WIN64)
-typedef union MDBX_srwlock {
- __anonymous_struct_extension__ struct {
- long volatile readerCount;
- long volatile writerCount;
- };
- RTL_SRWLOCK native;
-} MDBX_srwlock;
-#endif /* Windows */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny);
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny);
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-#if defined(__cplusplus) && !defined(__STDC_NO_ATOMICS__) && (__has_include(<cstdatomic>) || __has_extension(cxx_atomic))
-#include <cstdatomic>
-#define MDBX_HAVE_C11ATOMICS
-#elif !defined(__cplusplus) && \
- (__STDC_VERSION__ >= 201112L || __has_extension(c_atomic)) && \
- !defined(__STDC_NO_ATOMICS__) && \
- (__GNUC_PREREQ(4, 9) || __CLANG_PREREQ(3, 8) || \
- !(defined(__GNUC__) || defined(__clang__)))
-#include <stdatomic.h>
-#define MDBX_HAVE_C11ATOMICS
-#elif defined(__GNUC__) || defined(__clang__)
-#elif defined(_MSC_VER)
-#pragma warning(disable : 4163) /* 'xyz': not available as an intrinsic */
-#pragma warning(disable : 4133) /* 'function': incompatible types - from \
- 'size_t' to 'LONGLONG' */
-#pragma warning(disable : 4244) /* 'return': conversion from 'LONGLONG' to \
- 'std::size_t', possible loss of data */
-#pragma warning(disable : 4267) /* 'function': conversion from 'size_t' to \
- 'long', possible loss of data */
-#pragma intrinsic(_InterlockedExchangeAdd, _InterlockedCompareExchange)
-#pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedCompareExchange64)
-#elif defined(__APPLE__)
-#include <libkern/OSAtomic.h>
-#else
-#error FIXME atomic-ops
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Memory/Compiler barriers, cache coherence */
-
-#if __has_include(<sys/cachectl.h>)
-#include <sys/cachectl.h>
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS should have explicit cache control */
-#include <sys/cachectl.h>
-#endif
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_compiler_barrier(void) {
-#if defined(__clang__) || defined(__GNUC__)
- __asm__ __volatile__("" ::: "memory");
-#elif defined(_MSC_VER)
- _ReadWriteBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
- __memory_barrier();
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __compiler_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_sched_fence(/* LY: no-arg meaning 'all expect ALU', e.g. 0x3D3D */);
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __fence();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_memory_barrier(void) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(memory_order_seq_cst);
-#elif defined(__ATOMIC_SEQ_CST)
-#ifdef __clang__
- __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
- __atomic_thread_fence(__ATOMIC_SEQ_CST);
-#endif
-#elif defined(__clang__) || defined(__GNUC__)
- __sync_synchronize();
-#elif defined(_WIN32) || defined(_WIN64)
- MemoryBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
-#if defined(__ia32__)
- _mm_mfence();
-#else
- __mf();
-#endif
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __machine_rw_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_mf();
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __lwsync();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* libc compatibility stuff */
-
-#if (!defined(__GLIBC__) && __GLIBC_PREREQ(2, 1)) && \
- (defined(_GNU_SOURCE) || defined(_BSD_SOURCE))
-#define mdbx_asprintf asprintf
-#define mdbx_vasprintf vasprintf
-#else
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC
- MDBX_PRINTF_ARGS(2, 3) int mdbx_asprintf(char **strp, const char *fmt, ...);
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt, va_list ap);
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* OS abstraction layer stuff */
-
-/* max bytes to write in one call */
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_WRITE UINT32_C(0x01000000)
-#else
-#define MAX_WRITE UINT32_C(0x3fff0000)
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-MDBX_INTERNAL_VAR uint32_t mdbx_linux_kernel_version;
-MDBX_INTERNAL_VAR bool mdbx_RunningOnWSL1 /* Windows Subsystem 1 for Linux */;
-#endif /* Linux */
-
-#ifndef mdbx_strdup
-LIBMDBX_API char *mdbx_strdup(const char *str);
-#endif
-
-MDBX_MAYBE_UNUSED static __inline int mdbx_get_errno(void) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD rc = GetLastError();
-#else
- int rc = errno;
-#endif
- return rc;
-}
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result);
-#endif
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr);
-#endif
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair, bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair);
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex);
-
-MDBX_INTERNAL_FUNC int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov,
- int iovcnt, uint64_t offset,
- size_t expected_written);
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t count,
- uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t count, uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t count);
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg);
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread);
-
-enum mdbx_syncmode_bits {
- MDBX_SYNC_NONE = 0,
- MDBX_SYNC_DATA = 1,
- MDBX_SYNC_SIZE = 2,
- MDBX_SYNC_IODQ = 4
-};
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- const enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length);
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos);
-MDBX_INTERNAL_FUNC int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length);
-
-enum mdbx_openfile_purpose {
- MDBX_OPEN_DXB_READ = 0,
- MDBX_OPEN_DXB_LAZY = 1,
- MDBX_OPEN_DXB_DSYNC = 2,
- MDBX_OPEN_LCK = 3,
- MDBX_OPEN_COPY = 4,
- MDBX_OPEN_DELETE = 5
-};
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait);
-
-#define MMAP_OPTION_TRUNCATE 1
-#define MMAP_OPTION_SEMAPHORE 2
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t must, const size_t limit,
- const unsigned options);
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map);
-#define MDBX_MRESIZE_MAY_MOVE 0x00000100
-#define MDBX_MRESIZE_MAY_UNMAP 0x00000200
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit);
-#if defined(_WIN32) || defined(_WIN64)
-typedef struct {
- unsigned limit, count;
- HANDLE handles[31];
-} mdbx_handle_array_t;
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array);
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array);
-#endif /* Windows */
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err);
-
-MDBX_MAYBE_UNUSED static __inline uint32_t mdbx_getpid(void) {
- STATIC_ASSERT(sizeof(mdbx_pid_t) <= sizeof(uint32_t));
-#if defined(_WIN32) || defined(_WIN64)
- return GetCurrentProcessId();
-#else
- STATIC_ASSERT(sizeof(pid_t) <= sizeof(uint32_t));
- return getpid();
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline uintptr_t mdbx_thread_self(void) {
- mdbx_tid_t thunk;
- STATIC_ASSERT(sizeof(uintptr_t) >= sizeof(thunk));
-#if defined(_WIN32) || defined(_WIN64)
- thunk = GetCurrentThreadId();
-#else
- thunk = pthread_self();
-#endif
- return (uintptr_t)thunk;
-}
-
-#if !defined(_WIN32) && !defined(_WIN64)
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void);
-#else
-static __inline int mdbx_check_tid4bionic(void) { return 0; }
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-MDBX_MAYBE_UNUSED static __inline int
-mdbx_pthread_mutex_lock(pthread_mutex_t *mutex) {
- int err = mdbx_check_tid4bionic();
- return unlikely(err) ? err : pthread_mutex_lock(mutex);
-}
-#endif /* !Windows */
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void);
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16);
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime);
-
-MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void);
-/*----------------------------------------------------------------------------*/
-/* lck stuff */
-
-/// \brief Initialization of synchronization primitives linked with MDBX_env
-/// instance both in LCK-file and within the current process.
-/// \param
-/// global_uniqueness_flag = true - denotes that there are no other processes
-/// working with DB and LCK-file. Thus the function MUST initialize
-/// shared synchronization objects in memory-mapped LCK-file.
-/// global_uniqueness_flag = false - denotes that at least one process is
-/// already working with DB and LCK-file, including the case when DB
-/// has already been opened in the current process. Thus the function
-/// MUST NOT initialize shared synchronization objects in memory-mapped
-/// LCK-file that are already in use.
-/// \return Error code or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag);
-
-/// \brief Disconnects from shared interprocess objects and destructs
-/// synchronization objects linked with MDBX_env instance
-/// within the current process.
-/// \param
-/// inprocess_neighbor = NULL - if the current process does not have other
-/// instances of MDBX_env linked with the DB being closed.
-/// Thus the function MUST check for other processes working with DB or
-/// LCK-file, and keep or destroy shared synchronization objects in
-/// memory-mapped LCK-file depending on the result.
-/// inprocess_neighbor = not-NULL - pointer to another instance of MDBX_env
-/// (anyone of there is several) working with DB or LCK-file within the
-/// current process. Thus the function MUST NOT try to acquire exclusive
-/// lock and/or try to destruct shared synchronization objects linked with
-/// DB or LCK-file. Moreover, the implementation MUST ensure correct work
-/// of other instances of MDBX_env within the current process, e.g.
-/// restore POSIX-fcntl locks after the closing of file descriptors.
-/// \return Error code (MDBX_PANIC) or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor);
-
-/// \brief Connects to shared interprocess locking objects and tries to acquire
-/// the maximum lock level (shared if exclusive is not available)
-/// Depending on implementation or/and platform (Windows) this function may
-/// acquire the non-OS super-level lock (e.g. for shared synchronization
-/// objects initialization), which will be downgraded to OS-exclusive or
-/// shared via explicit calling of mdbx_lck_downgrade().
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if an exclusive lock was acquired and thus
-/// the current process is the first and only after the last use of DB.
-/// MDBX_RESULT_FALSE (0) - if a shared lock was acquired and thus
-/// DB has already been opened and now is used by other processes.
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env);
-
-/// \brief Downgrades the level of initially acquired lock to
-/// operational level specified by argument. The reson for such downgrade:
-/// - unblocking of other processes that are waiting for access, i.e.
-/// if (env->me_flags & MDBX_EXCLUSIVE) != 0, then other processes
-/// should be made aware that access is unavailable rather than
-/// wait for it.
-/// - freeing locks that interfere file operation (especially for Windows)
-/// (env->me_flags & MDBX_EXCLUSIVE) == 0 - downgrade to shared lock.
-/// (env->me_flags & MDBX_EXCLUSIVE) != 0 - downgrade to exclusive
-/// operational lock.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env);
-
-/// \brief Locks LCK-file or/and table of readers for (de)registering.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env);
-
-/// \brief Unlocks LCK-file or/and table of readers after (de)registering.
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env);
-
-/// \brief Acquires lock for DB change (on writing transaction start)
-/// Reading transactions will not be blocked.
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-/// \return Error code or zero on success
-LIBMDBX_API int mdbx_txn_lock(MDBX_env *env, bool dont_wait);
-
-/// \brief Releases lock once DB changes is made (after writing transaction
-/// has finished).
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-LIBMDBX_API void mdbx_txn_unlock(MDBX_env *env);
-
-/// \brief Sets alive-flag of reader presence (indicative lock) for PID of
-/// the current process. The function does no more than needed for
-/// the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env);
-
-/// \brief Resets alive-flag of reader presence (indicative lock)
-/// for PID of the current process. The function does no more than needed
-/// for the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env);
-
-/// \brief Checks for reading process status with the given pid with help of
-/// alive-flag of presence (indicative lock) or using another way.
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if the reader process with the given PID is alive
-/// and working with DB (indicative lock is present).
-/// MDBX_RESULT_FALSE (0) - if the reader process with the given PID is absent
-/// or not working with DB (indicative lock is not present).
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid);
-
-#if defined(_WIN32) || defined(_WIN64)
-
-typedef void(WINAPI *MDBX_srwlock_function)(MDBX_srwlock *);
-MDBX_INTERNAL_VAR MDBX_srwlock_function mdbx_srwlock_Init,
- mdbx_srwlock_AcquireShared, mdbx_srwlock_ReleaseShared,
- mdbx_srwlock_AcquireExclusive, mdbx_srwlock_ReleaseExclusive;
-
-#if _WIN32_WINNT < 0x0600 /* prior to Windows Vista */
-typedef enum _FILE_INFO_BY_HANDLE_CLASS {
- FileBasicInfo,
- FileStandardInfo,
- FileNameInfo,
- FileRenameInfo,
- FileDispositionInfo,
- FileAllocationInfo,
- FileEndOfFileInfo,
- FileStreamInfo,
- FileCompressionInfo,
- FileAttributeTagInfo,
- FileIdBothDirectoryInfo,
- FileIdBothDirectoryRestartInfo,
- FileIoPriorityHintInfo,
- FileRemoteProtocolInfo,
- MaximumFileInfoByHandleClass
-} FILE_INFO_BY_HANDLE_CLASS,
- *PFILE_INFO_BY_HANDLE_CLASS;
-
-typedef struct _FILE_END_OF_FILE_INFO {
- LARGE_INTEGER EndOfFile;
-} FILE_END_OF_FILE_INFO, *PFILE_END_OF_FILE_INFO;
-
-#define REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK 0x00000001
-#define REMOTE_PROTOCOL_INFO_FLAG_OFFLINE 0x00000002
-
-typedef struct _FILE_REMOTE_PROTOCOL_INFO {
- USHORT StructureVersion;
- USHORT StructureSize;
- DWORD Protocol;
- USHORT ProtocolMajorVersion;
- USHORT ProtocolMinorVersion;
- USHORT ProtocolRevision;
- USHORT Reserved;
- DWORD Flags;
- struct {
- DWORD Reserved[8];
- } GenericReserved;
- struct {
- DWORD Reserved[16];
- } ProtocolSpecificReserved;
-} FILE_REMOTE_PROTOCOL_INFO, *PFILE_REMOTE_PROTOCOL_INFO;
-
-#endif /* _WIN32_WINNT < 0x0600 (prior to Windows Vista) */
-
-typedef BOOL(WINAPI *MDBX_GetFileInformationByHandleEx)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_GetFileInformationByHandleEx
- mdbx_GetFileInformationByHandleEx;
-
-typedef BOOL(WINAPI *MDBX_GetVolumeInformationByHandleW)(
- _In_ HANDLE hFile, _Out_opt_ LPWSTR lpVolumeNameBuffer,
- _In_ DWORD nVolumeNameSize, _Out_opt_ LPDWORD lpVolumeSerialNumber,
- _Out_opt_ LPDWORD lpMaximumComponentLength,
- _Out_opt_ LPDWORD lpFileSystemFlags,
- _Out_opt_ LPWSTR lpFileSystemNameBuffer, _In_ DWORD nFileSystemNameSize);
-MDBX_INTERNAL_VAR MDBX_GetVolumeInformationByHandleW
- mdbx_GetVolumeInformationByHandleW;
-
-typedef DWORD(WINAPI *MDBX_GetFinalPathNameByHandleW)(_In_ HANDLE hFile,
- _Out_ LPWSTR lpszFilePath,
- _In_ DWORD cchFilePath,
- _In_ DWORD dwFlags);
-MDBX_INTERNAL_VAR MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-
-typedef BOOL(WINAPI *MDBX_SetFileInformationByHandle)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_SetFileInformationByHandle
- mdbx_SetFileInformationByHandle;
-
-typedef NTSTATUS(NTAPI *MDBX_NtFsControlFile)(
- IN HANDLE FileHandle, IN OUT HANDLE Event,
- IN OUT PVOID /* PIO_APC_ROUTINE */ ApcRoutine, IN OUT PVOID ApcContext,
- OUT PIO_STATUS_BLOCK IoStatusBlock, IN ULONG FsControlCode,
- IN OUT PVOID InputBuffer, IN ULONG InputBufferLength,
- OUT OPTIONAL PVOID OutputBuffer, IN ULONG OutputBufferLength);
-MDBX_INTERNAL_VAR MDBX_NtFsControlFile mdbx_NtFsControlFile;
-
-typedef uint64_t(WINAPI *MDBX_GetTickCount64)(void);
-MDBX_INTERNAL_VAR MDBX_GetTickCount64 mdbx_GetTickCount64;
-
-#if !defined(_WIN32_WINNT_WIN8) || _WIN32_WINNT < _WIN32_WINNT_WIN8
-typedef struct _WIN32_MEMORY_RANGE_ENTRY {
- PVOID VirtualAddress;
- SIZE_T NumberOfBytes;
-} WIN32_MEMORY_RANGE_ENTRY, *PWIN32_MEMORY_RANGE_ENTRY;
-#endif /* Windows 8.x */
-
-typedef BOOL(WINAPI *MDBX_PrefetchVirtualMemory)(
- HANDLE hProcess, ULONG_PTR NumberOfEntries,
- PWIN32_MEMORY_RANGE_ENTRY VirtualAddresses, ULONG Flags);
-MDBX_INTERNAL_VAR MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-
-typedef enum _SECTION_INHERIT { ViewShare = 1, ViewUnmap = 2 } SECTION_INHERIT;
-
-typedef NTSTATUS(NTAPI *MDBX_NtExtendSection)(IN HANDLE SectionHandle,
- IN PLARGE_INTEGER NewSectionSize);
-MDBX_INTERNAL_VAR MDBX_NtExtendSection mdbx_NtExtendSection;
-
-static __inline bool mdbx_RunningUnderWine(void) {
- return !mdbx_NtExtendSection;
-}
-
-typedef LSTATUS(WINAPI *MDBX_RegGetValueA)(HKEY hkey, LPCSTR lpSubKey,
- LPCSTR lpValue, DWORD dwFlags,
- LPDWORD pdwType, PVOID pvData,
- LPDWORD pcbData);
-MDBX_INTERNAL_VAR MDBX_RegGetValueA mdbx_RegGetValueA;
-
-NTSYSAPI ULONG RtlRandomEx(PULONG Seed);
-
-#endif /* Windows */
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_MSC_VER) && _MSC_VER >= 1900
-/* LY: MSVC 2015/2017/2019 has buggy/inconsistent PRIuPTR/PRIxPTR macros
- * for internal format-args checker. */
-#undef PRIuPTR
-#undef PRIiPTR
-#undef PRIdPTR
-#undef PRIxPTR
-#define PRIuPTR "Iu"
-#define PRIiPTR "Ii"
-#define PRIdPTR "Id"
-#define PRIxPTR "Ix"
-#define PRIuSIZE "zu"
-#define PRIiSIZE "zi"
-#define PRIdSIZE "zd"
-#define PRIxSIZE "zx"
-#endif /* fix PRI*PTR for _MSC_VER */
-
-#ifndef PRIuSIZE
-#define PRIuSIZE PRIuPTR
-#define PRIiSIZE PRIiPTR
-#define PRIdSIZE PRIdPTR
-#define PRIxSIZE PRIxPTR
-#endif /* PRI*SIZE macros for MSVC */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-#define mdbx_sourcery_anchor XCONCAT(mdbx_sourcery_, MDBX_BUILD_SOURCERY)
-#if defined(xMDBX_TOOLS)
-extern LIBMDBX_API const char *const mdbx_sourcery_anchor;
-#endif
-
-/*******************************************************************************
- *******************************************************************************
- *******************************************************************************
- *
- *
- * #### ##### ##### # #### # # ####
- * # # # # # # # # ## # #
- * # # # # # # # # # # # ####
- * # # ##### # # # # # # # #
- * # # # # # # # # ## # #
- * #### # # # #### # # ####
- *
- *
- */
-
-/** \defgroup build_option Build options
- * The libmdbx build options.
- @{ */
-
-/** Using fcntl(F_FULLFSYNC) with 5-10 times slowdown */
-#define MDBX_OSX_WANNA_DURABILITY 0
-/** Using fsync() with chance of data lost on power failure */
-#define MDBX_OSX_WANNA_SPEED 1
-
-#ifndef MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-/** Choices \ref MDBX_OSX_WANNA_DURABILITY or \ref MDBX_OSX_WANNA_SPEED
- * for OSX & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY MDBX_OSX_WANNA_DURABILITY
-#endif /* MDBX_OSX_SPEED_INSTEADOF_DURABILITY */
-
-/** Controls checking PID against reuse DB environment after the fork() */
-#ifndef MDBX_ENV_CHECKPID
-#if defined(MADV_DONTFORK) || defined(_WIN32) || defined(_WIN64)
-/* PID check could be omitted:
- * - on Linux when madvise(MADV_DONTFORK) is available, i.e. after the fork()
- * mapped pages will not be available for child process.
- * - in Windows where fork() not available. */
-#define MDBX_ENV_CHECKPID 0
-#else
-#define MDBX_ENV_CHECKPID 1
-#endif
-#define MDBX_ENV_CHECKPID_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#else
-#define MDBX_ENV_CHECKPID_CONFIG MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#endif /* MDBX_ENV_CHECKPID */
-
-/** Controls checking transaction owner thread against misuse transactions from
- * other threads. */
-#ifndef MDBX_TXN_CHECKOWNER
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_TXN_CHECKOWNER_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#else
-#define MDBX_TXN_CHECKOWNER_CONFIG MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#endif /* MDBX_TXN_CHECKOWNER */
-
-/** Does a system have battery-backed Real-Time Clock or just a fake. */
-#ifndef MDBX_TRUST_RTC
-#if defined(__linux__) || defined(__gnu_linux__) || defined(__NetBSD__) || \
- defined(__OpenBSD__)
-#define MDBX_TRUST_RTC 0 /* a lot of embedded systems have a fake RTC */
-#else
-#define MDBX_TRUST_RTC 1
-#endif
-#define MDBX_TRUST_RTC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#else
-#define MDBX_TRUST_RTC_CONFIG MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#endif /* MDBX_TRUST_RTC */
-
-/** Controls online database auto-compactification during write-transactions. */
-#ifndef MDBX_ENABLE_REFUND
-#define MDBX_ENABLE_REFUND 1
-#elif !(MDBX_ENABLE_REFUND == 0 || MDBX_ENABLE_REFUND == 1)
-#error MDBX_ENABLE_REFUND must be defined as 0 or 1
-#endif /* MDBX_ENABLE_REFUND */
-
-/** Controls gathering statistics for page operations. */
-#ifndef MDBX_ENABLE_PGOP_STAT
-#define MDBX_ENABLE_PGOP_STAT 1
-#elif !(MDBX_ENABLE_PGOP_STAT == 0 || MDBX_ENABLE_PGOP_STAT == 1)
-#error MDBX_ENABLE_PGOP_STAT must be defined as 0 or 1
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-/** Controls use of POSIX madvise() hints and friends. */
-#ifndef MDBX_ENABLE_MADVISE
-#define MDBX_ENABLE_MADVISE 1
-#elif !(MDBX_ENABLE_MADVISE == 0 || MDBX_ENABLE_MADVISE == 1)
-#error MDBX_ENABLE_MADVISE must be defined as 0 or 1
-#endif /* MDBX_ENABLE_MADVISE */
-
-/** Disable some checks to reduce an overhead and detection probability of
- * database corruption to a values closer to the LMDB. */
-#ifndef MDBX_DISABLE_PAGECHECKS
-#define MDBX_DISABLE_PAGECHECKS 0
-#elif !(MDBX_DISABLE_PAGECHECKS == 0 || MDBX_DISABLE_PAGECHECKS == 1)
-#error MDBX_DISABLE_PAGECHECKS must be defined as 0 or 1
-#endif /* MDBX_DISABLE_PAGECHECKS */
-
-#ifndef MDBX_PNL_PREALLOC_FOR_RADIXSORT
-#define MDBX_PNL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_PNL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_PNL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_PNL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
-
-#ifndef MDBX_DPL_PREALLOC_FOR_RADIXSORT
-#define MDBX_DPL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_DPL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_DPL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_DPL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
-
-/* Basically, this build-option is for TODO. Guess it should be replaced
- * with MDBX_ENABLE_WRITEMAP_SPILLING with the three variants:
- * 0/OFF = Don't track dirty pages at all and don't spilling ones.
- * This should be by-default on Linux and may-be other systems
- * (not sure: Darwin/OSX, FreeBSD, Windows 10) where kernel provides
- * properly LRU tracking and async writing on-demand.
- * 1/ON = Lite tracking of dirty pages but with LRU labels and explicit
- * spilling with msync(MS_ASYNC). */
-#ifndef MDBX_FAKE_SPILL_WRITEMAP
-#if defined(__linux__) || defined(__gnu_linux__)
-#define MDBX_FAKE_SPILL_WRITEMAP 1 /* msync(MS_ASYNC) is no-op on Linux */
-#else
-#define MDBX_FAKE_SPILL_WRITEMAP 0
-#endif
-#elif !(MDBX_FAKE_SPILL_WRITEMAP == 0 || MDBX_FAKE_SPILL_WRITEMAP == 1)
-#error MDBX_FAKE_SPILL_WRITEMAP must be defined as 0 or 1
-#endif /* MDBX_FAKE_SPILL_WRITEMAP */
-
-/** Controls sort order of internal page number lists.
- * This mostly experimental/advanced option with not for regular MDBX users.
- * \warning The database format depend on this option and libmdbx builded with
- * different option value are incompatible. */
-#ifndef MDBX_PNL_ASCENDING
-#define MDBX_PNL_ASCENDING 0
-#elif !(MDBX_PNL_ASCENDING == 0 || MDBX_PNL_ASCENDING == 1)
-#error MDBX_PNL_ASCENDING must be defined as 0 or 1
-#endif /* MDBX_PNL_ASCENDING */
-
-/** Avoid dependence from MSVC CRT and use ntdll.dll instead. */
-#ifndef MDBX_WITHOUT_MSVC_CRT
-#define MDBX_WITHOUT_MSVC_CRT 1
-#elif !(MDBX_WITHOUT_MSVC_CRT == 0 || MDBX_WITHOUT_MSVC_CRT == 1)
-#error MDBX_WITHOUT_MSVC_CRT must be defined as 0 or 1
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-/** Size of buffer used during copying a environment/database file. */
-#ifndef MDBX_ENVCOPY_WRITEBUF
-#define MDBX_ENVCOPY_WRITEBUF 1048576u
-#elif MDBX_ENVCOPY_WRITEBUF < 65536u || MDBX_ENVCOPY_WRITEBUF > 1073741824u || \
- MDBX_ENVCOPY_WRITEBUF % 65536u
-#error MDBX_ENVCOPY_WRITEBUF must be defined in range 65536..1073741824 and be multiple of 65536
-#endif /* MDBX_ENVCOPY_WRITEBUF */
-
-/** Forces assertion checking */
-#ifndef MDBX_FORCE_ASSERTIONS
-#define MDBX_FORCE_ASSERTIONS 0
-#elif !(MDBX_FORCE_ASSERTIONS == 0 || MDBX_FORCE_ASSERTIONS == 1)
-#error MDBX_FORCE_ASSERTIONS must be defined as 0 or 1
-#endif /* MDBX_FORCE_ASSERTIONS */
-
-/** Presumed malloc size overhead for each allocation
- * to adjust allocations to be more aligned. */
-#ifndef MDBX_ASSUME_MALLOC_OVERHEAD
-#ifdef __SIZEOF_POINTER__
-#define MDBX_ASSUME_MALLOC_OVERHEAD (__SIZEOF_POINTER__ * 2u)
-#else
-#define MDBX_ASSUME_MALLOC_OVERHEAD (sizeof(void *) * 2u)
-#endif
-#elif MDBX_ASSUME_MALLOC_OVERHEAD < 0 || MDBX_ASSUME_MALLOC_OVERHEAD > 64 || \
- MDBX_ASSUME_MALLOC_OVERHEAD % 4
-#error MDBX_ASSUME_MALLOC_OVERHEAD must be defined in range 0..64 and be multiple of 4
-#endif /* MDBX_ASSUME_MALLOC_OVERHEAD */
-
-/** In case the MDBX_DEBUG is undefined set it corresponding to NDEBUG */
-#ifndef MDBX_DEBUG
-#ifdef NDEBUG
-#define MDBX_DEBUG 0
-#else
-#define MDBX_DEBUG 1
-#endif
-#endif /* MDBX_DEBUG */
-
-/** If defined then enables integration with Valgrind,
- * a memory analyzing tool. */
-#ifndef MDBX_USE_VALGRIND
-#endif /* MDBX_USE_VALGRIND */
-
-/** If defined then enables use C11 atomics,
- * otherwise detects ones availability automatically. */
-#ifndef MDBX_HAVE_C11ATOMICS
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-//------------------------------------------------------------------------------
-
-/** Win32 File Locking API for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_WIN32FILES -1
-
-/** SystemV IPC semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_SYSV 5
-
-/** POSIX-1 Shared anonymous semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX1988 1988
-
-/** POSIX-2001 Shared Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2001 2001
-
-/** POSIX-2008 Robust Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2008 2008
-
-/** BeOS Benaphores, aka Futexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_BENAPHORE 1995
-
-/** Advanced: Choices the locking implementation (autodetection by default). */
-#if defined(_WIN32) || defined(_WIN64)
-#define MDBX_LOCKING MDBX_LOCKING_WIN32FILES
-#else
-#ifndef MDBX_LOCKING
-#if defined(_POSIX_THREAD_PROCESS_SHARED) && \
- _POSIX_THREAD_PROCESS_SHARED >= 200112L && !defined(__FreeBSD__)
-
-/* Some platforms define the EOWNERDEAD error code even though they
- * don't support Robust Mutexes. If doubt compile with -MDBX_LOCKING=2001. */
-#if defined(EOWNERDEAD) && _POSIX_THREAD_PROCESS_SHARED >= 200809L && \
- ((defined(_POSIX_THREAD_ROBUST_PRIO_INHERIT) && \
- _POSIX_THREAD_ROBUST_PRIO_INHERIT > 0) || \
- (defined(_POSIX_THREAD_ROBUST_PRIO_PROTECT) && \
- _POSIX_THREAD_ROBUST_PRIO_PROTECT > 0) || \
- defined(PTHREAD_MUTEX_ROBUST) || defined(PTHREAD_MUTEX_ROBUST_NP)) && \
- (!defined(__GLIBC__) || \
- __GLIBC_PREREQ(2, 10) /* troubles with Robust mutexes before 2.10 */)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2008
-#else
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2001
-#endif
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX1988
-#else
-#define MDBX_LOCKING MDBX_LOCKING_SYSV
-#endif
-#define MDBX_LOCKING_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_LOCKING)
-#else
-#define MDBX_LOCKING_CONFIG MDBX_STRINGIFY(MDBX_LOCKING)
-#endif /* MDBX_LOCKING */
-#endif /* !Windows */
-
-/** Advanced: Using POSIX OFD-locks (autodetection by default). */
-#ifndef MDBX_USE_OFDLOCKS
-#if defined(F_OFD_SETLK) && defined(F_OFD_SETLKW) && defined(F_OFD_GETLK) && \
- !defined(MDBX_SAFE4QEMU) && \
- !defined(__sun) /* OFD-lock are broken on Solaris */
-#define MDBX_USE_OFDLOCKS 1
-#else
-#define MDBX_USE_OFDLOCKS 0
-#endif
-#define MDBX_USE_OFDLOCKS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#else
-#define MDBX_USE_OFDLOCKS_CONFIG MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#endif /* MDBX_USE_OFDLOCKS */
-
-/** Advanced: Using sendfile() syscall (autodetection by default). */
-#ifndef MDBX_USE_SENDFILE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 21)
-#define MDBX_USE_SENDFILE 1
-#else
-#define MDBX_USE_SENDFILE 0
-#endif
-#endif /* MDBX_USE_SENDFILE */
-
-/** Advanced: Using copy_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_COPYFILERANGE
-#if __GLIBC_PREREQ(2, 27) && defined(_GNU_SOURCE)
-#define MDBX_USE_COPYFILERANGE 1
-#else
-#define MDBX_USE_COPYFILERANGE 0
-#endif
-#endif /* MDBX_USE_COPYFILERANGE */
-
-/** Advanced: Using sync_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_SYNCFILERANGE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- defined(SYNC_FILE_RANGE_WRITE) && !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 26)
-#define MDBX_USE_SYNCFILERANGE 1
-#else
-#define MDBX_USE_SYNCFILERANGE 0
-#endif
-#endif /* MDBX_USE_SYNCFILERANGE */
-
-//------------------------------------------------------------------------------
-
-#ifndef MDBX_CPU_WRITEBACK_INCOHERENT
-#if defined(__ia32__) || defined(__e2k__) || defined(__hppa) || \
- defined(__hppa__) || defined(DOXYGEN)
-#define MDBX_CPU_WRITEBACK_INCOHERENT 0
-#else
-#define MDBX_CPU_WRITEBACK_INCOHERENT 1
-#endif
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-#ifndef MDBX_MMAP_INCOHERENT_FILE_WRITE
-#ifdef __OpenBSD__
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 1
-#else
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#ifndef MDBX_MMAP_INCOHERENT_CPU_CACHE
-#if defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS has cache coherency issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 1
-#else
-/* LY: assume no relevant mmap/dcache issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#ifndef MDBX_64BIT_ATOMIC
-#if MDBX_WORDBITS >= 64 || defined(DOXYGEN)
-#define MDBX_64BIT_ATOMIC 1
-#else
-#define MDBX_64BIT_ATOMIC 0
-#endif
-#define MDBX_64BIT_ATOMIC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#else
-#define MDBX_64BIT_ATOMIC_CONFIG MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#endif /* MDBX_64BIT_ATOMIC */
-
-#ifndef MDBX_64BIT_CAS
-#if defined(ATOMIC_LLONG_LOCK_FREE)
-#if ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__GCC_ATOMIC_LLONG_LOCK_FREE)
-#if __GCC_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__CLANG_ATOMIC_LLONG_LOCK_FREE)
-#if __CLANG_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(_MSC_VER) || defined(__APPLE__) || defined(DOXYGEN)
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS MDBX_64BIT_ATOMIC
-#endif
-#define MDBX_64BIT_CAS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#else
-#define MDBX_64BIT_CAS_CONFIG MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#endif /* MDBX_64BIT_CAS */
-
-#ifndef MDBX_UNALIGNED_OK
-#if defined(__ALIGNED__) || defined(__SANITIZE_UNDEFINED__)
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#elif defined(__ARM_FEATURE_UNALIGNED)
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#elif __CLANG_PREREQ(5, 0) || __GNUC_PREREQ(5, 0)
-/* expecting an optimization will well done, also this
- * hushes false-positives from UBSAN (undefined behaviour sanitizer) */
-#define MDBX_UNALIGNED_OK 0
-#elif defined(__e2k__) || defined(__elbrus__)
-#if __iset__ > 4
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#endif
-#elif defined(__ia32__)
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#endif
-#elif MDBX_UNALIGNED_OK == 1
-#undef MDBX_UNALIGNED_OK
-#define MDBX_UNALIGNED_OK 32 /* any unaligned access allowed */
-#endif /* MDBX_UNALIGNED_OK */
-
-#ifndef MDBX_CACHELINE_SIZE
-#if defined(SYSTEM_CACHE_ALIGNMENT_SIZE)
-#define MDBX_CACHELINE_SIZE SYSTEM_CACHE_ALIGNMENT_SIZE
-#elif defined(__ia64__) || defined(__ia64) || defined(_M_IA64)
-#define MDBX_CACHELINE_SIZE 128
-#else
-#define MDBX_CACHELINE_SIZE 64
-#endif
-#endif /* MDBX_CACHELINE_SIZE */
-
-/** @} end of build options */
-/*******************************************************************************
- *******************************************************************************
- ******************************************************************************/
-
-#ifdef DOXYGEN
-/* !!! Actually this is a fake definitions !!!
- * !!! for documentation generation by Doxygen !!! */
-
-/** Controls enabling of debugging features.
- *
- * - `MDBX_DEBUG = 0` (by default) Disables any debugging features at all,
- * including logging and assertion controls.
- * Logging level and corresponding debug flags changing
- * by \ref mdbx_setup_debug() will not have effect.
- * - `MDBX_DEBUG > 0` Enables code for the debugging features (logging,
- * assertions checking and internal audit).
- * Simultaneously sets the default logging level
- * to the `MDBX_DEBUG` value.
- * Also enables \ref MDBX_DBG_AUDIT if `MDBX_DEBUG >= 2`.
- *
- * \ingroup build_option */
-#define MDBX_DEBUG 0...7
-
-/** Disables using of GNU libc extensions. */
-#define MDBX_DISABLE_GNU_SOURCE 0 or 1
-
-#endif /* DOXYGEN */
-
-/* Undefine the NDEBUG if debugging is enforced by MDBX_DEBUG */
-#if MDBX_DEBUG
-#undef NDEBUG
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-enum MDBX_memory_order {
- mo_Relaxed,
- mo_AcquireRelease,
- mo_SequentialConsistency
-};
-
-typedef union {
- volatile uint32_t weak;
-#ifdef MDBX_HAVE_C11ATOMICS
- volatile _Atomic uint32_t c11a;
-#endif /* MDBX_HAVE_C11ATOMICS */
-} MDBX_atomic_uint32_t;
-
-typedef union {
- volatile uint64_t weak;
-#if defined(MDBX_HAVE_C11ATOMICS) && (MDBX_64BIT_CAS || MDBX_64BIT_ATOMIC)
- volatile _Atomic uint64_t c11a;
-#endif
-#if !defined(MDBX_HAVE_C11ATOMICS) || !MDBX_64BIT_CAS || !MDBX_64BIT_ATOMIC
- __anonymous_struct_extension__ struct {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- MDBX_atomic_uint32_t low, high;
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- MDBX_atomic_uint32_t high, low;
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- };
-#endif
-} MDBX_atomic_uint64_t;
-
-#ifdef MDBX_HAVE_C11ATOMICS
-
-/* Crutches for C11 atomic compiler's bugs */
-#if defined(__e2k__) && defined(__LCC__) && __LCC__ < /* FIXME */ 127
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->weak)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->weak)
-#elif defined(__clang__) && __clang__ < 8
-#define MDBX_c11a_ro(type, ptr) ((volatile _Atomic(type) *)&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#else
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#endif /* Crutches for C11 atomic compiler's bugs */
-
-static __always_inline memory_order mo_c11_store(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_release;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-
-static __always_inline memory_order mo_c11_load(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_acquire;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED static __always_inline void
-mdbx_memory_fence(enum MDBX_memory_order order, bool write) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(write ? mo_c11_store(order) : mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_compiler_barrier();
- if (write &&
- order > (MDBX_CPU_WRITEBACK_INCOHERENT ? mo_Relaxed : mo_AcquireRelease))
- mdbx_memory_barrier();
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_store32(MDBX_atomic_uint32_t *p, const uint32_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint32_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_load32(const MDBX_atomic_uint32_t *p, enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint32_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint32_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint32_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-/* Basic constants and types */
-
-/* A stamp that identifies a file as an MDBX file.
- * There's nothing special about this value other than that it is easily
- * recognizable, and it will reflect any byte order mismatches. */
-#define MDBX_MAGIC UINT64_C(/* 56-bit prime */ 0x59659DBDEF4C11)
-
-/* FROZEN: The version number for a database's datafile format. */
-#define MDBX_DATA_VERSION 3
-/* The version number for a database's lockfile format. */
-#define MDBX_LOCK_VERSION 4
-
-/* handle for the DB used to track free pages. */
-#define FREE_DBI 0
-/* handle for the default DB. */
-#define MAIN_DBI 1
-/* Number of DBs in metapage (free and main) - also hardcoded elsewhere */
-#define CORE_DBS 2
-
-/* Number of meta pages - also hardcoded elsewhere */
-#define NUM_METAS 3
-
-/* A page number in the database.
- *
- * MDBX uses 32 bit for page numbers. This limits database
- * size up to 2^44 bytes, in case of 4K pages. */
-typedef uint32_t pgno_t;
-typedef MDBX_atomic_uint32_t atomic_pgno_t;
-#define PRIaPGNO PRIu32
-#define MAX_PAGENO UINT32_C(0x7FFFffff)
-#define MIN_PAGENO NUM_METAS
-
-#define SAFE64_INVALID_THRESHOLD UINT64_C(0xffffFFFF00000000)
-
-/* A transaction ID. */
-typedef uint64_t txnid_t;
-typedef MDBX_atomic_uint64_t atomic_txnid_t;
-#define PRIaTXN PRIi64
-#define MIN_TXNID UINT64_C(1)
-#define MAX_TXNID (SAFE64_INVALID_THRESHOLD - 1)
-#define INITIAL_TXNID (MIN_TXNID + NUM_METAS - 1)
-#define INVALID_TXNID UINT64_MAX
-/* LY: for testing non-atomic 64-bit txnid on 32-bit arches.
- * #define xMDBX_TXNID_STEP (UINT32_MAX / 3) */
-#ifndef xMDBX_TXNID_STEP
-#if MDBX_64BIT_CAS
-#define xMDBX_TXNID_STEP 1u
-#else
-#define xMDBX_TXNID_STEP 2u
-#endif
-#endif /* xMDBX_TXNID_STEP */
-
-/* Used for offsets within a single page.
- * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
- * this is plenty. */
-typedef uint16_t indx_t;
-
-#define MEGABYTE ((size_t)1 << 20)
-
-/*----------------------------------------------------------------------------*/
-/* Core structures for database and shared memory (i.e. format definition) */
-#pragma pack(push, 4)
-
-/* Information about a single database in the environment. */
-typedef struct MDBX_db {
- uint16_t md_flags; /* see mdbx_dbi_open */
- uint16_t md_depth; /* depth of this tree */
- uint32_t md_xsize; /* key-size for MDBX_DUPFIXED (LEAF2 pages) */
- pgno_t md_root; /* the root page of this tree */
- pgno_t md_branch_pages; /* number of internal pages */
- pgno_t md_leaf_pages; /* number of leaf pages */
- pgno_t md_overflow_pages; /* number of overflow pages */
- uint64_t md_seq; /* table sequence counter */
- uint64_t md_entries; /* number of data items */
- uint64_t md_mod_txnid; /* txnid of last committed modification */
-} MDBX_db;
-
-/* database size-related parameters */
-typedef struct MDBX_geo {
- uint16_t grow_pv; /* datafile growth step as a 16-bit packed (exponential
- quantized) value */
- uint16_t shrink_pv; /* datafile shrink threshold as a 16-bit packed
- (exponential quantized) value */
- pgno_t lower; /* minimal size of datafile in pages */
- pgno_t upper; /* maximal size of datafile in pages */
- pgno_t now; /* current size of datafile in pages */
- pgno_t next; /* first unused page in the datafile,
- but actually the file may be shorter. */
-} MDBX_geo;
-
-/* Meta page content.
- * A meta page is the start point for accessing a database snapshot.
- * Pages 0-1 are meta pages. Transaction N writes meta page (N % 2). */
-typedef struct MDBX_meta {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with MDBX_DATA_VERSION. */
- uint32_t mm_magic_and_version[2];
-
- /* txnid that committed this page, the first of a two-phase-update pair */
- uint32_t mm_txnid_a[2];
-
- uint16_t mm_extra_flags; /* extra DB flags, zero (nothing) for now */
- uint8_t mm_validator_id; /* ID of checksum and page validation method,
- * zero (nothing) for now */
- uint8_t mm_extra_pagehdr; /* extra bytes in the page header,
- * zero (nothing) for now */
-
- MDBX_geo mm_geo; /* database size-related parameters */
-
- MDBX_db mm_dbs[CORE_DBS]; /* first is free space, 2nd is main db */
- /* The size of pages used in this DB */
-#define mm_psize mm_dbs[FREE_DBI].md_xsize
- MDBX_canary mm_canary;
-
-#define MDBX_DATASIGN_NONE 0u
-#define MDBX_DATASIGN_WEAK 1u
-#define SIGN_IS_STEADY(sign) ((sign) > MDBX_DATASIGN_WEAK)
-#define META_IS_STEADY(meta) \
- SIGN_IS_STEADY(unaligned_peek_u64_volatile(4, (meta)->mm_datasync_sign))
- uint32_t mm_datasync_sign[2];
-
- /* txnid that committed this page, the second of a two-phase-update pair */
- uint32_t mm_txnid_b[2];
-
- /* Number of non-meta pages which were put in GC after COW. May be 0 in case
- * DB was previously handled by libmdbx without corresponding feature.
- * This value in couple with mr_snapshot_pages_retired allows fast estimation
- * of "how much reader is restraining GC recycling". */
- uint32_t mm_pages_retired[2];
-
- /* The analogue /proc/sys/kernel/random/boot_id or similar to determine
- * whether the system was rebooted after the last use of the database files.
- * If there was no reboot, but there is no need to rollback to the last
- * steady sync point. Zeros mean that no relevant information is available
- * from the system. */
- bin128_t mm_bootid;
-
-} MDBX_meta;
-
-#pragma pack(1)
-
-/* Common header for all page types. The page type depends on mp_flags.
- *
- * P_BRANCH and P_LEAF pages have unsorted 'MDBX_node's at the end, with
- * sorted mp_ptrs[] entries referring to them. Exception: P_LEAF2 pages
- * omit mp_ptrs and pack sorted MDBX_DUPFIXED values after the page header.
- *
- * P_OVERFLOW records occupy one or more contiguous pages where only the
- * first has a page header. They hold the real data of F_BIGDATA nodes.
- *
- * P_SUBP sub-pages are small leaf "pages" with duplicate data.
- * A node with flag F_DUPDATA but not F_SUBDATA contains a sub-page.
- * (Duplicate data can also go in sub-databases, which use normal pages.)
- *
- * P_META pages contain MDBX_meta, the start point of an MDBX snapshot.
- *
- * Each non-metapage up to MDBX_meta.mm_last_pg is reachable exactly once
- * in the snapshot: Either used by a database or listed in a GC record. */
-typedef struct MDBX_page {
- union {
-#define IS_FROZEN(txn, p) ((p)->mp_txnid < (txn)->mt_txnid)
-#define IS_SPILLED(txn, p) ((p)->mp_txnid == (txn)->mt_txnid)
-#define IS_SHADOWED(txn, p) ((p)->mp_txnid > (txn)->mt_txnid)
-#define IS_VALID(txn, p) ((p)->mp_txnid <= (txn)->mt_front)
-#define IS_MODIFIABLE(txn, p) ((p)->mp_txnid == (txn)->mt_front)
- uint64_t mp_txnid;
- struct MDBX_page *mp_next; /* for in-memory list of freed pages */
- };
- uint16_t mp_leaf2_ksize; /* key size if this is a LEAF2 page */
-#define P_BRANCH 0x01 /* branch page */
-#define P_LEAF 0x02 /* leaf page */
-#define P_OVERFLOW 0x04 /* overflow page */
-#define P_META 0x08 /* meta page */
-#define P_BAD 0x10 /* explicit flag for invalid/bad page */
-#define P_LEAF2 0x20 /* for MDBX_DUPFIXED records */
-#define P_SUBP 0x40 /* for MDBX_DUPSORT sub-pages */
-#define P_SPILLED 0x2000 /* spilled in parent txn */
-#define P_LOOSE 0x4000 /* page was dirtied then freed, can be reused */
-#define P_FROZEN 0x8000 /* used for retire page with known status */
-#define P_ILL_BITS (~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_SPILLED))
- uint16_t mp_flags;
- union {
- uint32_t mp_pages; /* number of overflow pages */
- __anonymous_struct_extension__ struct {
- indx_t mp_lower; /* lower bound of free space */
- indx_t mp_upper; /* upper bound of free space */
- };
- };
- pgno_t mp_pgno; /* page number */
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- indx_t mp_ptrs[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_page;
-
-/* Size of the page header, excluding dynamic data at the end */
-#define PAGEHDRSZ ((unsigned)offsetof(MDBX_page, mp_ptrs))
-
-#pragma pack(pop)
-
-#if MDBX_ENABLE_PGOP_STAT
-/* Statistics of page operations overall of all (running, completed and aborted)
- * transactions */
-typedef struct {
- MDBX_atomic_uint64_t newly; /* Quantity of a new pages added */
- MDBX_atomic_uint64_t cow; /* Quantity of pages copied for update */
- MDBX_atomic_uint64_t clone; /* Quantity of parent's dirty pages clones
- for nested transactions */
- MDBX_atomic_uint64_t split; /* Page splits */
- MDBX_atomic_uint64_t merge; /* Page merges */
- MDBX_atomic_uint64_t spill; /* Quantity of spilled dirty pages */
- MDBX_atomic_uint64_t unspill; /* Quantity of unspilled/reloaded pages */
- MDBX_atomic_uint64_t
- wops; /* Number of explicit write operations (not a pages) to a disk */
-} MDBX_pgop_stat_t;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-#if MDBX_LOCKING == MDBX_LOCKING_WIN32FILES
-#define MDBX_CLOCK_SIGN UINT32_C(0xF10C)
-typedef void mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
-
-#define MDBX_CLOCK_SIGN UINT32_C(0xF18D)
-typedef mdbx_pid_t mdbx_ipclock_t;
-#ifndef EOWNERDEAD
-#define EOWNERDEAD MDBX_RESULT_TRUE
-#endif
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#define MDBX_CLOCK_SIGN UINT32_C(0x8017)
-typedef pthread_mutex_t mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-#define MDBX_CLOCK_SIGN UINT32_C(0xFC29)
-typedef sem_t mdbx_ipclock_t;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV && !defined(__cplusplus)
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc);
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc);
-#endif /* MDBX_LOCKING */
-
-/* Reader Lock Table
- *
- * Readers don't acquire any locks for their data access. Instead, they
- * simply record their transaction ID in the reader table. The reader
- * mutex is needed just to find an empty slot in the reader table. The
- * slot's address is saved in thread-specific data so that subsequent
- * read transactions started by the same thread need no further locking to
- * proceed.
- *
- * If MDBX_NOTLS is set, the slot address is not saved in thread-specific data.
- * No reader table is used if the database is on a read-only filesystem.
- *
- * Since the database uses multi-version concurrency control, readers don't
- * actually need any locking. This table is used to keep track of which
- * readers are using data from which old transactions, so that we'll know
- * when a particular old transaction is no longer in use. Old transactions
- * that have discarded any data pages can then have those pages reclaimed
- * for use by a later write transaction.
- *
- * The lock table is constructed such that reader slots are aligned with the
- * processor's cache line size. Any slot is only ever used by one thread.
- * This alignment guarantees that there will be no contention or cache
- * thrashing as threads update their own slot info, and also eliminates
- * any need for locking when accessing a slot.
- *
- * A writer thread will scan every slot in the table to determine the oldest
- * outstanding reader transaction. Any freed pages older than this will be
- * reclaimed by the writer. The writer doesn't use any locks when scanning
- * this table. This means that there's no guarantee that the writer will
- * see the most up-to-date reader info, but that's not required for correct
- * operation - all we need is to know the upper bound on the oldest reader,
- * we don't care at all about the newest reader. So the only consequence of
- * reading stale information here is that old pages might hang around a
- * while longer before being reclaimed. That's actually good anyway, because
- * the longer we delay reclaiming old pages, the more likely it is that a
- * string of contiguous pages can be found after coalescing old pages from
- * many old transactions together. */
-
-/* The actual reader record, with cacheline padding. */
-typedef struct MDBX_reader {
- /* Current Transaction ID when this transaction began, or (txnid_t)-1.
- * Multiple readers that start at the same time will probably have the
- * same ID here. Again, it's not important to exclude them from
- * anything; all we need to know is which version of the DB they
- * started from so we can avoid overwriting any data used in that
- * particular version. */
- MDBX_atomic_uint64_t /* txnid_t */ mr_txnid;
-
- /* The information we store in a single slot of the reader table.
- * In addition to a transaction ID, we also record the process and
- * thread ID that owns a slot, so that we can detect stale information,
- * e.g. threads or processes that went away without cleaning up.
- *
- * NOTE: We currently don't check for stale records.
- * We simply re-init the table when we know that we're the only process
- * opening the lock file. */
-
- /* The thread ID of the thread owning this txn. */
- MDBX_atomic_uint64_t mr_tid;
-
- /* The process ID of the process owning this reader txn. */
- MDBX_atomic_uint32_t mr_pid;
-
- /* The number of pages used in the reader's MVCC snapshot,
- * i.e. the value of meta->mm_geo.next and txn->mt_next_pgno */
- atomic_pgno_t mr_snapshot_pages_used;
- /* Number of retired pages at the time this reader starts transaction. So,
- * at any time the difference mm_pages_retired - mr_snapshot_pages_retired
- * will give the number of pages which this reader restraining from reuse. */
- MDBX_atomic_uint64_t mr_snapshot_pages_retired;
-} MDBX_reader;
-
-/* The header for the reader table (a memory-mapped lock file). */
-typedef struct MDBX_lockinfo {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with with MDBX_LOCK_VERSION. */
- uint64_t mti_magic_and_version;
-
- /* Format of this lock file. Must be set to MDBX_LOCK_FORMAT. */
- uint32_t mti_os_and_format;
-
- /* Flags which environment was opened. */
- MDBX_atomic_uint32_t mti_envmode;
-
- /* Threshold of un-synced-with-disk pages for auto-sync feature,
- * zero means no-threshold, i.e. auto-sync is disabled. */
- atomic_pgno_t mti_autosync_threshold;
-
- /* Low 32-bit of txnid with which meta-pages was synced,
- * i.e. for sync-polling in the MDBX_NOMETASYNC mode. */
- MDBX_atomic_uint32_t mti_meta_sync_txnid;
-
- /* Period for timed auto-sync feature, i.e. at the every steady checkpoint
- * the mti_unsynced_timeout sets to the current_time + mti_autosync_period.
- * The time value is represented in a suitable system-dependent form, for
- * example clock_gettime(CLOCK_BOOTTIME) or clock_gettime(CLOCK_MONOTONIC).
- * Zero means timed auto-sync is disabled. */
- MDBX_atomic_uint64_t mti_autosync_period;
-
- /* Marker to distinguish uniqueness of DB/CLK. */
- MDBX_atomic_uint64_t mti_bait_uniqueness;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
-#if MDBX_ENABLE_PGOP_STAT
- /* Statistics of costly ops of all (running, completed and aborted)
- * transactions */
- MDBX_pgop_stat_t mti_pgop_stat;
-#endif /* MDBX_ENABLE_PGOP_STAT*/
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Write transaction lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_wlock;
-#endif /* MDBX_LOCKING > 0 */
-
- atomic_txnid_t mti_oldest_reader;
-
- /* Timestamp of the last steady sync. Value is represented in a suitable
- * system-dependent form, for example clock_gettime(CLOCK_BOOTTIME) or
- * clock_gettime(CLOCK_MONOTONIC). */
- MDBX_atomic_uint64_t mti_sync_timestamp;
-
- /* Number un-synced-with-disk pages for auto-sync feature. */
- atomic_pgno_t mti_unsynced_pages;
-
- /* Number of page which was discarded last time by madvise(MADV_FREE). */
- atomic_pgno_t mti_discarded_tail;
-
- /* Timestamp of the last readers check. */
- MDBX_atomic_uint64_t mti_reader_check_timestamp;
-
- /* Shared anchor for tracking readahead edge and enabled/disabled status. */
- pgno_t mti_readahead_anchor;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Readeaders registration lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_rlock;
-#endif /* MDBX_LOCKING > 0 */
-
- /* The number of slots that have been used in the reader table.
- * This always records the maximum count, it is not decremented
- * when readers release their slots. */
- MDBX_atomic_uint32_t mti_numreaders;
- MDBX_atomic_uint32_t mti_readers_refresh_flag;
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
- MDBX_reader mti_readers[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_lockinfo;
-
-/* Lockfile format signature: version, features and field layout */
-#define MDBX_LOCK_FORMAT \
- (MDBX_CLOCK_SIGN * 27733 + (unsigned)sizeof(MDBX_reader) * 13 + \
- (unsigned)offsetof(MDBX_reader, mr_snapshot_pages_used) * 251 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_oldest_reader) * 83 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_numreaders) * 37 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_readers) * 29)
-
-#define MDBX_DATA_MAGIC \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + MDBX_DATA_VERSION)
-
-#define MDBX_DATA_MAGIC_LEGACY_COMPAT \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + 2)
-
-#define MDBX_DATA_MAGIC_LEGACY_DEVEL ((MDBX_MAGIC << 8) + 255)
-
-#define MDBX_LOCK_MAGIC ((MDBX_MAGIC << 8) + MDBX_LOCK_VERSION)
-
-/* The maximum size of a database page.
- *
- * It is 64K, but value-PAGEHDRSZ must fit in MDBX_page.mp_upper.
- *
- * MDBX will use database pages < OS pages if needed.
- * That causes more I/O in write transactions: The OS must
- * know (read) the whole page before writing a partial page.
- *
- * Note that we don't currently support Huge pages. On Linux,
- * regular data files cannot use Huge pages, and in general
- * Huge pages aren't actually pageable. We rely on the OS
- * demand-pager to read our data and page it out when memory
- * pressure from other processes is high. So until OSs have
- * actual paging support for Huge pages, they're not viable. */
-#define MAX_PAGESIZE MDBX_MAX_PAGESIZE
-#define MIN_PAGESIZE MDBX_MIN_PAGESIZE
-
-#define MIN_MAPSIZE (MIN_PAGESIZE * MIN_PAGENO)
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_MAPSIZE32 UINT32_C(0x38000000)
-#else
-#define MAX_MAPSIZE32 UINT32_C(0x7f000000)
-#endif
-#define MAX_MAPSIZE64 ((MAX_PAGENO + 1) * (uint64_t)MAX_PAGESIZE)
-
-#if MDBX_WORDBITS >= 64
-#define MAX_MAPSIZE MAX_MAPSIZE64
-#define MDBX_PGL_LIMIT ((size_t)MAX_PAGENO)
-#else
-#define MAX_MAPSIZE MAX_MAPSIZE32
-#define MDBX_PGL_LIMIT (MAX_MAPSIZE32 / MIN_PAGESIZE)
-#endif /* MDBX_WORDBITS */
-
-#define MDBX_READERS_LIMIT 32767
-#define MDBX_RADIXSORT_THRESHOLD 333
-
-/*----------------------------------------------------------------------------*/
-
-/* An PNL is an Page Number List, a sorted array of IDs.
- * The first element of the array is a counter for how many actual page-numbers
- * are in the list. By default PNLs are sorted in descending order, this allow
- * cut off a page with lowest pgno (at the tail) just truncating the list. The
- * sort order of PNLs is controlled by the MDBX_PNL_ASCENDING build option. */
-typedef pgno_t *MDBX_PNL;
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_ORDERED(first, last) ((first) < (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) >= (last))
-#else
-#define MDBX_PNL_ORDERED(first, last) ((first) > (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) <= (last))
-#endif
-
-/* List of txnid, only for MDBX_txn.tw.lifo_reclaimed */
-typedef txnid_t *MDBX_TXL;
-
-/* An Dirty-Page list item is an pgno/pointer pair. */
-typedef struct MDBX_dp {
- MDBX_page *ptr;
- pgno_t pgno;
- union {
- unsigned extra;
- __anonymous_struct_extension__ struct {
- unsigned multi : 1;
- unsigned lru : 31;
- };
- };
-} MDBX_dp;
-
-/* An DPL (dirty-page list) is a sorted array of MDBX_DPs. */
-typedef struct MDBX_dpl {
- unsigned sorted;
- unsigned length;
- unsigned detent; /* allocated size excluding the MDBX_DPL_RESERVE_GAP */
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_dp items[] /* dynamic size with holes at zero and after the last */;
-#endif
-} MDBX_dpl;
-
-/* PNL sizes */
-#define MDBX_PNL_GRANULATE 1024
-#define MDBX_PNL_INITIAL \
- (MDBX_PNL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(pgno_t))
-
-#define MDBX_TXL_GRANULATE 32
-#define MDBX_TXL_INITIAL \
- (MDBX_TXL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-#define MDBX_TXL_MAX \
- ((1u << 17) - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-
-#define MDBX_PNL_ALLOCLEN(pl) ((pl)[-1])
-#define MDBX_PNL_SIZE(pl) ((pl)[0])
-#define MDBX_PNL_FIRST(pl) ((pl)[1])
-#define MDBX_PNL_LAST(pl) ((pl)[MDBX_PNL_SIZE(pl)])
-#define MDBX_PNL_BEGIN(pl) (&(pl)[1])
-#define MDBX_PNL_END(pl) (&(pl)[MDBX_PNL_SIZE(pl) + 1])
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_FIRST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_LAST(pl)
-#else
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_LAST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_FIRST(pl)
-#endif
-
-#define MDBX_PNL_SIZEOF(pl) ((MDBX_PNL_SIZE(pl) + 1) * sizeof(pgno_t))
-#define MDBX_PNL_IS_EMPTY(pl) (MDBX_PNL_SIZE(pl) == 0)
-
-/*----------------------------------------------------------------------------*/
-/* Internal structures */
-
-/* Auxiliary DB info.
- * The information here is mostly static/read-only. There is
- * only a single copy of this record in the environment. */
-typedef struct MDBX_dbx {
- MDBX_val md_name; /* name of the database */
- MDBX_cmp_func *md_cmp; /* function for comparing keys */
- MDBX_cmp_func *md_dcmp; /* function for comparing data items */
- size_t md_klen_min, md_klen_max; /* min/max key length for the database */
- size_t md_vlen_min,
- md_vlen_max; /* min/max value/data length for the database */
-} MDBX_dbx;
-
-/* A database transaction.
- * Every operation requires a transaction handle. */
-struct MDBX_txn {
-#define MDBX_MT_SIGNATURE UINT32_C(0x93D53A31)
- uint32_t mt_signature;
-
- /* Transaction Flags */
- /* mdbx_txn_begin() flags */
-#define MDBX_TXN_RO_BEGIN_FLAGS (MDBX_TXN_RDONLY | MDBX_TXN_RDONLY_PREPARE)
-#define MDBX_TXN_RW_BEGIN_FLAGS \
- (MDBX_TXN_NOMETASYNC | MDBX_TXN_NOSYNC | MDBX_TXN_TRY)
- /* Additional flag for mdbx_sync_locked() */
-#define MDBX_SHRINK_ALLOWED UINT32_C(0x40000000)
-
- /* internal txn flags */
-#define MDBX_TXN_FINISHED 0x01 /* txn is finished or never began */
-#define MDBX_TXN_ERROR 0x02 /* txn is unusable after an error */
-#define MDBX_TXN_DIRTY 0x04 /* must write, even if dirty list is empty */
-#define MDBX_TXN_SPILLS 0x08 /* txn or a parent has spilled pages */
-#define MDBX_TXN_HAS_CHILD 0x10 /* txn has an MDBX_txn.mt_child */
- /* most operations on the txn are currently illegal */
-#define MDBX_TXN_BLOCKED \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_HAS_CHILD)
-
-#define TXN_FLAGS \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_DIRTY | MDBX_TXN_SPILLS | \
- MDBX_TXN_HAS_CHILD)
-
-#if (TXN_FLAGS & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS)) || \
- ((MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS | TXN_FLAGS) & \
- MDBX_SHRINK_ALLOWED)
-#error "Oops, some flags overlapped or wrong"
-#endif
- uint32_t mt_flags;
-
- MDBX_txn *mt_parent; /* parent of a nested txn */
- /* Nested txn under this txn, set together with flag MDBX_TXN_HAS_CHILD */
- MDBX_txn *mt_child;
- MDBX_geo mt_geo;
- /* next unallocated page */
-#define mt_next_pgno mt_geo.next
- /* corresponding to the current size of datafile */
-#define mt_end_pgno mt_geo.now
-
- /* The ID of this transaction. IDs are integers incrementing from 1.
- * Only committed write transactions increment the ID. If a transaction
- * aborts, the ID may be re-used by the next writer. */
- txnid_t mt_txnid;
- txnid_t mt_front;
-
- MDBX_env *mt_env; /* the DB environment */
- /* Array of records for each DB known in the environment. */
- MDBX_dbx *mt_dbxs;
- /* Array of MDBX_db records for each known DB */
- MDBX_db *mt_dbs;
- /* Array of sequence numbers for each DB handle */
- unsigned *mt_dbiseqs;
-
- /* Transaction DBI Flags */
-#define DBI_DIRTY MDBX_DBI_DIRTY /* DB was written in this txn */
-#define DBI_STALE MDBX_DBI_STALE /* Named-DB record is older than txnID */
-#define DBI_FRESH MDBX_DBI_FRESH /* Named-DB handle opened in this txn */
-#define DBI_CREAT MDBX_DBI_CREAT /* Named-DB handle created in this txn */
-#define DBI_VALID 0x10 /* DB handle is valid, see also DB_VALID */
-#define DBI_USRVALID 0x20 /* As DB_VALID, but not set for FREE_DBI */
-#define DBI_AUDITED 0x40 /* Internal flag for accounting during audit */
- /* Array of flags for each DB */
- uint8_t *mt_dbistate;
- /* Number of DB records in use, or 0 when the txn is finished.
- * This number only ever increments until the txn finishes; we
- * don't decrement it when individual DB handles are closed. */
- MDBX_dbi mt_numdbs;
- size_t mt_owner; /* thread ID that owns this transaction */
- MDBX_canary mt_canary;
- void *mt_userctx; /* User-settable context */
- MDBX_cursor **mt_cursors;
-
- union {
- struct {
- /* For read txns: This thread/txn's reader table slot, or NULL. */
- MDBX_reader *reader;
- } to;
- struct {
- /* In write txns, array of cursors for each DB */
- pgno_t *reclaimed_pglist; /* Reclaimed GC pages */
- txnid_t last_reclaimed; /* ID of last used record */
-#if MDBX_ENABLE_REFUND
- pgno_t loose_refund_wl /* FIXME: describe */;
-#endif /* MDBX_ENABLE_REFUND */
- /* dirtylist room: Dirty array size - dirty pages visible to this txn.
- * Includes ancestor txns' dirty pages not hidden by other txns'
- * dirty/spilled pages. Thus commit(nested txn) has room to merge
- * dirtylist into mt_parent after freeing hidden mt_parent pages. */
- unsigned dirtyroom;
- /* a sequence to spilling dirty page with LRU policy */
- unsigned dirtylru;
- /* For write txns: Modified pages. Sorted when not MDBX_WRITEMAP. */
- MDBX_dpl *dirtylist;
- /* The list of reclaimed txns from GC */
- MDBX_TXL lifo_reclaimed;
- /* The list of pages that became unused during this transaction. */
- MDBX_PNL retired_pages;
- /* The list of loose pages that became unused and may be reused
- * in this transaction, linked through `mp_next`. */
- MDBX_page *loose_pages;
- /* Number of loose pages (tw.loose_pages) */
- unsigned loose_count;
- /* The sorted list of dirty pages we temporarily wrote to disk
- * because the dirty list was full. page numbers in here are
- * shifted left by 1, deleted slots have the LSB set. */
- MDBX_PNL spill_pages;
- unsigned spill_least_removed;
- } tw;
- };
-};
-
-#if MDBX_WORDBITS >= 64
-#define CURSOR_STACK 32
-#else
-#define CURSOR_STACK 24
-#endif
-
-struct MDBX_xcursor;
-
-/* Cursors are used for all DB operations.
- * A cursor holds a path of (page pointer, key index) from the DB
- * root to a position in the DB, plus other state. MDBX_DUPSORT
- * cursors include an xcursor to the current data item. Write txns
- * track their cursors and keep them up to date when data moves.
- * Exception: An xcursor's pointer to a P_SUBP page can be stale.
- * (A node with F_DUPDATA but no F_SUBDATA contains a subpage). */
-struct MDBX_cursor {
-#define MDBX_MC_LIVE UINT32_C(0xFE05D5B1)
-#define MDBX_MC_READY4CLOSE UINT32_C(0x2817A047)
-#define MDBX_MC_WAIT4EOT UINT32_C(0x90E297A7)
- uint32_t mc_signature;
- /* The database handle this cursor operates on */
- MDBX_dbi mc_dbi;
- /* Next cursor on this DB in this txn */
- MDBX_cursor *mc_next;
- /* Backup of the original cursor if this cursor is a shadow */
- MDBX_cursor *mc_backup;
- /* Context used for databases with MDBX_DUPSORT, otherwise NULL */
- struct MDBX_xcursor *mc_xcursor;
- /* The transaction that owns this cursor */
- MDBX_txn *mc_txn;
- /* The database record for this cursor */
- MDBX_db *mc_db;
- /* The database auxiliary record for this cursor */
- MDBX_dbx *mc_dbx;
- /* The mt_dbistate for this database */
- uint8_t *mc_dbistate;
- unsigned mc_snum; /* number of pushed pages */
- unsigned mc_top; /* index of top page, normally mc_snum-1 */
-
- /* Cursor state flags. */
-#define C_INITIALIZED 0x01 /* cursor has been initialized and is valid */
-#define C_EOF 0x02 /* No more data */
-#define C_SUB 0x04 /* Cursor is a sub-cursor */
-#define C_DEL 0x08 /* last op was a cursor_del */
-#define C_UNTRACK 0x10 /* Un-track cursor when closing */
-#define C_RECLAIMING 0x20 /* GC lookup is prohibited */
-#define C_GCFREEZE 0x40 /* reclaimed_pglist must not be updated */
-
- /* Cursor checking flags. */
-#define C_COPYING 0x100 /* skip key-value length check (copying simplify) */
-#define C_UPDATING 0x200 /* update/rebalance pending */
-#define C_RETIRING 0x400 /* refs to child pages may be invalid */
-#define C_SKIPORD 0x800 /* don't check keys ordering */
-
- unsigned mc_flags; /* see mdbx_cursor */
- MDBX_page *mc_pg[CURSOR_STACK]; /* stack of pushed pages */
- indx_t mc_ki[CURSOR_STACK]; /* stack of page indices */
-};
-
-/* Context for sorted-dup records.
- * We could have gone to a fully recursive design, with arbitrarily
- * deep nesting of sub-databases. But for now we only handle these
- * levels - main DB, optional sub-DB, sorted-duplicate DB. */
-typedef struct MDBX_xcursor {
- /* A sub-cursor for traversing the Dup DB */
- MDBX_cursor mx_cursor;
- /* The database record for this Dup DB */
- MDBX_db mx_db;
- /* The auxiliary DB record for this Dup DB */
- MDBX_dbx mx_dbx;
-} MDBX_xcursor;
-
-typedef struct MDBX_cursor_couple {
- MDBX_cursor outer;
- void *mc_userctx; /* User-settable context */
- MDBX_xcursor inner;
-} MDBX_cursor_couple;
-
-/* The database environment. */
-struct MDBX_env {
- /* ----------------------------------------------------- mostly static part */
-#define MDBX_ME_SIGNATURE UINT32_C(0x9A899641)
- MDBX_atomic_uint32_t me_signature;
- /* Failed to update the meta page. Probably an I/O error. */
-#define MDBX_FATAL_ERROR UINT32_C(0x80000000)
- /* Some fields are initialized. */
-#define MDBX_ENV_ACTIVE UINT32_C(0x20000000)
- /* me_txkey is set */
-#define MDBX_ENV_TXKEY UINT32_C(0x10000000)
- /* Legacy MDBX_MAPASYNC (prior v0.9) */
-#define MDBX_DEPRECATED_MAPASYNC UINT32_C(0x100000)
-#define ENV_INTERNAL_FLAGS (MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE | MDBX_ENV_TXKEY)
- uint32_t me_flags;
- mdbx_mmap_t me_dxb_mmap; /* The main data file */
-#define me_map me_dxb_mmap.dxb
-#define me_lazy_fd me_dxb_mmap.fd
- mdbx_filehandle_t me_dsync_fd;
- mdbx_mmap_t me_lck_mmap; /* The lock file */
-#define me_lfd me_lck_mmap.fd
- struct MDBX_lockinfo *me_lck;
-
- unsigned me_psize; /* DB page size, initialized from me_os_psize */
- unsigned me_leaf_nodemax; /* max size of a leaf-node */
- uint8_t me_psize2log; /* log2 of DB page size */
- int8_t me_stuck_meta; /* recovery-only: target meta page or less that zero */
- uint16_t me_merge_threshold,
- me_merge_threshold_gc; /* pages emptier than this are candidates for
- merging */
- unsigned me_os_psize; /* OS page size, from mdbx_syspagesize() */
- unsigned me_maxreaders; /* size of the reader table */
- MDBX_dbi me_maxdbs; /* size of the DB table */
- uint32_t me_pid; /* process ID of this env */
- mdbx_thread_key_t me_txkey; /* thread-key for readers */
- char *me_pathname; /* path to the DB files */
- void *me_pbuf; /* scratch area for DUPSORT put() */
- MDBX_txn *me_txn0; /* preallocated write transaction */
-
- MDBX_dbx *me_dbxs; /* array of static DB info */
- uint16_t *me_dbflags; /* array of flags from MDBX_db.md_flags */
- unsigned *me_dbiseqs; /* array of dbi sequence numbers */
- unsigned
- me_maxgc_ov1page; /* Number of pgno_t fit in a single overflow page */
- uint32_t me_live_reader; /* have liveness lock in reader table */
- void *me_userctx; /* User-settable context */
- MDBX_hsr_func *me_hsr_callback; /* Callback for kicking laggard readers */
-
- struct {
- unsigned dp_reserve_limit;
- unsigned rp_augment_limit;
- unsigned dp_limit;
- unsigned dp_initial;
- uint8_t dp_loose_limit;
- uint8_t spill_max_denominator;
- uint8_t spill_min_denominator;
- uint8_t spill_parent4child_denominator;
- unsigned merge_threshold_16dot16_percent;
- union {
- unsigned all;
- /* tracks options with non-auto values but tuned by user */
- struct {
- unsigned dp_limit : 1;
- } non_auto;
- } flags;
- } me_options;
-
- /* struct me_dbgeo used for accepting db-geo params from user for the new
- * database creation, i.e. when mdbx_env_set_geometry() was called before
- * mdbx_env_open(). */
- struct {
- size_t lower; /* minimal size of datafile */
- size_t upper; /* maximal size of datafile */
- size_t now; /* current size of datafile */
- size_t grow; /* step to grow datafile */
- size_t shrink; /* threshold to shrink datafile */
- } me_dbgeo;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- union {
- key_t key;
- int semid;
- } me_sysv_ipc;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
- MDBX_env *me_lcklist_next;
-
- /* --------------------------------------------------- mostly volatile part */
-
- MDBX_txn *me_txn; /* current write transaction */
- mdbx_fastmutex_t me_dbi_lock;
- MDBX_dbi me_numdbs; /* number of DBs opened */
-
- MDBX_page *me_dp_reserve; /* list of malloc'ed blocks for re-use */
- unsigned me_dp_reserve_len;
- /* PNL of pages that became unused in a write txn */
- MDBX_PNL me_retired_pages;
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_srwlock me_remap_guard;
- /* Workaround for LockFileEx and WriteFile multithread bug */
- CRITICAL_SECTION me_windowsbug_lock;
-#else
- mdbx_fastmutex_t me_remap_guard;
-#endif
-
- /* -------------------------------------------------------------- debugging */
-
-#if MDBX_DEBUG
- MDBX_assert_func *me_assert_func; /* Callback for assertion failures */
-#endif
-#ifdef MDBX_USE_VALGRIND
- int me_valgrind_handle;
-#endif
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- pgno_t me_poison_edge;
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-#ifndef xMDBX_DEBUG_SPILLING
-#define xMDBX_DEBUG_SPILLING 0
-#endif
-#if xMDBX_DEBUG_SPILLING == 2
- unsigned debug_dirtied_est, debug_dirtied_act;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- /* ------------------------------------------------- stub for lck-less mode */
- MDBX_atomic_uint64_t
- x_lckless_stub[(sizeof(MDBX_lockinfo) + MDBX_CACHELINE_SIZE - 1) /
- sizeof(MDBX_atomic_uint64_t)];
-};
-
-#ifndef __cplusplus
-/*----------------------------------------------------------------------------*/
-/* Debug and Logging stuff */
-
-#define MDBX_RUNTIME_FLAGS_INIT \
- ((MDBX_DEBUG) > 0) * MDBX_DBG_ASSERT + ((MDBX_DEBUG) > 1) * MDBX_DBG_AUDIT
-
-extern uint8_t mdbx_runtime_flags;
-extern uint8_t mdbx_loglevel;
-extern MDBX_debug_func *mdbx_debug_logger;
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny) {
-#if MDBX_DEBUG
- if (MDBX_DBG_JITTER & mdbx_runtime_flags)
- mdbx_osal_jitter(tiny);
-#else
- (void)tiny;
-#endif
-}
-
-MDBX_INTERNAL_FUNC void MDBX_PRINTF_ARGS(4, 5)
- mdbx_debug_log(int level, const char *function, int line, const char *fmt,
- ...) MDBX_PRINTF_ARGS(4, 5);
-MDBX_INTERNAL_FUNC void mdbx_debug_log_va(int level, const char *function,
- int line, const char *fmt,
- va_list args);
-
-#if MDBX_DEBUG
-#define mdbx_log_enabled(msg) unlikely(msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() unlikely((mdbx_runtime_flags & MDBX_DBG_AUDIT))
-#else /* MDBX_DEBUG */
-#define mdbx_log_enabled(msg) (msg < MDBX_LOG_VERBOSE && msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() (0)
-#endif /* MDBX_DEBUG */
-
-#if MDBX_FORCE_ASSERTIONS
-#define mdbx_assert_enabled() (1)
-#elif MDBX_DEBUG
-#define mdbx_assert_enabled() likely((mdbx_runtime_flags & MDBX_DBG_ASSERT))
-#else
-#define mdbx_assert_enabled() (0)
-#endif /* assertions */
-
-#if !MDBX_DEBUG && defined(__ANDROID_API__)
-#define mdbx_assert_fail(env, msg, func, line) \
- __android_log_assert(msg, "mdbx", "%s:%u", func, line)
-#else
-void mdbx_assert_fail(const MDBX_env *env, const char *msg, const char *func,
- int line);
-#endif
-
-#define mdbx_debug_extra(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, __func__, __LINE__, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug_extra_print(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, NULL, 0, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_trace(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_TRACE)) \
- mdbx_debug_log(MDBX_LOG_TRACE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_DEBUG)) \
- mdbx_debug_log(MDBX_LOG_DEBUG, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_verbose(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_VERBOSE)) \
- mdbx_debug_log(MDBX_LOG_VERBOSE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_notice(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_NOTICE)) \
- mdbx_debug_log(MDBX_LOG_NOTICE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_warning(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_WARN)) \
- mdbx_debug_log(MDBX_LOG_WARN, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_error(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) \
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_fatal(fmt, ...) \
- mdbx_debug_log(MDBX_LOG_FATAL, __func__, __LINE__, fmt "\n", __VA_ARGS__);
-
-#define mdbx_ensure_msg(env, expr, msg) \
- do { \
- if (unlikely(!(expr))) \
- mdbx_assert_fail(env, msg, __func__, __LINE__); \
- } while (0)
-
-#define mdbx_ensure(env, expr) mdbx_ensure_msg(env, expr, #expr)
-
-/* assert(3) variant in environment context */
-#define mdbx_assert(env, expr) \
- do { \
- if (mdbx_assert_enabled()) \
- mdbx_ensure(env, expr); \
- } while (0)
-
-/* assert(3) variant in cursor context */
-#define mdbx_cassert(mc, expr) mdbx_assert((mc)->mc_txn->mt_env, expr)
-
-/* assert(3) variant in transaction context */
-#define mdbx_tassert(txn, expr) mdbx_assert((txn)->mt_env, expr)
-
-#ifndef xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-#undef assert
-#define assert(expr) mdbx_assert(NULL, expr)
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Cache coherence and mmap invalidation */
-
-#if MDBX_CPU_WRITEBACK_INCOHERENT
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_memory_barrier()
-#else
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_compiler_barrier()
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-MDBX_MAYBE_UNUSED static __inline void
-mdbx_flush_incoherent_mmap(void *addr, size_t nbytes, const intptr_t pagesize) {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- char *const begin = (char *)(-pagesize & (intptr_t)addr);
- char *const end =
- (char *)(-pagesize & (intptr_t)((char *)addr + nbytes + pagesize - 1));
- int err = msync(begin, end - begin, MS_SYNC | MS_INVALIDATE) ? errno : 0;
- mdbx_assert(nullptr, err == 0);
- (void)err;
-#else
- (void)pagesize;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#if MDBX_MMAP_INCOHERENT_CPU_CACHE
-#ifdef DCACHE
- /* MIPS has cache coherency issues.
- * Note: for any nbytes >= on-chip cache size, entire is flushed. */
- cacheflush(addr, nbytes, DCACHE);
-#else
-#error "Oops, cacheflush() not available"
-#endif /* DCACHE */
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#if !MDBX_MMAP_INCOHERENT_FILE_WRITE && !MDBX_MMAP_INCOHERENT_CPU_CACHE
- (void)addr;
- (void)nbytes;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* Internal prototypes */
-
-MDBX_INTERNAL_FUNC int mdbx_cleanup_dead_readers(MDBX_env *env, int rlocked,
- int *dead);
-MDBX_INTERNAL_FUNC int mdbx_rthc_alloc(mdbx_thread_key_t *key,
- MDBX_reader *begin, MDBX_reader *end);
-MDBX_INTERNAL_FUNC void mdbx_rthc_remove(const mdbx_thread_key_t key);
-
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_init(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_dtor(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_thread_dtor(void *ptr);
-
-#endif /* !__cplusplus */
-
-#define MDBX_IS_ERROR(rc) \
- ((rc) != MDBX_RESULT_TRUE && (rc) != MDBX_RESULT_FALSE)
-
-/* Internal error codes, not exposed outside libmdbx */
-#define MDBX_NO_ROOT (MDBX_LAST_ADDED_ERRCODE + 10)
-
-/* Debugging output value of a cursor DBI: Negative in a sub-cursor. */
-#define DDBI(mc) \
- (((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
-
-/* Key size which fits in a DKBUF (debug key buffer). */
-#define DKBUF_MAX 511
-#define DKBUF char _kbuf[DKBUF_MAX * 4 + 2]
-#define DKEY(x) mdbx_dump_val(x, _kbuf, DKBUF_MAX * 2 + 1)
-#define DVAL(x) mdbx_dump_val(x, _kbuf + DKBUF_MAX * 2 + 1, DKBUF_MAX * 2 + 1)
-
-#if MDBX_DEBUG
-#define DKBUF_DEBUG DKBUF
-#define DKEY_DEBUG(x) DKEY(x)
-#define DVAL_DEBUG(x) DVAL(x)
-#else
-#define DKBUF_DEBUG ((void)(0))
-#define DKEY_DEBUG(x) ("-")
-#define DVAL_DEBUG(x) ("-")
-#endif
-
-/* An invalid page number.
- * Mainly used to denote an empty tree. */
-#define P_INVALID (~(pgno_t)0)
-
-/* Test if the flags f are set in a flag word w. */
-#define F_ISSET(w, f) (((w) & (f)) == (f))
-
-/* Round n up to an even number. */
-#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
-
-/* Default size of memory map.
- * This is certainly too small for any actual applications. Apps should
- * always set the size explicitly using mdbx_env_set_geometry(). */
-#define DEFAULT_MAPSIZE MEGABYTE
-
-/* Number of slots in the reader table.
- * This value was chosen somewhat arbitrarily. The 61 is a prime number,
- * and such readers plus a couple mutexes fit into single 4KB page.
- * Applications should set the table size using mdbx_env_set_maxreaders(). */
-#define DEFAULT_READERS 61
-
-/* Test if a page is a leaf page */
-#define IS_LEAF(p) (((p)->mp_flags & P_LEAF) != 0)
-/* Test if a page is a LEAF2 page */
-#define IS_LEAF2(p) unlikely(((p)->mp_flags & P_LEAF2) != 0)
-/* Test if a page is a branch page */
-#define IS_BRANCH(p) (((p)->mp_flags & P_BRANCH) != 0)
-/* Test if a page is an overflow page */
-#define IS_OVERFLOW(p) unlikely(((p)->mp_flags & P_OVERFLOW) != 0)
-/* Test if a page is a sub page */
-#define IS_SUBP(p) (((p)->mp_flags & P_SUBP) != 0)
-
-#define PAGETYPE(p) ((p)->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW))
-
-/* Header for a single key/data pair within a page.
- * Used in pages of type P_BRANCH and P_LEAF without P_LEAF2.
- * We guarantee 2-byte alignment for 'MDBX_node's.
- *
- * Leaf node flags describe node contents. F_BIGDATA says the node's
- * data part is the page number of an overflow page with actual data.
- * F_DUPDATA and F_SUBDATA can be combined giving duplicate data in
- * a sub-page/sub-database, and named databases (just F_SUBDATA). */
-typedef struct MDBX_node {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- union {
- uint32_t mn_dsize;
- uint32_t mn_pgno32;
- };
- uint8_t mn_flags; /* see mdbx_node flags */
- uint8_t mn_extra;
- uint16_t mn_ksize; /* key size */
-#else
- uint16_t mn_ksize; /* key size */
- uint8_t mn_extra;
- uint8_t mn_flags; /* see mdbx_node flags */
- union {
- uint32_t mn_pgno32;
- uint32_t mn_dsize;
- };
-#endif /* __BYTE_ORDER__ */
-
- /* mdbx_node Flags */
-#define F_BIGDATA 0x01 /* data put on overflow page */
-#define F_SUBDATA 0x02 /* data is a sub-database */
-#define F_DUPDATA 0x04 /* data has duplicates */
-
- /* valid flags for mdbx_node_add() */
-#define NODE_ADD_FLAGS (F_DUPDATA | F_SUBDATA | MDBX_RESERVE | MDBX_APPEND)
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- uint8_t mn_data[] /* key and data are appended here */;
-#endif /* C99 */
-} MDBX_node;
-
-#define DB_PERSISTENT_FLAGS \
- (MDBX_REVERSEKEY | MDBX_DUPSORT | MDBX_INTEGERKEY | MDBX_DUPFIXED | \
- MDBX_INTEGERDUP | MDBX_REVERSEDUP)
-
-/* mdbx_dbi_open() flags */
-#define DB_USABLE_FLAGS (DB_PERSISTENT_FLAGS | MDBX_CREATE | MDBX_DB_ACCEDE)
-
-#define DB_VALID 0x8000 /* DB handle is valid, for me_dbflags */
-#define DB_INTERNAL_FLAGS DB_VALID
-
-#if DB_INTERNAL_FLAGS & DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-#if DB_PERSISTENT_FLAGS & ~DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-
-/* max number of pages to commit in one writev() call */
-#define MDBX_COMMIT_PAGES 64
-#if defined(IOV_MAX) && IOV_MAX < MDBX_COMMIT_PAGES /* sysconf(_SC_IOV_MAX) */
-#undef MDBX_COMMIT_PAGES
-#define MDBX_COMMIT_PAGES IOV_MAX
-#endif
-
-/*
- * /
- * | -1, a < b
- * CMP2INT(a,b) = < 0, a == b
- * | 1, a > b
- * \
- */
-#ifndef __e2k__
-/* LY: fast enough on most systems */
-#define CMP2INT(a, b) (((b) > (a)) ? -1 : (a) > (b))
-#else
-/* LY: more parallelable on VLIW Elbrus */
-#define CMP2INT(a, b) (((a) > (b)) - ((b) > (a)))
-#endif
-
-/* Do not spill pages to disk if txn is getting full, may fail instead */
-#define MDBX_NOSPILL 0x8000
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-int64pgno(int64_t i64) {
- if (likely(i64 >= (int64_t)MIN_PAGENO && i64 <= (int64_t)MAX_PAGENO + 1))
- return (pgno_t)i64;
- return (i64 < (int64_t)MIN_PAGENO) ? MIN_PAGENO : MAX_PAGENO;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_add(size_t base, size_t augend) {
- assert(base <= MAX_PAGENO + 1 && augend < MAX_PAGENO);
- return int64pgno(base + augend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_sub(size_t base, size_t subtrahend) {
- assert(base >= MIN_PAGENO && base <= MAX_PAGENO + 1 &&
- subtrahend < MAX_PAGENO);
- return int64pgno(base - subtrahend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline bool
-is_powerof2(size_t x) {
- return (x & (x - 1)) == 0;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-floor_powerof2(size_t value, size_t granularity) {
- assert(is_powerof2(granularity));
- return value & ~(granularity - 1);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-ceil_powerof2(size_t value, size_t granularity) {
- return floor_powerof2(value + granularity - 1, granularity);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static unsigned
-log2n_powerof2(size_t value) {
- assert(value > 0 && value < INT32_MAX && is_powerof2(value));
- assert((value & -(int32_t)value) == value);
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_ctzl)
- return __builtin_ctzl(value);
-#elif defined(_MSC_VER)
- unsigned long index;
- _BitScanForward(&index, (unsigned long)value);
- return index;
-#else
- static const uint8_t debruijn_ctz32[32] = {
- 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
- 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9};
- return debruijn_ctz32[(uint32_t)(value * 0x077CB531u) >> 27];
-#endif
-}
-
-/* Only a subset of the mdbx_env flags can be changed
- * at runtime. Changing other flags requires closing the
- * environment and re-opening it with the new flags. */
-#define ENV_CHANGEABLE_FLAGS \
- (MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC | MDBX_DEPRECATED_MAPASYNC | \
- MDBX_NOMEMINIT | MDBX_COALESCE | MDBX_PAGEPERTURB | MDBX_ACCEDE)
-#define ENV_CHANGELESS_FLAGS \
- (MDBX_NOSUBDIR | MDBX_RDONLY | MDBX_WRITEMAP | MDBX_NOTLS | MDBX_NORDAHEAD | \
- MDBX_LIFORECLAIM | MDBX_EXCLUSIVE)
-#define ENV_USABLE_FLAGS (ENV_CHANGEABLE_FLAGS | ENV_CHANGELESS_FLAGS)
-
-#if !defined(__cplusplus) || CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_MAYBE_UNUSED static void static_checks(void) {
- STATIC_ASSERT_MSG(INT16_MAX - CORE_DBS == MDBX_MAX_DBI,
- "Oops, MDBX_MAX_DBI or CORE_DBS?");
- STATIC_ASSERT_MSG((unsigned)(MDBX_DB_ACCEDE | MDBX_CREATE) ==
- ((DB_USABLE_FLAGS | DB_INTERNAL_FLAGS) &
- (ENV_USABLE_FLAGS | ENV_INTERNAL_FLAGS)),
- "Oops, some flags overlapped or wrong");
- STATIC_ASSERT_MSG((ENV_INTERNAL_FLAGS & ENV_USABLE_FLAGS) == 0,
- "Oops, some flags overlapped or wrong");
-}
-#endif /* Disabled for MSVC 19.0 (VisualStudio 2015) */
-
-#ifdef __cplusplus
-}
-#endif
-
-#define MDBX_ASAN_POISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("POISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_POISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#define MDBX_ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("UNPOISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_UNPOISON_MEMORY_REGION(addr, size); \
- } while (0)
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>.
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * This code is derived from "LMDB engine" written by
- * Howard Chu (Symas Corporation), which itself derived from btree.c
- * written by Martin Hedenfalk.
- *
- * ---
- *
- * Portions Copyright 2011-2015 Howard Chu, Symas Corp. All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- *
- * ---
- *
- * Portions Copyright (c) 2009, 2010 Martin Hedenfalk <martin@bzero.se>
- *
- * Permission to use, copy, modify, and distribute this software for any
- * purpose with or without fee is hereby granted, provided that the above
- * copyright notice and this permission notice appear in all copies.
- *
- * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
- * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
- * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
- * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
- * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
- * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
- * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. */
-
-
-/*------------------------------------------------------------------------------
- * Internal inline functions */
-
-MDBX_NOTHROW_CONST_FUNCTION static unsigned branchless_abs(int value) {
- assert(value > INT_MIN);
- const unsigned expanded_sign =
- (unsigned)(value >> (sizeof(value) * CHAR_BIT - 1));
- return ((unsigned)value + expanded_sign) ^ expanded_sign;
-}
-
-/* Pack/Unpack 16-bit values for Grow step & Shrink threshold */
-MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t me2v(unsigned m,
- unsigned e) {
- assert(m < 2048 && e < 8);
- return (pgno_t)(32768 + ((m + 1) << (e + 8)));
-}
-
-MDBX_NOTHROW_CONST_FUNCTION static __inline uint16_t v2me(size_t v,
- unsigned e) {
- assert(v > (e ? me2v(2047, e - 1) : 32768));
- assert(v <= me2v(2047, e));
- size_t m = (v - 32768 + ((size_t)1 << (e + 8)) - 1) >> (e + 8);
- m -= m > 0;
- assert(m < 2048 && e < 8);
- // f e d c b a 9 8 7 6 5 4 3 2 1 0
- // 1 e e e m m m m m m m m m m m 1
- const uint16_t pv = (uint16_t)(0x8001 + (e << 12) + (m << 1));
- assert(pv != 65535);
- return pv;
-}
-
-/* Convert 16-bit packed (exponential quantized) value to number of pages */
-MDBX_NOTHROW_CONST_FUNCTION static pgno_t pv2pages(uint16_t pv) {
- if ((pv & 0x8001) != 0x8001)
- return pv;
- if (pv == 65535)
- return 65536;
- // f e d c b a 9 8 7 6 5 4 3 2 1 0
- // 1 e e e m m m m m m m m m m m 1
- return me2v((pv >> 1) & 2047, (pv >> 12) & 7);
-}
-
-/* Convert number of pages to 16-bit packed (exponential quantized) value */
-MDBX_NOTHROW_CONST_FUNCTION static uint16_t pages2pv(size_t pages) {
- if (pages < 32769 || (pages < 65536 && (pages & 1) == 0))
- return (uint16_t)pages;
- if (pages <= me2v(2047, 0))
- return v2me(pages, 0);
- if (pages <= me2v(2047, 1))
- return v2me(pages, 1);
- if (pages <= me2v(2047, 2))
- return v2me(pages, 2);
- if (pages <= me2v(2047, 3))
- return v2me(pages, 3);
- if (pages <= me2v(2047, 4))
- return v2me(pages, 4);
- if (pages <= me2v(2047, 5))
- return v2me(pages, 5);
- if (pages <= me2v(2047, 6))
- return v2me(pages, 6);
- return (pages < me2v(2046, 7)) ? v2me(pages, 7) : 65533;
-}
-
-/*------------------------------------------------------------------------------
- * Unaligned access */
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline unsigned
-field_alignment(unsigned alignment_baseline, size_t field_offset) {
- unsigned merge = alignment_baseline | (unsigned)field_offset;
- return merge & -(int)merge;
-}
-
-/* read-thunk for UB-sanitizer */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline uint8_t
-peek_u8(const uint8_t *const __restrict ptr) {
- return *ptr;
-}
-
-/* write-thunk for UB-sanitizer */
-static __always_inline void poke_u8(uint8_t *const __restrict ptr,
- const uint8_t v) {
- *ptr = v;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline uint16_t
-unaligned_peek_u16(const unsigned expected_alignment, const void *const ptr) {
- assert((uintptr_t)ptr % expected_alignment == 0);
- if (MDBX_UNALIGNED_OK >= 2 || (expected_alignment % sizeof(uint16_t)) == 0)
- return *(const uint16_t *)ptr;
- else {
-#if defined(__unaligned) || defined(_M_ARM) || defined(_M_ARM64) || \
- defined(_M_X64) || defined(_M_IA64)
- return *(const __unaligned uint16_t *)ptr;
-#else
- uint16_t v;
- memcpy(&v, ptr, sizeof(v));
- return v;
-#endif /* _MSC_VER || __unaligned */
- }
-}
-
-static __always_inline void
-unaligned_poke_u16(const unsigned expected_alignment,
- void *const __restrict ptr, const uint16_t v) {
- assert((uintptr_t)ptr % expected_alignment == 0);
- if (MDBX_UNALIGNED_OK >= 2 || (expected_alignment % sizeof(v)) == 0)
- *(uint16_t *)ptr = v;
- else {
-#if defined(__unaligned) || defined(_M_ARM) || defined(_M_ARM64) || \
- defined(_M_X64) || defined(_M_IA64)
- *((uint16_t __unaligned *)ptr) = v;
-#else
- memcpy(ptr, &v, sizeof(v));
-#endif /* _MSC_VER || __unaligned */
- }
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline uint32_t unaligned_peek_u32(
- const unsigned expected_alignment, const void *const __restrict ptr) {
- assert((uintptr_t)ptr % expected_alignment == 0);
- if (MDBX_UNALIGNED_OK >= 4 || (expected_alignment % sizeof(uint32_t)) == 0)
- return *(const uint32_t *)ptr;
- else if ((expected_alignment % sizeof(uint16_t)) == 0) {
- const uint16_t lo =
- ((const uint16_t *)ptr)[__BYTE_ORDER__ != __ORDER_LITTLE_ENDIAN__];
- const uint16_t hi =
- ((const uint16_t *)ptr)[__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__];
- return lo | (uint32_t)hi << 16;
- } else {
-#if defined(__unaligned) || defined(_M_ARM) || defined(_M_ARM64) || \
- defined(_M_X64) || defined(_M_IA64)
- return *(const __unaligned uint32_t *)ptr;
-#else
- uint32_t v;
- memcpy(&v, ptr, sizeof(v));
- return v;
-#endif /* _MSC_VER || __unaligned */
- }
-}
-
-static __always_inline void
-unaligned_poke_u32(const unsigned expected_alignment,
- void *const __restrict ptr, const uint32_t v) {
- assert((uintptr_t)ptr % expected_alignment == 0);
- if (MDBX_UNALIGNED_OK >= 4 || (expected_alignment % sizeof(v)) == 0)
- *(uint32_t *)ptr = v;
- else if ((expected_alignment % sizeof(uint16_t)) == 0) {
- ((uint16_t *)ptr)[__BYTE_ORDER__ != __ORDER_LITTLE_ENDIAN__] = (uint16_t)v;
- ((uint16_t *)ptr)[__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__] =
- (uint16_t)(v >> 16);
- } else {
-#if defined(__unaligned) || defined(_M_ARM) || defined(_M_ARM64) || \
- defined(_M_X64) || defined(_M_IA64)
- *((uint32_t __unaligned *)ptr) = v;
-#else
- memcpy(ptr, &v, sizeof(v));
-#endif /* _MSC_VER || __unaligned */
- }
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline uint64_t unaligned_peek_u64(
- const unsigned expected_alignment, const void *const __restrict ptr) {
- assert((uintptr_t)ptr % expected_alignment == 0);
- if (MDBX_UNALIGNED_OK >= 8 || (expected_alignment % sizeof(uint64_t)) == 0)
- return *(const uint64_t *)ptr;
- else if ((expected_alignment % sizeof(uint32_t)) == 0) {
- const uint32_t lo =
- ((const uint32_t *)ptr)[__BYTE_ORDER__ != __ORDER_LITTLE_ENDIAN__];
- const uint32_t hi =
- ((const uint32_t *)ptr)[__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__];
- return lo | (uint64_t)hi << 32;
- } else {
-#if defined(__unaligned) || defined(_M_ARM) || defined(_M_ARM64) || \
- defined(_M_X64) || defined(_M_IA64)
- return *(const __unaligned uint64_t *)ptr;
-#else
- uint64_t v;
- memcpy(&v, ptr, sizeof(v));
- return v;
-#endif /* _MSC_VER || __unaligned */
- }
-}
-
-static __always_inline uint64_t
-unaligned_peek_u64_volatile(const unsigned expected_alignment,
- volatile const void *const __restrict ptr) {
- assert((uintptr_t)ptr % expected_alignment == 0);
- assert(expected_alignment % sizeof(uint32_t) == 0);
- if (MDBX_UNALIGNED_OK >= 8 || (expected_alignment % sizeof(uint64_t)) == 0)
- return *(volatile const uint64_t *)ptr;
- else {
-#if defined(__unaligned) || defined(_M_ARM) || defined(_M_ARM64) || \
- defined(_M_X64) || defined(_M_IA64)
- return *(volatile const __unaligned uint64_t *)ptr;
-#else
- const uint32_t lo = ((volatile const uint32_t *)
- ptr)[__BYTE_ORDER__ != __ORDER_LITTLE_ENDIAN__];
- const uint32_t hi = ((volatile const uint32_t *)
- ptr)[__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__];
- return lo | (uint64_t)hi << 32;
-#endif /* _MSC_VER || __unaligned */
- }
-}
-
-static __always_inline void
-unaligned_poke_u64(const unsigned expected_alignment,
- void *const __restrict ptr, const uint64_t v) {
- assert((uintptr_t)ptr % expected_alignment == 0);
- if (MDBX_UNALIGNED_OK >= 8 || (expected_alignment % sizeof(v)) == 0)
- *(uint64_t *)ptr = v;
- else if ((expected_alignment % sizeof(uint32_t)) == 0) {
- ((uint32_t *)ptr)[__BYTE_ORDER__ != __ORDER_LITTLE_ENDIAN__] = (uint32_t)v;
- ((uint32_t *)ptr)[__BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__] =
- (uint32_t)(v >> 32);
- } else {
-#if defined(__unaligned) || defined(_M_ARM) || defined(_M_ARM64) || \
- defined(_M_X64) || defined(_M_IA64)
- *((uint64_t __unaligned *)ptr) = v;
-#else
- memcpy(ptr, &v, sizeof(v));
-#endif /* _MSC_VER || __unaligned */
- }
-}
-
-#define UNALIGNED_PEEK_8(ptr, struct, field) \
- peek_u8((const uint8_t *)(ptr) + offsetof(struct, field))
-#define UNALIGNED_POKE_8(ptr, struct, field, value) \
- poke_u8((uint8_t *)(ptr) + offsetof(struct, field), value)
-
-#define UNALIGNED_PEEK_16(ptr, struct, field) \
- unaligned_peek_u16(1, (const char *)(ptr) + offsetof(struct, field))
-#define UNALIGNED_POKE_16(ptr, struct, field, value) \
- unaligned_poke_u16(1, (char *)(ptr) + offsetof(struct, field), value)
-
-#define UNALIGNED_PEEK_32(ptr, struct, field) \
- unaligned_peek_u32(1, (const char *)(ptr) + offsetof(struct, field))
-#define UNALIGNED_POKE_32(ptr, struct, field, value) \
- unaligned_poke_u32(1, (char *)(ptr) + offsetof(struct, field), value)
-
-#define UNALIGNED_PEEK_64(ptr, struct, field) \
- unaligned_peek_u64(1, (const char *)(ptr) + offsetof(struct, field))
-#define UNALIGNED_POKE_64(ptr, struct, field, value) \
- unaligned_poke_u64(1, (char *)(ptr) + offsetof(struct, field), value)
-
-/* Get the page number pointed to by a branch node */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline pgno_t
-node_pgno(const MDBX_node *const __restrict node) {
- pgno_t pgno = UNALIGNED_PEEK_32(node, MDBX_node, mn_pgno32);
- if (sizeof(pgno) > 4)
- pgno |= ((uint64_t)UNALIGNED_PEEK_8(node, MDBX_node, mn_extra)) << 32;
- return pgno;
-}
-
-/* Set the page number in a branch node */
-static __always_inline void node_set_pgno(MDBX_node *const __restrict node,
- pgno_t pgno) {
- assert(pgno >= MIN_PAGENO && pgno <= MAX_PAGENO);
-
- UNALIGNED_POKE_32(node, MDBX_node, mn_pgno32, (uint32_t)pgno);
- if (sizeof(pgno) > 4)
- UNALIGNED_POKE_8(node, MDBX_node, mn_extra,
- (uint8_t)((uint64_t)pgno >> 32));
-}
-
-/* Get the size of the data in a leaf node */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline size_t
-node_ds(const MDBX_node *const __restrict node) {
- return UNALIGNED_PEEK_32(node, MDBX_node, mn_dsize);
-}
-
-/* Set the size of the data for a leaf node */
-static __always_inline void node_set_ds(MDBX_node *const __restrict node,
- size_t size) {
- assert(size < INT_MAX);
- UNALIGNED_POKE_32(node, MDBX_node, mn_dsize, (uint32_t)size);
-}
-
-/* The size of a key in a node */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline size_t
-node_ks(const MDBX_node *const __restrict node) {
- return UNALIGNED_PEEK_16(node, MDBX_node, mn_ksize);
-}
-
-/* Set the size of the key for a leaf node */
-static __always_inline void node_set_ks(MDBX_node *const __restrict node,
- size_t size) {
- assert(size < INT16_MAX);
- UNALIGNED_POKE_16(node, MDBX_node, mn_ksize, (uint16_t)size);
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline uint8_t
-node_flags(const MDBX_node *const __restrict node) {
- return UNALIGNED_PEEK_8(node, MDBX_node, mn_flags);
-}
-
-static __always_inline void node_set_flags(MDBX_node *const __restrict node,
- uint8_t flags) {
- UNALIGNED_POKE_8(node, MDBX_node, mn_flags, flags);
-}
-
-/* Size of the node header, excluding dynamic data at the end */
-#define NODESIZE offsetof(MDBX_node, mn_data)
-
-/* Address of the key for the node */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline void *
-node_key(const MDBX_node *const __restrict node) {
- return (char *)node + NODESIZE;
-}
-
-/* Address of the data for a node */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline void *
-node_data(const MDBX_node *const __restrict node) {
- return (char *)node_key(node) + node_ks(node);
-}
-
-/* Size of a node in a leaf page with a given key and data.
- * This is node header plus key plus data size. */
-MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-node_size_len(const size_t key_len, const size_t value_len) {
- return NODESIZE + EVEN(key_len + value_len);
-}
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline size_t
-node_size(const MDBX_val *key, const MDBX_val *value) {
- return node_size_len(key ? key->iov_len : 0, value ? value->iov_len : 0);
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline pgno_t
-peek_pgno(const void *const __restrict ptr) {
- if (sizeof(pgno_t) == sizeof(uint32_t))
- return (pgno_t)unaligned_peek_u32(1, ptr);
- else if (sizeof(pgno_t) == sizeof(uint64_t))
- return (pgno_t)unaligned_peek_u64(1, ptr);
- else {
- pgno_t pgno;
- memcpy(&pgno, ptr, sizeof(pgno));
- return pgno;
- }
-}
-
-static __always_inline void poke_pgno(void *const __restrict ptr,
- const pgno_t pgno) {
- if (sizeof(pgno) == sizeof(uint32_t))
- unaligned_poke_u32(1, ptr, pgno);
- else if (sizeof(pgno) == sizeof(uint64_t))
- unaligned_poke_u64(1, ptr, pgno);
- else
- memcpy(ptr, &pgno, sizeof(pgno));
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline pgno_t
-node_largedata_pgno(const MDBX_node *const __restrict node) {
- assert(node_flags(node) & F_BIGDATA);
- return peek_pgno(node_data(node));
-}
-
-/*------------------------------------------------------------------------------
- * Nodes, Keys & Values length limitation factors:
- *
- * BRANCH_NODE_MAX
- * Branch-page must contain at least two nodes, within each a key and a child
- * page number. But page can't be splitted if it contains less that 4 keys,
- * i.e. a page should not overflow before adding the fourth key. Therefore,
- * at least 3 branch-node should fit in the single branch-page. Further, the
- * first node of a branch-page doesn't contain a key, i.e. the first node
- * is always require space just for itself. Thus:
- * PAGEROOM = pagesize - page_hdr_len;
- * BRANCH_NODE_MAX = even_floor(
- * (PAGEROOM - sizeof(indx_t) - NODESIZE) / (3 - 1) - sizeof(indx_t));
- * KEYLEN_MAX = BRANCH_NODE_MAX - node_hdr_len;
- *
- * LEAF_NODE_MAX
- * Leaf-node must fit into single leaf-page, where a value could be placed on
- * a large/overflow page. However, may require to insert a nearly page-sized
- * node between two large nodes are already fill-up a page. In this case the
- * page must be splitted to two if some pair of nodes fits on one page, or
- * otherwise the page should be splitted to the THREE with a single node
- * per each of ones. Such 1-into-3 page splitting is costly and complex since
- * requires TWO insertion into the parent page, that could lead to split it
- * and so on up to the root. Therefore double-splitting is avoided here and
- * the maximum node size is half of a leaf page space:
- * LEAF_NODE_MAX = even_floor(PAGEROOM / 2 - sizeof(indx_t));
- * DATALEN_NO_OVERFLOW = LEAF_NODE_MAX - KEYLEN_MAX;
- *
- * - SubDatabase-node must fit into one leaf-page:
- * SUBDB_NAME_MAX = LEAF_NODE_MAX - node_hdr_len - sizeof(MDBX_db);
- *
- * - Dupsort values itself are a keys in a dupsort-subdb and couldn't be longer
- * than the KEYLEN_MAX. But dupsort node must not great than LEAF_NODE_MAX,
- * since dupsort value couldn't be placed on a large/overflow page:
- * DUPSORT_DATALEN_MAX = min(KEYLEN_MAX,
- * max(DATALEN_NO_OVERFLOW, sizeof(MDBX_db));
- */
-
-#define PAGEROOM(pagesize) ((pagesize)-PAGEHDRSZ)
-#define EVEN_FLOOR(n) ((n) & ~(size_t)1)
-#define BRANCH_NODE_MAX(pagesize) \
- (EVEN_FLOOR((PAGEROOM(pagesize) - sizeof(indx_t) - NODESIZE) / (3 - 1) - \
- sizeof(indx_t)))
-#define LEAF_NODE_MAX(pagesize) \
- (EVEN_FLOOR(PAGEROOM(pagesize) / 2) - sizeof(indx_t))
-#define MAX_GC1OVPAGE(pagesize) (PAGEROOM(pagesize) / sizeof(pgno_t) - 1)
-
-static __inline unsigned keysize_max(size_t pagesize, MDBX_db_flags_t flags) {
- assert(pagesize >= MIN_PAGESIZE && pagesize <= MAX_PAGESIZE &&
- is_powerof2(pagesize));
- STATIC_ASSERT(BRANCH_NODE_MAX(MIN_PAGESIZE) - NODESIZE >= 8);
- if (flags & MDBX_INTEGERKEY)
- return 8 /* sizeof(uint64_t) */;
-
- const intptr_t max_branch_key = BRANCH_NODE_MAX(pagesize) - NODESIZE;
- STATIC_ASSERT(LEAF_NODE_MAX(MIN_PAGESIZE) - NODESIZE -
- /* sizeof(uint64) as a key */ 8 >
- sizeof(MDBX_db));
- if (flags &
- (MDBX_DUPSORT | MDBX_DUPFIXED | MDBX_REVERSEDUP | MDBX_INTEGERDUP)) {
- const intptr_t max_dupsort_leaf_key =
- LEAF_NODE_MAX(pagesize) - NODESIZE - sizeof(MDBX_db);
- return (max_branch_key < max_dupsort_leaf_key)
- ? (unsigned)max_branch_key
- : (unsigned)max_dupsort_leaf_key;
- }
- return (unsigned)max_branch_key;
-}
-
-static __inline size_t valsize_max(size_t pagesize, MDBX_db_flags_t flags) {
- assert(pagesize >= MIN_PAGESIZE && pagesize <= MAX_PAGESIZE &&
- is_powerof2(pagesize));
-
- if (flags & MDBX_INTEGERDUP)
- return 8 /* sizeof(uint64_t) */;
-
- if (flags & (MDBX_DUPSORT | MDBX_DUPFIXED | MDBX_REVERSEDUP))
- return keysize_max(pagesize, 0);
-
- const unsigned page_ln2 = log2n_powerof2(pagesize);
- const size_t hard = 0x7FF00000ul;
- const size_t hard_pages = hard >> page_ln2;
- STATIC_ASSERT(MDBX_PGL_LIMIT <= MAX_PAGENO);
- const size_t pages_limit = MDBX_PGL_LIMIT / 4;
- const size_t limit =
- (hard_pages < pages_limit) ? hard : (pages_limit << page_ln2);
- return (limit < MAX_MAPSIZE / 2) ? limit : MAX_MAPSIZE / 2;
-}
-
-__cold int mdbx_env_get_maxkeysize(const MDBX_env *env) {
- return mdbx_env_get_maxkeysize_ex(env, MDBX_DUPSORT);
-}
-
-__cold int mdbx_env_get_maxkeysize_ex(const MDBX_env *env,
- MDBX_db_flags_t flags) {
- if (unlikely(!env || env->me_signature.weak != MDBX_ME_SIGNATURE))
- return -1;
-
- return (int)mdbx_limits_keysize_max((intptr_t)env->me_psize, flags);
-}
-
-size_t mdbx_default_pagesize(void) {
- size_t pagesize = mdbx_syspagesize();
- mdbx_ensure(nullptr, is_powerof2(pagesize));
- pagesize = (pagesize >= MIN_PAGESIZE) ? pagesize : MIN_PAGESIZE;
- pagesize = (pagesize <= MAX_PAGESIZE) ? pagesize : MAX_PAGESIZE;
- return pagesize;
-}
-
-__cold intptr_t mdbx_limits_keysize_max(intptr_t pagesize,
- MDBX_db_flags_t flags) {
- if (pagesize < 1)
- pagesize = (intptr_t)mdbx_default_pagesize();
- if (unlikely(pagesize < (intptr_t)MIN_PAGESIZE ||
- pagesize > (intptr_t)MAX_PAGESIZE ||
- !is_powerof2((size_t)pagesize)))
- return -1;
-
- return keysize_max(pagesize, flags);
-}
-
-__cold int mdbx_env_get_maxvalsize_ex(const MDBX_env *env,
- MDBX_db_flags_t flags) {
- if (unlikely(!env || env->me_signature.weak != MDBX_ME_SIGNATURE))
- return -1;
-
- return (int)mdbx_limits_valsize_max((intptr_t)env->me_psize, flags);
-}
-
-__cold intptr_t mdbx_limits_valsize_max(intptr_t pagesize,
- MDBX_db_flags_t flags) {
- if (pagesize < 1)
- pagesize = (intptr_t)mdbx_default_pagesize();
- if (unlikely(pagesize < (intptr_t)MIN_PAGESIZE ||
- pagesize > (intptr_t)MAX_PAGESIZE ||
- !is_powerof2((size_t)pagesize)))
- return -1;
-
- return valsize_max(pagesize, flags);
-}
-
-/* Calculate the size of a leaf node.
- *
- * The size depends on the environment's page size; if a data item
- * is too large it will be put onto an overflow page and the node
- * size will only include the key and not the data. Sizes are always
- * rounded up to an even number of bytes, to guarantee 2-byte alignment
- * of the MDBX_node headers. */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline size_t
-leaf_size(const MDBX_env *env, const MDBX_val *key, const MDBX_val *data) {
- size_t node_bytes = node_size(key, data);
- if (node_bytes > env->me_leaf_nodemax) {
- /* put on overflow page */
- node_bytes = node_size_len(key->iov_len, 0) + sizeof(pgno_t);
- }
-
- return node_bytes + sizeof(indx_t);
-}
-
-/* Calculate the size of a branch node.
- *
- * The size should depend on the environment's page size but since
- * we currently don't support spilling large keys onto overflow
- * pages, it's simply the size of the MDBX_node header plus the
- * size of the key. Sizes are always rounded up to an even number
- * of bytes, to guarantee 2-byte alignment of the MDBX_node headers.
- *
- * [in] env The environment handle.
- * [in] key The key for the node.
- *
- * Returns The number of bytes needed to store the node. */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline size_t
-branch_size(const MDBX_env *env, const MDBX_val *key) {
- /* Size of a node in a branch page with a given key.
- * This is just the node header plus the key, there is no data. */
- size_t node_bytes = node_size(key, nullptr);
- if (unlikely(node_bytes > env->me_leaf_nodemax)) {
- /* put on overflow page */
- /* not implemented */
- mdbx_assert_fail(env, "INDXSIZE(key) <= env->me_nodemax", __func__,
- __LINE__);
- node_bytes = node_size(key, nullptr) + sizeof(pgno_t);
- }
-
- return node_bytes + sizeof(indx_t);
-}
-
-MDBX_NOTHROW_CONST_FUNCTION static __always_inline uint16_t
-flags_db2sub(uint16_t db_flags) {
- uint16_t sub_flags = db_flags & MDBX_DUPFIXED;
-
- /* MDBX_INTEGERDUP => MDBX_INTEGERKEY */
-#define SHIFT_INTEGERDUP_TO_INTEGERKEY 2
- STATIC_ASSERT((MDBX_INTEGERDUP >> SHIFT_INTEGERDUP_TO_INTEGERKEY) ==
- MDBX_INTEGERKEY);
- sub_flags |= (db_flags & MDBX_INTEGERDUP) >> SHIFT_INTEGERDUP_TO_INTEGERKEY;
-
- /* MDBX_REVERSEDUP => MDBX_REVERSEKEY */
-#define SHIFT_REVERSEDUP_TO_REVERSEKEY 5
- STATIC_ASSERT((MDBX_REVERSEDUP >> SHIFT_REVERSEDUP_TO_REVERSEKEY) ==
- MDBX_REVERSEKEY);
- sub_flags |= (db_flags & MDBX_REVERSEDUP) >> SHIFT_REVERSEDUP_TO_REVERSEKEY;
-
- return sub_flags;
-}
-
-/*----------------------------------------------------------------------------*/
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline size_t
-pgno2bytes(const MDBX_env *env, pgno_t pgno) {
- mdbx_assert(env, (1u << env->me_psize2log) == env->me_psize);
- return ((size_t)pgno) << env->me_psize2log;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline MDBX_page *
-pgno2page(const MDBX_env *env, pgno_t pgno) {
- return (MDBX_page *)(env->me_map + pgno2bytes(env, pgno));
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline pgno_t
-bytes2pgno(const MDBX_env *env, size_t bytes) {
- mdbx_assert(env, (env->me_psize >> env->me_psize2log) == 1);
- return (pgno_t)(bytes >> env->me_psize2log);
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static size_t
-pgno_align2os_bytes(const MDBX_env *env, pgno_t pgno) {
- return ceil_powerof2(pgno2bytes(env, pgno), env->me_os_psize);
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static pgno_t pgno_align2os_pgno(const MDBX_env *env,
- pgno_t pgno) {
- return bytes2pgno(env, pgno_align2os_bytes(env, pgno));
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static size_t
-bytes_align2os_bytes(const MDBX_env *env, size_t bytes) {
- return ceil_powerof2(ceil_powerof2(bytes, env->me_psize), env->me_os_psize);
-}
-
-/* Address of first usable data byte in a page, after the header */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline void *
-page_data(const MDBX_page *mp) {
- return (char *)mp + PAGEHDRSZ;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline const MDBX_page *
-data_page(const void *data) {
- return container_of(data, MDBX_page, mp_ptrs);
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline MDBX_meta *
-page_meta(MDBX_page *mp) {
- return (MDBX_meta *)page_data(mp);
-}
-
-/* Number of nodes on a page */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline unsigned
-page_numkeys(const MDBX_page *mp) {
- return mp->mp_lower >> 1;
-}
-
-/* The amount of space remaining in the page */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline unsigned
-page_room(const MDBX_page *mp) {
- return mp->mp_upper - mp->mp_lower;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline unsigned
-page_space(const MDBX_env *env) {
- STATIC_ASSERT(PAGEHDRSZ % 2 == 0);
- return env->me_psize - PAGEHDRSZ;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline unsigned
-page_used(const MDBX_env *env, const MDBX_page *mp) {
- return page_space(env) - page_room(mp);
-}
-
-/* The percentage of space used in the page, in a percents. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_PURE_FUNCTION static __inline double
-page_fill(const MDBX_env *env, const MDBX_page *mp) {
- return page_used(env, mp) * 100.0 / page_space(env);
-}
-
-/* The number of overflow pages needed to store the given size. */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline pgno_t
-number_of_ovpages(const MDBX_env *env, size_t bytes) {
- return bytes2pgno(env, PAGEHDRSZ - 1 + bytes) + 1;
-}
-
-__cold static int MDBX_PRINTF_ARGS(2, 3)
- bad_page(const MDBX_page *mp, const char *fmt, ...) {
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) {
- static const MDBX_page *prev;
- if (prev != mp) {
- prev = mp;
- const char *type;
- switch (mp->mp_flags & (P_BRANCH | P_LEAF | P_OVERFLOW | P_META |
- P_LEAF2 | P_BAD | P_SUBP)) {
- case P_BRANCH:
- type = "branch";
- break;
- case P_LEAF:
- type = "leaf";
- break;
- case P_LEAF | P_SUBP:
- type = "subleaf";
- break;
- case P_LEAF | P_LEAF2:
- type = "dupfixed-leaf";
- break;
- case P_LEAF | P_LEAF2 | P_SUBP:
- type = "dupfixed-subleaf";
- break;
- case P_OVERFLOW:
- type = "large";
- break;
- default:
- type = "broken";
- }
- mdbx_debug_log(MDBX_LOG_ERROR, "badpage", 0,
- "corrupted %s-page #%u, mod-txnid %" PRIaTXN "\n", type,
- mp->mp_pgno, mp->mp_txnid);
- }
-
- va_list args;
- va_start(args, fmt);
- mdbx_debug_log_va(MDBX_LOG_ERROR, "badpage", 0, fmt, args);
- va_end(args);
- }
- return MDBX_CORRUPTED;
-}
-
-/* Address of node i in page p */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline MDBX_node *
-page_node(const MDBX_page *mp, unsigned i) {
- assert((mp->mp_flags & (P_LEAF2 | P_OVERFLOW | P_META)) == 0);
- assert(page_numkeys(mp) > (unsigned)(i));
- assert(mp->mp_ptrs[i] % 2 == 0);
- return (MDBX_node *)((char *)mp + mp->mp_ptrs[i] + PAGEHDRSZ);
-}
-
-/* The address of a key in a LEAF2 page.
- * LEAF2 pages are used for MDBX_DUPFIXED sorted-duplicate sub-DBs.
- * There are no node headers, keys are stored contiguously. */
-MDBX_NOTHROW_PURE_FUNCTION static __always_inline void *
-page_leaf2key(const MDBX_page *mp, unsigned i, size_t keysize) {
- assert((mp->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_META)) ==
- (P_LEAF | P_LEAF2));
- assert(mp->mp_leaf2_ksize == keysize);
- (void)keysize;
- return (char *)mp + PAGEHDRSZ + (i * mp->mp_leaf2_ksize);
-}
-
-/* Set the node's key into keyptr. */
-static __always_inline void get_key(const MDBX_node *node, MDBX_val *keyptr) {
- keyptr->iov_len = node_ks(node);
- keyptr->iov_base = node_key(node);
-}
-
-/* Set the node's key into keyptr, if requested. */
-static __always_inline void
-get_key_optional(const MDBX_node *node, MDBX_val *keyptr /* __may_null */) {
- if (keyptr)
- get_key(node, keyptr);
-}
-
-/*------------------------------------------------------------------------------
- * Workaround for mmaped-lookahead-cross-page-boundary bug
- * in an obsolete versions of Elbrus's libc and kernels. */
-#if defined(__e2k__) && defined(MDBX_E2K_MLHCPB_WORKAROUND) && \
- MDBX_E2K_MLHCPB_WORKAROUND
-int __hot mdbx_e2k_memcmp_bug_workaround(const void *s1, const void *s2,
- size_t n) {
- if (unlikely(n > 42
- /* LY: align followed access if reasonable possible */
- && (((uintptr_t)s1) & 7) != 0 &&
- (((uintptr_t)s1) & 7) == (((uintptr_t)s2) & 7))) {
- if (((uintptr_t)s1) & 1) {
- const int diff = *(uint8_t *)s1 - *(uint8_t *)s2;
- if (diff)
- return diff;
- s1 = (char *)s1 + 1;
- s2 = (char *)s2 + 1;
- n -= 1;
- }
-
- if (((uintptr_t)s1) & 2) {
- const uint16_t a = *(uint16_t *)s1;
- const uint16_t b = *(uint16_t *)s2;
- if (likely(a != b))
- return (__builtin_bswap16(a) > __builtin_bswap16(b)) ? 1 : -1;
- s1 = (char *)s1 + 2;
- s2 = (char *)s2 + 2;
- n -= 2;
- }
-
- if (((uintptr_t)s1) & 4) {
- const uint32_t a = *(uint32_t *)s1;
- const uint32_t b = *(uint32_t *)s2;
- if (likely(a != b))
- return (__builtin_bswap32(a) > __builtin_bswap32(b)) ? 1 : -1;
- s1 = (char *)s1 + 4;
- s2 = (char *)s2 + 4;
- n -= 4;
- }
- }
-
- while (n >= 8) {
- const uint64_t a = *(uint64_t *)s1;
- const uint64_t b = *(uint64_t *)s2;
- if (likely(a != b))
- return (__builtin_bswap64(a) > __builtin_bswap64(b)) ? 1 : -1;
- s1 = (char *)s1 + 8;
- s2 = (char *)s2 + 8;
- n -= 8;
- }
-
- if (n & 4) {
- const uint32_t a = *(uint32_t *)s1;
- const uint32_t b = *(uint32_t *)s2;
- if (likely(a != b))
- return (__builtin_bswap32(a) > __builtin_bswap32(b)) ? 1 : -1;
- s1 = (char *)s1 + 4;
- s2 = (char *)s2 + 4;
- }
-
- if (n & 2) {
- const uint16_t a = *(uint16_t *)s1;
- const uint16_t b = *(uint16_t *)s2;
- if (likely(a != b))
- return (__builtin_bswap16(a) > __builtin_bswap16(b)) ? 1 : -1;
- s1 = (char *)s1 + 2;
- s2 = (char *)s2 + 2;
- }
-
- return (n & 1) ? *(uint8_t *)s1 - *(uint8_t *)s2 : 0;
-}
-
-int __hot mdbx_e2k_strcmp_bug_workaround(const char *s1, const char *s2) {
- while (true) {
- int diff = *(uint8_t *)s1 - *(uint8_t *)s2;
- if (likely(diff != 0) || *s1 == '\0')
- return diff;
- s1 += 1;
- s2 += 1;
- }
-}
-
-int __hot mdbx_e2k_strncmp_bug_workaround(const char *s1, const char *s2,
- size_t n) {
- while (n > 0) {
- int diff = *(uint8_t *)s1 - *(uint8_t *)s2;
- if (likely(diff != 0) || *s1 == '\0')
- return diff;
- s1 += 1;
- s2 += 1;
- n -= 1;
- }
- return 0;
-}
-
-size_t __hot mdbx_e2k_strlen_bug_workaround(const char *s) {
- size_t n = 0;
- while (*s) {
- s += 1;
- n += 1;
- }
- return n;
-}
-
-size_t __hot mdbx_e2k_strnlen_bug_workaround(const char *s, size_t maxlen) {
- size_t n = 0;
- while (maxlen > n && *s) {
- s += 1;
- n += 1;
- }
- return n;
-}
-#endif /* MDBX_E2K_MLHCPB_WORKAROUND */
-
-/*------------------------------------------------------------------------------
- * safe read/write volatile 64-bit fields on 32-bit architectures. */
-
-MDBX_MAYBE_UNUSED static __always_inline uint64_t
-atomic_store64(MDBX_atomic_uint64_t *p, const uint64_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint64_t) == 8);
-#if MDBX_64BIT_ATOMIC
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint64_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint64_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
-#else /* !MDBX_64BIT_ATOMIC */
- mdbx_compiler_barrier();
- atomic_store32(&p->low, (uint32_t)value, mo_Relaxed);
- mdbx_jitter4testing(true);
- atomic_store32(&p->high, (uint32_t)(value >> 32), order);
- mdbx_jitter4testing(true);
-#endif /* !MDBX_64BIT_ATOMIC */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static
-#if MDBX_64BIT_ATOMIC
- __always_inline
-#endif /* MDBX_64BIT_ATOMIC */
- uint64_t
- atomic_load64(const MDBX_atomic_uint64_t *p,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint64_t) == 8);
-#if MDBX_64BIT_ATOMIC
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint64_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint64_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint64_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-#else /* !MDBX_64BIT_ATOMIC */
- mdbx_compiler_barrier();
- uint64_t value = (uint64_t)atomic_load32(&p->high, order) << 32;
- mdbx_jitter4testing(true);
- value |= atomic_load32(&p->low, (order == mo_Relaxed) ? mo_Relaxed
- : mo_AcquireRelease);
- mdbx_jitter4testing(true);
- for (;;) {
- mdbx_compiler_barrier();
- uint64_t again = (uint64_t)atomic_load32(&p->high, order) << 32;
- mdbx_jitter4testing(true);
- again |= atomic_load32(&p->low, (order == mo_Relaxed) ? mo_Relaxed
- : mo_AcquireRelease);
- mdbx_jitter4testing(true);
- if (likely(value == again))
- return value;
- value = again;
- }
-#endif /* !MDBX_64BIT_ATOMIC */
-}
-
-static __always_inline void atomic_yield(void) {
-#if defined(_WIN32) || defined(_WIN64)
- YieldProcessor();
-#elif defined(__ia32__) || defined(__e2k__)
- __builtin_ia32_pause();
-#elif defined(__ia64__)
-#if defined(__HP_cc__) || defined(__HP_aCC__)
- _Asm_hint(_HINT_PAUSE);
-#else
- __asm__ __volatile__("hint @pause");
-#endif
-#elif defined(__aarch64__) || (defined(__ARM_ARCH) && __ARM_ARCH > 6) || \
- defined(__ARM_ARCH_6K__)
-#ifdef __CC_ARM
- __yield();
-#else
- __asm__ __volatile__("yield");
-#endif
-#elif (defined(__mips64) || defined(__mips64__)) && defined(__mips_isa_rev) && \
- __mips_isa_rev >= 2
- __asm__ __volatile__("pause");
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
- __asm__ __volatile__(".word 0x00000140");
-#elif defined(__linux__) || defined(__gnu_linux__) || defined(_UNIX03_SOURCE)
- sched_yield();
-#elif (defined(_GNU_SOURCE) && __GLIBC_PREREQ(2, 1)) || defined(_OPEN_THREADS)
- pthread_yield();
-#endif
-}
-
-#if MDBX_64BIT_CAS
-static __always_inline bool atomic_cas64(MDBX_atomic_uint64_t *p, uint64_t c,
- uint64_t v) {
-#ifdef MDBX_HAVE_C11ATOMICS
- STATIC_ASSERT(sizeof(long long) >= sizeof(uint64_t));
-#ifdef ATOMIC_LLONG_LOCK_FREE
- STATIC_ASSERT(ATOMIC_LLONG_LOCK_FREE > 0);
-#if ATOMIC_LLONG_LOCK_FREE < 2
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint64_t, p)));
-#endif /* ATOMIC_LLONG_LOCK_FREE < 2 */
-#else /* defined(ATOMIC_LLONG_LOCK_FREE) */
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint64_t, p)));
-#endif
- return atomic_compare_exchange_strong(MDBX_c11a_rw(uint64_t, p), &c, v);
-#elif defined(__GNUC__) || defined(__clang__)
- return __sync_bool_compare_and_swap(&p->weak, c, v);
-#elif defined(_MSC_VER)
- return c == (uint64_t)_InterlockedCompareExchange64(
- (volatile __int64 *)&p->weak, v, c);
-#elif defined(__APPLE__)
- return OSAtomicCompareAndSwap64Barrier(c, v, &p->weak);
-#else
-#error FIXME: Unsupported compiler
-#endif
-}
-#endif /* MDBX_64BIT_CAS */
-
-static __always_inline bool atomic_cas32(MDBX_atomic_uint32_t *p, uint32_t c,
- uint32_t v) {
-#ifdef MDBX_HAVE_C11ATOMICS
- STATIC_ASSERT(sizeof(int) >= sizeof(uint32_t));
-#ifdef ATOMIC_INT_LOCK_FREE
- STATIC_ASSERT(ATOMIC_INT_LOCK_FREE > 0);
-#if ATOMIC_INT_LOCK_FREE < 2
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
-#endif
-#else
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
-#endif
- return atomic_compare_exchange_strong(MDBX_c11a_rw(uint32_t, p), &c, v);
-#elif defined(__GNUC__) || defined(__clang__)
- return __sync_bool_compare_and_swap(&p->weak, c, v);
-#elif defined(_MSC_VER)
- STATIC_ASSERT(sizeof(volatile long) == sizeof(volatile uint32_t));
- return c ==
- (uint32_t)_InterlockedCompareExchange((volatile long *)&p->weak, v, c);
-#elif defined(__APPLE__)
- return OSAtomicCompareAndSwap32Barrier(c, v, &p->weak);
-#else
-#error FIXME: Unsupported compiler
-#endif
-}
-
-static __always_inline uint32_t atomic_add32(MDBX_atomic_uint32_t *p,
- uint32_t v) {
-#ifdef MDBX_HAVE_C11ATOMICS
- STATIC_ASSERT(sizeof(int) >= sizeof(uint32_t));
-#ifdef ATOMIC_INT_LOCK_FREE
- STATIC_ASSERT(ATOMIC_INT_LOCK_FREE > 0);
-#if ATOMIC_INT_LOCK_FREE < 2
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
-#endif
-#else
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
-#endif
- return atomic_fetch_add(MDBX_c11a_rw(uint32_t, p), v);
-#elif defined(__GNUC__) || defined(__clang__)
- return __sync_fetch_and_add(&p->weak, v);
-#elif defined(_MSC_VER)
- STATIC_ASSERT(sizeof(volatile long) == sizeof(volatile uint32_t));
- return (uint32_t)_InterlockedExchangeAdd((volatile long *)&p->weak, v);
-#elif defined(__APPLE__)
- return OSAtomicAdd32Barrier(v, &p->weak);
-#else
-#error FIXME: Unsupported compiler
-#endif
-}
-
-#define atomic_sub32(p, v) atomic_add32(p, 0 - (v))
-
-static __always_inline uint64_t safe64_txnid_next(uint64_t txnid) {
- txnid += xMDBX_TXNID_STEP;
-#if !MDBX_64BIT_CAS
- /* avoid overflow of low-part in safe64_reset() */
- txnid += (UINT32_MAX == (uint32_t)txnid);
-#endif
- return txnid;
-}
-
-static __always_inline void safe64_reset(MDBX_atomic_uint64_t *p,
- bool single_writer) {
-#if !MDBX_64BIT_CAS
- if (!single_writer) {
- STATIC_ASSERT(xMDBX_TXNID_STEP > 1);
- /* it is safe to increment low-part to avoid ABA, since xMDBX_TXNID_STEP > 1
- * and overflow was preserved in safe64_txnid_next() */
- atomic_add32(&p->low, 1) /* avoid ABA in safe64_reset_compare() */;
- atomic_store32(
- &p->high, UINT32_MAX,
- mo_Relaxed) /* atomically make >= SAFE64_INVALID_THRESHOLD */;
- atomic_add32(&p->low, 1) /* avoid ABA in safe64_reset_compare() */;
- } else
-#endif /* !MDBX_64BIT_CAS */
-#if MDBX_64BIT_ATOMIC
- /* atomically make value >= SAFE64_INVALID_THRESHOLD by 64-bit operation */
- atomic_store64(p, UINT64_MAX,
- single_writer ? mo_AcquireRelease
- : mo_SequentialConsistency);
-#else
- /* atomically make value >= SAFE64_INVALID_THRESHOLD by 32-bit operation */
- atomic_store32(&p->high, UINT32_MAX,
- single_writer ? mo_AcquireRelease : mo_SequentialConsistency);
-#endif /* MDBX_64BIT_ATOMIC */
- assert(p->weak >= SAFE64_INVALID_THRESHOLD);
- mdbx_jitter4testing(true);
-}
-
-static __always_inline bool safe64_reset_compare(MDBX_atomic_uint64_t *p,
- txnid_t compare) {
- /* LY: This function is used to reset `mr_txnid` from hsr-handler in case
- * the asynchronously cancellation of read transaction. Therefore,
- * there may be a collision between the cleanup performed here and
- * asynchronous termination and restarting of the read transaction
- * in another proces/thread. In general we MUST NOT reset the `mr_txnid`
- * if a new transaction was started (i.e. if `mr_txnid` was changed). */
-#if MDBX_64BIT_CAS
- bool rc = atomic_cas64(p, compare, UINT64_MAX);
-#else
- /* LY: There is no gold ratio here since shared mutex is too costly,
- * in such way we must acquire/release it for every update of mr_txnid,
- * i.e. twice for each read transaction). */
- bool rc = false;
- if (likely(atomic_load32(&p->low, mo_AcquireRelease) == (uint32_t)compare &&
- atomic_cas32(&p->high, (uint32_t)(compare >> 32), UINT32_MAX))) {
- if (unlikely(atomic_load32(&p->low, mo_AcquireRelease) !=
- (uint32_t)compare))
- atomic_cas32(&p->high, UINT32_MAX, (uint32_t)(compare >> 32));
- else
- rc = true;
- }
-#endif /* MDBX_64BIT_CAS */
- mdbx_jitter4testing(true);
- return rc;
-}
-
-static __always_inline void safe64_write(MDBX_atomic_uint64_t *p,
- const uint64_t v) {
- assert(p->weak >= SAFE64_INVALID_THRESHOLD);
-#if MDBX_64BIT_ATOMIC
- atomic_store64(p, v, mo_AcquireRelease);
-#else /* MDBX_64BIT_ATOMIC */
- mdbx_compiler_barrier();
- /* update low-part but still value >= SAFE64_INVALID_THRESHOLD */
- atomic_store32(&p->low, (uint32_t)v, mo_Relaxed);
- assert(p->weak >= SAFE64_INVALID_THRESHOLD);
- mdbx_jitter4testing(true);
- /* update high-part from SAFE64_INVALID_THRESHOLD to actual value */
- atomic_store32(&p->high, (uint32_t)(v >> 32), mo_AcquireRelease);
-#endif /* MDBX_64BIT_ATOMIC */
- assert(p->weak == v);
- mdbx_jitter4testing(true);
-}
-
-static __always_inline uint64_t safe64_read(const MDBX_atomic_uint64_t *p) {
- mdbx_jitter4testing(true);
- uint64_t v = atomic_load64(p, mo_AcquireRelease);
- mdbx_jitter4testing(true);
- return v;
-}
-
-#if 0 /* unused for now */
-MDBX_MAYBE_UNUSED static __always_inline bool safe64_is_valid(uint64_t v) {
-#if MDBX_WORDBITS >= 64
- return v < SAFE64_INVALID_THRESHOLD;
-#else
- return (v >> 32) != UINT32_MAX;
-#endif /* MDBX_WORDBITS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline bool
- safe64_is_valid_ptr(const MDBX_atomic_uint64_t *p) {
-#if MDBX_64BIT_ATOMIC
- return atomic_load64(p, mo_AcquireRelease) < SAFE64_INVALID_THRESHOLD;
-#else
- return atomic_load32(&p->high, mo_AcquireRelease) != UINT32_MAX;
-#endif /* MDBX_64BIT_ATOMIC */
-}
-#endif /* unused for now */
-
-/* non-atomic write with safety for reading a half-updated value */
-static __always_inline void safe64_update(MDBX_atomic_uint64_t *p,
- const uint64_t v) {
-#if MDBX_64BIT_ATOMIC
- atomic_store64(p, v, mo_Relaxed);
-#else
- safe64_reset(p, true);
- safe64_write(p, v);
-#endif /* MDBX_64BIT_ATOMIC */
-}
-
-/* non-atomic increment with safety for reading a half-updated value */
-MDBX_MAYBE_UNUSED static
-#if MDBX_64BIT_ATOMIC
- __always_inline
-#endif /* MDBX_64BIT_ATOMIC */
- void
- safe64_inc(MDBX_atomic_uint64_t *p, const uint64_t v) {
- assert(v > 0);
- safe64_update(p, atomic_load64(p, mo_Relaxed) + v);
-}
-
-/*----------------------------------------------------------------------------*/
-/* rthc (tls keys and destructors) */
-
-typedef struct rthc_entry_t {
- MDBX_reader *begin;
- MDBX_reader *end;
- mdbx_thread_key_t thr_tls_key;
- bool key_valid;
-} rthc_entry_t;
-
-#if MDBX_DEBUG
-#define RTHC_INITIAL_LIMIT 1
-#else
-#define RTHC_INITIAL_LIMIT 16
-#endif
-
-static bin128_t bootid;
-
-#if defined(_WIN32) || defined(_WIN64)
-static CRITICAL_SECTION rthc_critical_section;
-static CRITICAL_SECTION lcklist_critical_section;
-#else
-int __cxa_thread_atexit_impl(void (*dtor)(void *), void *obj, void *dso_symbol)
- __attribute__((__weak__));
-#ifdef __APPLE__ /* FIXME: Thread-Local Storage destructors & DSO-unloading */
-int __cxa_thread_atexit_impl(void (*dtor)(void *), void *obj,
- void *dso_symbol) {
- (void)dtor;
- (void)obj;
- (void)dso_symbol;
- return -1;
-}
-#endif /* __APPLE__ */
-
-static pthread_mutex_t lcklist_mutex = PTHREAD_MUTEX_INITIALIZER;
-static pthread_mutex_t rthc_mutex = PTHREAD_MUTEX_INITIALIZER;
-static pthread_cond_t rthc_cond = PTHREAD_COND_INITIALIZER;
-static mdbx_thread_key_t rthc_key;
-static MDBX_atomic_uint32_t rthc_pending;
-
-__cold static void workaround_glibc_bug21031(void) {
- /* Workaround for https://sourceware.org/bugzilla/show_bug.cgi?id=21031
- *
- * Due race between pthread_key_delete() and __nptl_deallocate_tsd()
- * The destructor(s) of thread-local-storage object(s) may be running
- * in another thread(s) and be blocked or not finished yet.
- * In such case we get a SEGFAULT after unload this library DSO.
- *
- * So just by yielding a few timeslices we give a chance
- * to such destructor(s) for completion and avoids segfault. */
- sched_yield();
- sched_yield();
- sched_yield();
-}
-#endif
-
-static unsigned rthc_count, rthc_limit;
-static rthc_entry_t *rthc_table;
-static rthc_entry_t rthc_table_static[RTHC_INITIAL_LIMIT];
-
-static __inline void rthc_lock(void) {
-#if defined(_WIN32) || defined(_WIN64)
- EnterCriticalSection(&rthc_critical_section);
-#else
- mdbx_ensure(nullptr, mdbx_pthread_mutex_lock(&rthc_mutex) == 0);
-#endif
-}
-
-static __inline void rthc_unlock(void) {
-#if defined(_WIN32) || defined(_WIN64)
- LeaveCriticalSection(&rthc_critical_section);
-#else
- mdbx_ensure(nullptr, pthread_mutex_unlock(&rthc_mutex) == 0);
-#endif
-}
-
-static __inline int thread_key_create(mdbx_thread_key_t *key) {
- int rc;
-#if defined(_WIN32) || defined(_WIN64)
- *key = TlsAlloc();
- rc = (*key != TLS_OUT_OF_INDEXES) ? MDBX_SUCCESS : GetLastError();
-#else
- rc = pthread_key_create(key, nullptr);
-#endif
- mdbx_trace("&key = %p, value %" PRIuPTR ", rc %d",
- __Wpedantic_format_voidptr(key), (uintptr_t)*key, rc);
- return rc;
-}
-
-static __inline void thread_key_delete(mdbx_thread_key_t key) {
- mdbx_trace("key = %" PRIuPTR, (uintptr_t)key);
-#if defined(_WIN32) || defined(_WIN64)
- mdbx_ensure(nullptr, TlsFree(key));
-#else
- mdbx_ensure(nullptr, pthread_key_delete(key) == 0);
- workaround_glibc_bug21031();
-#endif
-}
-
-static __inline void *thread_rthc_get(mdbx_thread_key_t key) {
-#if defined(_WIN32) || defined(_WIN64)
- return TlsGetValue(key);
-#else
- return pthread_getspecific(key);
-#endif
-}
-
-static void thread_rthc_set(mdbx_thread_key_t key, const void *value) {
-#if defined(_WIN32) || defined(_WIN64)
- mdbx_ensure(nullptr, TlsSetValue(key, (void *)value));
-#else
-#define MDBX_THREAD_RTHC_ZERO 0
-#define MDBX_THREAD_RTHC_REGISTERED 1
-#define MDBX_THREAD_RTHC_COUNTED 2
- static __thread char thread_registration_state;
- if (value && unlikely(thread_registration_state == MDBX_THREAD_RTHC_ZERO)) {
- thread_registration_state = MDBX_THREAD_RTHC_REGISTERED;
- mdbx_trace("thread registered 0x%" PRIxPTR, mdbx_thread_self());
- if (&__cxa_thread_atexit_impl == nullptr ||
- __cxa_thread_atexit_impl(mdbx_rthc_thread_dtor,
- &thread_registration_state,
- (void *)&mdbx_version /* dso_anchor */)) {
- mdbx_ensure(nullptr, pthread_setspecific(
- rthc_key, &thread_registration_state) == 0);
- thread_registration_state = MDBX_THREAD_RTHC_COUNTED;
- const unsigned count_before = atomic_add32(&rthc_pending, 1);
- mdbx_ensure(nullptr, count_before < INT_MAX);
- mdbx_trace("fallback to pthreads' tsd, key %" PRIuPTR ", count %u",
- (uintptr_t)rthc_key, count_before);
- (void)count_before;
- }
- }
- mdbx_ensure(nullptr, pthread_setspecific(key, value) == 0);
-#endif
-}
-
-__cold void mdbx_rthc_global_init(void) {
- rthc_limit = RTHC_INITIAL_LIMIT;
- rthc_table = rthc_table_static;
-#if defined(_WIN32) || defined(_WIN64)
- InitializeCriticalSection(&rthc_critical_section);
- InitializeCriticalSection(&lcklist_critical_section);
-#else
- mdbx_ensure(nullptr,
- pthread_key_create(&rthc_key, mdbx_rthc_thread_dtor) == 0);
- mdbx_trace("pid %d, &mdbx_rthc_key = %p, value 0x%x", mdbx_getpid(),
- __Wpedantic_format_voidptr(&rthc_key), (unsigned)rthc_key);
-#endif
- /* checking time conversion, this also avoids racing on 32-bit architectures
- * during writing calculated 64-bit ratio(s) into memory. */
- uint32_t proba = UINT32_MAX;
- while (true) {
- unsigned time_conversion_checkup =
- mdbx_osal_monotime_to_16dot16(mdbx_osal_16dot16_to_monotime(proba));
- unsigned one_more = (proba < UINT32_MAX) ? proba + 1 : proba;
- unsigned one_less = (proba > 0) ? proba - 1 : proba;
- mdbx_ensure(nullptr, time_conversion_checkup >= one_less &&
- time_conversion_checkup <= one_more);
- if (proba == 0)
- break;
- proba >>= 1;
- }
-
- bootid = mdbx_osal_bootid();
-#if 0 /* debug */
- for (unsigned i = 0; i < 65536; ++i) {
- size_t pages = pv2pages(i);
- unsigned x = pages2pv(pages);
- size_t xp = pv2pages(x);
- if (!(x == i || (x % 2 == 0 && x < 65536)) || pages != xp)
- printf("%u => %zu => %u => %zu\n", i, pages, x, xp);
- assert(pages == xp);
- }
- fflush(stdout);
-#endif
-}
-
-/* dtor called for thread, i.e. for all mdbx's environment objects */
-__cold void mdbx_rthc_thread_dtor(void *ptr) {
- rthc_lock();
- mdbx_trace(">> pid %d, thread 0x%" PRIxPTR ", rthc %p", mdbx_getpid(),
- mdbx_thread_self(), ptr);
-
- const uint32_t self_pid = mdbx_getpid();
- for (unsigned i = 0; i < rthc_count; ++i) {
- if (!rthc_table[i].key_valid)
- continue;
- const mdbx_thread_key_t key = rthc_table[i].thr_tls_key;
- MDBX_reader *const rthc = thread_rthc_get(key);
- if (rthc < rthc_table[i].begin || rthc >= rthc_table[i].end)
- continue;
-#if !defined(_WIN32) && !defined(_WIN64)
- if (pthread_setspecific(key, nullptr) != 0) {
- mdbx_trace("== thread 0x%" PRIxPTR
- ", rthc %p: ignore race with tsd-key deletion",
- mdbx_thread_self(), ptr);
- continue /* ignore race with tsd-key deletion by mdbx_env_close() */;
- }
-#endif
-
- mdbx_trace("== thread 0x%" PRIxPTR
- ", rthc %p, [%i], %p ... %p (%+i), rtch-pid %i, "
- "current-pid %i",
- mdbx_thread_self(), __Wpedantic_format_voidptr(rthc), i,
- __Wpedantic_format_voidptr(rthc_table[i].begin),
- __Wpedantic_format_voidptr(rthc_table[i].end),
- (int)(rthc - rthc_table[i].begin), rthc->mr_pid.weak, self_pid);
- if (atomic_load32(&rthc->mr_pid, mo_Relaxed) == self_pid) {
- mdbx_trace("==== thread 0x%" PRIxPTR ", rthc %p, cleanup",
- mdbx_thread_self(), __Wpedantic_format_voidptr(rthc));
- atomic_store32(&rthc->mr_pid, 0, mo_AcquireRelease);
- }
- }
-
-#if defined(_WIN32) || defined(_WIN64)
- mdbx_trace("<< thread 0x%" PRIxPTR ", rthc %p", mdbx_thread_self(), ptr);
- rthc_unlock();
-#else
- const char self_registration = *(volatile char *)ptr;
- *(volatile char *)ptr = MDBX_THREAD_RTHC_ZERO;
- mdbx_trace("== thread 0x%" PRIxPTR ", rthc %p, pid %d, self-status %d",
- mdbx_thread_self(), ptr, mdbx_getpid(), self_registration);
- if (self_registration == MDBX_THREAD_RTHC_COUNTED)
- mdbx_ensure(nullptr, atomic_sub32(&rthc_pending, 1) > 0);
-
- if (atomic_load32(&rthc_pending, mo_AcquireRelease) == 0) {
- mdbx_trace("== thread 0x%" PRIxPTR ", rthc %p, pid %d, wake",
- mdbx_thread_self(), ptr, mdbx_getpid());
- mdbx_ensure(nullptr, pthread_cond_broadcast(&rthc_cond) == 0);
- }
-
- mdbx_trace("<< thread 0x%" PRIxPTR ", rthc %p", mdbx_thread_self(), ptr);
- /* Allow tail call optimization, i.e. gcc should generate the jmp instruction
- * instead of a call for pthread_mutex_unlock() and therefore CPU could not
- * return to current DSO's code section, which may be unloaded immediately
- * after the mutex got released. */
- pthread_mutex_unlock(&rthc_mutex);
-#endif
-}
-
-__cold void mdbx_rthc_global_dtor(void) {
- mdbx_trace(">> pid %d", mdbx_getpid());
-
- rthc_lock();
-#if !defined(_WIN32) && !defined(_WIN64)
- char *rthc = pthread_getspecific(rthc_key);
- mdbx_trace(
- "== thread 0x%" PRIxPTR ", rthc %p, pid %d, self-status %d, left %d",
- mdbx_thread_self(), __Wpedantic_format_voidptr(rthc), mdbx_getpid(),
- rthc ? *rthc : -1, atomic_load32(&rthc_pending, mo_Relaxed));
- if (rthc) {
- const char self_registration = *rthc;
- *rthc = MDBX_THREAD_RTHC_ZERO;
- if (self_registration == MDBX_THREAD_RTHC_COUNTED)
- mdbx_ensure(nullptr, atomic_sub32(&rthc_pending, 1) > 0);
- }
-
- struct timespec abstime;
- mdbx_ensure(nullptr, clock_gettime(CLOCK_REALTIME, &abstime) == 0);
- abstime.tv_nsec += 1000000000l / 10;
- if (abstime.tv_nsec >= 1000000000l) {
- abstime.tv_nsec -= 1000000000l;
- abstime.tv_sec += 1;
- }
-#if MDBX_DEBUG > 0
- abstime.tv_sec += 600;
-#endif
-
- for (unsigned left;
- (left = atomic_load32(&rthc_pending, mo_AcquireRelease)) > 0;) {
- mdbx_trace("pid %d, pending %u, wait for...", mdbx_getpid(), left);
- const int rc = pthread_cond_timedwait(&rthc_cond, &rthc_mutex, &abstime);
- if (rc && rc != EINTR)
- break;
- }
- thread_key_delete(rthc_key);
-#endif
-
- const uint32_t self_pid = mdbx_getpid();
- for (unsigned i = 0; i < rthc_count; ++i) {
- if (!rthc_table[i].key_valid)
- continue;
- const mdbx_thread_key_t key = rthc_table[i].thr_tls_key;
- thread_key_delete(key);
- for (MDBX_reader *rthc = rthc_table[i].begin; rthc < rthc_table[i].end;
- ++rthc) {
- mdbx_trace(
- "== [%i] = key %" PRIuPTR ", %p ... %p, rthc %p (%+i), "
- "rthc-pid %i, current-pid %i",
- i, (uintptr_t)key, __Wpedantic_format_voidptr(rthc_table[i].begin),
- __Wpedantic_format_voidptr(rthc_table[i].end),
- __Wpedantic_format_voidptr(rthc), (int)(rthc - rthc_table[i].begin),
- rthc->mr_pid.weak, self_pid);
- if (atomic_load32(&rthc->mr_pid, mo_Relaxed) == self_pid) {
- atomic_store32(&rthc->mr_pid, 0, mo_AcquireRelease);
- mdbx_trace("== cleanup %p", __Wpedantic_format_voidptr(rthc));
- }
- }
- }
-
- rthc_limit = rthc_count = 0;
- if (rthc_table != rthc_table_static)
- mdbx_free(rthc_table);
- rthc_table = nullptr;
- rthc_unlock();
-
-#if defined(_WIN32) || defined(_WIN64)
- DeleteCriticalSection(&lcklist_critical_section);
- DeleteCriticalSection(&rthc_critical_section);
-#else
- /* LY: yielding a few timeslices to give a more chance
- * to racing destructor(s) for completion. */
- workaround_glibc_bug21031();
-#endif
-
- mdbx_trace("<< pid %d\n", mdbx_getpid());
-}
-
-__cold int mdbx_rthc_alloc(mdbx_thread_key_t *key, MDBX_reader *begin,
- MDBX_reader *end) {
- int rc;
- if (key) {
-#ifndef NDEBUG
- *key = (mdbx_thread_key_t)0xBADBADBAD;
-#endif /* NDEBUG */
- rc = thread_key_create(key);
- if (rc != MDBX_SUCCESS)
- return rc;
- }
-
- rthc_lock();
- const mdbx_thread_key_t new_key = key ? *key : 0;
- mdbx_trace(">> key %" PRIuPTR ", rthc_count %u, rthc_limit %u",
- (uintptr_t)new_key, rthc_count, rthc_limit);
- if (rthc_count == rthc_limit) {
- rthc_entry_t *new_table =
- mdbx_realloc((rthc_table == rthc_table_static) ? nullptr : rthc_table,
- sizeof(rthc_entry_t) * rthc_limit * 2);
- if (new_table == nullptr) {
- rc = MDBX_ENOMEM;
- goto bailout;
- }
- if (rthc_table == rthc_table_static)
- memcpy(new_table, rthc_table_static, sizeof(rthc_table_static));
- rthc_table = new_table;
- rthc_limit *= 2;
- }
- mdbx_trace("== [%i] = key %" PRIuPTR ", %p ... %p", rthc_count,
- (uintptr_t)new_key, __Wpedantic_format_voidptr(begin),
- __Wpedantic_format_voidptr(end));
- rthc_table[rthc_count].key_valid = key ? true : false;
- rthc_table[rthc_count].thr_tls_key = key ? new_key : 0;
- rthc_table[rthc_count].begin = begin;
- rthc_table[rthc_count].end = end;
- ++rthc_count;
- mdbx_trace("<< key %" PRIuPTR ", rthc_count %u, rthc_limit %u",
- (uintptr_t)new_key, rthc_count, rthc_limit);
- rthc_unlock();
- return MDBX_SUCCESS;
-
-bailout:
- if (key)
- thread_key_delete(*key);
- rthc_unlock();
- return rc;
-}
-
-__cold void mdbx_rthc_remove(const mdbx_thread_key_t key) {
- thread_key_delete(key);
- rthc_lock();
- mdbx_trace(">> key %zu, rthc_count %u, rthc_limit %u", (size_t)key,
- rthc_count, rthc_limit);
-
- for (unsigned i = 0; i < rthc_count; ++i) {
- if (rthc_table[i].key_valid && key == rthc_table[i].thr_tls_key) {
- const uint32_t self_pid = mdbx_getpid();
- mdbx_trace("== [%i], %p ...%p, current-pid %d", i,
- __Wpedantic_format_voidptr(rthc_table[i].begin),
- __Wpedantic_format_voidptr(rthc_table[i].end), self_pid);
-
- for (MDBX_reader *rthc = rthc_table[i].begin; rthc < rthc_table[i].end;
- ++rthc) {
- if (atomic_load32(&rthc->mr_pid, mo_Relaxed) == self_pid) {
- atomic_store32(&rthc->mr_pid, 0, mo_AcquireRelease);
- mdbx_trace("== cleanup %p", __Wpedantic_format_voidptr(rthc));
- }
- }
- if (--rthc_count > 0)
- rthc_table[i] = rthc_table[rthc_count];
- else if (rthc_table != rthc_table_static) {
- mdbx_free(rthc_table);
- rthc_table = rthc_table_static;
- rthc_limit = RTHC_INITIAL_LIMIT;
- }
- break;
- }
- }
-
- mdbx_trace("<< key %zu, rthc_count %u, rthc_limit %u", (size_t)key,
- rthc_count, rthc_limit);
- rthc_unlock();
-}
-
-//------------------------------------------------------------------------------
-
-#define RTHC_ENVLIST_END ((MDBX_env *)((uintptr_t)50459))
-static MDBX_env *inprocess_lcklist_head = RTHC_ENVLIST_END;
-
-static __inline void lcklist_lock(void) {
-#if defined(_WIN32) || defined(_WIN64)
- EnterCriticalSection(&lcklist_critical_section);
-#else
- mdbx_ensure(nullptr, mdbx_pthread_mutex_lock(&lcklist_mutex) == 0);
-#endif
-}
-
-static __inline void lcklist_unlock(void) {
-#if defined(_WIN32) || defined(_WIN64)
- LeaveCriticalSection(&lcklist_critical_section);
-#else
- mdbx_ensure(nullptr, pthread_mutex_unlock(&lcklist_mutex) == 0);
-#endif
-}
-
-MDBX_NOTHROW_CONST_FUNCTION static uint64_t rrxmrrxmsx_0(uint64_t v) {
- /* Pelle Evensen's mixer, https://bit.ly/2HOfynt */
- v ^= (v << 39 | v >> 25) ^ (v << 14 | v >> 50);
- v *= UINT64_C(0xA24BAED4963EE407);
- v ^= (v << 40 | v >> 24) ^ (v << 15 | v >> 49);
- v *= UINT64_C(0x9FB21C651E98DF25);
- return v ^ v >> 28;
-}
-
-static int uniq_peek(const mdbx_mmap_t *pending, mdbx_mmap_t *scan) {
- int rc;
- uint64_t bait;
- MDBX_lockinfo *const pending_lck = pending->lck;
- MDBX_lockinfo *const scan_lck = scan->lck;
- if (pending_lck) {
- bait = atomic_load64(&pending_lck->mti_bait_uniqueness, mo_AcquireRelease);
- rc = MDBX_SUCCESS;
- } else {
- bait = 0 /* hush MSVC warning */;
- rc = mdbx_msync(scan, 0, sizeof(MDBX_lockinfo), MDBX_SYNC_DATA);
- if (rc == MDBX_SUCCESS)
- rc = mdbx_pread(pending->fd, &bait, sizeof(scan_lck->mti_bait_uniqueness),
- offsetof(MDBX_lockinfo, mti_bait_uniqueness));
- }
- if (likely(rc == MDBX_SUCCESS) &&
- bait == atomic_load64(&scan_lck->mti_bait_uniqueness, mo_AcquireRelease))
- rc = MDBX_RESULT_TRUE;
-
- mdbx_trace("uniq-peek: %s, bait 0x%016" PRIx64 ",%s rc %d",
- pending_lck ? "mem" : "file", bait,
- (rc == MDBX_RESULT_TRUE) ? " found," : (rc ? " FAILED," : ""), rc);
- return rc;
-}
-
-static int uniq_poke(const mdbx_mmap_t *pending, mdbx_mmap_t *scan,
- uint64_t *abra) {
- if (*abra == 0) {
- const uintptr_t tid = mdbx_thread_self();
- uintptr_t uit = 0;
- memcpy(&uit, &tid, (sizeof(tid) < sizeof(uit)) ? sizeof(tid) : sizeof(uit));
- *abra =
- rrxmrrxmsx_0(mdbx_osal_monotime() + UINT64_C(5873865991930747) * uit);
- }
- const uint64_t cadabra =
- rrxmrrxmsx_0(*abra + UINT64_C(7680760450171793) * (unsigned)mdbx_getpid())
- << 24 |
- *abra >> 40;
- MDBX_lockinfo *const scan_lck = scan->lck;
- atomic_store64(&scan_lck->mti_bait_uniqueness, cadabra,
- mo_SequentialConsistency);
- *abra = *abra * UINT64_C(6364136223846793005) + 1;
- return uniq_peek(pending, scan);
-}
-
-__cold static int uniq_check(const mdbx_mmap_t *pending, MDBX_env **found) {
- *found = nullptr;
- uint64_t salt = 0;
- for (MDBX_env *scan = inprocess_lcklist_head; scan != RTHC_ENVLIST_END;
- scan = scan->me_lcklist_next) {
- MDBX_lockinfo *const scan_lck = scan->me_lck_mmap.lck;
- int err = atomic_load64(&scan_lck->mti_bait_uniqueness, mo_AcquireRelease)
- ? uniq_peek(pending, &scan->me_lck_mmap)
- : uniq_poke(pending, &scan->me_lck_mmap, &salt);
- if (err == MDBX_ENODATA) {
- uint64_t length;
- if (likely(mdbx_filesize(pending->fd, &length) == MDBX_SUCCESS &&
- length == 0)) {
- /* LY: skip checking since LCK-file is empty, i.e. just created. */
- mdbx_debug("uniq-probe: %s", "unique (new/empty lck)");
- return MDBX_RESULT_TRUE;
- }
- }
- if (err == MDBX_RESULT_TRUE)
- err = uniq_poke(pending, &scan->me_lck_mmap, &salt);
- if (err == MDBX_RESULT_TRUE) {
- (void)mdbx_msync(&scan->me_lck_mmap, 0, sizeof(MDBX_lockinfo),
- MDBX_SYNC_NONE);
- err = uniq_poke(pending, &scan->me_lck_mmap, &salt);
- }
- if (err == MDBX_RESULT_TRUE) {
- err = uniq_poke(pending, &scan->me_lck_mmap, &salt);
- *found = scan;
- mdbx_debug("uniq-probe: found %p", __Wpedantic_format_voidptr(*found));
- return MDBX_RESULT_FALSE;
- }
- if (unlikely(err != MDBX_SUCCESS)) {
- mdbx_debug("uniq-probe: failed rc %d", err);
- return err;
- }
- }
-
- mdbx_debug("uniq-probe: %s", "unique");
- return MDBX_RESULT_TRUE;
-}
-
-static int lcklist_detach_locked(MDBX_env *env) {
- MDBX_env *inprocess_neighbor = nullptr;
- int rc = MDBX_SUCCESS;
- if (env->me_lcklist_next != nullptr) {
- mdbx_ensure(env, env->me_lcklist_next != nullptr);
- mdbx_ensure(env, inprocess_lcklist_head != RTHC_ENVLIST_END);
- for (MDBX_env **ptr = &inprocess_lcklist_head; *ptr != RTHC_ENVLIST_END;
- ptr = &(*ptr)->me_lcklist_next) {
- if (*ptr == env) {
- *ptr = env->me_lcklist_next;
- env->me_lcklist_next = nullptr;
- break;
- }
- }
- mdbx_ensure(env, env->me_lcklist_next == nullptr);
- }
-
- rc = likely(mdbx_getpid() == env->me_pid)
- ? uniq_check(&env->me_lck_mmap, &inprocess_neighbor)
- : MDBX_PANIC;
- if (!inprocess_neighbor && env->me_live_reader)
- (void)mdbx_rpid_clear(env);
- if (!MDBX_IS_ERROR(rc))
- rc = mdbx_lck_destroy(env, inprocess_neighbor);
- return rc;
-}
-
-/*------------------------------------------------------------------------------
- * LY: State of the art quicksort-based sorting, with internal stack
- * and network-sort for small chunks.
- * Thanks to John M. Gamble for the http://pages.ripco.net/~jgamble/nw.html */
-
-#define SORT_CMP_SWAP(TYPE, CMP, a, b) \
- do { \
- const TYPE swap_tmp = (a); \
- const bool swap_cmp = CMP(swap_tmp, b); \
- (a) = swap_cmp ? swap_tmp : b; \
- (b) = swap_cmp ? b : swap_tmp; \
- } while (0)
-
-// 3 comparators, 3 parallel operations
-// o-----^--^--o
-// | |
-// o--^--|--v--o
-// | |
-// o--v--v-----o
-//
-// [[1,2]]
-// [[0,2]]
-// [[0,1]]
-#define SORT_NETWORK_3(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- } while (0)
-
-// 5 comparators, 3 parallel operations
-// o--^--^--------o
-// | |
-// o--v--|--^--^--o
-// | | |
-// o--^--v--|--v--o
-// | |
-// o--v-----v-----o
-//
-// [[0,1],[2,3]]
-// [[0,2],[1,3]]
-// [[1,2]]
-#define SORT_NETWORK_4(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- } while (0)
-
-// 9 comparators, 5 parallel operations
-// o--^--^-----^-----------o
-// | | |
-// o--|--|--^--v-----^--^--o
-// | | | | |
-// o--|--v--|--^--^--|--v--o
-// | | | | |
-// o--|-----v--|--v--|--^--o
-// | | | |
-// o--v--------v-----v--v--o
-//
-// [[0,4],[1,3]]
-// [[0,2]]
-// [[2,4],[0,1]]
-// [[2,3],[1,4]]
-// [[1,2],[3,4]]
-#define SORT_NETWORK_5(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- } while (0)
-
-// 12 comparators, 6 parallel operations
-// o-----^--^--^-----------------o
-// | | |
-// o--^--|--v--|--^--------^-----o
-// | | | | |
-// o--v--v-----|--|--^--^--|--^--o
-// | | | | | |
-// o-----^--^--v--|--|--|--v--v--o
-// | | | | |
-// o--^--|--v-----v--|--v--------o
-// | | |
-// o--v--v-----------v-----------o
-//
-// [[1,2],[4,5]]
-// [[0,2],[3,5]]
-// [[0,1],[3,4],[2,5]]
-// [[0,3],[1,4]]
-// [[2,4],[1,3]]
-// [[2,3]]
-#define SORT_NETWORK_6(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- } while (0)
-
-// 16 comparators, 6 parallel operations
-// o--^--------^-----^-----------------o
-// | | |
-// o--|--^-----|--^--v--------^--^-----o
-// | | | | | |
-// o--|--|--^--v--|--^-----^--|--v-----o
-// | | | | | | |
-// o--|--|--|-----v--|--^--v--|--^--^--o
-// | | | | | | | |
-// o--v--|--|--^-----v--|--^--v--|--v--o
-// | | | | | |
-// o-----v--|--|--------v--v-----|--^--o
-// | | | |
-// o--------v--v-----------------v--v--o
-//
-// [[0,4],[1,5],[2,6]]
-// [[0,2],[1,3],[4,6]]
-// [[2,4],[3,5],[0,1]]
-// [[2,3],[4,5]]
-// [[1,4],[3,6]]
-// [[1,2],[3,4],[5,6]]
-#define SORT_NETWORK_7(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- } while (0)
-
-// 19 comparators, 6 parallel operations
-// o--^--------^-----^-----------------o
-// | | |
-// o--|--^-----|--^--v--------^--^-----o
-// | | | | | |
-// o--|--|--^--v--|--^-----^--|--v-----o
-// | | | | | | |
-// o--|--|--|--^--v--|--^--v--|--^--^--o
-// | | | | | | | | |
-// o--v--|--|--|--^--v--|--^--v--|--v--o
-// | | | | | | |
-// o-----v--|--|--|--^--v--v-----|--^--o
-// | | | | | |
-// o--------v--|--v--|--^--------v--v--o
-// | | |
-// o-----------v-----v--v--------------o
-//
-// [[0,4],[1,5],[2,6],[3,7]]
-// [[0,2],[1,3],[4,6],[5,7]]
-// [[2,4],[3,5],[0,1],[6,7]]
-// [[2,3],[4,5]]
-// [[1,4],[3,6]]
-// [[1,2],[3,4],[5,6]]
-#define SORT_NETWORK_8(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- } while (0)
-
-// 25 comparators, 9 parallel operations
-// o--^-----^--^-----^-----------------------------------o
-// | | | |
-// o--v--^--v--|-----|--^-----^-----------^--------------o
-// | | | | | |
-// o-----v-----|-----|--|-----|--^-----^--|--^-----^--^--o
-// | | | | | | | | | |
-// o--^-----^--v--^--v--|-----|--|-----|--v--|-----|--v--o
-// | | | | | | | | |
-// o--v--^--v-----|-----v--^--v--|-----|-----|--^--v-----o
-// | | | | | | |
-// o-----v--------|--------|-----v--^--v--^--|--|--^-----o
-// | | | | | | |
-// o--^-----^-----v--------|--------|-----|--v--v--v-----o
-// | | | | |
-// o--v--^--v--------------v--------|-----v--------------o
-// | |
-// o-----v--------------------------v--------------------o
-//
-// [[0,1],[3,4],[6,7]]
-// [[1,2],[4,5],[7,8]]
-// [[0,1],[3,4],[6,7],[2,5]]
-// [[0,3],[1,4],[5,8]]
-// [[3,6],[4,7],[2,5]]
-// [[0,3],[1,4],[5,7],[2,6]]
-// [[1,3],[4,6]]
-// [[2,4],[5,6]]
-// [[2,3]]
-#define SORT_NETWORK_9(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- } while (0)
-
-// 29 comparators, 9 parallel operations
-// o--------------^-----^--^--^-----------------------o
-// | | | |
-// o-----------^--|--^--|--|--v--^--------^-----------o
-// | | | | | | |
-// o--------^--|--|--|--|--v--^--v-----^--|--^--------o
-// | | | | | | | | |
-// o-----^--|--|--|--|--v--^--|-----^--|--v--v--^-----o
-// | | | | | | | | | |
-// o--^--|--|--|--|--v-----|--v--^--|--|--^-----v--^--o
-// | | | | | | | | | | |
-// o--|--|--|--|--v--^-----|--^--|--v--v--|-----^--v--o
-// | | | | | | | | | |
-// o--|--|--|--v--^--|-----v--|--v--^-----|--^--v-----o
-// | | | | | | | | |
-// o--|--|--v-----|--|--^-----v--^--|-----v--v--------o
-// | | | | | | |
-// o--|--v--------|--v--|--^-----v--v-----------------o
-// | | | |
-// o--v-----------v-----v--v--------------------------o
-//
-// [[4,9],[3,8],[2,7],[1,6],[0,5]]
-// [[1,4],[6,9],[0,3],[5,8]]
-// [[0,2],[3,6],[7,9]]
-// [[0,1],[2,4],[5,7],[8,9]]
-// [[1,2],[4,6],[7,8],[3,5]]
-// [[2,5],[6,8],[1,3],[4,7]]
-// [[2,3],[6,7]]
-// [[3,4],[5,6]]
-// [[4,5]]
-#define SORT_NETWORK_10(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- } while (0)
-
-// 35 comparators, 9 parallel operations
-// o--^-----^-----------------^--------^--------------------o
-// | | | |
-// o--v--^--|--^--^--------^--|--------|--^-----------------o
-// | | | | | | | |
-// o--^--|--v--v--|-----^--|--|--------|--|-----^--^--------o
-// | | | | | | | | | |
-// o--v--v--------|-----|--|--|--^-----|--|--^--v--|--^--^--o
-// | | | | | | | | | | |
-// o--^-----^-----|-----|--|--v--|--^--v--v--|-----v--|--v--o
-// | | | | | | | | |
-// o--v--^--|--^--v--^--|--v-----|--|--------|--------v--^--o
-// | | | | | | | | |
-// o--^--|--v--v--^--|--v--^-----|--|--------|--------^--v--o
-// | | | | | | | | |
-// o--v--v--------|--|-----|-----v--|--^-----|-----^--|--^--o
-// | | | | | | | | |
-// o--^--^--------|--|-----|--------v--|-----v--^--|--v--v--o
-// | | | | | | | |
-// o--v--|--^-----|--v-----|-----------|--------v--v--------o
-// | | | | |
-// o-----v--v-----v--------v-----------v--------------------o
-//
-// [[0,1],[2,3],[4,5],[6,7],[8,9]]
-// [[1,3],[5,7],[0,2],[4,6],[8,10]]
-// [[1,2],[5,6],[9,10],[0,4],[3,7]]
-// [[1,5],[6,10],[4,8]]
-// [[5,9],[2,6],[0,4],[3,8]]
-// [[1,5],[6,10],[2,3],[8,9]]
-// [[1,4],[7,10],[3,5],[6,8]]
-// [[2,4],[7,9],[5,6]]
-// [[3,4],[7,8]]
-#define SORT_NETWORK_11(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- } while (0)
-
-// 39 comparators, parallel operations
-// o--^-----^-----------------^--------^--------------------o
-// | | | |
-// o--v--^--|--^--^--------^--|--------|--^-----------------o
-// | | | | | | | |
-// o--^--|--v--v--|-----^--|--|--------|--|-----^--^--------o
-// | | | | | | | | | |
-// o--v--v--------|-----|--|--|--^-----|--|--^--v--|--^--^--o
-// | | | | | | | | | | |
-// o--^-----^-----|-----|--|--v--|--^--v--v--|-----v--|--v--o
-// | | | | | | | | |
-// o--v--^--|--^--v--^--|--v-----|--|--------|--------v--^--o
-// | | | | | | | | |
-// o--^--|--v--v--^--|--v--^-----|--|--------|--------^--v--o
-// | | | | | | | | |
-// o--v--v--------|--|-----|--^--v--|--^--^--|-----^--|--^--o
-// | | | | | | | | | | |
-// o--^-----^-----|--|-----|--|-----v--|--|--v--^--|--v--v--o
-// | | | | | | | | | |
-// o--v--^--|--^--|--v-----|--|--------|--|-----v--v--------o
-// | | | | | | | |
-// o--^--|--v--v--v--------v--|--------|--v-----------------o
-// | | | |
-// o--v--v--------------------v--------v--------------------o
-//
-// [[0,1],[2,3],[4,5],[6,7],[8,9],[10,11]]
-// [[1,3],[5,7],[9,11],[0,2],[4,6],[8,10]]
-// [[1,2],[5,6],[9,10],[0,4],[7,11]]
-// [[1,5],[6,10],[3,7],[4,8]]
-// [[5,9],[2,6],[0,4],[7,11],[3,8]]
-// [[1,5],[6,10],[2,3],[8,9]]
-// [[1,4],[7,10],[3,5],[6,8]]
-// [[2,4],[7,9],[5,6]]
-// [[3,4],[7,8]]
-#define SORT_NETWORK_12(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- } while (0)
-
-// 45 comparators, 10 parallel operations
-// o--------^--^-----^-----------------------------^-----------------o
-// | | | |
-// o--^-----|--v-----|-----^--------------^-----^--|-----^-----------o
-// | | | | | | | |
-// o--|-----|--^--^--v-----|--------------|--^--|--|--^--v--^--------o
-// | | | | | | | | | | |
-// o--|--^--|--|--v-----^--|--------^-----|--|--v--|--|--^--v-----^--o
-// | | | | | | | | | | | | |
-// o--|--v--|--|--^-----|--v-----^--v-----|--|--^--|--|--|--^--^--v--o
-// | | | | | | | | | | | | | |
-// o--|--^--|--|--|--^--|--------|-----^--|--|--|--v--v--v--|--v--^--o
-// | | | | | | | | | | | | | |
-// o--|--|--|--v--v--|--|--^-----|--^--v--|--v--|--^--------v--^--v--o
-// | | | | | | | | | | | |
-// o--v--|--|-----^--|--v--|--^--|--|-----v-----v--|--^--------v-----o
-// | | | | | | | | | |
-// o-----v--|--^--|--|-----|--v--|--|--^-----^-----v--v--^-----------o
-// | | | | | | | | | |
-// o--^-----|--|--|--v-----|-----v--|--v--^--|--^--------v-----------o
-// | | | | | | | | |
-// o--|-----|--|--|--^-----|--------v--^--|--v--v--------------------o
-// | | | | | | | |
-// o--v-----|--v--|--v-----|--^--------v--v--------------------------o
-// | | | |
-// o--------v-----v--------v--v--------------------------------------o
-//
-// [[1,7],[9,11],[3,4],[5,8],[0,12],[2,6]]
-// [[0,1],[2,3],[4,6],[8,11],[7,12],[5,9]]
-// [[0,2],[3,7],[10,11],[1,4],[6,12]]
-// [[7,8],[11,12],[4,9],[6,10]]
-// [[3,4],[5,6],[8,9],[10,11],[1,7]]
-// [[2,6],[9,11],[1,3],[4,7],[8,10],[0,5]]
-// [[2,5],[6,8],[9,10]]
-// [[1,2],[3,5],[7,8],[4,6]]
-// [[2,3],[4,5],[6,7],[8,9]]
-// [[3,4],[5,6]]
-#define SORT_NETWORK_13(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- } while (0)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-// 51 comparators, 10 parallel operations
-// o--^--^-----^-----------^-----------------------------------------------------------o
-// | | | |
-// o--v--|--^--|--^--------|--^-----^-----------------------^--------------------------o
-// | | | | | | | |
-// o--^--v--|--|--|--^-----|--|--^--v-----------------------|--^--^--------------------o
-// | | | | | | | | | | |
-// o--v-----v--|--|--|--^--|--|--|--^--------------^--------|--|--|--^--^--^-----------o
-// | | | | | | | | | | | | | | |
-// o--^--^-----v--|--|--|--|--|--|--|--^-----------|-----^--v--|--v--|--|--v-----------o
-// | | | | | | | | | | | | | | |
-// o--v--|--^-----v--|--|--|--|--|--|--|--^--^-----|-----|-----|--^--|--v-----^--------o
-// | | | | | | | | | | | | | | | | |
-// o--^--v--|--------v--|--|--|--|--|--|--|--|--^--|-----|-----|--v--|-----^--v-----^--o
-// | | | | | | | | | | | | | | | | |
-// o--v-----v-----------v--|--|--|--|--|--|--|--|--|--^--|--^--|-----|--^--|--^--^--v--o
-// | | | | | | | | | | | | | | | | | |
-// o--^--^-----^-----------v--|--|--|--|--|--|--|--|--|--v--|--v-----v--|--v--|--v--^--o
-// | | | | | | | | | | | | | | | |
-// o--v--|--^--|--^-----------v--|--|--|--|--|--v--|--|-----|--^--------|-----v--^--v--o
-// | | | | | | | | | | | | | | |
-// o--^--v--|--|--|--------------v--|--|--|--v-----|--|-----|--v--------|--^-----v-----o
-// | | | | | | | | | | | |
-// o--v-----v--|--|-----------------v--|--|--------|--v-----|--^--------|--|--^--------o
-// | | | | | | | | | |
-// o--^--------v--|--------------------v--|--------v--------|--|--------v--v--v--------o
-// | | | | |
-// o--v-----------v-----------------------v-----------------v--v-----------------------o
-//
-// [[0,1],[2,3],[4,5],[6,7],[8,9],[10,11],[12,13]]
-// [[0,2],[4,6],[8,10],[1,3],[5,7],[9,11]]
-// [[0,4],[8,12],[1,5],[9,13],[2,6],[3,7]]
-// [[0,8],[1,9],[2,10],[3,11],[4,12],[5,13]]
-// [[5,10],[6,9],[3,12],[7,11],[1,2],[4,8]]
-// [[1,4],[7,13],[2,8],[5,6],[9,10]]
-// [[2,4],[11,13],[3,8],[7,12]]
-// [[6,8],[10,12],[3,5],[7,9]]
-// [[3,4],[5,6],[7,8],[9,10],[11,12]]
-// [[6,7],[8,9]]
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#define SORT_NETWORK_14(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[12], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- } while (0)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-// 56 comparators, 10 parallel operations
-// o--^--^-----^-----------^--------------------------------------------------------------o
-// | | | |
-// o--v--|--^--|--^--------|--^-----^--------------------------^--------------------------o
-// | | | | | | | |
-// o--^--v--|--|--|--^-----|--|--^--v--------------------------|--^--^--------------------o
-// | | | | | | | | | | |
-// o--v-----v--|--|--|--^--|--|--|--^-----------------^--------|--|--|--^--^--^-----------o
-// | | | | | | | | | | | | | | |
-// o--^--^-----v--|--|--|--|--|--|--|--^--------------|-----^--v--|--v--|--|--v-----------o
-// | | | | | | | | | | | | | | |
-// o--v--|--^-----v--|--|--|--|--|--|--|--^-----^-----|-----|-----|--^--|--v-----^--------o
-// | | | | | | | | | | | | | | | | |
-// o--^--v--|--------v--|--|--|--|--|--|--|--^--|--^--|-----|-----|--v--|-----^--v-----^--o
-// | | | | | | | | | | | | | | | | | |
-// o--v-----v-----------v--|--|--|--|--|--|--|--|--|--|--^--|--^--|-----|--^--|--^--^--v--o
-// | | | | | | | | | | | | | | | | | | |
-// o--^--^-----^-----------v--|--|--|--|--|--|--|--|--|--|--v--|--v-----v--|--v--|--v--^--o
-// | | | | | | | | | | | | | | | | |
-// o--v--|--^--|--^-----------v--|--|--|--|--|--|--v--|--|-----|--^--------|-----v--^--v--o
-// | | | | | | | | | | | | | | | |
-// o--^--v--|--|--|--^-----------v--|--|--|--|--v-----|--|-----|--v--------|--^-----v-----o
-// | | | | | | | | | | | | | |
-// o--v-----v--|--|--|--------------v--|--|--|--------|--v-----|--^--^-----|--|--^--------o
-// | | | | | | | | | | | | |
-// o--^--^-----v--|--|-----------------v--|--|--------v--------|--|--|-----v--v--v--------o
-// | | | | | | | | |
-// o--v--|--------v--|--------------------v--|--^--------------v--|--v--------------------o
-// | | | | |
-// o-----v-----------v-----------------------v--v-----------------v-----------------------o
-//
-// [[0,1],[2,3],[4,5],[6,7],[8,9],[10,11],[12,13]]
-// [[0,2],[4,6],[8,10],[12,14],[1,3],[5,7],[9,11]]
-// [[0,4],[8,12],[1,5],[9,13],[2,6],[10,14],[3,7]]
-// [[0,8],[1,9],[2,10],[3,11],[4,12],[5,13],[6,14]]
-// [[5,10],[6,9],[3,12],[13,14],[7,11],[1,2],[4,8]]
-// [[1,4],[7,13],[2,8],[11,14],[5,6],[9,10]]
-// [[2,4],[11,13],[3,8],[7,12]]
-// [[6,8],[10,12],[3,5],[7,9]]
-// [[3,4],[5,6],[7,8],[9,10],[11,12]]
-// [[6,7],[8,9]]
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#define SORT_NETWORK_15(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[12], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[12], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[13], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- } while (0)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-// 60 comparators, 10 parallel operations
-// o--^--^-----^-----------^-----------------------------------------------------------------o
-// | | | |
-// o--v--|--^--|--^--------|--^-----^-----------------------------^--------------------------o
-// | | | | | | | |
-// o--^--v--|--|--|--^-----|--|--^--v-----------------------------|--^--^--------------------o
-// | | | | | | | | | | |
-// o--v-----v--|--|--|--^--|--|--|--^--------------------^--------|--|--|--^--^--^-----------o
-// | | | | | | | | | | | | | | |
-// o--^--^-----v--|--|--|--|--|--|--|--^-----------------|-----^--v--|--v--|--|--v-----------o
-// | | | | | | | | | | | | | | |
-// o--v--|--^-----v--|--|--|--|--|--|--|--^--------^-----|-----|-----|--^--|--v-----^--------o
-// | | | | | | | | | | | | | | | | |
-// o--^--v--|--------v--|--|--|--|--|--|--|--^-----|--^--|-----|-----|--v--|-----^--v-----^--o
-// | | | | | | | | | | | | | | | | | |
-// o--v-----v-----------v--|--|--|--|--|--|--|--^--|--|--|--^--|--^--|-----|--^--|--^--^--v--o
-// | | | | | | | | | | | | | | | | | | | |
-// o--^--^-----^-----------v--|--|--|--|--|--|--|--|--|--|--|--v--|--v-----v--|--v--|--v--^--o
-// | | | | | | | | | | | | | | | | | |
-// o--v--|--^--|--^-----------v--|--|--|--|--|--|--|--v--|--|-----|--^--------|-----v--^--v--o
-// | | | | | | | | | | | | | | | | |
-// o--^--v--|--|--|--^-----------v--|--|--|--|--|--v-----|--|-----|--v--------|--^-----v-----o
-// | | | | | | | | | | | | | | |
-// o--v-----v--|--|--|--^-----------v--|--|--|--|--------|--v-----|--^--^-----|--|--^--------o
-// | | | | | | | | | | | | | | |
-// o--^--^-----v--|--|--|--------------v--|--|--|--------v--------|--|--|-----v--v--v--------o
-// | | | | | | | | | | |
-// o--v--|--^-----v--|--|-----------------v--|--|--^--------------v--|--v--------------------o
-// | | | | | | | |
-// o--^--v--|--------v--|--------------------v--|--v-----------------v-----------------------o
-// | | | |
-// o--v-----v-----------v-----------------------v--------------------------------------------o
-//
-// [[0,1],[2,3],[4,5],[6,7],[8,9],[10,11],[12,13],[14,15]]
-// [[0,2],[4,6],[8,10],[12,14],[1,3],[5,7],[9,11],[13,15]]
-// [[0,4],[8,12],[1,5],[9,13],[2,6],[10,14],[3,7],[11,15]]
-// [[0,8],[1,9],[2,10],[3,11],[4,12],[5,13],[6,14],[7,15]]
-// [[5,10],[6,9],[3,12],[13,14],[7,11],[1,2],[4,8]]
-// [[1,4],[7,13],[2,8],[11,14],[5,6],[9,10]]
-// [[2,4],[11,13],[3,8],[7,12]]
-// [[6,8],[10,12],[3,5],[7,9]]
-// [[3,4],[5,6],[7,8],[9,10],[11,12]]
-// [[6,7],[8,9]]
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#define SORT_NETWORK_16(TYPE, CMP, begin) \
- do { \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[12], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[14], begin[15]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[12], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[3]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[13], begin[15]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[15]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[15]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[13], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[11]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[2]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[4], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[1], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[14]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[2], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[13]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[10], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[5]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[9]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[3], begin[4]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[5], begin[6]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[7], begin[8]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[9], begin[10]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[11], begin[12]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[6], begin[7]); \
- SORT_CMP_SWAP(TYPE, CMP, begin[8], begin[9]); \
- } while (0)
-
-#define SORT_INNER(TYPE, CMP, begin, end, len) \
- switch (len) { \
- default: \
- __unreachable(); \
- case 0: \
- case 1: \
- break; \
- case 2: \
- SORT_CMP_SWAP(TYPE, CMP, begin[0], begin[1]); \
- break; \
- case 3: \
- SORT_NETWORK_3(TYPE, CMP, begin); \
- break; \
- case 4: \
- SORT_NETWORK_4(TYPE, CMP, begin); \
- break; \
- case 5: \
- SORT_NETWORK_5(TYPE, CMP, begin); \
- break; \
- case 6: \
- SORT_NETWORK_6(TYPE, CMP, begin); \
- break; \
- case 7: \
- SORT_NETWORK_7(TYPE, CMP, begin); \
- break; \
- case 8: \
- SORT_NETWORK_8(TYPE, CMP, begin); \
- break; \
- case 9: \
- SORT_NETWORK_9(TYPE, CMP, begin); \
- break; \
- case 10: \
- SORT_NETWORK_10(TYPE, CMP, begin); \
- break; \
- case 11: \
- SORT_NETWORK_11(TYPE, CMP, begin); \
- break; \
- case 12: \
- SORT_NETWORK_12(TYPE, CMP, begin); \
- break; \
- case 13: \
- SORT_NETWORK_13(TYPE, CMP, begin); \
- break; \
- case 14: \
- SORT_NETWORK_14(TYPE, CMP, begin); \
- break; \
- case 15: \
- SORT_NETWORK_15(TYPE, CMP, begin); \
- break; \
- case 16: \
- SORT_NETWORK_16(TYPE, CMP, begin); \
- break; \
- }
-
-#define SORT_SWAP(TYPE, a, b) \
- do { \
- const TYPE swap_tmp = (a); \
- (a) = (b); \
- (b) = swap_tmp; \
- } while (0)
-
-#define SORT_PUSH(low, high) \
- do { \
- top->lo = (low); \
- top->hi = (high); \
- ++top; \
- } while (0)
-
-#define SORT_POP(low, high) \
- do { \
- --top; \
- low = top->lo; \
- high = top->hi; \
- } while (0)
-
-#define SORT_IMPL(NAME, EXPECT_LOW_CARDINALITY_OR_PRESORTED, TYPE, CMP) \
- \
- static __inline bool NAME##_is_sorted(const TYPE *first, const TYPE *last) { \
- while (++first <= last) \
- if (CMP(first[0], first[-1])) \
- return false; \
- return true; \
- } \
- \
- typedef struct { \
- TYPE *lo, *hi; \
- } NAME##_stack; \
- \
- static __hot void NAME(TYPE *const begin, TYPE *const end) { \
- NAME##_stack stack[sizeof(unsigned) * CHAR_BIT], *top = stack; \
- \
- TYPE *hi = end - 1; \
- TYPE *lo = begin; \
- while (true) { \
- const ptrdiff_t len = hi - lo; \
- if (len < 16) { \
- SORT_INNER(TYPE, CMP, lo, hi + 1, len + 1); \
- if (unlikely(top == stack)) \
- break; \
- SORT_POP(lo, hi); \
- continue; \
- } \
- \
- TYPE *mid = lo + (len >> 1); \
- SORT_CMP_SWAP(TYPE, CMP, *lo, *mid); \
- SORT_CMP_SWAP(TYPE, CMP, *mid, *hi); \
- SORT_CMP_SWAP(TYPE, CMP, *lo, *mid); \
- \
- TYPE *right = hi - 1; \
- TYPE *left = lo + 1; \
- while (1) { \
- while (CMP(*left, *mid)) \
- ++left; \
- while (CMP(*mid, *right)) \
- --right; \
- if (unlikely(left > right)) { \
- if (EXPECT_LOW_CARDINALITY_OR_PRESORTED) { \
- if (NAME##_is_sorted(lo, right)) \
- lo = right + 1; \
- if (NAME##_is_sorted(left, hi)) \
- hi = left; \
- } \
- break; \
- } \
- SORT_SWAP(TYPE, *left, *right); \
- mid = (mid == left) ? right : (mid == right) ? left : mid; \
- ++left; \
- --right; \
- } \
- \
- if (right - lo > hi - left) { \
- SORT_PUSH(lo, right); \
- lo = left; \
- } else { \
- SORT_PUSH(left, hi); \
- hi = right; \
- } \
- } \
- \
- if (mdbx_audit_enabled()) { \
- for (TYPE *scan = begin + 1; scan < end; ++scan) \
- assert(CMP(scan[-1], scan[0])); \
- } \
- }
-
-/*------------------------------------------------------------------------------
- * LY: radix sort for large chunks */
-
-#define RADIXSORT_IMPL(NAME, TYPE, EXTRACT_KEY, BUFFER_PREALLOCATED, END_GAP) \
- \
- __hot static bool NAME##_radixsort(TYPE *const begin, \
- const unsigned length) { \
- TYPE *tmp; \
- if (BUFFER_PREALLOCATED) { \
- tmp = begin + length + END_GAP; \
- /* memset(tmp, 0xDeadBeef, sizeof(TYPE) * length); */ \
- } else { \
- tmp = mdbx_malloc(sizeof(TYPE) * length); \
- if (unlikely(!tmp)) \
- return false; \
- } \
- \
- unsigned key_shift = 0, key_diff_mask; \
- do { \
- struct { \
- unsigned a[256], b[256]; \
- } counters; \
- memset(&counters, 0, sizeof(counters)); \
- \
- key_diff_mask = 0; \
- unsigned prev_key = EXTRACT_KEY(begin) >> key_shift; \
- TYPE *r = begin, *end = begin + length; \
- do { \
- const unsigned key = EXTRACT_KEY(r) >> key_shift; \
- counters.a[key & 255]++; \
- counters.b[(key >> 8) & 255]++; \
- key_diff_mask |= prev_key ^ key; \
- prev_key = key; \
- } while (++r != end); \
- \
- unsigned ta = 0, tb = 0; \
- for (unsigned i = 0; i < 256; ++i) { \
- const unsigned ia = counters.a[i]; \
- counters.a[i] = ta; \
- ta += ia; \
- const unsigned ib = counters.b[i]; \
- counters.b[i] = tb; \
- tb += ib; \
- } \
- \
- r = begin; \
- do { \
- const unsigned key = EXTRACT_KEY(r) >> key_shift; \
- tmp[counters.a[key & 255]++] = *r; \
- } while (++r != end); \
- \
- if (unlikely(key_diff_mask < 256)) { \
- memcpy(begin, tmp, (char *)end - (char *)begin); \
- break; \
- } \
- end = (r = tmp) + length; \
- do { \
- const unsigned key = EXTRACT_KEY(r) >> key_shift; \
- begin[counters.b[(key >> 8) & 255]++] = *r; \
- } while (++r != end); \
- \
- key_shift += 16; \
- } while (key_diff_mask >> 16); \
- \
- if (!(BUFFER_PREALLOCATED)) \
- mdbx_free(tmp); \
- return true; \
- }
-
-/*------------------------------------------------------------------------------
- * LY: Binary search */
-
-#define SEARCH_IMPL(NAME, TYPE_LIST, TYPE_ARG, CMP) \
- static __always_inline const TYPE_LIST *NAME( \
- const TYPE_LIST *first, unsigned length, const TYPE_ARG item) { \
- const TYPE_LIST *const begin = first, *const end = begin + length; \
- \
- while (length > 3) { \
- const unsigned whole = length; \
- length >>= 1; \
- const TYPE_LIST *const middle = first + length; \
- const unsigned left = whole - length - 1; \
- const bool cmp = CMP(*middle, item); \
- length = cmp ? left : length; \
- first = cmp ? middle + 1 : first; \
- } \
- \
- switch (length) { \
- case 3: \
- if (!CMP(*first, item)) \
- break; \
- ++first; \
- __fallthrough /* fall through */; \
- case 2: \
- if (!CMP(*first, item)) \
- break; \
- ++first; \
- __fallthrough /* fall through */; \
- case 1: \
- if (!CMP(*first, item)) \
- break; \
- ++first; \
- __fallthrough /* fall through */; \
- case 0: \
- break; \
- default: \
- __unreachable(); \
- } \
- \
- if (mdbx_audit_enabled()) { \
- for (const TYPE_LIST *scan = begin; scan < first; ++scan) \
- assert(CMP(*scan, item)); \
- for (const TYPE_LIST *scan = first; scan < end; ++scan) \
- assert(!CMP(*scan, item)); \
- (void)begin, (void)end; \
- } \
- \
- return first; \
- }
-
-/*----------------------------------------------------------------------------*/
-
-static __always_inline size_t pnl2bytes(size_t size) {
- assert(size > 0 && size <= MDBX_PGL_LIMIT);
-#if MDBX_PNL_PREALLOC_FOR_RADIXSORT
- size += size;
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
- STATIC_ASSERT(MDBX_ASSUME_MALLOC_OVERHEAD +
- (MDBX_PGL_LIMIT * (MDBX_PNL_PREALLOC_FOR_RADIXSORT + 1) +
- MDBX_PNL_GRANULATE + 2) *
- sizeof(pgno_t) <
- SIZE_MAX / 4 * 3);
- size_t bytes =
- ceil_powerof2(MDBX_ASSUME_MALLOC_OVERHEAD + sizeof(pgno_t) * (size + 2),
- MDBX_PNL_GRANULATE * sizeof(pgno_t)) -
- MDBX_ASSUME_MALLOC_OVERHEAD;
- return bytes;
-}
-
-static __always_inline pgno_t bytes2pnl(const size_t bytes) {
- size_t size = bytes / sizeof(pgno_t);
- assert(size > 2 && size <= MDBX_PGL_LIMIT + /* alignment gap */ 65536);
- size -= 2;
-#if MDBX_PNL_PREALLOC_FOR_RADIXSORT
- size >>= 1;
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
- return (pgno_t)size;
-}
-
-static MDBX_PNL mdbx_pnl_alloc(size_t size) {
- size_t bytes = pnl2bytes(size);
- MDBX_PNL pl = mdbx_malloc(bytes);
- if (likely(pl)) {
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
- bytes = malloc_usable_size(pl);
-#endif /* malloc_usable_size */
- pl[0] = bytes2pnl(bytes);
- assert(pl[0] >= size);
- pl[1] = 0;
- pl += 1;
- }
- return pl;
-}
-
-static void mdbx_pnl_free(MDBX_PNL pl) {
- if (likely(pl))
- mdbx_free(pl - 1);
-}
-
-/* Shrink the PNL to the default size if it has grown larger */
-static void mdbx_pnl_shrink(MDBX_PNL *ppl) {
- assert(bytes2pnl(pnl2bytes(MDBX_PNL_INITIAL)) >= MDBX_PNL_INITIAL &&
- bytes2pnl(pnl2bytes(MDBX_PNL_INITIAL)) < MDBX_PNL_INITIAL * 3 / 2);
- assert(MDBX_PNL_SIZE(*ppl) <= MDBX_PGL_LIMIT &&
- MDBX_PNL_ALLOCLEN(*ppl) >= MDBX_PNL_SIZE(*ppl));
- MDBX_PNL_SIZE(*ppl) = 0;
- if (unlikely(MDBX_PNL_ALLOCLEN(*ppl) >
- MDBX_PNL_INITIAL * 2 - MDBX_CACHELINE_SIZE / sizeof(pgno_t))) {
- size_t bytes = pnl2bytes(MDBX_PNL_INITIAL);
- MDBX_PNL pl = mdbx_realloc(*ppl - 1, bytes);
- if (likely(pl)) {
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
- bytes = malloc_usable_size(pl);
-#endif /* malloc_usable_size */
- *pl = bytes2pnl(bytes);
- *ppl = pl + 1;
- }
- }
-}
-
-/* Grow the PNL to the size growed to at least given size */
-static int mdbx_pnl_reserve(MDBX_PNL *ppl, const size_t wanna) {
- const size_t allocated = MDBX_PNL_ALLOCLEN(*ppl);
- assert(MDBX_PNL_SIZE(*ppl) <= MDBX_PGL_LIMIT &&
- MDBX_PNL_ALLOCLEN(*ppl) >= MDBX_PNL_SIZE(*ppl));
- if (likely(allocated >= wanna))
- return MDBX_SUCCESS;
-
- if (unlikely(wanna > /* paranoia */ MDBX_PGL_LIMIT)) {
- mdbx_error("PNL too long (%zu > %zu)", wanna, (size_t)MDBX_PGL_LIMIT);
- return MDBX_TXN_FULL;
- }
-
- const size_t size = (wanna + wanna - allocated < MDBX_PGL_LIMIT)
- ? wanna + wanna - allocated
- : MDBX_PGL_LIMIT;
- size_t bytes = pnl2bytes(size);
- MDBX_PNL pl = mdbx_realloc(*ppl - 1, bytes);
- if (likely(pl)) {
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
- bytes = malloc_usable_size(pl);
-#endif /* malloc_usable_size */
- *pl = bytes2pnl(bytes);
- assert(*pl >= wanna);
- *ppl = pl + 1;
- return MDBX_SUCCESS;
- }
- return MDBX_ENOMEM;
-}
-
-/* Make room for num additional elements in an PNL */
-static __always_inline int __must_check_result mdbx_pnl_need(MDBX_PNL *ppl,
- size_t num) {
- assert(MDBX_PNL_SIZE(*ppl) <= MDBX_PGL_LIMIT &&
- MDBX_PNL_ALLOCLEN(*ppl) >= MDBX_PNL_SIZE(*ppl));
- assert(num <= MDBX_PGL_LIMIT);
- const size_t wanna = MDBX_PNL_SIZE(*ppl) + num;
- return likely(MDBX_PNL_ALLOCLEN(*ppl) >= wanna)
- ? MDBX_SUCCESS
- : mdbx_pnl_reserve(ppl, wanna);
-}
-
-static __always_inline void mdbx_pnl_xappend(MDBX_PNL pl, pgno_t pgno) {
- assert(MDBX_PNL_SIZE(pl) < MDBX_PNL_ALLOCLEN(pl));
- if (mdbx_audit_enabled()) {
- for (unsigned i = MDBX_PNL_SIZE(pl); i > 0; --i)
- assert(pgno != pl[i]);
- }
- MDBX_PNL_SIZE(pl) += 1;
- MDBX_PNL_LAST(pl) = pgno;
-}
-
-/* Append an pgno range onto an unsorted PNL */
-__always_inline static int __must_check_result
-mdbx_pnl_append_range(bool spilled, MDBX_PNL *ppl, pgno_t pgno, unsigned n) {
- assert(n > 0);
- int rc = mdbx_pnl_need(ppl, n);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- const MDBX_PNL pnl = *ppl;
-#if MDBX_PNL_ASCENDING
- unsigned w = MDBX_PNL_SIZE(pnl);
- do {
- pnl[++w] = pgno;
- pgno += spilled ? 2 : 1;
- } while (--n);
- MDBX_PNL_SIZE(pnl) = w;
-#else
- unsigned w = MDBX_PNL_SIZE(pnl) + n;
- MDBX_PNL_SIZE(pnl) = w;
- do {
- pnl[w--] = pgno;
- pgno += spilled ? 2 : 1;
- } while (--n);
-#endif
-
- return MDBX_SUCCESS;
-}
-
-/* Append an pgno range into the sorted PNL */
-static __hot int __must_check_result mdbx_pnl_insert_range(MDBX_PNL *ppl,
- pgno_t pgno,
- unsigned n) {
- assert(n > 0);
- int rc = mdbx_pnl_need(ppl, n);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- const MDBX_PNL pnl = *ppl;
- unsigned r = MDBX_PNL_SIZE(pnl), w = r + n;
- MDBX_PNL_SIZE(pnl) = w;
- while (r && MDBX_PNL_DISORDERED(pnl[r], pgno))
- pnl[w--] = pnl[r--];
-
- for (pgno_t fill = MDBX_PNL_ASCENDING ? pgno + n : pgno; w > r; --w)
- pnl[w] = MDBX_PNL_ASCENDING ? --fill : fill++;
-
- return MDBX_SUCCESS;
-}
-
-static bool mdbx_pnl_check(const MDBX_PNL pl, const size_t limit) {
- assert(limit >= MIN_PAGENO - MDBX_ENABLE_REFUND);
- if (likely(MDBX_PNL_SIZE(pl))) {
- assert(MDBX_PNL_LEAST(pl) >= MIN_PAGENO);
- assert(MDBX_PNL_MOST(pl) < limit);
- assert(MDBX_PNL_SIZE(pl) <= MDBX_PGL_LIMIT);
- if (unlikely(MDBX_PNL_SIZE(pl) > MDBX_PGL_LIMIT))
- return false;
- if (unlikely(MDBX_PNL_LEAST(pl) < MIN_PAGENO))
- return false;
- if (unlikely(MDBX_PNL_MOST(pl) >= limit))
- return false;
- if (mdbx_audit_enabled()) {
- for (const pgno_t *scan = &MDBX_PNL_LAST(pl); --scan > pl;) {
- assert(MDBX_PNL_ORDERED(scan[0], scan[1]));
- if (unlikely(!MDBX_PNL_ORDERED(scan[0], scan[1])))
- return false;
- }
- }
- }
- return true;
-}
-
-static __always_inline bool mdbx_pnl_check4assert(const MDBX_PNL pl,
- const size_t limit) {
- if (unlikely(pl == nullptr))
- return true;
- assert(MDBX_PNL_ALLOCLEN(pl) >= MDBX_PNL_SIZE(pl));
- if (unlikely(MDBX_PNL_ALLOCLEN(pl) < MDBX_PNL_SIZE(pl)))
- return false;
- return mdbx_pnl_check(pl, limit);
-}
-
-/* Merge an PNL onto an PNL. The destination PNL must be big enough */
-static void __hot mdbx_pnl_xmerge(MDBX_PNL dst, const MDBX_PNL src) {
- assert(mdbx_pnl_check4assert(dst, MAX_PAGENO + 1));
- assert(mdbx_pnl_check(src, MAX_PAGENO + 1));
- const size_t total = MDBX_PNL_SIZE(dst) + MDBX_PNL_SIZE(src);
- assert(MDBX_PNL_ALLOCLEN(dst) >= total);
- pgno_t *w = dst + total;
- pgno_t *d = dst + MDBX_PNL_SIZE(dst);
- const pgno_t *s = src + MDBX_PNL_SIZE(src);
- dst[0] = /* detent for scan below */ (MDBX_PNL_ASCENDING ? 0 : ~(pgno_t)0);
- while (s > src) {
- while (MDBX_PNL_ORDERED(*s, *d))
- *w-- = *d--;
- *w-- = *s--;
- }
- MDBX_PNL_SIZE(dst) = (pgno_t)total;
- assert(mdbx_pnl_check4assert(dst, MAX_PAGENO + 1));
-}
-
-static void mdbx_spill_remove(MDBX_txn *txn, unsigned idx, unsigned npages) {
- mdbx_tassert(txn, idx > 0 && idx <= MDBX_PNL_SIZE(txn->tw.spill_pages) &&
- txn->tw.spill_least_removed > 0);
- txn->tw.spill_least_removed =
- (idx < txn->tw.spill_least_removed) ? idx : txn->tw.spill_least_removed;
- txn->tw.spill_pages[idx] |= 1;
- MDBX_PNL_SIZE(txn->tw.spill_pages) -=
- (idx == MDBX_PNL_SIZE(txn->tw.spill_pages));
-
- while (unlikely(npages > 1)) {
- const pgno_t pgno = (txn->tw.spill_pages[idx] >> 1) + 1;
- if (MDBX_PNL_ASCENDING) {
- if (++idx > MDBX_PNL_SIZE(txn->tw.spill_pages) ||
- (txn->tw.spill_pages[idx] >> 1) != pgno)
- return;
- } else {
- if (--idx < 1 || (txn->tw.spill_pages[idx] >> 1) != pgno)
- return;
- txn->tw.spill_least_removed = (idx < txn->tw.spill_least_removed)
- ? idx
- : txn->tw.spill_least_removed;
- }
- txn->tw.spill_pages[idx] |= 1;
- MDBX_PNL_SIZE(txn->tw.spill_pages) -=
- (idx == MDBX_PNL_SIZE(txn->tw.spill_pages));
- --npages;
- }
-}
-
-static MDBX_PNL mdbx_spill_purge(MDBX_txn *txn) {
- mdbx_tassert(txn, txn->tw.spill_least_removed > 0);
- const MDBX_PNL sl = txn->tw.spill_pages;
- if (txn->tw.spill_least_removed != INT_MAX) {
- unsigned len = MDBX_PNL_SIZE(sl), r, w;
- for (w = r = txn->tw.spill_least_removed; r <= len; ++r) {
- sl[w] = sl[r];
- w += 1 - (sl[r] & 1);
- }
- for (size_t i = 1; i < w; ++i)
- mdbx_tassert(txn, (sl[i] & 1) == 0);
- MDBX_PNL_SIZE(sl) = w - 1;
- txn->tw.spill_least_removed = INT_MAX;
- } else {
- for (size_t i = 1; i <= MDBX_PNL_SIZE(sl); ++i)
- mdbx_tassert(txn, (sl[i] & 1) == 0);
- }
- return sl;
-}
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_EXTRACT_KEY(ptr) (*(ptr))
-#else
-#define MDBX_PNL_EXTRACT_KEY(ptr) (P_INVALID - *(ptr))
-#endif
-RADIXSORT_IMPL(pgno, pgno_t, MDBX_PNL_EXTRACT_KEY,
- MDBX_PNL_PREALLOC_FOR_RADIXSORT, 0)
-
-SORT_IMPL(pgno_sort, false, pgno_t, MDBX_PNL_ORDERED)
-
-static __hot void mdbx_pnl_sort_nochk(MDBX_PNL pnl) {
- if (likely(MDBX_PNL_SIZE(pnl) < MDBX_RADIXSORT_THRESHOLD) ||
- unlikely(!pgno_radixsort(&MDBX_PNL_FIRST(pnl), MDBX_PNL_SIZE(pnl))))
- pgno_sort(MDBX_PNL_BEGIN(pnl), MDBX_PNL_END(pnl));
-}
-
-static __inline void mdbx_pnl_sort(MDBX_PNL pnl, size_t limit4check) {
- mdbx_pnl_sort_nochk(pnl);
- assert(mdbx_pnl_check(pnl, limit4check));
- (void)limit4check;
-}
-
-/* Search for an pgno in an PNL.
- * Returns The index of the first item greater than or equal to pgno. */
-SEARCH_IMPL(pgno_bsearch, pgno_t, pgno_t, MDBX_PNL_ORDERED)
-
-static __hot unsigned mdbx_pnl_search_nochk(const MDBX_PNL pnl, pgno_t pgno) {
- const pgno_t *begin = MDBX_PNL_BEGIN(pnl);
- const pgno_t *it = pgno_bsearch(begin, MDBX_PNL_SIZE(pnl), pgno);
- const pgno_t *end = begin + MDBX_PNL_SIZE(pnl);
- assert(it >= begin && it <= end);
- if (it != begin)
- assert(MDBX_PNL_ORDERED(it[-1], pgno));
- if (it != end)
- assert(!MDBX_PNL_ORDERED(it[0], pgno));
- return (unsigned)(it - begin + 1);
-}
-
-static __inline unsigned mdbx_pnl_search(const MDBX_PNL pnl, pgno_t pgno,
- size_t limit) {
- assert(mdbx_pnl_check4assert(pnl, limit));
- assert(pgno < limit);
- (void)limit;
- return mdbx_pnl_search_nochk(pnl, pgno);
-}
-
-static __inline unsigned mdbx_search_spilled(const MDBX_txn *txn, pgno_t pgno) {
- const MDBX_PNL pnl = txn->tw.spill_pages;
- if (likely(!pnl))
- return 0;
- pgno <<= 1;
- unsigned n = mdbx_pnl_search(pnl, pgno, (size_t)(MAX_PAGENO + 1) << 1);
- return (n <= MDBX_PNL_SIZE(pnl) && pnl[n] == pgno) ? n : 0;
-}
-
-static __inline bool mdbx_intersect_spilled(const MDBX_txn *txn, pgno_t pgno,
- unsigned npages) {
- const MDBX_PNL pnl = txn->tw.spill_pages;
- if (likely(!pnl))
- return false;
- const unsigned len = MDBX_PNL_SIZE(pnl);
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) {
- mdbx_debug_extra("PNL len %u [", len);
- for (unsigned i = 1; i <= len; ++i)
- mdbx_debug_extra_print(" %li", (pnl[i] & 1) ? -(long)(pnl[i] >> 1)
- : (long)(pnl[i] >> 1));
- mdbx_debug_extra_print("%s\n", "]");
- }
- const pgno_t spilled_range_begin = pgno << 1;
- const pgno_t spilled_range_last = ((pgno + npages) << 1) - 1;
-#if MDBX_PNL_ASCENDING
- const unsigned n =
- mdbx_pnl_search(pnl, spilled_range_begin, (size_t)(MAX_PAGENO + 1) << 1);
- assert(n && (n == MDBX_PNL_SIZE(pnl) + 1 || spilled_range_begin <= pnl[n]));
- const bool rc = n <= MDBX_PNL_SIZE(pnl) && pnl[n] <= spilled_range_last;
-#else
- const unsigned n =
- mdbx_pnl_search(pnl, spilled_range_last, (size_t)(MAX_PAGENO + 1) << 1);
- assert(n && (n == MDBX_PNL_SIZE(pnl) + 1 || spilled_range_last >= pnl[n]));
- const bool rc = n <= MDBX_PNL_SIZE(pnl) && pnl[n] >= spilled_range_begin;
-#endif
- if (mdbx_assert_enabled()) {
- bool check = false;
- for (unsigned i = 0; i < npages; ++i)
- check |= mdbx_search_spilled(txn, pgno + i) != 0;
- assert(check == rc);
- }
- return rc;
-}
-
-/*----------------------------------------------------------------------------*/
-
-static __always_inline size_t txl2bytes(const size_t size) {
- assert(size > 0 && size <= MDBX_TXL_MAX * 2);
- size_t bytes =
- ceil_powerof2(MDBX_ASSUME_MALLOC_OVERHEAD + sizeof(txnid_t) * (size + 2),
- MDBX_TXL_GRANULATE * sizeof(txnid_t)) -
- MDBX_ASSUME_MALLOC_OVERHEAD;
- return bytes;
-}
-
-static __always_inline size_t bytes2txl(const size_t bytes) {
- size_t size = bytes / sizeof(txnid_t);
- assert(size > 2 && size <= MDBX_TXL_MAX * 2);
- return size - 2;
-}
-
-static MDBX_TXL mdbx_txl_alloc(void) {
- size_t bytes = txl2bytes(MDBX_TXL_INITIAL);
- MDBX_TXL tl = mdbx_malloc(bytes);
- if (likely(tl)) {
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
- bytes = malloc_usable_size(tl);
-#endif /* malloc_usable_size */
- tl[0] = bytes2txl(bytes);
- assert(tl[0] >= MDBX_TXL_INITIAL);
- tl[1] = 0;
- tl += 1;
- }
- return tl;
-}
-
-static void mdbx_txl_free(MDBX_TXL tl) {
- if (likely(tl))
- mdbx_free(tl - 1);
-}
-
-static int mdbx_txl_reserve(MDBX_TXL *ptl, const size_t wanna) {
- const size_t allocated = (size_t)MDBX_PNL_ALLOCLEN(*ptl);
- assert(MDBX_PNL_SIZE(*ptl) <= MDBX_TXL_MAX &&
- MDBX_PNL_ALLOCLEN(*ptl) >= MDBX_PNL_SIZE(*ptl));
- if (likely(allocated >= wanna))
- return MDBX_SUCCESS;
-
- if (unlikely(wanna > /* paranoia */ MDBX_TXL_MAX)) {
- mdbx_error("TXL too long (%zu > %zu)", wanna, (size_t)MDBX_TXL_MAX);
- return MDBX_TXN_FULL;
- }
-
- const size_t size = (wanna + wanna - allocated < MDBX_TXL_MAX)
- ? wanna + wanna - allocated
- : MDBX_TXL_MAX;
- size_t bytes = txl2bytes(size);
- MDBX_TXL tl = mdbx_realloc(*ptl - 1, bytes);
- if (likely(tl)) {
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
- bytes = malloc_usable_size(tl);
-#endif /* malloc_usable_size */
- *tl = bytes2txl(bytes);
- assert(*tl >= wanna);
- *ptl = tl + 1;
- return MDBX_SUCCESS;
- }
- return MDBX_ENOMEM;
-}
-
-static __always_inline int __must_check_result mdbx_txl_need(MDBX_TXL *ptl,
- size_t num) {
- assert(MDBX_PNL_SIZE(*ptl) <= MDBX_TXL_MAX &&
- MDBX_PNL_ALLOCLEN(*ptl) >= MDBX_PNL_SIZE(*ptl));
- assert(num <= MDBX_PGL_LIMIT);
- const size_t wanna = (size_t)MDBX_PNL_SIZE(*ptl) + num;
- return likely(MDBX_PNL_ALLOCLEN(*ptl) >= wanna)
- ? MDBX_SUCCESS
- : mdbx_txl_reserve(ptl, wanna);
-}
-
-static __always_inline void mdbx_txl_xappend(MDBX_TXL tl, txnid_t id) {
- assert(MDBX_PNL_SIZE(tl) < MDBX_PNL_ALLOCLEN(tl));
- MDBX_PNL_SIZE(tl) += 1;
- MDBX_PNL_LAST(tl) = id;
-}
-
-#define TXNID_SORT_CMP(first, last) ((first) > (last))
-SORT_IMPL(txnid_sort, false, txnid_t, TXNID_SORT_CMP)
-static void mdbx_txl_sort(MDBX_TXL tl) {
- txnid_sort(MDBX_PNL_BEGIN(tl), MDBX_PNL_END(tl));
-}
-
-static int __must_check_result mdbx_txl_append(MDBX_TXL *ptl, txnid_t id) {
- if (unlikely(MDBX_PNL_SIZE(*ptl) == MDBX_PNL_ALLOCLEN(*ptl))) {
- int rc = mdbx_txl_need(ptl, MDBX_TXL_GRANULATE);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
- mdbx_txl_xappend(*ptl, id);
- return MDBX_SUCCESS;
-}
-
-/*----------------------------------------------------------------------------*/
-
-#define MDBX_DPL_UNSORTED_BACKLOG 16
-#define MDBX_DPL_GAP_FOR_MERGESORT MDBX_DPL_UNSORTED_BACKLOG
-#define MDBX_DPL_GAP_FOR_EDGING 2
-#define MDBX_DPL_RESERVE_GAP \
- (MDBX_DPL_GAP_FOR_MERGESORT + MDBX_DPL_GAP_FOR_EDGING)
-
-static __always_inline size_t dpl2bytes(ptrdiff_t size) {
- assert(size > CURSOR_STACK && (size_t)size <= MDBX_PGL_LIMIT);
-#if MDBX_DPL_PREALLOC_FOR_RADIXSORT
- size += size;
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
- STATIC_ASSERT(MDBX_ASSUME_MALLOC_OVERHEAD + sizeof(MDBX_dpl) +
- (MDBX_PGL_LIMIT * (MDBX_DPL_PREALLOC_FOR_RADIXSORT + 1) +
- MDBX_DPL_RESERVE_GAP) *
- sizeof(MDBX_dp) +
- MDBX_PNL_GRANULATE * sizeof(void *) * 2 <
- SIZE_MAX / 4 * 3);
- size_t bytes =
- ceil_powerof2(MDBX_ASSUME_MALLOC_OVERHEAD + sizeof(MDBX_dpl) +
- ((size_t)size + MDBX_DPL_RESERVE_GAP) * sizeof(MDBX_dp),
- MDBX_PNL_GRANULATE * sizeof(void *) * 2) -
- MDBX_ASSUME_MALLOC_OVERHEAD;
- return bytes;
-}
-
-static __always_inline unsigned bytes2dpl(const ptrdiff_t bytes) {
- size_t size = (bytes - sizeof(MDBX_dpl)) / sizeof(MDBX_dp);
- assert(size > CURSOR_STACK + MDBX_DPL_RESERVE_GAP &&
- size <= MDBX_PGL_LIMIT + MDBX_PNL_GRANULATE);
- size -= MDBX_DPL_RESERVE_GAP;
-#if MDBX_DPL_PREALLOC_FOR_RADIXSORT
- size >>= 1;
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
- return (unsigned)size;
-}
-
-static __always_inline unsigned dpl_setlen(MDBX_dpl *dl, unsigned len) {
- static const MDBX_page dpl_stub_pageE = {
- {0}, 0, P_BAD, {0}, /* pgno */ ~(pgno_t)0};
- assert(dpl_stub_pageE.mp_flags == P_BAD &&
- dpl_stub_pageE.mp_pgno == P_INVALID);
- dl->length = len;
- dl->items[len + 1].ptr = (MDBX_page *)&dpl_stub_pageE;
- dl->items[len + 1].pgno = P_INVALID;
- dl->items[len + 1].extra = 0;
- return len;
-}
-
-static __always_inline void dpl_clear(MDBX_dpl *dl) {
- static const MDBX_page dpl_stub_pageB = {{0}, 0, P_BAD, {0}, /* pgno */ 0};
- assert(dpl_stub_pageB.mp_flags == P_BAD && dpl_stub_pageB.mp_pgno == 0);
- dl->sorted = dpl_setlen(dl, 0);
- dl->items[0].ptr = (MDBX_page *)&dpl_stub_pageB;
- dl->items[0].pgno = 0;
- dl->items[0].extra = 0;
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
-}
-
-static void mdbx_dpl_free(MDBX_txn *txn) {
- if (likely(txn->tw.dirtylist)) {
- mdbx_free(txn->tw.dirtylist);
- txn->tw.dirtylist = NULL;
- }
-}
-
-static MDBX_dpl *mdbx_dpl_reserve(MDBX_txn *txn, size_t size) {
- size_t bytes = dpl2bytes((size < MDBX_PGL_LIMIT) ? size : MDBX_PGL_LIMIT);
- MDBX_dpl *const dl = mdbx_realloc(txn->tw.dirtylist, bytes);
- if (likely(dl)) {
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
- bytes = malloc_usable_size(dl);
-#endif /* malloc_usable_size */
- dl->detent = bytes2dpl(bytes);
- mdbx_tassert(txn, txn->tw.dirtylist == NULL || dl->length <= dl->detent);
- txn->tw.dirtylist = dl;
- }
- return dl;
-}
-
-static int mdbx_dpl_alloc(MDBX_txn *txn) {
- mdbx_tassert(txn, (txn->mt_flags & MDBX_TXN_RDONLY) == 0);
- const int wanna = (txn->mt_env->me_options.dp_initial < txn->mt_geo.upper)
- ? txn->mt_env->me_options.dp_initial
- : txn->mt_geo.upper;
- if (txn->tw.dirtylist) {
- dpl_clear(txn->tw.dirtylist);
- const int realloc_threshold = 64;
- if (likely(
- !((int)(txn->tw.dirtylist->detent - wanna) > realloc_threshold ||
- (int)(txn->tw.dirtylist->detent - wanna) < -realloc_threshold)))
- return MDBX_SUCCESS;
- }
- if (unlikely(!mdbx_dpl_reserve(txn, wanna)))
- return MDBX_ENOMEM;
- dpl_clear(txn->tw.dirtylist);
- return MDBX_SUCCESS;
-}
-
-#define MDBX_DPL_EXTRACT_KEY(ptr) ((ptr)->pgno)
-RADIXSORT_IMPL(dpl, MDBX_dp, MDBX_DPL_EXTRACT_KEY,
- MDBX_DPL_PREALLOC_FOR_RADIXSORT, 1)
-
-#define DP_SORT_CMP(first, last) ((first).pgno < (last).pgno)
-SORT_IMPL(dp_sort, false, MDBX_dp, DP_SORT_CMP)
-
-__hot __noinline static MDBX_dpl *mdbx_dpl_sort_slowpath(const MDBX_txn *txn) {
- MDBX_dpl *dl = txn->tw.dirtylist;
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- const unsigned unsorted = dl->length - dl->sorted;
- if (likely(unsorted < MDBX_RADIXSORT_THRESHOLD) ||
- unlikely(!dpl_radixsort(dl->items + 1, dl->length))) {
- if (dl->sorted > unsorted / 4 + 4 &&
- (MDBX_DPL_PREALLOC_FOR_RADIXSORT ||
- dl->length + unsorted < dl->detent + MDBX_DPL_GAP_FOR_MERGESORT)) {
- MDBX_dp *const sorted_begin = dl->items + 1;
- MDBX_dp *const sorted_end = sorted_begin + dl->sorted;
- MDBX_dp *const end =
- dl->items + (MDBX_DPL_PREALLOC_FOR_RADIXSORT
- ? dl->length + dl->length + 1
- : dl->detent + MDBX_DPL_RESERVE_GAP);
- MDBX_dp *const tmp = end - unsorted;
- assert(dl->items + dl->length + 1 < tmp);
- /* copy unsorted to the end of allocated space and sort it */
- memcpy(tmp, sorted_end, unsorted * sizeof(MDBX_dp));
- dp_sort(tmp, tmp + unsorted);
- /* merge two parts from end to begin */
- MDBX_dp *w = dl->items + dl->length;
- MDBX_dp *l = dl->items + dl->sorted;
- MDBX_dp *r = end - 1;
- do {
- const bool cmp = l->pgno > r->pgno;
- *w = cmp ? *l : *r;
- l -= cmp;
- r += cmp - 1;
- } while (likely(--w > l));
- assert(r == tmp - 1);
- assert(dl->items[0].pgno == 0 &&
- dl->items[dl->length + 1].pgno == P_INVALID);
- if (mdbx_assert_enabled())
- for (unsigned i = 0; i <= dl->length; ++i)
- assert(dl->items[i].pgno < dl->items[i + 1].pgno);
- } else {
- dp_sort(dl->items + 1, dl->items + dl->length + 1);
- assert(dl->items[0].pgno == 0 &&
- dl->items[dl->length + 1].pgno == P_INVALID);
- }
- } else {
- assert(dl->items[0].pgno == 0 &&
- dl->items[dl->length + 1].pgno == P_INVALID);
- }
- dl->sorted = dl->length;
- return dl;
-}
-
-static __always_inline MDBX_dpl *mdbx_dpl_sort(const MDBX_txn *txn) {
- MDBX_dpl *dl = txn->tw.dirtylist;
- assert(dl->length <= MDBX_PGL_LIMIT);
- assert(dl->sorted <= dl->length);
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- return likely(dl->sorted == dl->length) ? dl : mdbx_dpl_sort_slowpath(txn);
-}
-
-/* Returns the index of the first dirty-page whose pgno
- * member is greater than or equal to id. */
-#define DP_SEARCH_CMP(dp, id) ((dp).pgno < (id))
-SEARCH_IMPL(dp_bsearch, MDBX_dp, pgno_t, DP_SEARCH_CMP)
-
-static unsigned __hot mdbx_dpl_search(const MDBX_txn *txn, pgno_t pgno) {
- MDBX_dpl *dl = txn->tw.dirtylist;
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- if (mdbx_audit_enabled()) {
- for (const MDBX_dp *ptr = dl->items + dl->sorted; --ptr > dl->items;) {
- assert(ptr[0].pgno < ptr[1].pgno);
- assert(ptr[0].pgno >= NUM_METAS);
- }
- }
-
- switch (dl->length - dl->sorted) {
- default:
- /* sort a whole */
- mdbx_dpl_sort_slowpath(txn);
- break;
- case 0:
- /* whole sorted cases */
- break;
-
-#define LINEAR_SEARCH_CASE(N) \
- case N: \
- if (dl->items[dl->length - N + 1].pgno == pgno) \
- return dl->length - N + 1; \
- __fallthrough
-
- /* try linear search until the threshold */
- LINEAR_SEARCH_CASE(16); /* fall through */
- LINEAR_SEARCH_CASE(15); /* fall through */
- LINEAR_SEARCH_CASE(14); /* fall through */
- LINEAR_SEARCH_CASE(13); /* fall through */
- LINEAR_SEARCH_CASE(12); /* fall through */
- LINEAR_SEARCH_CASE(11); /* fall through */
- LINEAR_SEARCH_CASE(10); /* fall through */
- LINEAR_SEARCH_CASE(9); /* fall through */
- LINEAR_SEARCH_CASE(8); /* fall through */
- LINEAR_SEARCH_CASE(7); /* fall through */
- LINEAR_SEARCH_CASE(6); /* fall through */
- LINEAR_SEARCH_CASE(5); /* fall through */
- LINEAR_SEARCH_CASE(4); /* fall through */
- LINEAR_SEARCH_CASE(3); /* fall through */
- LINEAR_SEARCH_CASE(2); /* fall through */
- case 1:
- if (dl->items[dl->length].pgno == pgno)
- return dl->length;
- /* continue bsearch on the sorted part */
- break;
- }
- return (unsigned)(dp_bsearch(dl->items + 1, dl->sorted, pgno) - dl->items);
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __inline unsigned
-dpl_npages(const MDBX_dpl *dl, unsigned i) {
- assert(0 <= (int)i && i <= dl->length);
- unsigned n = likely(!dl->items[i].multi) ? 1 : dl->items[i].ptr->mp_pages;
- assert(n == (IS_OVERFLOW(dl->items[i].ptr) ? dl->items[i].ptr->mp_pages : 1));
- return n;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static __inline unsigned
-dpl_endpgno(const MDBX_dpl *dl, unsigned i) {
- return dpl_npages(dl, i) + dl->items[i].pgno;
-}
-
-static __inline bool mdbx_dpl_intersect(const MDBX_txn *txn, pgno_t pgno,
- unsigned npages) {
- MDBX_dpl *dl = txn->tw.dirtylist;
- assert(dl->sorted == dl->length);
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- unsigned const n = mdbx_dpl_search(txn, pgno);
- assert(n >= 1 && n <= dl->length + 1);
- assert(pgno <= dl->items[n].pgno);
- assert(pgno > dl->items[n - 1].pgno);
- const bool rc =
- /* intersection with founded */ pgno + npages > dl->items[n].pgno ||
- /* intersection with prev */ dpl_endpgno(dl, n - 1) > pgno;
- if (mdbx_assert_enabled()) {
- bool check = false;
- for (unsigned i = 1; i <= dl->length; ++i) {
- const MDBX_page *const dp = dl->items[i].ptr;
- if (!(dp->mp_pgno /* begin */ >= /* end */ pgno + npages ||
- dpl_endpgno(dl, i) /* end */ <= /* begin */ pgno))
- check |= true;
- }
- assert(check == rc);
- }
- return rc;
-}
-
-static __always_inline unsigned mdbx_dpl_exist(MDBX_txn *txn, pgno_t pgno) {
- MDBX_dpl *dl = txn->tw.dirtylist;
- unsigned i = mdbx_dpl_search(txn, pgno);
- assert((int)i > 0);
- return (dl->items[i].pgno == pgno) ? i : 0;
-}
-
-MDBX_MAYBE_UNUSED static const MDBX_page *debug_dpl_find(const MDBX_txn *txn,
- const pgno_t pgno) {
- const MDBX_dpl *dl = txn->tw.dirtylist;
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- for (unsigned i = dl->length; i > dl->sorted; --i)
- if (dl->items[i].pgno == pgno)
- return dl->items[i].ptr;
-
- if (dl->sorted) {
- const unsigned i =
- (unsigned)(dp_bsearch(dl->items + 1, dl->sorted, pgno) - dl->items);
- if (dl->items[i].pgno == pgno)
- return dl->items[i].ptr;
- }
- return nullptr;
-}
-
-static void mdbx_dpl_remove(const MDBX_txn *txn, unsigned i) {
- MDBX_dpl *dl = txn->tw.dirtylist;
- assert((int)i > 0 && i <= dl->length);
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- dl->sorted -= dl->sorted >= i;
- dl->length -= 1;
- memmove(dl->items + i, dl->items + i + 1,
- (dl->length - i + 2) * sizeof(dl->items[0]));
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
-}
-
-static __always_inline int __must_check_result
-mdbx_dpl_append(MDBX_txn *txn, pgno_t pgno, MDBX_page *page, unsigned npages) {
- MDBX_dpl *dl = txn->tw.dirtylist;
- assert(dl->length <= MDBX_PGL_LIMIT + MDBX_PNL_GRANULATE);
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- if (mdbx_audit_enabled()) {
- for (unsigned i = dl->length; i > 0; --i) {
- assert(dl->items[i].pgno != pgno);
- if (unlikely(dl->items[i].pgno == pgno)) {
- mdbx_error("Page %u already exist in the DPL at %u", pgno, i);
- return MDBX_PROBLEM;
- }
- }
- }
-
- const unsigned length = dl->length + 1;
- const unsigned sorted =
- (dl->sorted == dl->length && dl->items[dl->length].pgno < pgno)
- ? length
- : dl->sorted;
-
- if (unlikely(dl->length == dl->detent)) {
- if (unlikely(dl->detent >= MDBX_PGL_LIMIT)) {
- mdbx_error("DPL is full (MDBX_PGL_LIMIT %zu)", MDBX_PGL_LIMIT);
- return MDBX_TXN_FULL;
- }
- const size_t size = (dl->detent < MDBX_PNL_INITIAL * 42)
- ? dl->detent + dl->detent
- : dl->detent + dl->detent / 2;
- dl = mdbx_dpl_reserve(txn, size);
- if (unlikely(!dl))
- return MDBX_ENOMEM;
- mdbx_tassert(txn, dl->length < dl->detent);
- }
-
- /* copy the stub beyond the end */
- dl->items[length + 1] = dl->items[length];
- /* append page */
- dl->items[length].ptr = page;
- dl->items[length].pgno = pgno;
- dl->items[length].multi = npages > 1;
- dl->items[length].lru = txn->tw.dirtylru++;
- dl->length = length;
- dl->sorted = sorted;
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- return MDBX_SUCCESS;
-}
-
-static __inline uint32_t mdbx_dpl_age(const MDBX_txn *txn, unsigned i) {
- const MDBX_dpl *dl = txn->tw.dirtylist;
- assert((int)i > 0 && i <= dl->length);
- /* overflow could be here */
- return (txn->tw.dirtylru - dl->items[i].lru) & UINT32_C(0x7fffFFFF);
-}
-
-/*----------------------------------------------------------------------------*/
-
-uint8_t mdbx_runtime_flags = MDBX_RUNTIME_FLAGS_INIT;
-uint8_t mdbx_loglevel = MDBX_LOG_FATAL;
-MDBX_debug_func *mdbx_debug_logger;
-
-static __must_check_result __inline int mdbx_page_retire(MDBX_cursor *mc,
- MDBX_page *mp);
-
-static int __must_check_result mdbx_page_dirty(MDBX_txn *txn, MDBX_page *mp,
- unsigned npages);
-struct page_result {
- MDBX_page *page;
- int err;
-};
-
-static struct page_result mdbx_page_alloc(MDBX_cursor *mc, const pgno_t num,
- int flags);
-static txnid_t mdbx_kick_longlived_readers(MDBX_env *env,
- const txnid_t laggard);
-
-static struct page_result mdbx_page_new(MDBX_cursor *mc, const unsigned flags,
- const unsigned npages);
-static int mdbx_page_touch(MDBX_cursor *mc);
-static int mdbx_cursor_touch(MDBX_cursor *mc);
-static int mdbx_touch_dbi(MDBX_cursor *mc);
-
-#define MDBX_END_NAMES \
- { \
- "committed", "empty-commit", "abort", "reset", "reset-tmp", "fail-begin", \
- "fail-beginchild" \
- }
-enum {
- /* mdbx_txn_end operation number, for logging */
- MDBX_END_COMMITTED,
- MDBX_END_PURE_COMMIT,
- MDBX_END_ABORT,
- MDBX_END_RESET,
- MDBX_END_RESET_TMP,
- MDBX_END_FAIL_BEGIN,
- MDBX_END_FAIL_BEGINCHILD
-};
-#define MDBX_END_OPMASK 0x0F /* mask for mdbx_txn_end() operation number */
-#define MDBX_END_UPDATE 0x10 /* update env state (DBIs) */
-#define MDBX_END_FREE 0x20 /* free txn unless it is MDBX_env.me_txn0 */
-#define MDBX_END_EOTDONE 0x40 /* txn's cursors already closed */
-#define MDBX_END_SLOT 0x80 /* release any reader slot if MDBX_NOTLS */
-static int mdbx_txn_end(MDBX_txn *txn, const unsigned mode);
-
-__hot static struct page_result __must_check_result
-mdbx_page_get_ex(MDBX_cursor *const mc, const pgno_t pgno, txnid_t front);
-static __always_inline int __must_check_result mdbx_page_get(MDBX_cursor *mc,
- pgno_t pgno,
- MDBX_page **mp,
- txnid_t front) {
-
- struct page_result ret = mdbx_page_get_ex(mc, pgno, front);
- *mp = ret.page;
- return ret.err;
-}
-
-static int __must_check_result mdbx_page_search_root(MDBX_cursor *mc,
- const MDBX_val *key,
- int flags);
-
-#define MDBX_PS_MODIFY 1
-#define MDBX_PS_ROOTONLY 2
-#define MDBX_PS_FIRST 4
-#define MDBX_PS_LAST 8
-static int __must_check_result mdbx_page_search(MDBX_cursor *mc,
- const MDBX_val *key, int flags);
-static int __must_check_result mdbx_page_merge(MDBX_cursor *csrc,
- MDBX_cursor *cdst);
-
-#define MDBX_SPLIT_REPLACE MDBX_APPENDDUP /* newkey is not new */
-static int __must_check_result mdbx_page_split(MDBX_cursor *mc,
- const MDBX_val *const newkey,
- MDBX_val *const newdata,
- pgno_t newpgno, unsigned nflags);
-
-static bool meta_checktxnid(const MDBX_env *env, const MDBX_meta *meta,
- bool report);
-static int __must_check_result mdbx_validate_meta_copy(MDBX_env *env,
- const MDBX_meta *meta,
- MDBX_meta *dest);
-static int __must_check_result mdbx_override_meta(MDBX_env *env,
- unsigned target,
- txnid_t txnid,
- const MDBX_meta *shape);
-static int __must_check_result mdbx_read_header(MDBX_env *env, MDBX_meta *meta,
- const int lck_exclusive,
- const mdbx_mode_t mode_bits);
-static int __must_check_result mdbx_sync_locked(MDBX_env *env, unsigned flags,
- MDBX_meta *const pending);
-static int mdbx_env_close0(MDBX_env *env);
-
-struct node_result {
- MDBX_node *node;
- bool exact;
-};
-
-static struct node_result mdbx_node_search(MDBX_cursor *mc,
- const MDBX_val *key);
-
-static int __must_check_result mdbx_node_add_branch(MDBX_cursor *mc,
- unsigned indx,
- const MDBX_val *key,
- pgno_t pgno);
-static int __must_check_result mdbx_node_add_leaf(MDBX_cursor *mc,
- unsigned indx,
- const MDBX_val *key,
- MDBX_val *data,
- unsigned flags);
-static int __must_check_result mdbx_node_add_leaf2(MDBX_cursor *mc,
- unsigned indx,
- const MDBX_val *key);
-
-static void mdbx_node_del(MDBX_cursor *mc, size_t ksize);
-static void mdbx_node_shrink(MDBX_page *mp, unsigned indx);
-static int __must_check_result mdbx_node_move(MDBX_cursor *csrc,
- MDBX_cursor *cdst, bool fromleft);
-static int __must_check_result mdbx_node_read(MDBX_cursor *mc,
- const MDBX_node *leaf,
- MDBX_val *data,
- const txnid_t front);
-static int __must_check_result mdbx_rebalance(MDBX_cursor *mc);
-static int __must_check_result mdbx_update_key(MDBX_cursor *mc,
- const MDBX_val *key);
-
-static void mdbx_cursor_pop(MDBX_cursor *mc);
-static int __must_check_result mdbx_cursor_push(MDBX_cursor *mc, MDBX_page *mp);
-
-static int __must_check_result mdbx_audit_ex(MDBX_txn *txn,
- unsigned retired_stored,
- bool dont_filter_gc);
-
-static int __must_check_result mdbx_page_check(MDBX_cursor *const mc,
- const MDBX_page *const mp,
- unsigned options);
-static int __must_check_result mdbx_cursor_check(MDBX_cursor *mc,
- unsigned options);
-static int __must_check_result mdbx_cursor_del0(MDBX_cursor *mc);
-static int __must_check_result mdbx_del0(MDBX_txn *txn, MDBX_dbi dbi,
- const MDBX_val *key,
- const MDBX_val *data, unsigned flags);
-#define SIBLING_LEFT 0
-#define SIBLING_RIGHT 2
-static int __must_check_result mdbx_cursor_sibling(MDBX_cursor *mc, int dir);
-static int __must_check_result mdbx_cursor_next(MDBX_cursor *mc, MDBX_val *key,
- MDBX_val *data,
- MDBX_cursor_op op);
-static int __must_check_result mdbx_cursor_prev(MDBX_cursor *mc, MDBX_val *key,
- MDBX_val *data,
- MDBX_cursor_op op);
-struct cursor_set_result {
- int err;
- bool exact;
-};
-
-static struct cursor_set_result mdbx_cursor_set(MDBX_cursor *mc, MDBX_val *key,
- MDBX_val *data,
- MDBX_cursor_op op);
-static int __must_check_result mdbx_cursor_first(MDBX_cursor *mc, MDBX_val *key,
- MDBX_val *data);
-static int __must_check_result mdbx_cursor_last(MDBX_cursor *mc, MDBX_val *key,
- MDBX_val *data);
-
-static int __must_check_result mdbx_cursor_init(MDBX_cursor *mc, MDBX_txn *txn,
- MDBX_dbi dbi);
-static int __must_check_result mdbx_xcursor_init0(MDBX_cursor *mc);
-static int __must_check_result mdbx_xcursor_init1(MDBX_cursor *mc,
- MDBX_node *node,
- const MDBX_page *mp);
-static int __must_check_result mdbx_xcursor_init2(MDBX_cursor *mc,
- MDBX_xcursor *src_mx,
- bool new_dupdata);
-static void cursor_copy(const MDBX_cursor *csrc, MDBX_cursor *cdst);
-
-static int __must_check_result mdbx_drop_tree(MDBX_cursor *mc,
- const bool may_have_subDBs);
-static int __must_check_result mdbx_fetch_sdb(MDBX_txn *txn, MDBX_dbi dbi);
-static int __must_check_result mdbx_setup_dbx(MDBX_dbx *const dbx,
- const MDBX_db *const db,
- const unsigned pagesize);
-
-static MDBX_cmp_func cmp_lexical, cmp_reverse, cmp_int_align4, cmp_int_align2,
- cmp_int_unaligned, cmp_lenfast;
-
-static __inline MDBX_cmp_func *get_default_keycmp(unsigned flags);
-static __inline MDBX_cmp_func *get_default_datacmp(unsigned flags);
-
-__cold const char *mdbx_liberr2str(int errnum) {
- /* Table of descriptions for MDBX errors */
- static const char *const tbl[] = {
- "MDBX_KEYEXIST: Key/data pair already exists",
- "MDBX_NOTFOUND: No matching key/data pair found",
- "MDBX_PAGE_NOTFOUND: Requested page not found",
- "MDBX_CORRUPTED: Database is corrupted",
- "MDBX_PANIC: Environment had fatal error",
- "MDBX_VERSION_MISMATCH: DB version mismatch libmdbx",
- "MDBX_INVALID: File is not an MDBX file",
- "MDBX_MAP_FULL: Environment mapsize limit reached",
- "MDBX_DBS_FULL: Too many DBI-handles (maxdbs reached)",
- "MDBX_READERS_FULL: Too many readers (maxreaders reached)",
- NULL /* MDBX_TLS_FULL (-30789): unused in MDBX */,
- "MDBX_TXN_FULL: Transaction has too many dirty pages,"
- " i.e transaction is too big",
- "MDBX_CURSOR_FULL: Cursor stack limit reachedn - this usually indicates"
- " corruption, i.e branch-pages loop",
- "MDBX_PAGE_FULL: Internal error - Page has no more space",
- "MDBX_UNABLE_EXTEND_MAPSIZE: Database engine was unable to extend"
- " mapping, e.g. since address space is unavailable or busy,"
- " or Operation system not supported such operations",
- "MDBX_INCOMPATIBLE: Environment or database is not compatible"
- " with the requested operation or the specified flags",
- "MDBX_BAD_RSLOT: Invalid reuse of reader locktable slot,"
- " e.g. read-transaction already run for current thread",
- "MDBX_BAD_TXN: Transaction is not valid for requested operation,"
- " e.g. had errored and be must aborted, has a child, or is invalid",
- "MDBX_BAD_VALSIZE: Invalid size or alignment of key or data"
- " for target database, either invalid subDB name",
- "MDBX_BAD_DBI: The specified DBI-handle is invalid"
- " or changed by another thread/transaction",
- "MDBX_PROBLEM: Unexpected internal error, transaction should be aborted",
- "MDBX_BUSY: Another write transaction is running,"
- " or environment is already used while opening with MDBX_EXCLUSIVE flag",
- };
-
- if (errnum >= MDBX_KEYEXIST && errnum <= MDBX_BUSY) {
- int i = errnum - MDBX_KEYEXIST;
- return tbl[i];
- }
-
- switch (errnum) {
- case MDBX_SUCCESS:
- return "MDBX_SUCCESS: Successful";
- case MDBX_EMULTIVAL:
- return "MDBX_EMULTIVAL: The specified key has"
- " more than one associated value";
- case MDBX_EBADSIGN:
- return "MDBX_EBADSIGN: Wrong signature of a runtime object(s),"
- " e.g. memory corruption or double-free";
- case MDBX_WANNA_RECOVERY:
- return "MDBX_WANNA_RECOVERY: Database should be recovered,"
- " but this could NOT be done automatically for now"
- " since it opened in read-only mode";
- case MDBX_EKEYMISMATCH:
- return "MDBX_EKEYMISMATCH: The given key value is mismatched to the"
- " current cursor position";
- case MDBX_TOO_LARGE:
- return "MDBX_TOO_LARGE: Database is too large for current system,"
- " e.g. could NOT be mapped into RAM";
- case MDBX_THREAD_MISMATCH:
- return "MDBX_THREAD_MISMATCH: A thread has attempted to use a not"
- " owned object, e.g. a transaction that started by another thread";
- case MDBX_TXN_OVERLAPPING:
- return "MDBX_TXN_OVERLAPPING: Overlapping read and write transactions for"
- " the current thread";
- default:
- return NULL;
- }
-}
-
-__cold const char *mdbx_strerror_r(int errnum, char *buf, size_t buflen) {
- const char *msg = mdbx_liberr2str(errnum);
- if (!msg && buflen > 0 && buflen < INT_MAX) {
-#if defined(_WIN32) || defined(_WIN64)
- const DWORD size = FormatMessageA(
- FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS, NULL,
- errnum, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), buf, (DWORD)buflen,
- NULL);
- return size ? buf : "FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM) failed";
-#elif defined(_GNU_SOURCE) && defined(__GLIBC__)
- /* GNU-specific */
- if (errnum > 0)
- msg = strerror_r(errnum, buf, buflen);
-#elif (_POSIX_C_SOURCE >= 200112L || _XOPEN_SOURCE >= 600)
- /* XSI-compliant */
- if (errnum > 0 && strerror_r(errnum, buf, buflen) == 0)
- msg = buf;
-#else
- if (errnum > 0) {
- msg = strerror(errnum);
- if (msg) {
- strncpy(buf, msg, buflen);
- msg = buf;
- }
- }
-#endif
- if (!msg) {
- (void)snprintf(buf, buflen, "error %d", errnum);
- msg = buf;
- }
- buf[buflen - 1] = '\0';
- }
- return msg;
-}
-
-__cold const char *mdbx_strerror(int errnum) {
-#if defined(_WIN32) || defined(_WIN64)
- static char buf[1024];
- return mdbx_strerror_r(errnum, buf, sizeof(buf));
-#else
- const char *msg = mdbx_liberr2str(errnum);
- if (!msg) {
- if (errnum > 0)
- msg = strerror(errnum);
- if (!msg) {
- static char buf[32];
- (void)snprintf(buf, sizeof(buf) - 1, "error %d", errnum);
- msg = buf;
- }
- }
- return msg;
-#endif
-}
-
-#if defined(_WIN32) || defined(_WIN64) /* Bit of madness for Windows */
-const char *mdbx_strerror_r_ANSI2OEM(int errnum, char *buf, size_t buflen) {
- const char *msg = mdbx_liberr2str(errnum);
- if (!msg && buflen > 0 && buflen < INT_MAX) {
- const DWORD size = FormatMessageA(
- FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_IGNORE_INSERTS, NULL,
- errnum, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), buf, (DWORD)buflen,
- NULL);
- if (!size)
- msg = "FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM) failed";
- else if (!CharToOemBuffA(buf, buf, size))
- msg = "CharToOemBuffA() failed";
- else
- msg = buf;
- }
- return msg;
-}
-
-const char *mdbx_strerror_ANSI2OEM(int errnum) {
- static char buf[1024];
- return mdbx_strerror_r_ANSI2OEM(errnum, buf, sizeof(buf));
-}
-#endif /* Bit of madness for Windows */
-
-__cold void mdbx_debug_log_va(int level, const char *function, int line,
- const char *fmt, va_list args) {
- if (mdbx_debug_logger)
- mdbx_debug_logger(level, function, line, fmt, args);
- else {
-#if defined(_WIN32) || defined(_WIN64)
- if (IsDebuggerPresent()) {
- int prefix_len = 0;
- char *prefix = nullptr;
- if (function && line > 0)
- prefix_len = mdbx_asprintf(&prefix, "%s:%d ", function, line);
- else if (function)
- prefix_len = mdbx_asprintf(&prefix, "%s: ", function);
- else if (line > 0)
- prefix_len = mdbx_asprintf(&prefix, "%d: ", line);
- if (prefix_len > 0 && prefix) {
- OutputDebugStringA(prefix);
- mdbx_free(prefix);
- }
- char *msg = nullptr;
- int msg_len = mdbx_vasprintf(&msg, fmt, args);
- if (msg_len > 0 && msg) {
- OutputDebugStringA(msg);
- mdbx_free(msg);
- }
- }
-#else
- if (function && line > 0)
- fprintf(stderr, "%s:%d ", function, line);
- else if (function)
- fprintf(stderr, "%s: ", function);
- else if (line > 0)
- fprintf(stderr, "%d: ", line);
- vfprintf(stderr, fmt, args);
- fflush(stderr);
-#endif
- }
-}
-
-__cold void mdbx_debug_log(int level, const char *function, int line,
- const char *fmt, ...) {
- va_list args;
- va_start(args, fmt);
- mdbx_debug_log_va(level, function, line, fmt, args);
- va_end(args);
-}
-
-/* Dump a key in ascii or hexadecimal. */
-const char *mdbx_dump_val(const MDBX_val *key, char *const buf,
- const size_t bufsize) {
- if (!key)
- return "<null>";
- if (!key->iov_len)
- return "<empty>";
- if (!buf || bufsize < 4)
- return nullptr;
-
- bool is_ascii = true;
- const uint8_t *const data = key->iov_base;
- for (unsigned i = 0; i < key->iov_len; i++)
- if (data[i] < ' ' || data[i] > '~') {
- is_ascii = false;
- break;
- }
-
- if (is_ascii) {
- int len =
- snprintf(buf, bufsize, "%.*s",
- (key->iov_len > INT_MAX) ? INT_MAX : (int)key->iov_len, data);
- assert(len > 0 && (unsigned)len < bufsize);
- (void)len;
- } else {
- char *const detent = buf + bufsize - 2;
- char *ptr = buf;
- *ptr++ = '<';
- for (unsigned i = 0; i < key->iov_len; i++) {
- const ptrdiff_t left = detent - ptr;
- assert(left > 0);
- int len = snprintf(ptr, left, "%02x", data[i]);
- if (len < 0 || len >= left)
- break;
- ptr += len;
- }
- if (ptr < detent) {
- ptr[0] = '>';
- ptr[1] = '\0';
- }
- }
- return buf;
-}
-
-/*------------------------------------------------------------------------------
- LY: debug stuff */
-
-static const char *mdbx_leafnode_type(MDBX_node *n) {
- static const char *const tp[2][2] = {{"", ": DB"},
- {": sub-page", ": sub-DB"}};
- return F_ISSET(node_flags(n), F_BIGDATA)
- ? ": overflow page"
- : tp[F_ISSET(node_flags(n), F_DUPDATA)]
- [F_ISSET(node_flags(n), F_SUBDATA)];
-}
-
-/* Display all the keys in the page. */
-MDBX_MAYBE_UNUSED static void mdbx_page_list(MDBX_page *mp) {
- pgno_t pgno = mp->mp_pgno;
- const char *type;
- MDBX_node *node;
- unsigned i, nkeys, nsize, total = 0;
- MDBX_val key;
- DKBUF;
-
- switch (mp->mp_flags &
- (P_BRANCH | P_LEAF | P_LEAF2 | P_META | P_OVERFLOW | P_SUBP)) {
- case P_BRANCH:
- type = "Branch page";
- break;
- case P_LEAF:
- type = "Leaf page";
- break;
- case P_LEAF | P_SUBP:
- type = "Leaf sub-page";
- break;
- case P_LEAF | P_LEAF2:
- type = "Leaf2 page";
- break;
- case P_LEAF | P_LEAF2 | P_SUBP:
- type = "Leaf2 sub-page";
- break;
- case P_OVERFLOW:
- mdbx_verbose("Overflow page %" PRIaPGNO " pages %u\n", pgno, mp->mp_pages);
- return;
- case P_META:
- mdbx_verbose("Meta-page %" PRIaPGNO " txnid %" PRIu64 "\n", pgno,
- unaligned_peek_u64(4, page_meta(mp)->mm_txnid_a));
- return;
- default:
- mdbx_verbose("Bad page %" PRIaPGNO " flags 0x%X\n", pgno, mp->mp_flags);
- return;
- }
-
- nkeys = page_numkeys(mp);
- mdbx_verbose("%s %" PRIaPGNO " numkeys %u\n", type, pgno, nkeys);
-
- for (i = 0; i < nkeys; i++) {
- if (IS_LEAF2(mp)) { /* LEAF2 pages have no mp_ptrs[] or node headers */
- key.iov_len = nsize = mp->mp_leaf2_ksize;
- key.iov_base = page_leaf2key(mp, i, nsize);
- total += nsize;
- mdbx_verbose("key %u: nsize %u, %s\n", i, nsize, DKEY(&key));
- continue;
- }
- node = page_node(mp, i);
- key.iov_len = node_ks(node);
- key.iov_base = node->mn_data;
- nsize = (unsigned)(NODESIZE + key.iov_len);
- if (IS_BRANCH(mp)) {
- mdbx_verbose("key %u: page %" PRIaPGNO ", %s\n", i, node_pgno(node),
- DKEY(&key));
- total += nsize;
- } else {
- if (F_ISSET(node_flags(node), F_BIGDATA))
- nsize += sizeof(pgno_t);
- else
- nsize += (unsigned)node_ds(node);
- total += nsize;
- nsize += sizeof(indx_t);
- mdbx_verbose("key %u: nsize %u, %s%s\n", i, nsize, DKEY(&key),
- mdbx_leafnode_type(node));
- }
- total = EVEN(total);
- }
- mdbx_verbose("Total: header %u + contents %u + unused %u\n",
- IS_LEAF2(mp) ? PAGEHDRSZ : PAGEHDRSZ + mp->mp_lower, total,
- page_room(mp));
-}
-
-/*----------------------------------------------------------------------------*/
-
-/* Check if there is an initialized xcursor, so XCURSOR_REFRESH() is proper */
-#define XCURSOR_INITED(mc) \
- ((mc)->mc_xcursor && ((mc)->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED))
-
-/* Update sub-page pointer, if any, in mc->mc_xcursor.
- * Needed when the node which contains the sub-page may have moved.
- * Called with mp = mc->mc_pg[mc->mc_top], ki = mc->mc_ki[mc->mc_top]. */
-#define XCURSOR_REFRESH(mc, mp, ki) \
- do { \
- MDBX_page *xr_pg = (mp); \
- MDBX_node *xr_node = page_node(xr_pg, ki); \
- if ((node_flags(xr_node) & (F_DUPDATA | F_SUBDATA)) == F_DUPDATA) \
- (mc)->mc_xcursor->mx_cursor.mc_pg[0] = node_data(xr_node); \
- } while (0)
-
-MDBX_MAYBE_UNUSED static bool cursor_is_tracked(const MDBX_cursor *mc) {
- for (MDBX_cursor *scan = mc->mc_txn->mt_cursors[mc->mc_dbi]; scan;
- scan = scan->mc_next)
- if (mc == ((mc->mc_flags & C_SUB) ? &scan->mc_xcursor->mx_cursor : scan))
- return true;
- return false;
-}
-
-/* Perform act while tracking temporary cursor mn */
-#define WITH_CURSOR_TRACKING(mn, act) \
- do { \
- mdbx_cassert(&(mn), \
- mn.mc_txn->mt_cursors != NULL /* must be not rdonly txt */); \
- mdbx_cassert(&(mn), !cursor_is_tracked(&(mn))); \
- MDBX_cursor mc_dummy; \
- MDBX_cursor **tracking_head = &(mn).mc_txn->mt_cursors[mn.mc_dbi]; \
- MDBX_cursor *tracked = &(mn); \
- if ((mn).mc_flags & C_SUB) { \
- mc_dummy.mc_flags = C_INITIALIZED; \
- mc_dummy.mc_top = 0; \
- mc_dummy.mc_snum = 0; \
- mc_dummy.mc_xcursor = (MDBX_xcursor *)&(mn); \
- tracked = &mc_dummy; \
- } \
- tracked->mc_next = *tracking_head; \
- *tracking_head = tracked; \
- { act; } \
- *tracking_head = tracked->mc_next; \
- } while (0)
-
-int mdbx_cmp(const MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *a,
- const MDBX_val *b) {
- mdbx_assert(NULL, txn->mt_signature == MDBX_MT_SIGNATURE);
- return txn->mt_dbxs[dbi].md_cmp(a, b);
-}
-
-int mdbx_dcmp(const MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *a,
- const MDBX_val *b) {
- mdbx_assert(NULL, txn->mt_signature == MDBX_MT_SIGNATURE);
- return txn->mt_dbxs[dbi].md_dcmp(a, b);
-}
-
-/* Allocate memory for a page.
- * Re-use old malloc'ed pages first for singletons, otherwise just malloc.
- * Set MDBX_TXN_ERROR on failure. */
-static MDBX_page *mdbx_page_malloc(MDBX_txn *txn, unsigned num) {
- MDBX_env *env = txn->mt_env;
- MDBX_page *np = env->me_dp_reserve;
- size_t size = env->me_psize;
- if (likely(num == 1 && np)) {
- mdbx_assert(env, env->me_dp_reserve_len > 0);
- MDBX_ASAN_UNPOISON_MEMORY_REGION(np, size);
- VALGRIND_MEMPOOL_ALLOC(env, np, size);
- VALGRIND_MAKE_MEM_DEFINED(&np->mp_next, sizeof(np->mp_next));
- env->me_dp_reserve = np->mp_next;
- env->me_dp_reserve_len -= 1;
- } else {
- size = pgno2bytes(env, num);
- np = mdbx_malloc(size);
- if (unlikely(!np)) {
- txn->mt_flags |= MDBX_TXN_ERROR;
- return np;
- }
- VALGRIND_MEMPOOL_ALLOC(env, np, size);
- }
-
- if ((env->me_flags & MDBX_NOMEMINIT) == 0) {
- /* For a single page alloc, we init everything after the page header.
- * For multi-page, we init the final page; if the caller needed that
- * many pages they will be filling in at least up to the last page. */
- size_t skip = PAGEHDRSZ;
- if (num > 1)
- skip += pgno2bytes(env, num - 1);
- memset((char *)np + skip, 0, size - skip);
- }
-#if MDBX_DEBUG
- np->mp_pgno = 0;
-#endif
- VALGRIND_MAKE_MEM_UNDEFINED(np, size);
- np->mp_flags = 0;
- np->mp_pages = num;
- return np;
-}
-
-/* Free a shadow dirty page */
-static void mdbx_dpage_free(MDBX_env *env, MDBX_page *dp, unsigned npages) {
- VALGRIND_MAKE_MEM_UNDEFINED(dp, pgno2bytes(env, npages));
- MDBX_ASAN_UNPOISON_MEMORY_REGION(dp, pgno2bytes(env, npages));
- if (MDBX_DEBUG != 0 || unlikely(env->me_flags & MDBX_PAGEPERTURB))
- memset(dp, -1, pgno2bytes(env, npages));
- if (npages == 1 &&
- env->me_dp_reserve_len < env->me_options.dp_reserve_limit) {
- MDBX_ASAN_POISON_MEMORY_REGION((char *)dp + sizeof(dp->mp_next),
- pgno2bytes(env, npages) -
- sizeof(dp->mp_next));
- dp->mp_next = env->me_dp_reserve;
- VALGRIND_MEMPOOL_FREE(env, dp);
- env->me_dp_reserve = dp;
- env->me_dp_reserve_len += 1;
- } else {
- /* large pages just get freed directly */
- VALGRIND_MEMPOOL_FREE(env, dp);
- mdbx_free(dp);
- }
-}
-
-/* Return all dirty pages to dpage list */
-static void mdbx_dlist_free(MDBX_txn *txn) {
- MDBX_env *env = txn->mt_env;
- MDBX_dpl *const dl = txn->tw.dirtylist;
-
- for (unsigned i = 1; i <= dl->length; i++) {
- MDBX_page *dp = dl->items[i].ptr;
- mdbx_dpage_free(env, dp, dpl_npages(dl, i));
- }
-
- dpl_clear(dl);
-}
-
-static __always_inline MDBX_db *mdbx_outer_db(MDBX_cursor *mc) {
- mdbx_cassert(mc, (mc->mc_flags & C_SUB) != 0);
- MDBX_xcursor *mx = container_of(mc->mc_db, MDBX_xcursor, mx_db);
- MDBX_cursor_couple *couple = container_of(mx, MDBX_cursor_couple, inner);
- mdbx_cassert(mc, mc->mc_db == &couple->outer.mc_xcursor->mx_db);
- mdbx_cassert(mc, mc->mc_dbx == &couple->outer.mc_xcursor->mx_dbx);
- return couple->outer.mc_db;
-}
-
-MDBX_MAYBE_UNUSED __cold static bool mdbx_dirtylist_check(MDBX_txn *txn) {
- const MDBX_dpl *const dl = txn->tw.dirtylist;
- assert(dl->items[0].pgno == 0 && dl->items[dl->length + 1].pgno == P_INVALID);
- mdbx_tassert(txn, txn->tw.dirtyroom + dl->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
-
- if (!mdbx_audit_enabled())
- return true;
-
- unsigned loose = 0;
- for (unsigned i = dl->length; i > 0; --i) {
- const MDBX_page *const dp = dl->items[i].ptr;
- if (!dp)
- continue;
-
- mdbx_tassert(txn, dp->mp_pgno == dl->items[i].pgno);
- if (unlikely(dp->mp_pgno != dl->items[i].pgno))
- return false;
-
- const uint32_t age = mdbx_dpl_age(txn, i);
- mdbx_tassert(txn, age < UINT32_MAX / 3);
- if (unlikely(age > UINT32_MAX / 3))
- return false;
-
- mdbx_tassert(txn, dp->mp_flags == P_LOOSE || IS_MODIFIABLE(txn, dp));
- if (dp->mp_flags == P_LOOSE) {
- loose += 1;
- } else if (unlikely(!IS_MODIFIABLE(txn, dp)))
- return false;
-
- const unsigned num = dpl_npages(dl, i);
- mdbx_tassert(txn, txn->mt_next_pgno >= dp->mp_pgno + num);
- if (unlikely(txn->mt_next_pgno < dp->mp_pgno + num))
- return false;
-
- if (i < dl->sorted) {
- mdbx_tassert(txn, dl->items[i + 1].pgno >= dp->mp_pgno + num);
- if (unlikely(dl->items[i + 1].pgno < dp->mp_pgno + num))
- return false;
- }
-
- const unsigned rpa = mdbx_pnl_search(txn->tw.reclaimed_pglist, dp->mp_pgno,
- txn->mt_next_pgno);
- mdbx_tassert(txn, rpa > MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) ||
- txn->tw.reclaimed_pglist[rpa] != dp->mp_pgno);
- if (rpa <= MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) &&
- unlikely(txn->tw.reclaimed_pglist[rpa] == dp->mp_pgno))
- return false;
- if (num > 1) {
- const unsigned rpb = mdbx_pnl_search(
- txn->tw.reclaimed_pglist, dp->mp_pgno + num - 1, txn->mt_next_pgno);
- mdbx_tassert(txn, rpa == rpb);
- if (unlikely(rpa != rpb))
- return false;
- }
- }
-
- mdbx_tassert(txn, loose == txn->tw.loose_count);
- if (unlikely(loose != txn->tw.loose_count))
- return false;
-
- for (unsigned i = 1; i <= MDBX_PNL_SIZE(txn->tw.retired_pages); ++i) {
- const MDBX_page *const dp = debug_dpl_find(txn, txn->tw.retired_pages[i]);
- mdbx_tassert(txn, !dp);
- if (unlikely(dp))
- return false;
- }
-
- return true;
-}
-
-#if MDBX_ENABLE_REFUND
-static void mdbx_refund_reclaimed(MDBX_txn *txn) {
- /* Scanning in descend order */
- pgno_t next_pgno = txn->mt_next_pgno;
- const MDBX_PNL pnl = txn->tw.reclaimed_pglist;
- mdbx_tassert(txn, MDBX_PNL_SIZE(pnl) && MDBX_PNL_MOST(pnl) == next_pgno - 1);
-#if MDBX_PNL_ASCENDING
- unsigned i = MDBX_PNL_SIZE(pnl);
- mdbx_tassert(txn, pnl[i] == next_pgno - 1);
- while (--next_pgno, --i > 0 && pnl[i] == next_pgno - 1)
- ;
- MDBX_PNL_SIZE(pnl) = i;
-#else
- unsigned i = 1;
- mdbx_tassert(txn, pnl[i] == next_pgno - 1);
- unsigned len = MDBX_PNL_SIZE(pnl);
- while (--next_pgno, ++i <= len && pnl[i] == next_pgno - 1)
- ;
- MDBX_PNL_SIZE(pnl) = len -= i - 1;
- for (unsigned move = 0; move < len; ++move)
- pnl[1 + move] = pnl[i + move];
-#endif
- mdbx_verbose("refunded %" PRIaPGNO " pages: %" PRIaPGNO " -> %" PRIaPGNO,
- txn->mt_next_pgno - next_pgno, txn->mt_next_pgno, next_pgno);
- txn->mt_next_pgno = next_pgno;
- mdbx_tassert(txn, mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - 1));
-}
-
-static void mdbx_refund_loose(MDBX_txn *txn) {
- mdbx_tassert(txn, txn->tw.loose_pages != nullptr);
- mdbx_tassert(txn, txn->tw.loose_count > 0);
-
- MDBX_dpl *const dl = txn->tw.dirtylist;
- mdbx_tassert(txn, dl->length >= txn->tw.loose_count);
-
- pgno_t onstack[MDBX_CACHELINE_SIZE * 8 / sizeof(pgno_t)];
- MDBX_PNL suitable = onstack;
-
- if (dl->length - dl->sorted > txn->tw.loose_count) {
- /* Dirty list is useless since unsorted. */
- if (bytes2pnl(sizeof(onstack)) < txn->tw.loose_count) {
- suitable = mdbx_pnl_alloc(txn->tw.loose_count);
- if (unlikely(!suitable))
- return /* this is not a reason for transaction fail */;
- }
-
- /* Collect loose-pages which may be refunded. */
- mdbx_tassert(txn, txn->mt_next_pgno >= MIN_PAGENO + txn->tw.loose_count);
- pgno_t most = MIN_PAGENO;
- unsigned w = 0;
- for (const MDBX_page *lp = txn->tw.loose_pages; lp; lp = lp->mp_next) {
- mdbx_tassert(txn, lp->mp_flags == P_LOOSE);
- mdbx_tassert(txn, txn->mt_next_pgno > lp->mp_pgno);
- if (likely(txn->mt_next_pgno - txn->tw.loose_count <= lp->mp_pgno)) {
- mdbx_tassert(txn,
- w < ((suitable == onstack) ? bytes2pnl(sizeof(onstack))
- : MDBX_PNL_ALLOCLEN(suitable)));
- suitable[++w] = lp->mp_pgno;
- most = (lp->mp_pgno > most) ? lp->mp_pgno : most;
- }
- }
-
- if (most + 1 == txn->mt_next_pgno) {
- /* Sort suitable list and refund pages at the tail. */
- MDBX_PNL_SIZE(suitable) = w;
- mdbx_pnl_sort(suitable, MAX_PAGENO + 1);
-
- /* Scanning in descend order */
- const int step = MDBX_PNL_ASCENDING ? -1 : 1;
- const int begin = MDBX_PNL_ASCENDING ? MDBX_PNL_SIZE(suitable) : 1;
- const int end = MDBX_PNL_ASCENDING ? 0 : MDBX_PNL_SIZE(suitable) + 1;
- mdbx_tassert(txn, suitable[begin] >= suitable[end - step]);
- mdbx_tassert(txn, most == suitable[begin]);
-
- for (int i = begin + step; i != end; i += step) {
- if (suitable[i] != most - 1)
- break;
- most -= 1;
- }
- const unsigned refunded = txn->mt_next_pgno - most;
- mdbx_debug("refund-suitable %u pages %" PRIaPGNO " -> %" PRIaPGNO,
- refunded, most, txn->mt_next_pgno);
- txn->tw.loose_count -= refunded;
- txn->tw.dirtyroom += refunded;
- assert(txn->tw.dirtyroom <= txn->mt_env->me_options.dp_limit);
- txn->mt_next_pgno = most;
-
- /* Filter-out dirty list */
- unsigned r = 0;
- w = 0;
- if (dl->sorted) {
- do {
- if (dl->items[++r].pgno < most) {
- if (++w != r)
- dl->items[w] = dl->items[r];
- }
- } while (r < dl->sorted);
- dl->sorted = w;
- }
- while (r < dl->length) {
- if (dl->items[++r].pgno < most) {
- if (++w != r)
- dl->items[w] = dl->items[r];
- }
- }
- dpl_setlen(dl, w);
- mdbx_tassert(txn,
- txn->tw.dirtyroom + txn->tw.dirtylist->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
-
- goto unlink_loose;
- }
- } else {
- /* Dirtylist is mostly sorted, just refund loose pages at the end. */
- mdbx_dpl_sort(txn);
- mdbx_tassert(txn, dl->length < 2 ||
- dl->items[1].pgno < dl->items[dl->length].pgno);
- mdbx_tassert(txn, dl->sorted == dl->length);
-
- /* Scan dirtylist tail-forward and cutoff suitable pages. */
- unsigned n;
- for (n = dl->length; dl->items[n].pgno == txn->mt_next_pgno - 1 &&
- dl->items[n].ptr->mp_flags == P_LOOSE;
- --n) {
- mdbx_tassert(txn, n > 0);
- MDBX_page *dp = dl->items[n].ptr;
- mdbx_debug("refund-sorted page %" PRIaPGNO, dp->mp_pgno);
- mdbx_tassert(txn, dp->mp_pgno == dl->items[n].pgno);
- txn->mt_next_pgno -= 1;
- }
- dpl_setlen(dl, n);
-
- if (dl->sorted != dl->length) {
- const unsigned refunded = dl->sorted - dl->length;
- dl->sorted = dl->length;
- txn->tw.loose_count -= refunded;
- txn->tw.dirtyroom += refunded;
- mdbx_tassert(txn,
- txn->tw.dirtyroom + txn->tw.dirtylist->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
-
- /* Filter-out loose chain & dispose refunded pages. */
- unlink_loose:
- for (MDBX_page **link = &txn->tw.loose_pages; *link;) {
- MDBX_page *dp = *link;
- mdbx_tassert(txn, dp->mp_flags == P_LOOSE);
- if (txn->mt_next_pgno > dp->mp_pgno) {
- link = &dp->mp_next;
- } else {
- *link = dp->mp_next;
- if ((txn->mt_flags & MDBX_WRITEMAP) == 0)
- mdbx_dpage_free(txn->mt_env, dp, 1);
- }
- }
- }
- }
-
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- if (suitable != onstack)
- mdbx_pnl_free(suitable);
- txn->tw.loose_refund_wl = txn->mt_next_pgno;
-}
-
-static bool mdbx_refund(MDBX_txn *txn) {
- const pgno_t before = txn->mt_next_pgno;
-
- if (txn->tw.loose_pages && txn->tw.loose_refund_wl > txn->mt_next_pgno)
- mdbx_refund_loose(txn);
-
- while (true) {
- if (MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) == 0 ||
- MDBX_PNL_MOST(txn->tw.reclaimed_pglist) != txn->mt_next_pgno - 1)
- break;
-
- mdbx_refund_reclaimed(txn);
- if (!txn->tw.loose_pages || txn->tw.loose_refund_wl <= txn->mt_next_pgno)
- break;
-
- const pgno_t memo = txn->mt_next_pgno;
- mdbx_refund_loose(txn);
- if (memo == txn->mt_next_pgno)
- break;
- }
-
- if (before == txn->mt_next_pgno)
- return false;
-
- if (txn->tw.spill_pages)
- /* Squash deleted pagenums if we refunded any */
- mdbx_spill_purge(txn);
-
- return true;
-}
-#else /* MDBX_ENABLE_REFUND */
-static __inline bool mdbx_refund(MDBX_txn *txn) {
- (void)txn;
- /* No online auto-compactification. */
- return false;
-}
-#endif /* MDBX_ENABLE_REFUND */
-
-__cold static void mdbx_kill_page(MDBX_txn *txn, MDBX_page *mp, pgno_t pgno,
- unsigned npages) {
- MDBX_env *const env = txn->mt_env;
- mdbx_debug("kill %u page(s) %" PRIaPGNO, npages, pgno);
- mdbx_assert(env, pgno >= NUM_METAS && npages);
- if (!IS_FROZEN(txn, mp)) {
- const size_t bytes = pgno2bytes(env, npages);
- memset(mp, -1, bytes);
- mp->mp_pgno = pgno;
- if ((env->me_flags & MDBX_WRITEMAP) == 0)
- mdbx_pwrite(env->me_lazy_fd, mp, bytes, pgno2bytes(env, pgno));
- } else {
- struct iovec iov[MDBX_COMMIT_PAGES];
- iov[0].iov_len = env->me_psize;
- iov[0].iov_base = (char *)env->me_pbuf + env->me_psize;
- size_t iov_off = pgno2bytes(env, pgno);
- unsigned n = 1;
- while (--npages) {
- iov[n] = iov[0];
- if (++n == MDBX_COMMIT_PAGES) {
- mdbx_pwritev(env->me_lazy_fd, iov, MDBX_COMMIT_PAGES, iov_off,
- pgno2bytes(env, MDBX_COMMIT_PAGES));
- iov_off += pgno2bytes(env, MDBX_COMMIT_PAGES);
- n = 0;
- }
- }
- mdbx_pwritev(env->me_lazy_fd, iov, n, iov_off, pgno2bytes(env, n));
- }
-}
-
-/* Remove page from dirty list */
-static __inline void mdbx_page_wash(MDBX_txn *txn, const unsigned di,
- MDBX_page *const mp,
- const unsigned npages) {
- mdbx_tassert(txn, di && di <= txn->tw.dirtylist->length &&
- txn->tw.dirtylist->items[di].ptr == mp);
- mdbx_dpl_remove(txn, di);
- txn->tw.dirtyroom++;
- mdbx_tassert(txn, txn->tw.dirtyroom + txn->tw.dirtylist->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
- mp->mp_txnid = INVALID_TXNID;
- mp->mp_flags = 0xFFFF;
- VALGRIND_MAKE_MEM_UNDEFINED(mp, PAGEHDRSZ);
- if (txn->mt_flags & MDBX_WRITEMAP) {
- VALGRIND_MAKE_MEM_NOACCESS(page_data(mp),
- pgno2bytes(txn->mt_env, npages) - PAGEHDRSZ);
- MDBX_ASAN_POISON_MEMORY_REGION(page_data(mp),
- pgno2bytes(txn->mt_env, npages) - PAGEHDRSZ);
- } else
- mdbx_dpage_free(txn->mt_env, mp, npages);
-}
-
-static __inline txnid_t pp_txnid4chk(const MDBX_page *mp, const MDBX_txn *txn) {
- (void)txn;
-#if MDBX_DISABLE_PAGECHECKS
- (void)mp;
- return 0;
-#else
- return /* maybe zero in legacy DB */ mp->mp_txnid;
-#endif /* !MDBX_DISABLE_PAGECHECKS */
-}
-
-/* Retire, loosen or free a single page.
- *
- * For dirty pages, saves single pages to a list for future reuse in this same
- * txn. It has been pulled from the GC and already resides on the dirty list,
- * but has been deleted. Use these pages first before pulling again from the GC.
- *
- * If the page wasn't dirtied in this txn, just add it
- * to this txn's free list. */
-static int mdbx_page_retire_ex(MDBX_cursor *mc, const pgno_t pgno,
- MDBX_page *mp /* maybe null */,
- int pagetype /* maybe unknown/zero */) {
- int rc;
- MDBX_txn *const txn = mc->mc_txn;
- mdbx_tassert(txn, !mp || (mp->mp_pgno == pgno && PAGETYPE(mp) == pagetype));
-
- /* During deleting entire subtrees, it is reasonable and possible to avoid
- * reading leaf pages, i.e. significantly reduce hard page-faults & IOPs:
- * - mp is null, i.e. the page has not yet been read;
- * - pagetype is known and the P_LEAF bit is set;
- * - we can determine the page status via scanning the lists
- * of dirty and spilled pages.
- *
- * On the other hand, this could be suboptimal for WRITEMAP mode, since
- * requires support the list of dirty pages and avoid explicit spilling.
- * So for flexibility and avoid extra internal dependencies we just
- * fallback to reading if dirty list was not allocated yet. */
- unsigned di = 0, si = 0, npages = 1;
- bool is_frozen = false, is_spilled = false, is_shadowed = false;
- if (unlikely(!mp)) {
- if (mdbx_assert_enabled() && pagetype) {
- MDBX_page *check;
- rc = mdbx_page_get(mc, pgno, &check, txn->mt_front);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- mdbx_tassert(txn, (PAGETYPE(check) & ~P_LEAF2) == (pagetype & ~P_FROZEN));
- mdbx_tassert(txn, !(pagetype & P_FROZEN) || IS_FROZEN(txn, check));
- }
- if (pagetype & P_FROZEN) {
- is_frozen = true;
- if (mdbx_assert_enabled()) {
- for (MDBX_txn *scan = txn; scan; scan = scan->mt_parent) {
- mdbx_tassert(txn, !mdbx_search_spilled(scan, pgno));
- mdbx_tassert(txn, !scan->tw.dirtylist || !debug_dpl_find(scan, pgno));
- }
- }
- goto status_done;
- } else if (pagetype && txn->tw.dirtylist) {
- if ((di = mdbx_dpl_exist(txn, pgno)) != 0) {
- mp = txn->tw.dirtylist->items[di].ptr;
- mdbx_tassert(txn, IS_MODIFIABLE(txn, mp));
- goto status_done;
- }
- if ((si = mdbx_search_spilled(txn, pgno)) != 0) {
- is_spilled = true;
- goto status_done;
- }
- for (MDBX_txn *parent = txn->mt_parent; parent;
- parent = parent->mt_parent) {
- if (mdbx_dpl_exist(parent, pgno)) {
- is_shadowed = true;
- goto status_done;
- }
- if (mdbx_search_spilled(parent, pgno)) {
- is_spilled = true;
- goto status_done;
- }
- }
- is_frozen = true;
- goto status_done;
- }
-
- rc = mdbx_page_get(mc, pgno, &mp, txn->mt_front);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- mdbx_tassert(txn, !pagetype || PAGETYPE(mp) == pagetype);
- pagetype = PAGETYPE(mp);
- }
-
- is_frozen = IS_FROZEN(txn, mp);
- if (!is_frozen) {
- const bool is_dirty = IS_MODIFIABLE(txn, mp);
- is_spilled = IS_SPILLED(txn, mp) && !(txn->mt_flags & MDBX_WRITEMAP);
- is_shadowed = IS_SHADOWED(txn, mp);
- if (is_dirty) {
- mdbx_tassert(txn, !is_spilled);
- mdbx_tassert(txn, !mdbx_search_spilled(txn, pgno));
- mdbx_tassert(txn, debug_dpl_find(txn, pgno) == mp || txn->mt_parent ||
- (txn->mt_flags & MDBX_WRITEMAP));
- } else {
- mdbx_tassert(txn, !debug_dpl_find(txn, pgno));
- }
-
- di = is_dirty ? mdbx_dpl_exist(txn, pgno) : 0;
- si = is_spilled ? mdbx_search_spilled(txn, pgno) : 0;
- mdbx_tassert(txn, !is_dirty || di || (txn->mt_flags & MDBX_WRITEMAP));
- } else {
- mdbx_tassert(txn, !IS_MODIFIABLE(txn, mp));
- mdbx_tassert(txn, !IS_SPILLED(txn, mp));
- mdbx_tassert(txn, !IS_SHADOWED(txn, mp));
- }
-
-status_done:
- if (likely((pagetype & P_OVERFLOW) == 0)) {
- STATIC_ASSERT(P_BRANCH == 1);
- const bool is_branch = pagetype & P_BRANCH;
- if (unlikely(mc->mc_flags & C_SUB)) {
- MDBX_db *outer = mdbx_outer_db(mc);
- mdbx_cassert(mc, !is_branch || outer->md_branch_pages > 0);
- outer->md_branch_pages -= is_branch;
- mdbx_cassert(mc, is_branch || outer->md_leaf_pages > 0);
- outer->md_leaf_pages -= 1 - is_branch;
- }
- mdbx_cassert(mc, !is_branch || mc->mc_db->md_branch_pages > 0);
- mc->mc_db->md_branch_pages -= is_branch;
- mdbx_cassert(mc, (pagetype & P_LEAF) == 0 || mc->mc_db->md_leaf_pages > 0);
- mc->mc_db->md_leaf_pages -= (pagetype & P_LEAF) != 0;
- } else {
- npages = mp->mp_pages;
- mdbx_cassert(mc, mc->mc_db->md_overflow_pages >= npages);
- mc->mc_db->md_overflow_pages -= npages;
- }
-
- if (is_frozen) {
- retire:
- mdbx_debug("retire %u page %" PRIaPGNO, npages, pgno);
- rc = mdbx_pnl_append_range(false, &txn->tw.retired_pages, pgno, npages);
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- return rc;
- }
-
- /* Возврат Ñтраниц в нераÑпределенный "хвоÑÑ‚" БД.
- * Содержимое Ñтраниц не уничтожаетÑÑ, а Ð´Ð»Ñ Ð²Ð»Ð¾Ð¶ÐµÐ½Ð½Ñ‹Ñ… транзакций граница
- * нераÑпределенного "хвоÑта" БД ÑдвигаетÑÑ Ñ‚Ð¾Ð»ÑŒÐºÐ¾ при их коммите. */
- if (MDBX_ENABLE_REFUND && unlikely(pgno + npages == txn->mt_next_pgno)) {
- const char *kind = nullptr;
- if (di) {
- /* Страница иÑпачкана в Ñтой транзакции, но до Ñтого могла быть
- * аллоцирована, иÑпачкана и пролита в одной из родительÑких транзакций.
- * Её МОЖÐО вытолкнуть в нераÑпределенный хвоÑÑ‚. */
- kind = "dirty";
- /* Remove from dirty list */
- mdbx_page_wash(txn, di, mp, npages);
- } else if (si) {
- /* Страница пролита в Ñтой транзакции, Ñ‚.е. она аллоцирована
- * и запачкана в Ñтой или одной из родительÑких транзакций.
- * Её МОЖÐО вытолкнуть в нераÑпределенный хвоÑÑ‚. */
- kind = "spilled";
- mdbx_spill_remove(txn, si, npages);
- } else if ((txn->mt_flags & MDBX_WRITEMAP)) {
- kind = "writemap";
- mdbx_tassert(txn, mp && IS_MODIFIABLE(txn, mp));
- } else {
- /* Страница аллоцирована, запачкана и возможно пролита в одной
- * из родительÑких транзакций.
- * Её МОЖÐО вытолкнуть в нераÑпределенный хвоÑÑ‚. */
- kind = "parent's";
- if (mdbx_assert_enabled() && mp) {
- kind = nullptr;
- for (MDBX_txn *parent = txn->mt_parent; parent;
- parent = parent->mt_parent) {
- if (mdbx_search_spilled(parent, pgno)) {
- kind = "parent-spilled";
- mdbx_tassert(txn, is_spilled);
- break;
- }
- if (mp == debug_dpl_find(parent, pgno)) {
- kind = "parent-dirty";
- mdbx_tassert(txn, !is_spilled);
- break;
- }
- }
- mdbx_tassert(txn, kind != nullptr);
- }
- mdbx_tassert(txn,
- is_spilled || is_shadowed || (mp && IS_SHADOWED(txn, mp)));
- }
- mdbx_debug("refunded %u %s page %" PRIaPGNO, npages, kind, pgno);
- txn->mt_next_pgno = pgno;
- mdbx_refund(txn);
- return MDBX_SUCCESS;
- }
-
- if (di) {
- /* Dirty page from this transaction */
- /* If suitable we can reuse it through loose list */
- if (likely(npages == 1 &&
- txn->tw.loose_count < txn->mt_env->me_options.dp_loose_limit &&
- (!MDBX_ENABLE_REFUND ||
- /* skip pages near to the end in favor of compactification */
- txn->mt_next_pgno >
- pgno + txn->mt_env->me_options.dp_loose_limit ||
- txn->mt_next_pgno <= txn->mt_env->me_options.dp_loose_limit))) {
- mdbx_debug("loosen dirty page %" PRIaPGNO, pgno);
- mp->mp_flags = P_LOOSE;
- mp->mp_next = txn->tw.loose_pages;
- txn->tw.loose_pages = mp;
- txn->tw.loose_count++;
-#if MDBX_ENABLE_REFUND
- txn->tw.loose_refund_wl = (pgno + 2 > txn->tw.loose_refund_wl)
- ? pgno + 2
- : txn->tw.loose_refund_wl;
-#endif /* MDBX_ENABLE_REFUND */
- if (MDBX_DEBUG != 0 || unlikely(txn->mt_env->me_flags & MDBX_PAGEPERTURB))
- memset(page_data(mp), -1, txn->mt_env->me_psize - PAGEHDRSZ);
- VALGRIND_MAKE_MEM_NOACCESS(page_data(mp),
- txn->mt_env->me_psize - PAGEHDRSZ);
- MDBX_ASAN_POISON_MEMORY_REGION(page_data(mp),
- txn->mt_env->me_psize - PAGEHDRSZ);
- return MDBX_SUCCESS;
- }
-
-#if !MDBX_DEBUG && !defined(MDBX_USE_VALGRIND) && !defined(__SANITIZE_ADDRESS__)
- if (unlikely(txn->mt_env->me_flags & MDBX_PAGEPERTURB))
-#endif
- {
- /* Страница могла быть изменена в одной из родительÑких транзакций,
- * в том чиÑле, позже выгружена и затем Ñнова загружена и изменена.
- * Ð’ обоих ÑлучаÑÑ… её Ð½ÐµÐ»ÑŒÐ·Ñ Ð·Ð°Ñ‚Ð¸Ñ€Ð°Ñ‚ÑŒ на диÑке и помечать недоÑтупной
- * в asan и/или valgrind */
- for (MDBX_txn *parent = txn->mt_parent;
- parent && (parent->mt_flags & MDBX_TXN_SPILLS);
- parent = parent->mt_parent) {
- if (mdbx_intersect_spilled(parent, pgno, npages))
- goto skip_invalidate;
- if (mdbx_dpl_intersect(parent, pgno, npages))
- goto skip_invalidate;
- }
-
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- if (MDBX_DEBUG != 0 || unlikely(txn->mt_env->me_flags & MDBX_PAGEPERTURB))
-#endif
- mdbx_kill_page(txn, mp, pgno, npages);
- if (!(txn->mt_flags & MDBX_WRITEMAP)) {
- VALGRIND_MAKE_MEM_NOACCESS(page_data(pgno2page(txn->mt_env, pgno)),
- pgno2bytes(txn->mt_env, npages) - PAGEHDRSZ);
- MDBX_ASAN_POISON_MEMORY_REGION(page_data(pgno2page(txn->mt_env, pgno)),
- pgno2bytes(txn->mt_env, npages) -
- PAGEHDRSZ);
- }
- }
- skip_invalidate:
- /* Remove from dirty list */
- mdbx_page_wash(txn, di, mp, npages);
-
- reclaim:
- mdbx_debug("reclaim %u %s page %" PRIaPGNO, npages, "dirty", pgno);
- rc = mdbx_pnl_insert_range(&txn->tw.reclaimed_pglist, pgno, npages);
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- return rc;
- }
-
- if (si) {
- /* Page ws spilled in this txn */
- mdbx_spill_remove(txn, si, npages);
- /* Страница могла быть выделена и затем пролита в Ñтой транзакции,
- * тогда её необходимо помеÑтить в reclaimed-ÑпиÑок.
- * Либо она могла быть выделена в одной из родительÑких транзакций и затем
- * пролита в Ñтой транзакции, тогда её необходимо помеÑтить в
- * retired-ÑпиÑок Ð´Ð»Ñ Ð¿Ð¾Ñледующей фильтрации при коммите. */
- for (MDBX_txn *parent = txn->mt_parent; parent;
- parent = parent->mt_parent) {
- if (mdbx_dpl_exist(parent, pgno))
- goto retire;
- }
- /* Страница точно была выделена в Ñтой транзакции
- * и теперь может быть иÑпользована повторно. */
- goto reclaim;
- }
-
- if (is_shadowed) {
- /* Dirty page MUST BE a clone from (one of) parent transaction(s). */
- if (mdbx_assert_enabled()) {
- const MDBX_page *parent_dp = nullptr;
- /* Check parent(s)'s dirty lists. */
- for (MDBX_txn *parent = txn->mt_parent; parent && !parent_dp;
- parent = parent->mt_parent) {
- mdbx_tassert(txn, !mdbx_search_spilled(parent, pgno));
- parent_dp = debug_dpl_find(parent, pgno);
- }
- mdbx_tassert(txn, parent_dp && (!mp || parent_dp == mp));
- }
- /* Страница была выделена в родительÑкой транзакции и теперь может быть
- * иÑпользована повторно, но только внутри Ñтой транзакции, либо дочерних.
- */
- goto reclaim;
- }
-
- /* Страница может входить в доÑтупный читателÑм MVCC-Ñнимок, либо же она
- * могла быть выделена, а затем пролита в одной из родительÑких
- * транзакций. ПоÑтому пока помещаем её в retired-ÑпиÑок, который будет
- * фильтроватьÑÑ Ð¾Ñ‚Ð½Ð¾Ñительно dirty- и spilled-ÑпиÑков родительÑких
- * транзакций при коммите дочерних транзакций, либо же будет запиÑан
- * в GC в неизменном виде. */
- goto retire;
-}
-
-static __inline int mdbx_page_retire(MDBX_cursor *mc, MDBX_page *mp) {
- return mdbx_page_retire_ex(mc, mp->mp_pgno, mp, PAGETYPE(mp));
-}
-
-struct mdbx_iov_ctx {
- unsigned iov_items;
- size_t iov_bytes;
- size_t iov_off;
- pgno_t flush_begin;
- pgno_t flush_end;
- struct iovec iov[MDBX_COMMIT_PAGES];
-};
-
-static __inline void mdbx_iov_init(MDBX_txn *const txn,
- struct mdbx_iov_ctx *ctx) {
- ctx->flush_begin = MAX_PAGENO;
- ctx->flush_end = MIN_PAGENO;
- ctx->iov_items = 0;
- ctx->iov_bytes = 0;
- ctx->iov_off = 0;
- (void)txn;
-}
-
-static __inline void mdbx_iov_done(MDBX_txn *const txn,
- struct mdbx_iov_ctx *ctx) {
- mdbx_tassert(txn, ctx->iov_items == 0);
-#if defined(__linux__) || defined(__gnu_linux__)
- MDBX_env *const env = txn->mt_env;
- if (!(txn->mt_flags & MDBX_WRITEMAP) &&
- mdbx_linux_kernel_version < 0x02060b00)
- /* Linux kernels older than version 2.6.11 ignore the addr and nbytes
- * arguments, making this function fairly expensive. Therefore, the
- * whole cache is always flushed. */
- mdbx_flush_incoherent_mmap(
- env->me_map + pgno2bytes(env, ctx->flush_begin),
- pgno2bytes(env, ctx->flush_end - ctx->flush_begin), env->me_os_psize);
-#endif /* Linux */
-}
-
-static int mdbx_iov_write(MDBX_txn *const txn, struct mdbx_iov_ctx *ctx) {
- mdbx_tassert(txn, !(txn->mt_flags & MDBX_WRITEMAP));
- mdbx_tassert(txn, ctx->iov_items > 0);
-
- MDBX_env *const env = txn->mt_env;
- int rc;
- if (likely(ctx->iov_items == 1)) {
- mdbx_assert(env, ctx->iov_bytes == (size_t)ctx->iov[0].iov_len);
- rc = mdbx_pwrite(env->me_lazy_fd, ctx->iov[0].iov_base, ctx->iov[0].iov_len,
- ctx->iov_off);
- } else {
- rc = mdbx_pwritev(env->me_lazy_fd, ctx->iov, ctx->iov_items, ctx->iov_off,
- ctx->iov_bytes);
- }
-
- if (unlikely(rc != MDBX_SUCCESS))
- mdbx_error("Write error: %s", mdbx_strerror(rc));
- else {
- VALGRIND_MAKE_MEM_DEFINED(txn->mt_env->me_map + ctx->iov_off,
- ctx->iov_bytes);
- MDBX_ASAN_UNPOISON_MEMORY_REGION(txn->mt_env->me_map + ctx->iov_off,
- ctx->iov_bytes);
- }
-
- unsigned iov_items = ctx->iov_items;
-#if MDBX_ENABLE_PGOP_STAT
- txn->mt_env->me_lck->mti_pgop_stat.wops.weak += iov_items;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- ctx->iov_items = 0;
- ctx->iov_bytes = 0;
-
- uint64_t timestamp = 0;
- for (unsigned i = 0; i < iov_items; i++) {
- MDBX_page *wp = (MDBX_page *)ctx->iov[i].iov_base;
- const MDBX_page *rp = pgno2page(txn->mt_env, wp->mp_pgno);
- /* check with timeout as the workaround
- * for todo4recovery://erased_by_github/libmdbx/issues/269 */
- while (likely(rc == MDBX_SUCCESS) &&
- unlikely(memcmp(wp, rp, ctx->iov[i].iov_len) != 0)) {
- if (!timestamp) {
- timestamp = mdbx_osal_monotime();
- mdbx_iov_done(txn, ctx);
- mdbx_warning(
- "catch delayed/non-arrived page %" PRIaPGNO " %s", wp->mp_pgno,
- "(workaround for incoherent flaw of unified page/buffer cache)");
- } else if (unlikely(mdbx_osal_monotime() - timestamp > 65536 / 10)) {
- mdbx_error(
- "bailout waiting for %" PRIaPGNO " page arrival %s", wp->mp_pgno,
- "(workaround for incoherent flaw of unified page/buffer cache)");
- rc = MDBX_CORRUPTED;
- }
-#if defined(_WIN32) || defined(_WIN64)
- SwitchToThread();
-#elif defined(__linux__) || defined(__gnu_linux__) || defined(_UNIX03_SOURCE)
- sched_yield();
-#elif (defined(_GNU_SOURCE) && __GLIBC_PREREQ(2, 1)) || defined(_OPEN_THREADS)
- pthread_yield();
-#else
- usleep(42);
-#endif
- }
- mdbx_dpage_free(env, wp, bytes2pgno(env, ctx->iov[i].iov_len));
- }
- return rc;
-}
-
-static int iov_page(MDBX_txn *txn, struct mdbx_iov_ctx *ctx, MDBX_page *dp,
- unsigned npages) {
- MDBX_env *const env = txn->mt_env;
- mdbx_tassert(txn,
- dp->mp_pgno >= MIN_PAGENO && dp->mp_pgno < txn->mt_next_pgno);
- mdbx_tassert(txn, IS_MODIFIABLE(txn, dp));
- mdbx_tassert(txn,
- !(dp->mp_flags & ~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW)));
-
- ctx->flush_begin =
- (ctx->flush_begin < dp->mp_pgno) ? ctx->flush_begin : dp->mp_pgno;
- ctx->flush_end = (ctx->flush_end > dp->mp_pgno + npages)
- ? ctx->flush_end
- : dp->mp_pgno + npages;
- env->me_lck->mti_unsynced_pages.weak += npages;
-
- if (IS_SHADOWED(txn, dp)) {
- mdbx_tassert(txn, !(txn->mt_flags & MDBX_WRITEMAP));
- dp->mp_txnid = txn->mt_txnid;
- mdbx_tassert(txn, IS_SPILLED(txn, dp));
- const size_t size = pgno2bytes(env, npages);
- if (ctx->iov_off + ctx->iov_bytes != pgno2bytes(env, dp->mp_pgno) ||
- ctx->iov_items == ARRAY_LENGTH(ctx->iov) ||
- ctx->iov_bytes + size > MAX_WRITE) {
- if (ctx->iov_items) {
- int err = mdbx_iov_write(txn, ctx);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-#if defined(__linux__) || defined(__gnu_linux__)
- if (mdbx_linux_kernel_version >= 0x02060b00)
- /* Linux kernels older than version 2.6.11 ignore the addr and nbytes
- * arguments, making this function fairly expensive. Therefore, the
- * whole cache is always flushed. */
-#endif /* Linux */
- mdbx_flush_incoherent_mmap(env->me_map + ctx->iov_off, ctx->iov_bytes,
- env->me_os_psize);
- }
- ctx->iov_off = pgno2bytes(env, dp->mp_pgno);
- }
- ctx->iov[ctx->iov_items].iov_base = (void *)dp;
- ctx->iov[ctx->iov_items].iov_len = size;
- ctx->iov_items += 1;
- ctx->iov_bytes += size;
- } else {
- mdbx_tassert(txn, txn->mt_flags & MDBX_WRITEMAP);
- }
- return MDBX_SUCCESS;
-}
-
-static int spill_page(MDBX_txn *txn, struct mdbx_iov_ctx *ctx, MDBX_page *dp,
- unsigned npages) {
- mdbx_tassert(txn, !(txn->mt_flags & MDBX_WRITEMAP));
- pgno_t pgno = dp->mp_pgno;
- int err = iov_page(txn, ctx, dp, npages);
- if (likely(err == MDBX_SUCCESS)) {
- err = mdbx_pnl_append_range(true, &txn->tw.spill_pages, pgno << 1, npages);
-#if MDBX_ENABLE_PGOP_STAT
- if (likely(err == MDBX_SUCCESS))
- txn->mt_env->me_lck->mti_pgop_stat.spill.weak += npages;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- }
- return err;
-}
-
-/* Set unspillable LRU-label for dirty pages watched by txn.
- * Returns the number of pages marked as unspillable. */
-static unsigned mdbx_cursor_keep(MDBX_txn *txn, MDBX_cursor *mc) {
- unsigned keep = 0;
- while (mc->mc_flags & C_INITIALIZED) {
- for (unsigned i = 0; i < mc->mc_snum; ++i) {
- const MDBX_page *mp = mc->mc_pg[i];
- if (IS_MODIFIABLE(txn, mp) && !IS_SUBP(mp)) {
- unsigned const n = mdbx_dpl_search(txn, mp->mp_pgno);
- if (txn->tw.dirtylist->items[n].pgno == mp->mp_pgno &&
- mdbx_dpl_age(txn, n)) {
- txn->tw.dirtylist->items[n].lru = txn->tw.dirtylru;
- ++keep;
- }
- }
- }
- if (!mc->mc_xcursor)
- break;
- mc = &mc->mc_xcursor->mx_cursor;
- }
- return keep;
-}
-
-static unsigned mdbx_txn_keep(MDBX_txn *txn, MDBX_cursor *m0) {
- unsigned keep = m0 ? mdbx_cursor_keep(txn, m0) : 0;
- for (unsigned i = FREE_DBI; i < txn->mt_numdbs; ++i)
- if (F_ISSET(txn->mt_dbistate[i], DBI_DIRTY | DBI_VALID) &&
- txn->mt_dbs[i].md_root != P_INVALID)
- for (MDBX_cursor *mc = txn->mt_cursors[i]; mc; mc = mc->mc_next)
- if (mc != m0)
- keep += mdbx_cursor_keep(txn, mc);
- return keep;
-}
-
-/* Returns the spilling priority (0..255) for a dirty page:
- * 0 = should be spilled;
- * ...
- * > 255 = must not be spilled. */
-static unsigned spill_prio(const MDBX_txn *txn, const unsigned i,
- const uint32_t reciprocal) {
- MDBX_dpl *const dl = txn->tw.dirtylist;
- const uint32_t age = mdbx_dpl_age(txn, i);
- const unsigned npages = dpl_npages(dl, i);
- const pgno_t pgno = dl->items[i].pgno;
- if (age == 0) {
- mdbx_debug("skip %s %u page %" PRIaPGNO, "keep", npages, pgno);
- return 256;
- }
-
- MDBX_page *const dp = dl->items[i].ptr;
- if (dp->mp_flags & (P_LOOSE | P_SPILLED)) {
- mdbx_debug("skip %s %u page %" PRIaPGNO,
- (dp->mp_flags & P_LOOSE) ? "loose"
- : (dp->mp_flags & P_LOOSE) ? "loose"
- : "parent-spilled",
- npages, pgno);
- return 256;
- }
-
- /* Can't spill twice,
- * make sure it's not already in a parent's spill list(s). */
- MDBX_txn *parent = txn->mt_parent;
- if (parent && (parent->mt_flags & MDBX_TXN_SPILLS)) {
- do
- if (mdbx_intersect_spilled(parent, pgno, npages)) {
- mdbx_debug("skip-2 parent-spilled %u page %" PRIaPGNO, npages, pgno);
- dp->mp_flags |= P_SPILLED;
- return 256;
- }
- while ((parent = parent->mt_parent) != nullptr);
- }
-
- mdbx_tassert(txn, age * (uint64_t)reciprocal < UINT32_MAX);
- unsigned prio = age * reciprocal >> 24;
- mdbx_tassert(txn, prio < 256);
- if (likely(npages == 1))
- return prio = 256 - prio;
-
- /* make a large/overflow pages be likely to spill */
- uint32_t factor = npages | npages >> 1;
- factor |= factor >> 2;
- factor |= factor >> 4;
- factor |= factor >> 8;
- factor |= factor >> 16;
- factor = prio * log2n_powerof2(factor + 1) + /* golden ratio */ 157;
- factor = (factor < 256) ? 255 - factor : 0;
- mdbx_tassert(txn, factor < 256 && factor < (256 - prio));
- return prio = factor;
-}
-
-/* Spill pages from the dirty list back to disk.
- * This is intended to prevent running into MDBX_TXN_FULL situations,
- * but note that they may still occur in a few cases:
- *
- * 1) our estimate of the txn size could be too small. Currently this
- * seems unlikely, except with a large number of MDBX_MULTIPLE items.
- *
- * 2) child txns may run out of space if their parents dirtied a
- * lot of pages and never spilled them. TODO: we probably should do
- * a preemptive spill during mdbx_txn_begin() of a child txn, if
- * the parent's dirtyroom is below a given threshold.
- *
- * Otherwise, if not using nested txns, it is expected that apps will
- * not run into MDBX_TXN_FULL any more. The pages are flushed to disk
- * the same way as for a txn commit, e.g. their dirty status is cleared.
- * If the txn never references them again, they can be left alone.
- * If the txn only reads them, they can be used without any fuss.
- * If the txn writes them again, they can be dirtied immediately without
- * going thru all of the work of mdbx_page_touch(). Such references are
- * handled by mdbx_page_unspill().
- *
- * Also note, we never spill DB root pages, nor pages of active cursors,
- * because we'll need these back again soon anyway. And in nested txns,
- * we can't spill a page in a child txn if it was already spilled in a
- * parent txn. That would alter the parent txns' data even though
- * the child hasn't committed yet, and we'd have no way to undo it if
- * the child aborted. */
-static int mdbx_txn_spill(MDBX_txn *const txn, MDBX_cursor *const m0,
- const unsigned need) {
-#if xMDBX_DEBUG_SPILLING != 1
- /* production mode */
- if (likely(txn->tw.dirtyroom + txn->tw.loose_count >= need))
- return MDBX_SUCCESS;
- unsigned wanna_spill = need - txn->tw.dirtyroom;
-#else
- /* debug mode: spill at least one page if xMDBX_DEBUG_SPILLING == 1 */
- unsigned wanna_spill =
- (need > txn->tw.dirtyroom) ? need - txn->tw.dirtyroom : 1;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- const unsigned dirty = txn->tw.dirtylist->length;
- const unsigned spill_min =
- txn->mt_env->me_options.spill_min_denominator
- ? dirty / txn->mt_env->me_options.spill_min_denominator
- : 0;
- const unsigned spill_max =
- dirty - (txn->mt_env->me_options.spill_max_denominator
- ? dirty / txn->mt_env->me_options.spill_max_denominator
- : 0);
- wanna_spill = (wanna_spill > spill_min) ? wanna_spill : spill_min;
- wanna_spill = (wanna_spill < spill_max) ? wanna_spill : spill_max;
- if (!wanna_spill)
- return MDBX_SUCCESS;
-
- mdbx_notice("spilling %u dirty-entries (have %u dirty-room, need %u)",
- wanna_spill, txn->tw.dirtyroom, need);
- mdbx_tassert(txn, txn->tw.dirtylist->length >= wanna_spill);
-
- struct mdbx_iov_ctx ctx;
- mdbx_iov_init(txn, &ctx);
- int rc = MDBX_SUCCESS;
- if (txn->mt_flags & MDBX_WRITEMAP) {
- MDBX_dpl *const dl = txn->tw.dirtylist;
- const unsigned span = dl->length - txn->tw.loose_count;
- txn->tw.dirtyroom += span;
- unsigned r, w;
- for (w = 0, r = 1; r <= dl->length; ++r) {
- MDBX_page *dp = dl->items[r].ptr;
- if (dp->mp_flags & P_LOOSE)
- dl->items[++w] = dl->items[r];
- else if (!MDBX_FAKE_SPILL_WRITEMAP) {
- rc = iov_page(txn, &ctx, dp, dpl_npages(dl, r));
- mdbx_tassert(txn, rc == MDBX_SUCCESS);
- }
- }
-
- mdbx_tassert(txn, span == r - 1 - w && w == txn->tw.loose_count);
- dl->sorted = (dl->sorted == dl->length) ? w : 0;
- dpl_setlen(dl, w);
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
-
- if (!MDBX_FAKE_SPILL_WRITEMAP && ctx.flush_end > ctx.flush_begin) {
- MDBX_env *const env = txn->mt_env;
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- rc = mdbx_msync(&env->me_dxb_mmap,
- pgno_align2os_bytes(env, ctx.flush_begin),
- pgno_align2os_bytes(env, ctx.flush_end - ctx.flush_begin),
- MDBX_SYNC_NONE);
- }
- return rc;
- }
-
- mdbx_tassert(txn, !(txn->mt_flags & MDBX_WRITEMAP));
- if (!txn->tw.spill_pages) {
- txn->tw.spill_least_removed = INT_MAX;
- txn->tw.spill_pages = mdbx_pnl_alloc(wanna_spill);
- if (unlikely(!txn->tw.spill_pages)) {
- rc = MDBX_ENOMEM;
- bailout:
- txn->mt_flags |= MDBX_TXN_ERROR;
- return rc;
- }
- } else {
- /* purge deleted slots */
- mdbx_spill_purge(txn);
- rc = mdbx_pnl_reserve(&txn->tw.spill_pages, wanna_spill);
- (void)rc /* ignore since the resulting list may be shorter
- and mdbx_pnl_append() will increase pnl on demand */
- ;
- }
-
- /* Сортируем чтобы запиÑÑŒ на диÑк была полее поÑледовательна */
- MDBX_dpl *const dl = mdbx_dpl_sort(txn);
-
- /* Preserve pages which may soon be dirtied again */
- const unsigned unspillable = mdbx_txn_keep(txn, m0);
- if (unspillable + txn->tw.loose_count >= dl->length) {
-#if xMDBX_DEBUG_SPILLING == 1 /* avoid false failure in debug mode */
- if (likely(txn->tw.dirtyroom + txn->tw.loose_count >= need))
- return MDBX_SUCCESS;
-#endif /* xMDBX_DEBUG_SPILLING */
- mdbx_error("all %u dirty pages are unspillable since referenced "
- "by a cursor(s), use fewer cursors or increase "
- "MDBX_opt_txn_dp_limit",
- unspillable);
- goto done;
- }
-
- /* Подзадача: Вытолкнуть чаÑÑ‚ÑŒ Ñтраниц на диÑк в ÑоответÑтвии Ñ LRU,
- * но при Ñтом учеÑÑ‚ÑŒ важные поправки:
- * - лучше выталкивать Ñтарые large/overflow Ñтраницы, так будет оÑвобождено
- * больше памÑти, а также так как они (в текущем понимании) гораздо реже
- * повторно изменÑÑŽÑ‚ÑÑ;
- * - при прочих равных лучше выталкивать Ñмежные Ñтраницы, так будет
- * меньше I/O операций;
- * - желательно потратить на Ñто меньше времени чем std::partial_sort_copy;
- *
- * Решение:
- * - Квантуем веÑÑŒ диапазон lru-меток до 256 значений и задейÑтвуем один
- * проход 8-битного radix-sort. В результате получаем 256 уровней
- * "ÑвежеÑти", в том чиÑле значение lru-метки, Ñтарее которой Ñтраницы
- * должны быть выгружены;
- * - ДвигаемÑÑ Ð¿Ð¾Ñледовательно в Ñторону ÑƒÐ²ÐµÐ»Ð¸Ñ‡ÐµÐ½Ð¸Ñ Ð½Ð¾Ð¼ÐµÑ€Ð¾Ð² Ñтраниц
- * и выталкиваем Ñтраницы Ñ lru-меткой Ñтарее отÑекающего значениÑ,
- * пока не вытолкнем доÑтаточно;
- * - Ð’ÑÑ‚Ñ€ÐµÑ‡Ð°Ñ Ñтраницы Ñмежные Ñ Ð²Ñ‹Ñ‚Ð°Ð»ÐºÐ¸Ð²Ð°ÐµÐ¼Ñ‹Ð¼Ð¸ Ð´Ð»Ñ ÑƒÐ¼ÐµÐ½ÑŒÑˆÐµÐ½Ð¸Ñ ÐºÐ¾Ð»-ва
- * I/O операций выталкиваем и их, еÑли они попадают в первую половину
- * между выталкиваемыми и Ñамыми Ñвежими lru-метками;
- * - дополнительно при Ñортировке умышленно Ñтарим large/overflow Ñтраницы,
- * тем Ñамым Ð¿Ð¾Ð²Ñ‹ÑˆÐ°Ñ Ð¸Ñ… шанÑÑ‹ на выталкивание. */
-
- /* get min/max of LRU-labels */
- uint32_t age_max = 0;
- for (unsigned i = 1; i <= dl->length; ++i) {
- const uint32_t age = mdbx_dpl_age(txn, i);
- age_max = (age_max >= age) ? age_max : age;
- }
-
- mdbx_verbose("lru-head %u, age-max %u", txn->tw.dirtylru, age_max);
-
- /* half of 8-bit radix-sort */
- unsigned radix_counters[256], spillable = 0, spilled = 0;
- memset(&radix_counters, 0, sizeof(radix_counters));
- const uint32_t reciprocal = (UINT32_C(255) << 24) / (age_max + 1);
- for (unsigned i = 1; i <= dl->length; ++i) {
- unsigned prio = spill_prio(txn, i, reciprocal);
- if (prio < 256) {
- radix_counters[prio] += 1;
- spillable += 1;
- }
- }
-
- if (likely(spillable > 0)) {
- unsigned prio2spill = 0, prio2adjacent = 128, amount = radix_counters[0];
- for (unsigned i = 1; i < 256; i++) {
- if (amount < wanna_spill) {
- prio2spill = i;
- prio2adjacent = i + (257 - i) / 2;
- amount += radix_counters[i];
- } else if (amount + amount < spillable + wanna_spill
- /* Ð ÐÐ’ÐОЗÐÐЧÐО: amount - wanna_spill < spillable - amount */) {
- prio2adjacent = i;
- amount += radix_counters[i];
- } else
- break;
- }
-
- mdbx_verbose("prio2spill %u, prio2adjacent %u, amount %u, spillable %u, "
- "wanna_spill %u",
- prio2spill, prio2adjacent, amount, spillable, wanna_spill);
- mdbx_tassert(txn, prio2spill < prio2adjacent && prio2adjacent <= 256);
-
- unsigned prev_prio = 256;
- unsigned r, w, prio;
- for (w = 0, r = 1; r <= dl->length && spilled < wanna_spill;
- prev_prio = prio, ++r) {
- prio = spill_prio(txn, r, reciprocal);
- MDBX_page *const dp = dl->items[r].ptr;
- if (prio < prio2adjacent) {
- const pgno_t pgno = dl->items[r].pgno;
- const unsigned npages = dpl_npages(dl, r);
- if (prio <= prio2spill) {
- if (prev_prio < prio2adjacent && prev_prio > prio2spill &&
- dpl_endpgno(dl, r - 1) == pgno) {
- mdbx_debug("co-spill %u prev-adjacent page %" PRIaPGNO
- " (age %d, prio %u)",
- dpl_npages(dl, w), dl->items[r - 1].pgno,
- mdbx_dpl_age(txn, r - 1), prev_prio);
- --w;
- rc = spill_page(txn, &ctx, dl->items[r - 1].ptr,
- dpl_npages(dl, r - 1));
- if (unlikely(rc != MDBX_SUCCESS))
- break;
- ++spilled;
- }
-
- mdbx_debug("spill %u page %" PRIaPGNO " (age %d, prio %u)", npages,
- dp->mp_pgno, mdbx_dpl_age(txn, r), prio);
- rc = spill_page(txn, &ctx, dp, npages);
- if (unlikely(rc != MDBX_SUCCESS))
- break;
- ++spilled;
- continue;
- }
-
- if (prev_prio <= prio2spill && dpl_endpgno(dl, r - 1) == pgno) {
- mdbx_debug("co-spill %u next-adjacent page %" PRIaPGNO
- " (age %d, prio %u)",
- npages, dp->mp_pgno, mdbx_dpl_age(txn, r), prio);
- rc = spill_page(txn, &ctx, dp, npages);
- if (unlikely(rc != MDBX_SUCCESS))
- break;
- prio = prev_prio /* to continue co-spilling next adjacent pages */;
- ++spilled;
- continue;
- }
- }
- dl->items[++w] = dl->items[r];
- }
-
- mdbx_tassert(txn, spillable == 0 || spilled > 0);
-
- while (r <= dl->length)
- dl->items[++w] = dl->items[r++];
- mdbx_tassert(txn, r - 1 - w == spilled);
-
- dl->sorted = dpl_setlen(dl, w);
- txn->tw.dirtyroom += spilled;
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
-
- if (ctx.iov_items)
- rc = mdbx_iov_write(txn, &ctx);
-
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
-
- mdbx_pnl_sort(txn->tw.spill_pages, (size_t)txn->mt_next_pgno << 1);
- txn->mt_flags |= MDBX_TXN_SPILLS;
- mdbx_notice("spilled %u dirty-entries, now have %u dirty-room", spilled,
- txn->tw.dirtyroom);
- mdbx_iov_done(txn, &ctx);
- } else {
- mdbx_tassert(txn, ctx.iov_items == 0 && rc == MDBX_SUCCESS);
- for (unsigned i = 1; i <= dl->length; ++i) {
- MDBX_page *dp = dl->items[i].ptr;
- mdbx_notice(
- "dirtylist[%u]: pgno %u, npages %u, flags 0x%04X, age %u, prio %u", i,
- dp->mp_pgno, dpl_npages(dl, i), dp->mp_flags, mdbx_dpl_age(txn, i),
- spill_prio(txn, i, reciprocal));
- }
- }
-
-#if xMDBX_DEBUG_SPILLING == 2
- if (txn->tw.loose_count + txn->tw.dirtyroom <= need / 2 + 1)
- mdbx_error("dirty-list length: before %u, after %u, parent %i, loose %u; "
- "needed %u, spillable %u; "
- "spilled %u dirty-entries, now have %u dirty-room",
- dl->length + spilled, dl->length,
- (txn->mt_parent && txn->mt_parent->tw.dirtylist)
- ? (int)txn->mt_parent->tw.dirtylist->length
- : -1,
- txn->tw.loose_count, need, spillable, spilled,
- txn->tw.dirtyroom);
- mdbx_ensure(txn->mt_env, txn->tw.loose_count + txn->tw.dirtyroom > need / 2);
-#endif /* xMDBX_DEBUG_SPILLING */
-
-done:
- return likely(txn->tw.dirtyroom + txn->tw.loose_count >
- ((need > CURSOR_STACK) ? CURSOR_STACK : need))
- ? MDBX_SUCCESS
- : MDBX_TXN_FULL;
-}
-
-static int mdbx_cursor_spill(MDBX_cursor *mc, const MDBX_val *key,
- const MDBX_val *data) {
- MDBX_txn *txn = mc->mc_txn;
- /* Estimate how much space this operation will take: */
- /* 1) Max b-tree height, reasonable enough with including dups' sub-tree */
- unsigned need = CURSOR_STACK + 3;
- /* 2) GC/FreeDB for any payload */
- if (mc->mc_dbi > FREE_DBI) {
- need += txn->mt_dbs[FREE_DBI].md_depth + 3;
- /* 3) Named DBs also dirty the main DB */
- if (mc->mc_dbi > MAIN_DBI)
- need += txn->mt_dbs[MAIN_DBI].md_depth + 3;
- }
-#if xMDBX_DEBUG_SPILLING != 2
- /* production mode */
- /* 4) Double the page chain estimation
- * for extensively splitting, rebalance and merging */
- need += need;
- /* 5) Factor the key+data which to be put in */
- need += bytes2pgno(txn->mt_env, node_size(key, data)) + 1;
-#else
- /* debug mode */
- (void)key;
- (void)data;
- mc->mc_txn->mt_env->debug_dirtied_est = ++need;
- mc->mc_txn->mt_env->debug_dirtied_act = 0;
-#endif /* xMDBX_DEBUG_SPILLING == 2 */
-
- return mdbx_txn_spill(txn, mc, need);
-}
-
-/*----------------------------------------------------------------------------*/
-
-static bool meta_bootid_match(const MDBX_meta *meta) {
- return memcmp(&meta->mm_bootid, &bootid, 16) == 0 &&
- (bootid.x | bootid.y) != 0;
-}
-
-static bool meta_weak_acceptable(const MDBX_env *env, const MDBX_meta *meta,
- const int lck_exclusive) {
- return lck_exclusive
- ? /* exclusive lock */ meta_bootid_match(meta)
- : /* db already opened */ env->me_lck_mmap.lck &&
- (env->me_lck_mmap.lck->mti_envmode.weak & MDBX_RDONLY) == 0;
-}
-
-#define METAPAGE(env, n) page_meta(pgno2page(env, n))
-#define METAPAGE_END(env) METAPAGE(env, NUM_METAS)
-
-MDBX_NOTHROW_PURE_FUNCTION static __inline txnid_t
-constmeta_txnid(const MDBX_env *env, const MDBX_meta *meta) {
- mdbx_memory_fence(mo_AcquireRelease, false);
- txnid_t a = unaligned_peek_u64(4, &meta->mm_txnid_a);
- txnid_t b = unaligned_peek_u64(4, &meta->mm_txnid_b);
- mdbx_assert(env, a == b);
- (void)env;
- return (a == b) ? a : 0;
-}
-
-static __inline txnid_t meta_txnid(const MDBX_env *env,
- volatile const MDBX_meta *meta) {
- (void)env;
- mdbx_memory_fence(mo_AcquireRelease, false);
- txnid_t a = unaligned_peek_u64_volatile(4, &meta->mm_txnid_a);
- txnid_t b = unaligned_peek_u64_volatile(4, &meta->mm_txnid_b);
- return (a == b) ? a : 0;
-}
-
-static __inline void meta_update_begin(const MDBX_env *env, MDBX_meta *meta,
- txnid_t txnid) {
- mdbx_assert(env, meta >= METAPAGE(env, 0) && meta < METAPAGE_END(env));
- mdbx_assert(env, unaligned_peek_u64(4, meta->mm_txnid_a) < txnid &&
- unaligned_peek_u64(4, meta->mm_txnid_b) < txnid);
- (void)env;
- unaligned_poke_u64(4, meta->mm_txnid_b, 0);
- mdbx_memory_fence(mo_AcquireRelease, true);
- unaligned_poke_u64(4, meta->mm_txnid_a, txnid);
-}
-
-static __inline void meta_update_end(const MDBX_env *env, MDBX_meta *meta,
- txnid_t txnid) {
- mdbx_assert(env, meta >= METAPAGE(env, 0) && meta < METAPAGE_END(env));
- mdbx_assert(env, unaligned_peek_u64(4, meta->mm_txnid_a) == txnid);
- mdbx_assert(env, unaligned_peek_u64(4, meta->mm_txnid_b) < txnid);
- (void)env;
- mdbx_jitter4testing(true);
- memcpy(&meta->mm_bootid, &bootid, 16);
- unaligned_poke_u64(4, meta->mm_txnid_b, txnid);
- mdbx_memory_fence(mo_AcquireRelease, true);
-}
-
-static __inline void meta_set_txnid(const MDBX_env *env, MDBX_meta *meta,
- const txnid_t txnid) {
- mdbx_assert(env, !env->me_map || meta < METAPAGE(env, 0) ||
- meta >= METAPAGE_END(env));
- (void)env;
- /* update inconsistently since this function used ONLY for filling meta-image
- * for writing, but not the actual meta-page */
- memcpy(&meta->mm_bootid, &bootid, 16);
- unaligned_poke_u64(4, meta->mm_txnid_a, txnid);
- unaligned_poke_u64(4, meta->mm_txnid_b, txnid);
-}
-
-static __inline uint64_t meta_sign(const MDBX_meta *meta) {
- uint64_t sign = MDBX_DATASIGN_NONE;
-#if 0 /* TODO */
- sign = hippeus_hash64(...);
-#else
- (void)meta;
-#endif
- /* LY: newer returns MDBX_DATASIGN_NONE or MDBX_DATASIGN_WEAK */
- return (sign > MDBX_DATASIGN_WEAK) ? sign : ~sign;
-}
-
-enum meta_choise_mode { prefer_last, prefer_steady };
-
-static __inline bool meta_ot(const enum meta_choise_mode mode,
- const MDBX_env *env, volatile const MDBX_meta *a,
- volatile const MDBX_meta *b) {
- mdbx_jitter4testing(true);
- const txnid_t txnid_a = meta_txnid(env, a);
- mdbx_jitter4testing(true);
- const txnid_t txnid_b = meta_txnid(env, b);
- mdbx_jitter4testing(true);
- const bool is_stead_b = META_IS_STEADY(b);
-
- if (mode == prefer_steady) {
- mdbx_jitter4testing(true);
- const bool is_stead_a = META_IS_STEADY(a);
- if (is_stead_a != is_stead_b)
- return is_stead_b;
- } else {
- mdbx_assert(env, mode == prefer_last);
- }
- if (txnid_a == txnid_b)
- return is_stead_b;
- return txnid_a < txnid_b;
-}
-
-static bool meta_eq(const MDBX_env *env, volatile const MDBX_meta *a,
- volatile const MDBX_meta *b) {
- mdbx_jitter4testing(true);
- const txnid_t txnid = meta_txnid(env, a);
- if (!txnid || txnid != meta_txnid(env, b))
- return false;
-
- mdbx_jitter4testing(true);
- if (META_IS_STEADY(a) != META_IS_STEADY(b))
- return false;
-
- mdbx_jitter4testing(true);
- return true;
-}
-
-static int meta_eq_mask(const MDBX_env *env) {
- volatile const MDBX_meta *m0 = METAPAGE(env, 0);
- volatile const MDBX_meta *m1 = METAPAGE(env, 1);
- volatile const MDBX_meta *m2 = METAPAGE(env, 2);
-
- int rc = meta_eq(env, m0, m1) ? 1 : 0;
- if (meta_eq(env, m1, m2))
- rc += 2;
- if (meta_eq(env, m2, m0))
- rc += 4;
- return rc;
-}
-
-static __inline volatile const MDBX_meta *
-meta_recent(const enum meta_choise_mode mode, const MDBX_env *env,
- volatile const MDBX_meta *a, volatile const MDBX_meta *b) {
- const bool a_older_that_b = meta_ot(mode, env, a, b);
- mdbx_assert(env, !meta_eq(env, a, b));
- return a_older_that_b ? b : a;
-}
-
-static const MDBX_meta *meta_ancient_prefer_weak(const MDBX_env *env,
- const MDBX_meta *a,
- const MDBX_meta *b) {
- const bool a_older_that_b = meta_ot(prefer_steady, env, a, b);
- mdbx_assert(env, !meta_eq(env, a, b));
- return a_older_that_b ? a : b;
-}
-
-static __inline volatile const MDBX_meta *
-meta_mostrecent(const enum meta_choise_mode mode, const MDBX_env *env) {
- volatile const MDBX_meta *m0 = METAPAGE(env, 0);
- volatile const MDBX_meta *m1 = METAPAGE(env, 1);
- volatile const MDBX_meta *m2 = METAPAGE(env, 2);
-
- volatile const MDBX_meta *head = meta_recent(mode, env, m0, m1);
- head = meta_recent(mode, env, head, m2);
- return head;
-}
-
-static volatile const MDBX_meta *meta_prefer_steady(const MDBX_env *env) {
- return meta_mostrecent(prefer_steady, env);
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static const MDBX_meta *
-constmeta_prefer_steady(const MDBX_env *env) {
- return (const MDBX_meta *)meta_mostrecent(prefer_steady, env);
-}
-
-static volatile const MDBX_meta *meta_prefer_last(const MDBX_env *env) {
- return meta_mostrecent(prefer_last, env);
-}
-
-MDBX_NOTHROW_PURE_FUNCTION static const MDBX_meta *
-constmeta_prefer_last(const MDBX_env *env) {
- return (const MDBX_meta *)meta_mostrecent(prefer_last, env);
-}
-
-static txnid_t mdbx_recent_committed_txnid(const MDBX_env *env) {
- while (true) {
- volatile const MDBX_meta *head = meta_prefer_last(env);
- const txnid_t recent = meta_txnid(env, head);
- mdbx_compiler_barrier();
- if (likely(head == meta_prefer_last(env) &&
- recent == meta_txnid(env, head)))
- return recent;
- }
-}
-
-static txnid_t mdbx_recent_steady_txnid(const MDBX_env *env) {
- while (true) {
- volatile const MDBX_meta *head = meta_prefer_steady(env);
- const txnid_t recent = meta_txnid(env, head);
- mdbx_compiler_barrier();
- if (likely(head == meta_prefer_steady(env) &&
- recent == meta_txnid(env, head)))
- return recent;
- }
-}
-
-static const char *mdbx_durable_str(volatile const MDBX_meta *const meta) {
- if (META_IS_STEADY(meta))
- return (unaligned_peek_u64_volatile(4, meta->mm_datasync_sign) ==
- meta_sign((const MDBX_meta *)meta))
- ? "Steady"
- : "Tainted";
- return "Weak";
-}
-
-/*----------------------------------------------------------------------------*/
-
-/* Find oldest txnid still referenced. */
-static txnid_t mdbx_find_oldest(const MDBX_txn *txn) {
- mdbx_tassert(txn, (txn->mt_flags & MDBX_TXN_RDONLY) == 0);
- MDBX_env *env = txn->mt_env;
- const txnid_t edge = mdbx_recent_steady_txnid(env);
- mdbx_tassert(txn, edge <= txn->mt_txnid);
-
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (unlikely(lck == NULL /* exclusive mode */))
- return atomic_store64(&lck->mti_oldest_reader, edge, mo_Relaxed);
-
- const txnid_t last_oldest =
- atomic_load64(&lck->mti_oldest_reader, mo_AcquireRelease);
- mdbx_tassert(txn, edge >= last_oldest);
- if (likely(last_oldest == edge))
- return edge;
-
- const uint32_t nothing_changed = MDBX_STRING_TETRAD("None");
- const uint32_t snap_readers_refresh_flag =
- atomic_load32(&lck->mti_readers_refresh_flag, mo_AcquireRelease);
- mdbx_jitter4testing(false);
- if (snap_readers_refresh_flag == nothing_changed)
- return last_oldest;
-
- txnid_t oldest = edge;
- atomic_store32(&lck->mti_readers_refresh_flag, nothing_changed, mo_Relaxed);
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- for (unsigned i = 0; i < snap_nreaders; ++i) {
- if (atomic_load32(&lck->mti_readers[i].mr_pid, mo_AcquireRelease)) {
- /* mdbx_jitter4testing(true); */
- const txnid_t snap = safe64_read(&lck->mti_readers[i].mr_txnid);
- if (oldest > snap && last_oldest <= /* ignore pending updates */ snap) {
- oldest = snap;
- if (oldest == last_oldest)
- return oldest;
- }
- }
- }
-
- if (oldest != last_oldest) {
- mdbx_verbose("update oldest %" PRIaTXN " -> %" PRIaTXN, last_oldest,
- oldest);
- mdbx_tassert(txn, oldest >= lck->mti_oldest_reader.weak);
- atomic_store64(&lck->mti_oldest_reader, oldest, mo_Relaxed);
- }
- return oldest;
-}
-
-/* Find largest mvcc-snapshot still referenced. */
-__cold static pgno_t mdbx_find_largest(MDBX_env *env, pgno_t largest) {
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (likely(lck != NULL /* exclusive mode */)) {
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- for (unsigned i = 0; i < snap_nreaders; ++i) {
- retry:
- if (atomic_load32(&lck->mti_readers[i].mr_pid, mo_AcquireRelease)) {
- /* mdbx_jitter4testing(true); */
- const pgno_t snap_pages = atomic_load32(
- &lck->mti_readers[i].mr_snapshot_pages_used, mo_Relaxed);
- const txnid_t snap_txnid = safe64_read(&lck->mti_readers[i].mr_txnid);
- if (unlikely(
- snap_pages !=
- atomic_load32(&lck->mti_readers[i].mr_snapshot_pages_used,
- mo_AcquireRelease) ||
- snap_txnid != safe64_read(&lck->mti_readers[i].mr_txnid)))
- goto retry;
- if (largest < snap_pages &&
- atomic_load64(&lck->mti_oldest_reader, mo_AcquireRelease) <=
- /* ignore pending updates */ snap_txnid &&
- snap_txnid <= env->me_txn0->mt_txnid)
- largest = snap_pages;
- }
- }
- }
-
- return largest;
-}
-
-/* Add a page to the txn's dirty list */
-static int __must_check_result mdbx_page_dirty(MDBX_txn *txn, MDBX_page *mp,
- unsigned npages) {
-#if xMDBX_DEBUG_SPILLING == 2
- txn->mt_env->debug_dirtied_act += 1;
- mdbx_ensure(txn->mt_env,
- txn->mt_env->debug_dirtied_act < txn->mt_env->debug_dirtied_est);
- mdbx_ensure(txn->mt_env, txn->tw.dirtyroom + txn->tw.loose_count > 0);
-#endif /* xMDBX_DEBUG_SPILLING == 2 */
-
- int rc;
- mp->mp_txnid = txn->mt_front;
- if (unlikely(txn->tw.dirtyroom == 0)) {
- if (txn->tw.loose_count) {
- MDBX_page *loose = txn->tw.loose_pages;
- mdbx_debug("purge-and-reclaim loose page %" PRIaPGNO, loose->mp_pgno);
- rc = mdbx_pnl_insert_range(&txn->tw.reclaimed_pglist, loose->mp_pgno, 1);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- unsigned di = mdbx_dpl_search(txn, loose->mp_pgno);
- mdbx_tassert(txn, txn->tw.dirtylist->items[di].ptr == loose);
- mdbx_dpl_remove(txn, di);
- txn->tw.loose_pages = loose->mp_next;
- txn->tw.loose_count--;
- txn->tw.dirtyroom++;
- if (!(txn->mt_flags & MDBX_WRITEMAP))
- mdbx_dpage_free(txn->mt_env, loose, 1);
- } else {
- mdbx_error("Dirtyroom is depleted, DPL length %u",
- txn->tw.dirtylist->length);
- if (!(txn->mt_flags & MDBX_WRITEMAP))
- mdbx_dpage_free(txn->mt_env, mp, npages);
- return MDBX_TXN_FULL;
- }
- }
-
- rc = mdbx_dpl_append(txn, mp->mp_pgno, mp, npages);
- if (unlikely(rc != MDBX_SUCCESS)) {
- bailout:
- txn->mt_flags |= MDBX_TXN_ERROR;
- return rc;
- }
- txn->tw.dirtyroom--;
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- return MDBX_SUCCESS;
-}
-
-#if !(defined(_WIN32) || defined(_WIN64))
-MDBX_MAYBE_UNUSED static __always_inline int ignore_enosys(int err) {
-#ifdef ENOSYS
- if (err == ENOSYS)
- return MDBX_RESULT_TRUE;
-#endif /* ENOSYS */
-#ifdef ENOIMPL
- if (err == ENOIMPL)
- return MDBX_RESULT_TRUE;
-#endif /* ENOIMPL */
-#ifdef ENOTSUP
- if (err == ENOTSUP)
- return MDBX_RESULT_TRUE;
-#endif /* ENOTSUP */
-#ifdef ENOSUPP
- if (err == ENOSUPP)
- return MDBX_RESULT_TRUE;
-#endif /* ENOSUPP */
-#ifdef EOPNOTSUPP
- if (err == EOPNOTSUPP)
- return MDBX_RESULT_TRUE;
-#endif /* EOPNOTSUPP */
- if (err == EAGAIN)
- return MDBX_RESULT_TRUE;
- return err;
-}
-#endif /* defined(_WIN32) || defined(_WIN64) */
-
-#if MDBX_ENABLE_MADVISE
-/* Turn on/off readahead. It's harmful when the DB is larger than RAM. */
-__cold static int mdbx_set_readahead(MDBX_env *env, const pgno_t edge,
- const bool enable,
- const bool force_whole) {
- mdbx_assert(env, edge >= NUM_METAS && edge <= MAX_PAGENO + 1);
- mdbx_assert(env, (enable & 1) == (enable != 0));
- const bool toggle = force_whole ||
- ((enable ^ env->me_lck->mti_readahead_anchor) & 1) ||
- !env->me_lck->mti_readahead_anchor;
- const pgno_t prev_edge = env->me_lck->mti_readahead_anchor >> 1;
- const size_t limit = env->me_dxb_mmap.limit;
- size_t offset =
- toggle ? 0
- : pgno_align2os_bytes(env, (prev_edge < edge) ? prev_edge : edge);
- offset = (offset < limit) ? offset : limit;
-
- size_t length =
- pgno_align2os_bytes(env, (prev_edge < edge) ? edge : prev_edge);
- length = (length < limit) ? length : limit;
- length -= offset;
-
- mdbx_assert(env, 0 <= (intptr_t)length);
- if (length == 0)
- return MDBX_SUCCESS;
-
- mdbx_notice("readahead %s %u..%u", enable ? "ON" : "OFF",
- bytes2pgno(env, offset), bytes2pgno(env, offset + length));
-
-#if defined(F_RDAHEAD)
- if (toggle && unlikely(fcntl(env->me_lazy_fd, F_RDAHEAD, enable) == -1))
- return errno;
-#endif /* F_RDAHEAD */
-
- int err;
- if (enable) {
-#if defined(MADV_NORMAL)
- err = madvise(env->me_map + offset, length, MADV_NORMAL)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(POSIX_MADV_NORMAL)
- err = ignore_enosys(
- posix_madvise(env->me_map + offset, length, POSIX_MADV_NORMAL));
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(POSIX_FADV_NORMAL) && defined(POSIX_FADV_WILLNEED)
- err = ignore_enosys(
- posix_fadvise(env->me_lazy_fd, offset, length, POSIX_FADV_NORMAL));
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(_WIN32) || defined(_WIN64)
- /* no madvise on Windows */
-#else
-#warning "FIXME"
-#endif
- if (toggle) {
- /* NOTE: Seems there is a bug in the Mach/Darwin/OSX kernel,
- * because MADV_WILLNEED with offset != 0 may cause SIGBUS
- * on following access to the hinted region.
- * 19.6.0 Darwin Kernel Version 19.6.0: Tue Jan 12 22:13:05 PST 2021;
- * root:xnu-6153.141.16~1/RELEASE_X86_64 x86_64 */
-#if defined(F_RDADVISE)
- struct radvisory hint;
- hint.ra_offset = offset;
- hint.ra_count = length;
- (void)/* Ignore ENOTTY for DB on the ram-disk and so on */ fcntl(
- env->me_lazy_fd, F_RDADVISE, &hint);
-#elif defined(MADV_WILLNEED)
- err = madvise(env->me_map + offset, length, MADV_WILLNEED)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(POSIX_MADV_WILLNEED)
- err = ignore_enosys(
- posix_madvise(env->me_map + offset, length, POSIX_MADV_WILLNEED));
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(_WIN32) || defined(_WIN64)
- if (mdbx_PrefetchVirtualMemory) {
- WIN32_MEMORY_RANGE_ENTRY hint;
- hint.VirtualAddress = env->me_map + offset;
- hint.NumberOfBytes = length;
- (void)mdbx_PrefetchVirtualMemory(GetCurrentProcess(), 1, &hint, 0);
- }
-#elif defined(POSIX_FADV_WILLNEED)
- err = ignore_enosys(
- posix_fadvise(env->me_lazy_fd, offset, length, POSIX_FADV_WILLNEED));
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#else
-#warning "FIXME"
-#endif
- }
- } else {
-#if defined(MADV_RANDOM)
- err = madvise(env->me_map + offset, length, MADV_RANDOM)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(POSIX_MADV_RANDOM)
- err = ignore_enosys(
- posix_madvise(env->me_map + offset, length, POSIX_MADV_RANDOM));
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(POSIX_FADV_RANDOM)
- err = ignore_enosys(
- posix_fadvise(env->me_lazy_fd, offset, length, POSIX_FADV_RANDOM));
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(_WIN32) || defined(_WIN64)
- /* no madvise on Windows */
-#else
-#warning "FIXME"
-#endif /* MADV_RANDOM */
- }
-
- env->me_lck->mti_readahead_anchor = (enable & 1) + (edge << 1);
- err = MDBX_SUCCESS;
- return err;
-}
-#endif /* MDBX_ENABLE_MADVISE */
-
-__cold static int mdbx_mapresize(MDBX_env *env, const pgno_t used_pgno,
- const pgno_t size_pgno,
- const pgno_t limit_pgno, const bool implicit) {
- const size_t limit_bytes = pgno_align2os_bytes(env, limit_pgno);
- const size_t size_bytes = pgno_align2os_bytes(env, size_pgno);
- const size_t prev_size = env->me_dxb_mmap.current;
- const size_t prev_limit = env->me_dxb_mmap.limit;
-#if MDBX_ENABLE_MADVISE || defined(MDBX_USE_VALGRIND)
- const void *const prev_addr = env->me_map;
-#endif /* MDBX_ENABLE_MADVISE || MDBX_USE_VALGRIND */
-
- mdbx_verbose("resize datafile/mapping: "
- "present %" PRIuPTR " -> %" PRIuPTR ", "
- "limit %" PRIuPTR " -> %" PRIuPTR,
- prev_size, size_bytes, prev_limit, limit_bytes);
-
- mdbx_assert(env, limit_bytes >= size_bytes);
- mdbx_assert(env, bytes2pgno(env, size_bytes) >= size_pgno);
- mdbx_assert(env, bytes2pgno(env, limit_bytes) >= limit_pgno);
-
- unsigned mresize_flags =
- env->me_flags & (MDBX_RDONLY | MDBX_WRITEMAP | MDBX_UTTERLY_NOSYNC);
-#if defined(_WIN32) || defined(_WIN64)
- /* Acquire guard in exclusive mode for:
- * - to avoid collision between read and write txns around env->me_dbgeo;
- * - to avoid attachment of new reading threads (see mdbx_rdt_lock); */
- mdbx_srwlock_AcquireExclusive(&env->me_remap_guard);
- mdbx_handle_array_t *suspended = NULL;
- mdbx_handle_array_t array_onstack;
- int rc = MDBX_SUCCESS;
- if (limit_bytes == env->me_dxb_mmap.limit &&
- size_bytes == env->me_dxb_mmap.current &&
- size_bytes == env->me_dxb_mmap.filesize)
- goto bailout;
-
- if ((env->me_flags & MDBX_NOTLS) == 0) {
- /* 1) Windows allows only extending a read-write section, but not a
- * corresponding mapped view. Therefore in other cases we must suspend
- * the local threads for safe remap.
- * 2) At least on Windows 10 1803 the entire mapped section is unavailable
- * for short time during NtExtendSection() or VirtualAlloc() execution.
- * 3) Under Wine runtime environment on Linux a section extending is not
- * supported.
- *
- * THEREFORE LOCAL THREADS SUSPENDING IS ALWAYS REQUIRED! */
- array_onstack.limit = ARRAY_LENGTH(array_onstack.handles);
- array_onstack.count = 0;
- suspended = &array_onstack;
- rc = mdbx_suspend_threads_before_remap(env, &suspended);
- if (rc != MDBX_SUCCESS) {
- mdbx_error("failed suspend-for-remap: errcode %d", rc);
- goto bailout;
- }
- mresize_flags |= implicit ? MDBX_MRESIZE_MAY_UNMAP
- : MDBX_MRESIZE_MAY_UNMAP | MDBX_MRESIZE_MAY_MOVE;
- }
-#else /* Windows */
- /* Acquire guard to avoid collision between read and write txns
- * around env->me_dbgeo */
- int rc = mdbx_fastmutex_acquire(&env->me_remap_guard);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- if (limit_bytes == env->me_dxb_mmap.limit &&
- size_bytes == env->me_dxb_mmap.current)
- goto bailout;
-
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (limit_bytes != env->me_dxb_mmap.limit && !(env->me_flags & MDBX_NOTLS) &&
- lck && !implicit) {
- int err = mdbx_rdt_lock(env) /* lock readers table until remap done */;
- if (unlikely(MDBX_IS_ERROR(err))) {
- rc = err;
- goto bailout;
- }
-
- /* looking for readers from this process */
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- mdbx_assert(env, !implicit);
- mresize_flags |= MDBX_MRESIZE_MAY_UNMAP | MDBX_MRESIZE_MAY_MOVE;
- for (unsigned i = 0; i < snap_nreaders; ++i) {
- if (lck->mti_readers[i].mr_pid.weak == env->me_pid &&
- lck->mti_readers[i].mr_tid.weak != mdbx_thread_self()) {
- /* the base address of the mapping can't be changed since
- * the other reader thread from this process exists. */
- mdbx_rdt_unlock(env);
- mresize_flags &= ~(MDBX_MRESIZE_MAY_UNMAP | MDBX_MRESIZE_MAY_MOVE);
- break;
- }
- }
- }
-#endif /* ! Windows */
-
- if ((env->me_flags & MDBX_WRITEMAP) && env->me_lck->mti_unsynced_pages.weak) {
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- rc = mdbx_msync(&env->me_dxb_mmap, 0, pgno_align2os_bytes(env, used_pgno),
- MDBX_SYNC_NONE);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
-
-#if MDBX_ENABLE_MADVISE
- if (size_bytes < prev_size) {
- mdbx_notice("resize-MADV_%s %u..%u",
- (env->me_flags & MDBX_WRITEMAP) ? "REMOVE" : "DONTNEED",
- size_pgno, bytes2pgno(env, prev_size));
- rc = MDBX_RESULT_TRUE;
-#if defined(MADV_REMOVE)
- if (env->me_flags & MDBX_WRITEMAP)
- rc =
- madvise(env->me_map + size_bytes, prev_size - size_bytes, MADV_REMOVE)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
-#endif /* MADV_REMOVE */
-#if defined(MADV_DONTNEED)
- if (rc == MDBX_RESULT_TRUE)
- rc = madvise(env->me_map + size_bytes, prev_size - size_bytes,
- MADV_DONTNEED)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
-#elif defined(POSIX_MADV_DONTNEED)
- if (rc == MDBX_RESULT_TRUE)
- rc = ignore_enosys(posix_madvise(env->me_map + size_bytes,
- prev_size - size_bytes,
- POSIX_MADV_DONTNEED));
-#elif defined(POSIX_FADV_DONTNEED)
- if (rc == MDBX_RESULT_TRUE)
- rc = ignore_enosys(posix_fadvise(env->me_lazy_fd, size_bytes,
- prev_size - size_bytes,
- POSIX_FADV_DONTNEED));
-#endif /* MADV_DONTNEED */
- if (unlikely(MDBX_IS_ERROR(rc)))
- goto bailout;
- if (env->me_lck->mti_discarded_tail.weak > size_pgno)
- env->me_lck->mti_discarded_tail.weak = size_pgno;
- }
-#endif /* MDBX_ENABLE_MADVISE */
-
- rc = mdbx_mresize(mresize_flags, &env->me_dxb_mmap, size_bytes, limit_bytes);
-
-#if MDBX_ENABLE_MADVISE
- if (rc == MDBX_SUCCESS) {
- env->me_lck->mti_discarded_tail.weak = size_pgno;
- const bool readahead =
- !(env->me_flags & MDBX_NORDAHEAD) &&
- mdbx_is_readahead_reasonable(size_bytes, -(intptr_t)prev_size);
- const bool force = limit_bytes != prev_limit ||
- env->me_dxb_mmap.address != prev_addr
-#if defined(_WIN32) || defined(_WIN64)
- || prev_size > size_bytes
-#endif /* Windows */
- ;
- rc = mdbx_set_readahead(env, size_pgno, readahead, force);
- }
-#endif /* MDBX_ENABLE_MADVISE */
-
-bailout:
- if (rc == MDBX_SUCCESS) {
- mdbx_assert(env, size_bytes == env->me_dxb_mmap.current);
- mdbx_assert(env, size_bytes <= env->me_dxb_mmap.filesize);
- mdbx_assert(env, limit_bytes == env->me_dxb_mmap.limit);
-#ifdef MDBX_USE_VALGRIND
- if (prev_limit != env->me_dxb_mmap.limit || prev_addr != env->me_map) {
- VALGRIND_DISCARD(env->me_valgrind_handle);
- env->me_valgrind_handle = 0;
- if (env->me_dxb_mmap.limit)
- env->me_valgrind_handle =
- VALGRIND_CREATE_BLOCK(env->me_map, env->me_dxb_mmap.limit, "mdbx");
- }
-#endif /* MDBX_USE_VALGRIND */
- } else {
- if (rc != MDBX_UNABLE_EXTEND_MAPSIZE && rc != MDBX_EPERM) {
- mdbx_error("failed resize datafile/mapping: "
- "present %" PRIuPTR " -> %" PRIuPTR ", "
- "limit %" PRIuPTR " -> %" PRIuPTR ", errcode %d",
- prev_size, size_bytes, prev_limit, limit_bytes, rc);
- } else {
- mdbx_warning("unable resize datafile/mapping: "
- "present %" PRIuPTR " -> %" PRIuPTR ", "
- "limit %" PRIuPTR " -> %" PRIuPTR ", errcode %d",
- prev_size, size_bytes, prev_limit, limit_bytes, rc);
- }
- if (!env->me_dxb_mmap.address) {
- env->me_flags |= MDBX_FATAL_ERROR;
- if (env->me_txn)
- env->me_txn->mt_flags |= MDBX_TXN_ERROR;
- rc = MDBX_PANIC;
- }
- }
-
-#if defined(_WIN32) || defined(_WIN64)
- int err = MDBX_SUCCESS;
- mdbx_srwlock_ReleaseExclusive(&env->me_remap_guard);
- if (suspended) {
- err = mdbx_resume_threads_after_remap(suspended);
- if (suspended != &array_onstack)
- mdbx_free(suspended);
- }
-#else
- if (env->me_lck_mmap.lck &&
- (mresize_flags & (MDBX_MRESIZE_MAY_UNMAP | MDBX_MRESIZE_MAY_MOVE)) != 0)
- mdbx_rdt_unlock(env);
- int err = mdbx_fastmutex_release(&env->me_remap_guard);
-#endif /* Windows */
- if (err != MDBX_SUCCESS) {
- mdbx_fatal("failed resume-after-remap: errcode %d", err);
- return MDBX_PANIC;
- }
- return rc;
-}
-
-__cold static int mdbx_mapresize_implicit(MDBX_env *env, const pgno_t used_pgno,
- const pgno_t size_pgno,
- const pgno_t limit_pgno) {
- const pgno_t mapped_pgno = bytes2pgno(env, env->me_dxb_mmap.limit);
- mdbx_assert(env, mapped_pgno >= used_pgno);
- return mdbx_mapresize(
- env, used_pgno, size_pgno,
- (size_pgno > mapped_pgno)
- ? limit_pgno
- : /* The actual mapsize may be less since the geo.upper may be changed
- by other process. So, avoids remapping until it necessary. */
- mapped_pgno,
- true);
-}
-
-static int mdbx_meta_unsteady(MDBX_env *env, const txnid_t last_steady,
- MDBX_meta *const meta, mdbx_filehandle_t fd) {
- const uint64_t wipe = MDBX_DATASIGN_NONE;
- if (unlikely(META_IS_STEADY(meta)) &&
- constmeta_txnid(env, meta) <= last_steady) {
- mdbx_warning("wipe txn #%" PRIaTXN ", meta %" PRIaPGNO, last_steady,
- data_page(meta)->mp_pgno);
- if (env->me_flags & MDBX_WRITEMAP)
- unaligned_poke_u64(4, meta->mm_datasync_sign, wipe);
- else
- return mdbx_pwrite(fd, &wipe, sizeof(meta->mm_datasync_sign),
- (uint8_t *)&meta->mm_datasync_sign - env->me_map);
- if (constmeta_txnid(env, meta) == last_steady)
- mdbx_assert(env, meta_checktxnid(env, meta, true));
- }
- return MDBX_SUCCESS;
-}
-
-__cold static int mdbx_wipe_steady(MDBX_env *env, const txnid_t last_steady) {
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- const mdbx_filehandle_t fd = (env->me_dsync_fd != INVALID_HANDLE_VALUE)
- ? env->me_dsync_fd
- : env->me_lazy_fd;
- int err = mdbx_meta_unsteady(env, last_steady, METAPAGE(env, 0), fd);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- err = mdbx_meta_unsteady(env, last_steady, METAPAGE(env, 1), fd);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- err = mdbx_meta_unsteady(env, last_steady, METAPAGE(env, 2), fd);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- if (env->me_flags & MDBX_WRITEMAP) {
- mdbx_flush_incoherent_cpu_writeback();
- err = mdbx_msync(&env->me_dxb_mmap, 0, pgno_align2os_bytes(env, NUM_METAS),
- MDBX_SYNC_DATA);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- } else {
- if (fd == env->me_lazy_fd) {
-#if MDBX_USE_SYNCFILERANGE
- static bool syncfilerange_unavailable;
- if (!syncfilerange_unavailable &&
- sync_file_range(env->me_lazy_fd, 0, pgno2bytes(env, NUM_METAS),
- SYNC_FILE_RANGE_WRITE | SYNC_FILE_RANGE_WAIT_AFTER)) {
- err = errno;
- if (ignore_enosys(err) == MDBX_RESULT_TRUE)
- syncfilerange_unavailable = true;
- }
- if (syncfilerange_unavailable)
-#endif /* MDBX_USE_SYNCFILERANGE */
- err = mdbx_fsync(env->me_lazy_fd, MDBX_SYNC_DATA);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- mdbx_flush_incoherent_mmap(env->me_map, pgno2bytes(env, NUM_METAS),
- env->me_os_psize);
- }
-
- /* force oldest refresh */
- atomic_store32(&env->me_lck->mti_readers_refresh_flag, true, mo_Relaxed);
- return MDBX_SUCCESS;
-}
-
-/* Allocate page numbers and memory for writing. Maintain mt_last_reclaimed,
- * mt_reclaimed_pglist and mt_next_pgno. Set MDBX_TXN_ERROR on failure.
- *
- * If there are free pages available from older transactions, they
- * are re-used first. Otherwise allocate a new page at mt_next_pgno.
- * Do not modify the GC, just merge GC records into mt_reclaimed_pglist
- * and move mt_last_reclaimed to say which records were consumed. Only this
- * function can create mt_reclaimed_pglist and move
- * mt_last_reclaimed/mt_next_pgno.
- *
- * [in] mc cursor A cursor handle identifying the transaction and
- * database for which we are allocating.
- * [in] num the number of pages to allocate.
- *
- * Returns 0 on success, non-zero on failure.*/
-
-#define MDBX_ALLOC_CACHE 1
-#define MDBX_ALLOC_GC 2
-#define MDBX_ALLOC_NEW 4
-#define MDBX_ALLOC_SLOT 8
-#define MDBX_ALLOC_FAKE 16
-#define MDBX_ALLOC_ALL (MDBX_ALLOC_CACHE | MDBX_ALLOC_GC | MDBX_ALLOC_NEW)
-
-__hot static struct page_result mdbx_page_alloc(MDBX_cursor *mc,
- const pgno_t num, int flags) {
- struct page_result ret;
- MDBX_txn *const txn = mc->mc_txn;
- MDBX_env *const env = txn->mt_env;
- mdbx_assert(env, num == 0 || !(flags & MDBX_ALLOC_SLOT));
- mdbx_assert(env, num > 0 || !(flags & MDBX_ALLOC_NEW));
-
- const unsigned coalesce_threshold =
- env->me_maxgc_ov1page - env->me_maxgc_ov1page / 4;
- if (likely(flags & MDBX_ALLOC_GC)) {
- flags |= env->me_flags & (MDBX_COALESCE | MDBX_LIFORECLAIM);
- if (MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) > coalesce_threshold)
- flags &= ~MDBX_COALESCE;
- if (unlikely(
- /* If mc is updating the GC, then the retired-list cannot play
- catch-up with itself by growing while trying to save it. */
- (mc->mc_flags & C_RECLAIMING) ||
- /* avoid (recursive) search inside empty tree and while tree is
- updating, todo4recovery://erased_by_github/libmdbx/issues/31 */
- txn->mt_dbs[FREE_DBI].md_entries == 0 ||
- /* If our dirty list is already full, we can't touch GC */
- (txn->tw.dirtyroom < txn->mt_dbs[FREE_DBI].md_depth &&
- !(txn->mt_dbistate[FREE_DBI] & DBI_DIRTY))))
- flags &= ~(MDBX_ALLOC_GC | MDBX_COALESCE);
- }
-
- if (likely(num == 1 && (flags & MDBX_ALLOC_CACHE) != 0)) {
- /* If there are any loose pages, just use them */
- mdbx_assert(env, (flags & MDBX_ALLOC_SLOT) == 0);
- if (likely(txn->tw.loose_pages)) {
-#if MDBX_ENABLE_REFUND
- if (txn->tw.loose_refund_wl > txn->mt_next_pgno) {
- mdbx_refund(txn);
- if (unlikely(!txn->tw.loose_pages))
- goto no_loose;
- }
-#endif /* MDBX_ENABLE_REFUND */
-
- ret.page = txn->tw.loose_pages;
- txn->tw.loose_pages = ret.page->mp_next;
- txn->tw.loose_count--;
- mdbx_debug_extra("db %d use loose page %" PRIaPGNO, DDBI(mc),
- ret.page->mp_pgno);
- mdbx_tassert(txn, ret.page->mp_pgno < txn->mt_next_pgno);
- mdbx_ensure(env, ret.page->mp_pgno >= NUM_METAS);
- VALGRIND_MAKE_MEM_UNDEFINED(page_data(ret.page), page_space(txn->mt_env));
- MDBX_ASAN_UNPOISON_MEMORY_REGION(page_data(ret.page),
- page_space(txn->mt_env));
- ret.page->mp_txnid = txn->mt_front;
- ret.err = MDBX_SUCCESS;
- return ret;
- }
- }
-#if MDBX_ENABLE_REFUND
-no_loose:
-#endif /* MDBX_ENABLE_REFUND */
-
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- pgno_t pgno, *re_list = txn->tw.reclaimed_pglist;
- unsigned range_begin = 0, re_len = MDBX_PNL_SIZE(re_list);
- txnid_t oldest = 0, last = 0;
-
- while (true) { /* hsr-kick retry loop */
- MDBX_cursor_couple recur;
- for (MDBX_cursor_op op = MDBX_FIRST;;
- op = (flags & MDBX_LIFORECLAIM) ? MDBX_PREV : MDBX_NEXT) {
- MDBX_val key, data;
-
- /* Seek a big enough contiguous page range.
- * Prefer pages with lower pgno. */
- mdbx_tassert(txn, mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno));
- if (!(flags & (MDBX_COALESCE | MDBX_ALLOC_SLOT)) && re_len >= num) {
- mdbx_tassert(txn, MDBX_PNL_LAST(re_list) < txn->mt_next_pgno &&
- MDBX_PNL_FIRST(re_list) < txn->mt_next_pgno);
- range_begin = MDBX_PNL_ASCENDING ? 1 : re_len;
- pgno = MDBX_PNL_LEAST(re_list);
- if (likely(num == 1))
- goto done;
-
- const unsigned wanna_range = num - 1;
-#if MDBX_PNL_ASCENDING
- mdbx_tassert(txn, pgno == re_list[1] && range_begin == 1);
- while (true) {
- unsigned range_end = range_begin + wanna_range;
- if (re_list[range_end] - pgno == wanna_range)
- goto done;
- if (range_end == re_len)
- break;
- pgno = re_list[++range_begin];
- }
-#else
- mdbx_tassert(txn, pgno == re_list[re_len] && range_begin == re_len);
- while (true) {
- if (re_list[range_begin - wanna_range] - pgno == wanna_range)
- goto done;
- if (range_begin == wanna_range)
- break;
- pgno = re_list[--range_begin];
- }
-#endif /* MDBX_PNL sort-order */
- }
-
- if (op == MDBX_FIRST) { /* 1st iteration, setup cursor, etc */
- if (unlikely(!(flags & MDBX_ALLOC_GC)))
- break /* reclaiming is prohibited for now */;
-
- /* Prepare to fetch more and coalesce */
- oldest = (flags & MDBX_LIFORECLAIM)
- ? mdbx_find_oldest(txn)
- : atomic_load64(&env->me_lck->mti_oldest_reader,
- mo_AcquireRelease);
- ret.err = mdbx_cursor_init(&recur.outer, txn, FREE_DBI);
- if (unlikely(ret.err != MDBX_SUCCESS))
- goto fail;
- if (flags & MDBX_LIFORECLAIM) {
- /* Begin from oldest reader if any */
- if (oldest > MIN_TXNID) {
- last = oldest - 1;
- op = MDBX_SET_RANGE;
- }
- } else if (txn->tw.last_reclaimed) {
- /* Continue lookup from txn->tw.last_reclaimed to oldest reader */
- last = txn->tw.last_reclaimed;
- op = MDBX_SET_RANGE;
- }
-
- key.iov_base = &last;
- key.iov_len = sizeof(last);
- }
-
- if (!(flags & MDBX_LIFORECLAIM)) {
- /* Do not try fetch more if the record will be too recent */
- if (op != MDBX_FIRST && ++last >= oldest) {
- oldest = mdbx_find_oldest(txn);
- if (oldest <= last)
- break;
- }
- }
-
- ret.err = mdbx_cursor_get(&recur.outer, &key, NULL, op);
- if (ret.err == MDBX_NOTFOUND && (flags & MDBX_LIFORECLAIM)) {
- if (op == MDBX_SET_RANGE)
- continue;
- txnid_t snap = mdbx_find_oldest(txn);
- if (oldest < snap) {
- oldest = snap;
- last = oldest - 1;
- key.iov_base = &last;
- key.iov_len = sizeof(last);
- op = MDBX_SET_RANGE;
- ret.err = mdbx_cursor_get(&recur.outer, &key, NULL, op);
- }
- }
- if (unlikely(ret.err)) {
- if (ret.err == MDBX_NOTFOUND)
- break;
- goto fail;
- }
-
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(key.iov_len != sizeof(txnid_t))) {
- ret.err = MDBX_CORRUPTED;
- goto fail;
- }
- last = unaligned_peek_u64(4, key.iov_base);
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(last < MIN_TXNID || last > MAX_TXNID)) {
- ret.err = MDBX_CORRUPTED;
- goto fail;
- }
- if (oldest <= last) {
- oldest = mdbx_find_oldest(txn);
- if (oldest <= last) {
- if (flags & MDBX_LIFORECLAIM)
- continue;
- break;
- }
- }
-
- if (flags & MDBX_LIFORECLAIM) {
- /* skip IDs of records that already reclaimed */
- if (txn->tw.lifo_reclaimed) {
- size_t i;
- for (i = (size_t)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed); i > 0; --i)
- if (txn->tw.lifo_reclaimed[i] == last)
- break;
- if (i)
- continue;
- }
- }
-
- /* Reading next GC record */
- MDBX_page *const mp = recur.outer.mc_pg[recur.outer.mc_top];
- if (unlikely((ret.err = mdbx_node_read(
- &recur.outer,
- page_node(mp, recur.outer.mc_ki[recur.outer.mc_top]),
- &data, pp_txnid4chk(mp, txn))) != MDBX_SUCCESS))
- goto fail;
-
- if ((flags & MDBX_LIFORECLAIM) && !txn->tw.lifo_reclaimed) {
- txn->tw.lifo_reclaimed = mdbx_txl_alloc();
- if (unlikely(!txn->tw.lifo_reclaimed)) {
- ret.err = MDBX_ENOMEM;
- goto fail;
- }
- }
-
- /* Append PNL from GC record to tw.reclaimed_pglist */
- mdbx_cassert(mc, (mc->mc_flags & C_GCFREEZE) == 0);
- pgno_t *gc_pnl = (pgno_t *)data.iov_base;
- mdbx_tassert(txn, data.iov_len >= MDBX_PNL_SIZEOF(gc_pnl));
- if (unlikely(data.iov_len < MDBX_PNL_SIZEOF(gc_pnl) ||
- !mdbx_pnl_check(gc_pnl, txn->mt_next_pgno))) {
- ret.err = MDBX_CORRUPTED;
- goto fail;
- }
- const unsigned gc_len = MDBX_PNL_SIZE(gc_pnl);
- if (unlikely(/* list is too long already */ MDBX_PNL_SIZE(
- txn->tw.reclaimed_pglist) >=
- env->me_options.rp_augment_limit) &&
- ((/* not a slot-request from gc-update */
- (flags & MDBX_ALLOC_SLOT) == 0 &&
- /* have enough unallocated space */ txn->mt_geo.upper >=
- txn->mt_next_pgno + (size_t)num) ||
- gc_len + MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) >=
- MDBX_PGL_LIMIT)) {
- /* Stop reclaiming to avoid overflow the page list.
- * This is a rare case while search for a continuously multi-page region
- * in a large database.
- * todo4recovery://erased_by_github/libmdbx/issues/123 */
- mdbx_notice("stop reclaiming to avoid PNL overflow: %u (current) + %u "
- "(chunk) -> %u",
- MDBX_PNL_SIZE(txn->tw.reclaimed_pglist), gc_len,
- gc_len + MDBX_PNL_SIZE(txn->tw.reclaimed_pglist));
- flags &= ~(MDBX_ALLOC_GC | MDBX_COALESCE);
- break;
- }
- ret.err = mdbx_pnl_need(&txn->tw.reclaimed_pglist, gc_len);
- if (unlikely(ret.err != MDBX_SUCCESS))
- goto fail;
- re_list = txn->tw.reclaimed_pglist;
-
- /* Remember ID of GC record */
- if (flags & MDBX_LIFORECLAIM) {
- ret.err = mdbx_txl_append(&txn->tw.lifo_reclaimed, last);
- if (unlikely(ret.err != MDBX_SUCCESS))
- goto fail;
- }
- txn->tw.last_reclaimed = last;
-
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) {
- mdbx_debug_extra("PNL read txn %" PRIaTXN " root %" PRIaPGNO
- " num %u, PNL",
- last, txn->mt_dbs[FREE_DBI].md_root, gc_len);
- for (unsigned i = gc_len; i; i--)
- mdbx_debug_extra_print(" %" PRIaPGNO, gc_pnl[i]);
- mdbx_debug_extra_print("%s\n", ".");
- }
-
- /* Merge in descending sorted order */
- const unsigned prev_re_len = MDBX_PNL_SIZE(re_list);
- mdbx_pnl_xmerge(re_list, gc_pnl);
- /* re-check to avoid duplicates */
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(!mdbx_pnl_check(re_list, txn->mt_next_pgno))) {
- ret.err = MDBX_CORRUPTED;
- goto fail;
- }
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
-
- re_len = MDBX_PNL_SIZE(re_list);
- mdbx_tassert(txn, re_len == 0 || re_list[re_len] < txn->mt_next_pgno);
- if (MDBX_ENABLE_REFUND && re_len &&
- unlikely(MDBX_PNL_MOST(re_list) == txn->mt_next_pgno - 1)) {
- /* Refund suitable pages into "unallocated" space */
- mdbx_refund(txn);
- re_list = txn->tw.reclaimed_pglist;
- re_len = MDBX_PNL_SIZE(re_list);
- }
-
- /* Done for a kick-reclaim mode, actually no page needed */
- if (unlikely(flags & MDBX_ALLOC_SLOT)) {
- mdbx_debug("early-return NULL-page for %s mode", "MDBX_ALLOC_SLOT");
- ret.err = MDBX_SUCCESS;
- ret.page = NULL;
- return ret;
- }
-
- /* Don't try to coalesce too much. */
- if (flags & MDBX_COALESCE) {
- if (re_len /* current size */ > coalesce_threshold ||
- (re_len > prev_re_len &&
- re_len - prev_re_len /* delta from prev */ >=
- coalesce_threshold / 2)) {
- mdbx_trace("clear %s %s", "MDBX_COALESCE", "since got threshold");
- flags &= ~MDBX_COALESCE;
- }
- }
- }
-
- if (F_ISSET(flags, MDBX_COALESCE | MDBX_ALLOC_GC)) {
- mdbx_debug_extra("clear %s and continue", "MDBX_COALESCE");
- flags &= ~MDBX_COALESCE;
- continue;
- }
-
- /* There is no suitable pages in the GC and to be able to allocate
- * we should CHOICE one of:
- * - make a new steady checkpoint if reclaiming was stopped by
- * the last steady-sync, or wipe it in the MDBX_UTTERLY_NOSYNC mode;
- * - kick lagging reader(s) if reclaiming was stopped by ones of it.
- * - extend the database file. */
-
- /* Will use new pages from the map if nothing is suitable in the GC. */
- range_begin = 0;
- pgno = txn->mt_next_pgno;
- const size_t next = (size_t)pgno + num;
-
- if (flags & MDBX_ALLOC_GC) {
- const MDBX_meta *const head = constmeta_prefer_last(env);
- const MDBX_meta *const steady = constmeta_prefer_steady(env);
- /* does reclaiming stopped at the last steady point? */
- if (head != steady && META_IS_STEADY(steady) &&
- oldest == constmeta_txnid(env, steady)) {
- mdbx_debug("gc-kick-steady: head %" PRIaTXN "-%s, tail %" PRIaTXN
- "-%s, oldest %" PRIaTXN,
- constmeta_txnid(env, head), mdbx_durable_str(head),
- constmeta_txnid(env, steady), mdbx_durable_str(steady),
- oldest);
- ret.err = MDBX_RESULT_TRUE;
- const pgno_t autosync_threshold =
- atomic_load32(&env->me_lck->mti_autosync_threshold, mo_Relaxed);
- const uint64_t autosync_period =
- atomic_load64(&env->me_lck->mti_autosync_period, mo_Relaxed);
- /* wipe the last steady-point if one of:
- * - UTTERLY_NOSYNC mode AND auto-sync threshold is NOT specified
- * - UTTERLY_NOSYNC mode AND free space at steady-point is exhausted
- * otherwise, make a new steady-point if one of:
- * - auto-sync threshold is specified and reached;
- * - upper limit of database size is reached;
- * - database is full (with the current file size)
- * AND auto-sync threshold it NOT specified */
- if (F_ISSET(env->me_flags, MDBX_UTTERLY_NOSYNC) &&
- ((autosync_threshold | autosync_period) == 0 ||
- next >= steady->mm_geo.now)) {
- /* wipe steady checkpoint in MDBX_UTTERLY_NOSYNC mode
- * without any auto-sync threshold(s). */
- ret.err = mdbx_wipe_steady(env, oldest);
- mdbx_debug("gc-wipe-steady, rc %d", ret.err);
- mdbx_assert(env, steady != meta_prefer_steady(env));
- } else if ((flags & MDBX_ALLOC_NEW) == 0 ||
- (autosync_threshold &&
- atomic_load32(&env->me_lck->mti_unsynced_pages,
- mo_Relaxed) >= autosync_threshold) ||
- (autosync_period &&
- mdbx_osal_monotime() -
- atomic_load64(&env->me_lck->mti_sync_timestamp,
- mo_Relaxed) >=
- autosync_period) ||
- next >= txn->mt_geo.upper ||
- (next >= txn->mt_end_pgno &&
- (autosync_threshold | autosync_period) == 0)) {
- /* make steady checkpoint. */
- MDBX_meta meta = *head;
- ret.err = mdbx_sync_locked(env, env->me_flags & MDBX_WRITEMAP, &meta);
- mdbx_debug("gc-make-steady, rc %d", ret.err);
- mdbx_assert(env, steady != meta_prefer_steady(env));
- }
- if (ret.err == MDBX_SUCCESS) {
- if (mdbx_find_oldest(txn) > oldest)
- continue;
- /* it is reasonable check/kick lagging reader(s) here,
- * since we made a new steady point or wipe the last. */
- if (oldest < txn->mt_txnid - xMDBX_TXNID_STEP &&
- mdbx_kick_longlived_readers(env, oldest) > oldest)
- continue;
- } else if (unlikely(ret.err != MDBX_RESULT_TRUE))
- goto fail;
- }
- }
-
- /* don't kick lagging reader(s) if is enough unallocated space
- * at the end of database file. */
- if ((flags & MDBX_ALLOC_NEW) && next <= txn->mt_end_pgno)
- goto done;
- if ((flags & MDBX_ALLOC_GC) && oldest < txn->mt_txnid - xMDBX_TXNID_STEP &&
- mdbx_kick_longlived_readers(env, oldest) > oldest)
- continue;
-
- ret.err = MDBX_NOTFOUND;
- if (flags & MDBX_ALLOC_NEW) {
- ret.err = MDBX_MAP_FULL;
- if (next < txn->mt_geo.upper && txn->mt_geo.grow_pv) {
- mdbx_assert(env, next > txn->mt_end_pgno);
- const pgno_t grow_step = pv2pages(txn->mt_geo.grow_pv);
- size_t aligned = pgno_align2os_pgno(
- env, (pgno_t)(next + grow_step - next % grow_step));
-
- if (aligned > txn->mt_geo.upper)
- aligned = txn->mt_geo.upper;
- mdbx_assert(env, aligned > txn->mt_end_pgno);
-
- mdbx_verbose("try growth datafile to %zu pages (+%zu)", aligned,
- aligned - txn->mt_end_pgno);
- ret.err = mdbx_mapresize_implicit(env, txn->mt_next_pgno,
- (pgno_t)aligned, txn->mt_geo.upper);
- if (ret.err == MDBX_SUCCESS) {
- env->me_txn->mt_end_pgno = (pgno_t)aligned;
- goto done;
- }
-
- mdbx_error("unable growth datafile to %zu pages (+%zu), errcode %d",
- aligned, aligned - txn->mt_end_pgno, ret.err);
- } else {
- mdbx_notice("gc-alloc: next %zu > upper %" PRIaPGNO, next,
- txn->mt_geo.upper);
- }
- }
-
- fail:
- mdbx_assert(env,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- if (likely(!(flags & MDBX_ALLOC_FAKE)))
- txn->mt_flags |= MDBX_TXN_ERROR;
- if (num != 1 || ret.err != MDBX_NOTFOUND)
- mdbx_notice("alloc %u pages failed, flags 0x%x, errcode %d", num, flags,
- ret.err);
- else
- mdbx_trace("alloc %u pages failed, flags 0x%x, errcode %d", num, flags,
- ret.err);
- mdbx_assert(env, ret.err != MDBX_SUCCESS);
- ret.page = NULL;
- return ret;
- }
-
-done:
- mdbx_assert(env, !(flags & MDBX_ALLOC_SLOT));
- mdbx_ensure(env, pgno >= NUM_METAS);
- if (unlikely(flags & MDBX_ALLOC_FAKE)) {
- mdbx_debug("return NULL-page for %u pages of %s mode", num,
- "MDBX_ALLOC_FAKE");
- ret.page = NULL;
- ret.err = MDBX_SUCCESS;
- return ret;
- }
-
- if (env->me_flags & MDBX_WRITEMAP) {
- ret.page = pgno2page(env, pgno);
- /* LY: reset no-access flag from mdbx_page_loose() */
- VALGRIND_MAKE_MEM_UNDEFINED(ret.page, pgno2bytes(env, num));
- MDBX_ASAN_UNPOISON_MEMORY_REGION(ret.page, pgno2bytes(env, num));
- } else {
- ret.page = mdbx_page_malloc(txn, num);
- if (unlikely(!ret.page)) {
- ret.err = MDBX_ENOMEM;
- goto fail;
- }
- }
-
- if (range_begin) {
- mdbx_cassert(mc, (mc->mc_flags & C_GCFREEZE) == 0);
- mdbx_tassert(txn, pgno < txn->mt_next_pgno);
- mdbx_tassert(txn, pgno == re_list[range_begin]);
- /* Cutoff allocated pages from tw.reclaimed_pglist */
-#if MDBX_PNL_ASCENDING
- for (unsigned i = range_begin + num; i <= re_len;)
- re_list[range_begin++] = re_list[i++];
- MDBX_PNL_SIZE(re_list) = re_len = range_begin - 1;
-#else
- MDBX_PNL_SIZE(re_list) = re_len -= num;
- for (unsigned i = range_begin - num; i < re_len;)
- re_list[++i] = re_list[++range_begin];
-#endif
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- } else {
- txn->mt_next_pgno = pgno + num;
- mdbx_assert(env, txn->mt_next_pgno <= txn->mt_end_pgno);
- }
-
- if (unlikely(env->me_flags & MDBX_PAGEPERTURB))
- memset(ret.page, -1, pgno2bytes(env, num));
- VALGRIND_MAKE_MEM_UNDEFINED(ret.page, pgno2bytes(env, num));
-
- ret.page->mp_pgno = pgno;
- ret.page->mp_leaf2_ksize = 0;
- ret.page->mp_flags = 0;
- if ((mdbx_assert_enabled() || mdbx_audit_enabled()) && num > 1) {
- ret.page->mp_pages = num;
- ret.page->mp_flags = P_OVERFLOW;
- }
- ret.err = mdbx_page_dirty(txn, ret.page, num);
- if (unlikely(ret.err != MDBX_SUCCESS))
- goto fail;
-
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- return ret;
-}
-
-/* Copy the used portions of a non-overflow page. */
-__hot static void mdbx_page_copy(MDBX_page *dst, const MDBX_page *src,
- size_t psize) {
- STATIC_ASSERT(UINT16_MAX > MAX_PAGESIZE - PAGEHDRSZ);
- STATIC_ASSERT(MIN_PAGESIZE > PAGEHDRSZ + NODESIZE * 4);
- if ((src->mp_flags & (P_LEAF2 | P_OVERFLOW)) == 0) {
- size_t upper = src->mp_upper, lower = src->mp_lower, unused = upper - lower;
-
- /* If page isn't full, just copy the used portion. Adjust
- * alignment so memcpy may copy words instead of bytes. */
- if (unused >= MDBX_CACHELINE_SIZE * 2) {
- lower = ceil_powerof2(lower + PAGEHDRSZ, sizeof(void *));
- upper = floor_powerof2(upper + PAGEHDRSZ, sizeof(void *));
- memcpy(dst, src, lower);
- dst = (void *)((char *)dst + upper);
- src = (void *)((char *)src + upper);
- psize -= upper;
- }
- }
- memcpy(dst, src, psize);
-}
-
-/* Pull a page off the txn's spill list, if present.
- *
- * If a page being referenced was spilled to disk in this txn, bring
- * it back and make it dirty/writable again. */
-static struct page_result __must_check_result
-mdbx_page_unspill(MDBX_txn *const txn, const MDBX_page *const mp) {
- mdbx_verbose("unspill page %" PRIaPGNO, mp->mp_pgno);
- mdbx_tassert(txn, (txn->mt_flags & MDBX_WRITEMAP) == 0);
- mdbx_tassert(txn, IS_SPILLED(txn, mp));
- const MDBX_txn *scan = txn;
- struct page_result ret;
- do {
- mdbx_tassert(txn, (scan->mt_flags & MDBX_TXN_SPILLS) != 0);
- const unsigned si = mdbx_search_spilled(scan, mp->mp_pgno);
- if (!si)
- continue;
- const unsigned npages = IS_OVERFLOW(mp) ? mp->mp_pages : 1;
- ret.page = mdbx_page_malloc(txn, npages);
- if (unlikely(!ret.page)) {
- ret.err = MDBX_ENOMEM;
- return ret;
- }
- mdbx_page_copy(ret.page, mp, pgno2bytes(txn->mt_env, npages));
- if (scan == txn) {
- /* If in current txn, this page is no longer spilled.
- * If it happens to be the last page, truncate the spill list.
- * Otherwise mark it as deleted by setting the LSB. */
- mdbx_spill_remove(txn, si, npages);
- } /* otherwise, if belonging to a parent txn, the
- * page remains spilled until child commits */
-
- ret.err = mdbx_page_dirty(txn, ret.page, npages);
- if (unlikely(ret.err != MDBX_SUCCESS))
- return ret;
-#if MDBX_ENABLE_PGOP_STAT
- txn->mt_env->me_lck->mti_pgop_stat.unspill.weak += npages;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- ret.page->mp_flags |= (scan == txn) ? 0 : P_SPILLED;
- ret.err = MDBX_SUCCESS;
- return ret;
- } while (likely((scan = scan->mt_parent) != nullptr &&
- (scan->mt_flags & MDBX_TXN_SPILLS) != 0));
- mdbx_error("Page %" PRIaPGNO " mod-txnid %" PRIaTXN
- " not found in the spill-list(s), current txn %" PRIaTXN
- " front %" PRIaTXN ", root txn %" PRIaTXN " front %" PRIaTXN,
- mp->mp_pgno, mp->mp_txnid, txn->mt_txnid, txn->mt_front,
- txn->mt_env->me_txn0->mt_txnid, txn->mt_env->me_txn0->mt_front);
- ret.err = MDBX_PROBLEM;
- ret.page = NULL;
- return ret;
-}
-
-/* Touch a page: make it dirty and re-insert into tree with updated pgno.
- * Set MDBX_TXN_ERROR on failure.
- *
- * [in] mc cursor pointing to the page to be touched
- *
- * Returns 0 on success, non-zero on failure. */
-__hot static int mdbx_page_touch(MDBX_cursor *mc) {
- const MDBX_page *const mp = mc->mc_pg[mc->mc_top];
- MDBX_page *np;
- MDBX_txn *txn = mc->mc_txn;
- int rc;
-
- if (mdbx_assert_enabled()) {
- if (mc->mc_flags & C_SUB) {
- MDBX_xcursor *mx = container_of(mc->mc_db, MDBX_xcursor, mx_db);
- MDBX_cursor_couple *couple = container_of(mx, MDBX_cursor_couple, inner);
- mdbx_tassert(txn, mc->mc_db == &couple->outer.mc_xcursor->mx_db);
- mdbx_tassert(txn, mc->mc_dbx == &couple->outer.mc_xcursor->mx_dbx);
- mdbx_tassert(txn, *couple->outer.mc_dbistate & DBI_DIRTY);
- } else {
- mdbx_tassert(txn, *mc->mc_dbistate & DBI_DIRTY);
- }
- mdbx_tassert(txn, mc->mc_txn->mt_flags & MDBX_TXN_DIRTY);
- mdbx_tassert(txn, !IS_OVERFLOW(mp));
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- }
-
- if (IS_MODIFIABLE(txn, mp) || IS_SUBP(mp))
- return MDBX_SUCCESS;
-
- if (IS_FROZEN(txn, mp)) {
- /* CoW the page */
- rc = mdbx_pnl_need(&txn->tw.retired_pages, 1);
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- const struct page_result par = mdbx_page_alloc(mc, 1, MDBX_ALLOC_ALL);
- rc = par.err;
- np = par.page;
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
-
- const pgno_t pgno = np->mp_pgno;
- mdbx_debug("touched db %d page %" PRIaPGNO " -> %" PRIaPGNO, DDBI(mc),
- mp->mp_pgno, pgno);
- mdbx_tassert(txn, mp->mp_pgno != pgno);
- mdbx_pnl_xappend(txn->tw.retired_pages, mp->mp_pgno);
- /* Update the parent page, if any, to point to the new page */
- if (mc->mc_top) {
- MDBX_page *parent = mc->mc_pg[mc->mc_top - 1];
- MDBX_node *node = page_node(parent, mc->mc_ki[mc->mc_top - 1]);
- node_set_pgno(node, pgno);
- } else {
- mc->mc_db->md_root = pgno;
- }
-
-#if MDBX_ENABLE_PGOP_STAT
- txn->mt_env->me_lck->mti_pgop_stat.cow.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- mdbx_page_copy(np, mp, txn->mt_env->me_psize);
- np->mp_pgno = pgno;
- np->mp_txnid = txn->mt_front;
- } else if (IS_SPILLED(txn, mp)) {
- struct page_result pur = mdbx_page_unspill(txn, mp);
- np = pur.page;
- rc = pur.err;
- if (likely(rc == MDBX_SUCCESS)) {
- mdbx_tassert(txn, np != nullptr);
- goto done;
- }
- goto fail;
- } else {
- if (unlikely(!txn->mt_parent)) {
- mdbx_error("Unexpected not frozen/modifiable/spilled but shadowed %s "
- "page %" PRIaPGNO " mod-txnid %" PRIaTXN ","
- " without parent transaction, current txn %" PRIaTXN
- " front %" PRIaTXN,
- IS_BRANCH(mp) ? "branch" : "leaf", mp->mp_pgno, mp->mp_txnid,
- mc->mc_txn->mt_txnid, mc->mc_txn->mt_front);
- rc = MDBX_PROBLEM;
- goto fail;
- }
-
- mdbx_debug("clone db %d page %" PRIaPGNO, DDBI(mc), mp->mp_pgno);
- mdbx_tassert(txn, txn->tw.dirtylist->length <=
- MDBX_PGL_LIMIT + MDBX_PNL_GRANULATE);
- /* No - copy it */
- np = mdbx_page_malloc(txn, 1);
- if (unlikely(!np)) {
- rc = MDBX_ENOMEM;
- goto fail;
- }
- mdbx_page_copy(np, mp, txn->mt_env->me_psize);
-
- /* insert a clone of parent's dirty page, so don't touch dirtyroom */
- rc = mdbx_page_dirty(txn, np, 1);
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
-
-#if MDBX_ENABLE_PGOP_STAT
- txn->mt_env->me_lck->mti_pgop_stat.clone.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- }
-
-done:
- /* Adjust cursors pointing to mp */
- mc->mc_pg[mc->mc_top] = np;
- MDBX_cursor *m2 = txn->mt_cursors[mc->mc_dbi];
- if (mc->mc_flags & C_SUB) {
- for (; m2; m2 = m2->mc_next) {
- MDBX_cursor *m3 = &m2->mc_xcursor->mx_cursor;
- if (m3->mc_snum < mc->mc_snum)
- continue;
- if (m3->mc_pg[mc->mc_top] == mp)
- m3->mc_pg[mc->mc_top] = np;
- }
- } else {
- for (; m2; m2 = m2->mc_next) {
- if (m2->mc_snum < mc->mc_snum)
- continue;
- if (m2 == mc)
- continue;
- if (m2->mc_pg[mc->mc_top] == mp) {
- m2->mc_pg[mc->mc_top] = np;
- if (XCURSOR_INITED(m2) && IS_LEAF(np))
- XCURSOR_REFRESH(m2, np, m2->mc_ki[mc->mc_top]);
- }
- }
- }
- return MDBX_SUCCESS;
-
-fail:
- txn->mt_flags |= MDBX_TXN_ERROR;
- return rc;
-}
-
-__cold static int mdbx_env_sync_internal(MDBX_env *env, bool force,
- bool nonblock) {
- bool locked = false;
- int rc = MDBX_RESULT_TRUE /* means "nothing to sync" */;
-
-retry:;
- unsigned flags = env->me_flags & ~(MDBX_NOMETASYNC | MDBX_SHRINK_ALLOWED);
- if (unlikely((flags & (MDBX_RDONLY | MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE)) !=
- MDBX_ENV_ACTIVE)) {
- rc = MDBX_EACCESS;
- if (!(flags & MDBX_ENV_ACTIVE))
- rc = MDBX_EPERM;
- if (flags & MDBX_FATAL_ERROR)
- rc = MDBX_PANIC;
- goto bailout;
- }
-
- const pgno_t unsynced_pages =
- atomic_load32(&env->me_lck->mti_unsynced_pages, mo_Relaxed);
- volatile const MDBX_meta *head = meta_prefer_last(env);
- const txnid_t head_txnid = meta_txnid(env, head);
- const uint32_t synched_meta_txnid_u32 =
- atomic_load32(&env->me_lck->mti_meta_sync_txnid, mo_Relaxed);
- if (unsynced_pages == 0 && synched_meta_txnid_u32 == (uint32_t)head_txnid &&
- META_IS_STEADY(head))
- goto bailout;
-
- const pgno_t autosync_threshold =
- atomic_load32(&env->me_lck->mti_autosync_threshold, mo_Relaxed);
- const uint64_t autosync_period =
- atomic_load64(&env->me_lck->mti_autosync_period, mo_Relaxed);
- if (force || (autosync_threshold && unsynced_pages >= autosync_threshold) ||
- (autosync_period &&
- mdbx_osal_monotime() -
- atomic_load64(&env->me_lck->mti_sync_timestamp, mo_Relaxed) >=
- autosync_period))
- flags &= MDBX_WRITEMAP /* clear flags for full steady sync */;
-
- const bool inside_txn = (env->me_txn0->mt_owner == mdbx_thread_self());
- if (!inside_txn) {
- if (!locked) {
- int err;
- unsigned wops = 0;
- /* pre-sync to avoid latency for writer */
- if (unsynced_pages > /* FIXME: define threshold */ 16 &&
- (flags & MDBX_SAFE_NOSYNC) == 0) {
- mdbx_assert(env, ((flags ^ env->me_flags) & MDBX_WRITEMAP) == 0);
- if (flags & MDBX_WRITEMAP) {
- /* Acquire guard to avoid collision with remap */
-#if defined(_WIN32) || defined(_WIN64)
- mdbx_srwlock_AcquireShared(&env->me_remap_guard);
-#else
- err = mdbx_fastmutex_acquire(&env->me_remap_guard);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-#endif
- const size_t usedbytes = pgno_align2os_bytes(env, head->mm_geo.next);
- err = mdbx_msync(&env->me_dxb_mmap, 0, usedbytes, MDBX_SYNC_DATA);
-#if defined(_WIN32) || defined(_WIN64)
- mdbx_srwlock_ReleaseShared(&env->me_remap_guard);
-#else
- int unlock_err = mdbx_fastmutex_release(&env->me_remap_guard);
- if (unlikely(unlock_err != MDBX_SUCCESS) && err == MDBX_SUCCESS)
- err = unlock_err;
-#endif
- } else
- err = mdbx_fsync(env->me_lazy_fd, MDBX_SYNC_DATA);
-
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- /* pre-sync done */
- wops = 1;
- rc = MDBX_SUCCESS /* means "some data was synced" */;
- }
-
- err = mdbx_txn_lock(env, nonblock);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- locked = true;
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += wops;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- goto retry;
- }
- env->me_txn0->mt_txnid = head_txnid;
- mdbx_assert(env, head_txnid == meta_txnid(env, head));
- mdbx_assert(env, head_txnid == mdbx_recent_committed_txnid(env));
- mdbx_find_oldest(env->me_txn0);
- flags |= MDBX_SHRINK_ALLOWED;
- }
-
- mdbx_assert(env, inside_txn || locked);
- mdbx_assert(env, !inside_txn || (flags & MDBX_SHRINK_ALLOWED) == 0);
-
- if (!META_IS_STEADY(head) ||
- ((flags & MDBX_SAFE_NOSYNC) == 0 && unsynced_pages)) {
- mdbx_debug("meta-head %" PRIaPGNO ", %s, sync_pending %" PRIaPGNO,
- data_page((const void *)head)->mp_pgno, mdbx_durable_str(head),
- unsynced_pages);
- MDBX_meta meta = *head;
- rc = mdbx_sync_locked(env, flags, &meta);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
-
- /* LY: sync meta-pages if MDBX_NOMETASYNC enabled
- * and someone was not synced above. */
- if (atomic_load32(&env->me_lck->mti_meta_sync_txnid, mo_Relaxed) !=
- (uint32_t)head_txnid) {
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- rc = (flags & MDBX_WRITEMAP)
- ? mdbx_msync(&env->me_dxb_mmap, 0,
- pgno_align2os_bytes(env, NUM_METAS),
- MDBX_SYNC_DATA | MDBX_SYNC_IODQ)
- : mdbx_fsync(env->me_lazy_fd, MDBX_SYNC_DATA | MDBX_SYNC_IODQ);
- if (likely(rc == MDBX_SUCCESS))
- atomic_store32(&env->me_lck->mti_meta_sync_txnid, (uint32_t)head_txnid,
- mo_Relaxed);
- }
-
-bailout:
- if (locked)
- mdbx_txn_unlock(env);
- return rc;
-}
-
-static __inline int check_env(const MDBX_env *env, const bool wanna_active) {
- if (unlikely(!env))
- return MDBX_EINVAL;
-
- if (unlikely(env->me_signature.weak != MDBX_ME_SIGNATURE))
- return MDBX_EBADSIGN;
-
-#if MDBX_ENV_CHECKPID
- if (unlikely(env->me_pid != mdbx_getpid())) {
- ((MDBX_env *)env)->me_flags |= MDBX_FATAL_ERROR;
- return MDBX_PANIC;
- }
-#endif /* MDBX_ENV_CHECKPID */
-
- if (unlikely(env->me_flags & MDBX_FATAL_ERROR))
- return MDBX_PANIC;
-
- if (wanna_active) {
- if (unlikely((env->me_flags & MDBX_ENV_ACTIVE) == 0))
- return MDBX_EPERM;
- mdbx_assert(env, env->me_map != nullptr);
- }
-
- return MDBX_SUCCESS;
-}
-
-__cold int mdbx_env_sync_ex(MDBX_env *env, bool force, bool nonblock) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- return mdbx_env_sync_internal(env, force, nonblock);
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-__cold int mdbx_env_sync(MDBX_env *env) { return __inline_mdbx_env_sync(env); }
-
-__cold int mdbx_env_sync_poll(MDBX_env *env) {
- return __inline_mdbx_env_sync_poll(env);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-/* Back up parent txn's cursors, then grab the originals for tracking */
-static int mdbx_cursor_shadow(MDBX_txn *parent, MDBX_txn *nested) {
- for (int i = parent->mt_numdbs; --i >= 0;) {
- nested->mt_cursors[i] = NULL;
- MDBX_cursor *mc = parent->mt_cursors[i];
- if (mc != NULL) {
- size_t size = mc->mc_xcursor ? sizeof(MDBX_cursor) + sizeof(MDBX_xcursor)
- : sizeof(MDBX_cursor);
- for (MDBX_cursor *bk; mc; mc = bk->mc_next) {
- bk = mc;
- if (mc->mc_signature != MDBX_MC_LIVE)
- continue;
- bk = mdbx_malloc(size);
- if (unlikely(!bk))
- return MDBX_ENOMEM;
-#if MDBX_DEBUG
- memset(bk, 0xCD, size);
- VALGRIND_MAKE_MEM_UNDEFINED(bk, size);
-#endif /* MDBX_DEBUG */
- *bk = *mc;
- mc->mc_backup = bk;
- /* Kill pointers into src to reduce abuse: The
- * user may not use mc until dst ends. But we need a valid
- * txn pointer here for cursor fixups to keep working. */
- mc->mc_txn = nested;
- mc->mc_db = &nested->mt_dbs[i];
- mc->mc_dbistate = &nested->mt_dbistate[i];
- MDBX_xcursor *mx = mc->mc_xcursor;
- if (mx != NULL) {
- *(MDBX_xcursor *)(bk + 1) = *mx;
- mx->mx_cursor.mc_txn = nested;
- }
- mc->mc_next = nested->mt_cursors[i];
- nested->mt_cursors[i] = mc;
- }
- }
- }
- return MDBX_SUCCESS;
-}
-
-/* Close this txn's cursors, give parent txn's cursors back to parent.
- *
- * [in] txn the transaction handle.
- * [in] merge true to keep changes to parent cursors, false to revert.
- *
- * Returns 0 on success, non-zero on failure. */
-static void mdbx_cursors_eot(MDBX_txn *txn, const bool merge) {
- for (int i = txn->mt_numdbs; --i >= 0;) {
- MDBX_cursor *next, *mc = txn->mt_cursors[i];
- if (!mc)
- continue;
- txn->mt_cursors[i] = NULL;
- do {
- const unsigned stage = mc->mc_signature;
- MDBX_cursor *bk = mc->mc_backup;
- next = mc->mc_next;
- mdbx_ensure(txn->mt_env,
- stage == MDBX_MC_LIVE || (stage == MDBX_MC_WAIT4EOT && bk));
- mdbx_cassert(mc, mc->mc_dbi == (unsigned)i);
- if (bk) {
- MDBX_xcursor *mx = mc->mc_xcursor;
- mdbx_cassert(mc, mx == bk->mc_xcursor);
- mdbx_tassert(txn, txn->mt_parent != NULL);
- mdbx_ensure(txn->mt_env, bk->mc_signature == MDBX_MC_LIVE);
- if (stage == MDBX_MC_WAIT4EOT /* Cursor was closed by user */)
- mc->mc_signature = stage /* Promote closed state to parent txn */;
- else if (merge) {
- /* Restore pointers to parent txn */
- mc->mc_next = bk->mc_next;
- mc->mc_backup = bk->mc_backup;
- mc->mc_txn = bk->mc_txn;
- mc->mc_db = bk->mc_db;
- mc->mc_dbistate = bk->mc_dbistate;
- if (mx) {
- if (mx != bk->mc_xcursor) {
- *bk->mc_xcursor = *mx;
- mx = bk->mc_xcursor;
- }
- mx->mx_cursor.mc_txn = bk->mc_txn;
- }
- } else {
- /* Restore from backup, i.e. rollback/abort nested txn */
- *mc = *bk;
- if (mx)
- *mx = *(MDBX_xcursor *)(bk + 1);
- }
- bk->mc_signature = 0;
- mdbx_free(bk);
- } else {
- mdbx_ensure(txn->mt_env, stage == MDBX_MC_LIVE);
- mc->mc_signature = MDBX_MC_READY4CLOSE /* Cursor may be reused */;
- mc->mc_flags = 0 /* reset C_UNTRACK */;
- }
- } while ((mc = next) != NULL);
- }
-}
-
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
-/* Find largest mvcc-snapshot still referenced by this process. */
-static pgno_t mdbx_find_largest_this(MDBX_env *env, pgno_t largest) {
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (likely(lck != NULL /* exclusive mode */)) {
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- for (unsigned i = 0; i < snap_nreaders; ++i) {
- retry:
- if (atomic_load32(&lck->mti_readers[i].mr_pid, mo_AcquireRelease) ==
- env->me_pid) {
- /* mdbx_jitter4testing(true); */
- const pgno_t snap_pages = atomic_load32(
- &lck->mti_readers[i].mr_snapshot_pages_used, mo_Relaxed);
- const txnid_t snap_txnid = safe64_read(&lck->mti_readers[i].mr_txnid);
- if (unlikely(
- snap_pages !=
- atomic_load32(&lck->mti_readers[i].mr_snapshot_pages_used,
- mo_AcquireRelease) ||
- snap_txnid != safe64_read(&lck->mti_readers[i].mr_txnid)))
- goto retry;
- if (largest < snap_pages &&
- atomic_load64(&lck->mti_oldest_reader, mo_AcquireRelease) <=
- /* ignore pending updates */ snap_txnid &&
- snap_txnid <= MAX_TXNID)
- largest = snap_pages;
- }
- }
- }
- return largest;
-}
-
-static void mdbx_txn_valgrind(MDBX_env *env, MDBX_txn *txn) {
-#if !defined(__SANITIZE_ADDRESS__)
- if (!RUNNING_ON_VALGRIND)
- return;
-#endif
-
- if (txn) { /* transaction start */
- if (env->me_poison_edge < txn->mt_next_pgno)
- env->me_poison_edge = txn->mt_next_pgno;
- VALGRIND_MAKE_MEM_DEFINED(env->me_map, pgno2bytes(env, txn->mt_next_pgno));
- MDBX_ASAN_UNPOISON_MEMORY_REGION(env->me_map,
- pgno2bytes(env, txn->mt_next_pgno));
- /* don't touch more, it should be already poisoned */
- } else { /* transaction end */
- bool should_unlock = false;
- pgno_t last = MAX_PAGENO + 1;
- if (env->me_txn0 && env->me_txn0->mt_owner == mdbx_thread_self()) {
- /* inside write-txn */
- const MDBX_meta *head = constmeta_prefer_last(env);
- last = head->mm_geo.next;
- } else if (env->me_flags & MDBX_RDONLY) {
- /* read-only mode, no write-txn, no wlock mutex */
- last = NUM_METAS;
- } else if (mdbx_txn_lock(env, true) == MDBX_SUCCESS) {
- /* no write-txn */
- last = NUM_METAS;
- should_unlock = true;
- } else {
- /* write txn is running, therefore shouldn't poison any memory range */
- return;
- }
-
- last = mdbx_find_largest_this(env, last);
- const pgno_t edge = env->me_poison_edge;
- if (edge > last) {
- mdbx_assert(env, last >= NUM_METAS);
- env->me_poison_edge = last;
- VALGRIND_MAKE_MEM_NOACCESS(env->me_map + pgno2bytes(env, last),
- pgno2bytes(env, edge - last));
- MDBX_ASAN_POISON_MEMORY_REGION(env->me_map + pgno2bytes(env, last),
- pgno2bytes(env, edge - last));
- }
- if (should_unlock)
- mdbx_txn_unlock(env);
- }
-}
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-typedef struct {
- int err;
- MDBX_reader *rslot;
-} bind_rslot_result;
-
-static bind_rslot_result bind_rslot(MDBX_env *env, const uintptr_t tid) {
- mdbx_assert(env, env->me_lck_mmap.lck);
- mdbx_assert(env, env->me_lck->mti_magic_and_version == MDBX_LOCK_MAGIC);
- mdbx_assert(env, env->me_lck->mti_os_and_format == MDBX_LOCK_FORMAT);
-
- bind_rslot_result result = {mdbx_rdt_lock(env), nullptr};
- if (unlikely(MDBX_IS_ERROR(result.err)))
- return result;
- if (unlikely(env->me_flags & MDBX_FATAL_ERROR)) {
- mdbx_rdt_unlock(env);
- result.err = MDBX_PANIC;
- return result;
- }
- if (unlikely(!env->me_map)) {
- mdbx_rdt_unlock(env);
- result.err = MDBX_EPERM;
- return result;
- }
-
- if (unlikely(env->me_live_reader != env->me_pid)) {
- result.err = mdbx_rpid_set(env);
- if (unlikely(result.err != MDBX_SUCCESS)) {
- mdbx_rdt_unlock(env);
- return result;
- }
- env->me_live_reader = env->me_pid;
- }
-
- result.err = MDBX_SUCCESS;
- unsigned slot, nreaders;
- while (1) {
- nreaders = atomic_load32(&env->me_lck->mti_numreaders, mo_Relaxed);
- for (slot = 0; slot < nreaders; slot++)
- if (atomic_load32(&env->me_lck->mti_readers[slot].mr_pid, mo_Relaxed) ==
- 0)
- break;
-
- if (likely(slot < env->me_maxreaders))
- break;
-
- result.err = mdbx_cleanup_dead_readers(env, true, NULL);
- if (result.err != MDBX_RESULT_TRUE) {
- mdbx_rdt_unlock(env);
- result.err =
- (result.err == MDBX_SUCCESS) ? MDBX_READERS_FULL : result.err;
- return result;
- }
- }
-
- result.rslot = &env->me_lck->mti_readers[slot];
- /* Claim the reader slot, carefully since other code
- * uses the reader table un-mutexed: First reset the
- * slot, next publish it in lck->mti_numreaders. After
- * that, it is safe for mdbx_env_close() to touch it.
- * When it will be closed, we can finally claim it. */
- atomic_store32(&result.rslot->mr_pid, 0, mo_Relaxed);
- safe64_reset(&result.rslot->mr_txnid, true);
- if (slot == nreaders)
- atomic_store32(&env->me_lck->mti_numreaders, ++nreaders, mo_Relaxed);
- atomic_store64(&result.rslot->mr_tid, (env->me_flags & MDBX_NOTLS) ? 0 : tid,
- mo_Relaxed);
- atomic_store32(&result.rslot->mr_pid, env->me_pid, mo_Relaxed);
- mdbx_rdt_unlock(env);
-
- if (likely(env->me_flags & MDBX_ENV_TXKEY)) {
- mdbx_assert(env, env->me_live_reader == env->me_pid);
- thread_rthc_set(env->me_txkey, result.rslot);
- }
- return result;
-}
-
-__cold int mdbx_thread_register(const MDBX_env *env) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!env->me_lck_mmap.lck))
- return (env->me_flags & MDBX_EXCLUSIVE) ? MDBX_EINVAL : MDBX_EPERM;
-
- if (unlikely((env->me_flags & MDBX_ENV_TXKEY) == 0)) {
- mdbx_assert(env, !env->me_lck_mmap.lck || (env->me_flags & MDBX_NOTLS));
- return MDBX_EINVAL /* MDBX_NOTLS mode */;
- }
-
- mdbx_assert(env, (env->me_flags & (MDBX_NOTLS | MDBX_ENV_TXKEY |
- MDBX_EXCLUSIVE)) == MDBX_ENV_TXKEY);
- MDBX_reader *r = thread_rthc_get(env->me_txkey);
- if (unlikely(r != NULL)) {
- mdbx_assert(env, r->mr_pid.weak == env->me_pid);
- mdbx_assert(env, r->mr_tid.weak == mdbx_thread_self());
- if (unlikely(r->mr_pid.weak != env->me_pid))
- return MDBX_BAD_RSLOT;
- return MDBX_RESULT_TRUE /* already registered */;
- }
-
- const uintptr_t tid = mdbx_thread_self();
- if (env->me_txn0 && unlikely(env->me_txn0->mt_owner == tid))
- return MDBX_TXN_OVERLAPPING;
- return bind_rslot((MDBX_env *)env, tid).err;
-}
-
-__cold int mdbx_thread_unregister(const MDBX_env *env) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!env->me_lck_mmap.lck))
- return MDBX_RESULT_TRUE;
-
- if (unlikely((env->me_flags & MDBX_ENV_TXKEY) == 0)) {
- mdbx_assert(env, !env->me_lck_mmap.lck || (env->me_flags & MDBX_NOTLS));
- return MDBX_RESULT_TRUE /* MDBX_NOTLS mode */;
- }
-
- mdbx_assert(env, (env->me_flags & (MDBX_NOTLS | MDBX_ENV_TXKEY |
- MDBX_EXCLUSIVE)) == MDBX_ENV_TXKEY);
- MDBX_reader *r = thread_rthc_get(env->me_txkey);
- if (unlikely(r == NULL))
- return MDBX_RESULT_TRUE /* not registered */;
-
- mdbx_assert(env, r->mr_pid.weak == env->me_pid);
- mdbx_assert(env, r->mr_tid.weak == mdbx_thread_self());
- if (unlikely(r->mr_pid.weak != env->me_pid ||
- r->mr_tid.weak != mdbx_thread_self()))
- return MDBX_BAD_RSLOT;
-
- if (unlikely(r->mr_txnid.weak < SAFE64_INVALID_THRESHOLD))
- return MDBX_BUSY /* transaction is still active */;
-
- atomic_store32(&r->mr_pid, 0, mo_Relaxed);
- atomic_store32(&env->me_lck->mti_readers_refresh_flag, true,
- mo_AcquireRelease);
- thread_rthc_set(env->me_txkey, nullptr);
- return MDBX_SUCCESS;
-}
-
-/* check against todo4recovery://erased_by_github/libmdbx/issues/269 */
-static bool meta_checktxnid(const MDBX_env *env, const MDBX_meta *meta,
- bool report) {
- const txnid_t meta_txnid = constmeta_txnid(env, meta);
- const txnid_t freedb_mod_txnid = meta->mm_dbs[FREE_DBI].md_mod_txnid;
- const txnid_t maindb_mod_txnid = meta->mm_dbs[MAIN_DBI].md_mod_txnid;
-
- const pgno_t freedb_root_pgno = meta->mm_dbs[FREE_DBI].md_root;
- const MDBX_page *freedb_root = (env->me_map && freedb_root_pgno != P_INVALID)
- ? pgno2page(env, freedb_root_pgno)
- : nullptr;
-
- const pgno_t maindb_root_pgno = meta->mm_dbs[MAIN_DBI].md_root;
- const MDBX_page *maindb_root = (env->me_map && maindb_root_pgno != P_INVALID)
- ? pgno2page(env, maindb_root_pgno)
- : nullptr;
-
- const uint64_t magic_and_version =
- unaligned_peek_u64(4, &meta->mm_magic_and_version);
- bool ok = true;
- if (unlikely(meta_txnid < freedb_mod_txnid ||
- (!freedb_mod_txnid && freedb_root &&
- likely(magic_and_version == MDBX_DATA_MAGIC)))) {
- if (report)
- mdbx_warning(
- "catch invalid %sdb_mod_txnid %" PRIaTXN " for meta_txnid %" PRIaTXN
- " %s",
- "free", freedb_mod_txnid, meta_txnid,
- "(workaround for incoherent flaw of unified page/buffer cache)");
- ok = false;
- }
- if (unlikely(meta_txnid < maindb_mod_txnid ||
- (!maindb_mod_txnid && maindb_root &&
- likely(magic_and_version == MDBX_DATA_MAGIC)))) {
- if (report)
- mdbx_warning(
- "catch invalid %sdb_mod_txnid %" PRIaTXN " for meta_txnid %" PRIaTXN
- " %s",
- "main", maindb_mod_txnid, meta_txnid,
- "(workaround for incoherent flaw of unified page/buffer cache)");
- ok = false;
- }
- if (likely(freedb_root && freedb_mod_txnid)) {
- VALGRIND_MAKE_MEM_DEFINED(freedb_root, sizeof(freedb_root->mp_txnid));
- MDBX_ASAN_UNPOISON_MEMORY_REGION(freedb_root,
- sizeof(freedb_root->mp_txnid));
- const txnid_t root_txnid = freedb_root->mp_txnid;
- if (unlikely(root_txnid != freedb_mod_txnid)) {
- if (report)
- mdbx_warning(
- "catch invalid root_page_txnid %" PRIaTXN
- " for %sdb_mod_txnid %" PRIaTXN " %s",
- root_txnid, "free", maindb_mod_txnid,
- "(workaround for incoherent flaw of unified page/buffer cache)");
- ok = false;
- }
- }
- if (likely(maindb_root && maindb_mod_txnid)) {
- VALGRIND_MAKE_MEM_DEFINED(maindb_root, sizeof(maindb_root->mp_txnid));
- MDBX_ASAN_UNPOISON_MEMORY_REGION(maindb_root,
- sizeof(maindb_root->mp_txnid));
- const txnid_t root_txnid = maindb_root->mp_txnid;
- if (unlikely(root_txnid != maindb_mod_txnid)) {
- if (report)
- mdbx_warning(
- "catch invalid root_page_txnid %" PRIaTXN
- " for %sdb_mod_txnid %" PRIaTXN " %s",
- root_txnid, "main", maindb_mod_txnid,
- "(workaround for incoherent flaw of unified page/buffer cache)");
- ok = false;
- }
- }
- return ok;
-}
-
-/* check with timeout as the workaround
- * for todo4recovery://erased_by_github/libmdbx/issues/269 */
-static int meta_waittxnid(const MDBX_env *env, const MDBX_meta *meta,
- uint64_t *timestamp) {
- if (likely(meta_checktxnid(env, (const MDBX_meta *)meta, !*timestamp)))
- return MDBX_SUCCESS;
-
- if (!*timestamp)
- *timestamp = mdbx_osal_monotime();
- else if (unlikely(mdbx_osal_monotime() - *timestamp > 65536 / 10)) {
- mdbx_error("bailout waiting for valid snapshot %s",
- "(workaround for incoherent flaw of unified page/buffer cache)");
- return MDBX_CORRUPTED;
- }
-
-#if defined(_WIN32) || defined(_WIN64)
- SwitchToThread();
-#elif defined(__linux__) || defined(__gnu_linux__) || defined(_UNIX03_SOURCE)
- sched_yield();
-#elif (defined(_GNU_SOURCE) && __GLIBC_PREREQ(2, 1)) || defined(_OPEN_THREADS)
- pthread_yield();
-#else
- usleep(42);
-#endif
- return MDBX_RESULT_TRUE;
-}
-
-/* Common code for mdbx_txn_begin() and mdbx_txn_renew(). */
-static int mdbx_txn_renew0(MDBX_txn *txn, const unsigned flags) {
- MDBX_env *env = txn->mt_env;
- int rc;
-
-#if MDBX_ENV_CHECKPID
- if (unlikely(env->me_pid != mdbx_getpid())) {
- env->me_flags |= MDBX_FATAL_ERROR;
- return MDBX_PANIC;
- }
-#endif /* MDBX_ENV_CHECKPID */
-
- STATIC_ASSERT(sizeof(MDBX_reader) == 32);
-#if MDBX_LOCKING > 0
- STATIC_ASSERT(offsetof(MDBX_lockinfo, mti_wlock) % MDBX_CACHELINE_SIZE == 0);
- STATIC_ASSERT(offsetof(MDBX_lockinfo, mti_rlock) % MDBX_CACHELINE_SIZE == 0);
-#else
- STATIC_ASSERT(
- offsetof(MDBX_lockinfo, mti_oldest_reader) % MDBX_CACHELINE_SIZE == 0);
- STATIC_ASSERT(offsetof(MDBX_lockinfo, mti_numreaders) % MDBX_CACHELINE_SIZE ==
- 0);
-#endif /* MDBX_LOCKING */
- STATIC_ASSERT(offsetof(MDBX_lockinfo, mti_readers) % MDBX_CACHELINE_SIZE ==
- 0);
-
- const uintptr_t tid = mdbx_thread_self();
- if (flags & MDBX_TXN_RDONLY) {
- mdbx_assert(env, (flags & ~(MDBX_TXN_RO_BEGIN_FLAGS | MDBX_WRITEMAP)) == 0);
- txn->mt_flags =
- MDBX_TXN_RDONLY | (env->me_flags & (MDBX_NOTLS | MDBX_WRITEMAP));
- MDBX_reader *r = txn->to.reader;
- STATIC_ASSERT(sizeof(uintptr_t) <= sizeof(r->mr_tid));
- if (likely(env->me_flags & MDBX_ENV_TXKEY)) {
- mdbx_assert(env, !(env->me_flags & MDBX_NOTLS));
- r = thread_rthc_get(env->me_txkey);
- if (likely(r)) {
- if (unlikely(!r->mr_pid.weak) &&
- (mdbx_runtime_flags & MDBX_DBG_LEGACY_MULTIOPEN)) {
- thread_rthc_set(env->me_txkey, nullptr);
- r = nullptr;
- } else {
- mdbx_assert(env, r->mr_pid.weak == env->me_pid);
- mdbx_assert(env, r->mr_tid.weak == mdbx_thread_self());
- }
- }
- } else {
- mdbx_assert(env, !env->me_lck_mmap.lck || (env->me_flags & MDBX_NOTLS));
- }
-
- if (likely(r)) {
- if (unlikely(r->mr_pid.weak != env->me_pid ||
- r->mr_txnid.weak < SAFE64_INVALID_THRESHOLD))
- return MDBX_BAD_RSLOT;
- } else if (env->me_lck_mmap.lck) {
- bind_rslot_result brs = bind_rslot(env, tid);
- if (unlikely(brs.err != MDBX_SUCCESS))
- return brs.err;
- r = brs.rslot;
- }
- txn->to.reader = r;
- if (flags & (MDBX_TXN_RDONLY_PREPARE - MDBX_TXN_RDONLY)) {
- mdbx_assert(env, txn->mt_txnid == 0);
- mdbx_assert(env, txn->mt_owner == 0);
- mdbx_assert(env, txn->mt_numdbs == 0);
- if (likely(r)) {
- mdbx_assert(env, r->mr_snapshot_pages_used.weak == 0);
- mdbx_assert(env, r->mr_txnid.weak >= SAFE64_INVALID_THRESHOLD);
- atomic_store32(&r->mr_snapshot_pages_used, 0, mo_Relaxed);
- }
- txn->mt_flags = MDBX_TXN_RDONLY | MDBX_TXN_FINISHED;
- return MDBX_SUCCESS;
- }
-
- /* Seek & fetch the last meta */
- if (likely(/* not recovery mode */ env->me_stuck_meta < 0)) {
- uint64_t timestamp = 0;
- while (1) {
- volatile const MDBX_meta *const meta = meta_prefer_last(env);
- mdbx_jitter4testing(false);
- const txnid_t snap = meta_txnid(env, meta);
- mdbx_jitter4testing(false);
- if (likely(r)) {
- safe64_reset(&r->mr_txnid, false);
- atomic_store32(&r->mr_snapshot_pages_used, meta->mm_geo.next,
- mo_Relaxed);
- atomic_store64(&r->mr_snapshot_pages_retired,
- unaligned_peek_u64_volatile(4, meta->mm_pages_retired),
- mo_Relaxed);
- safe64_write(&r->mr_txnid, snap);
- mdbx_jitter4testing(false);
- mdbx_assert(env, r->mr_pid.weak == mdbx_getpid());
- mdbx_assert(
- env, r->mr_tid.weak ==
- ((env->me_flags & MDBX_NOTLS) ? 0 : mdbx_thread_self()));
- mdbx_assert(env, r->mr_txnid.weak == snap);
- atomic_store32(&env->me_lck->mti_readers_refresh_flag, true,
- mo_AcquireRelease);
- } else {
- /* exclusive mode without lck */
- }
- mdbx_jitter4testing(true);
-
- /* Snap the state from current meta-head */
- txn->mt_txnid = snap;
- txn->mt_geo = meta->mm_geo;
- STATIC_ASSERT(CORE_DBS == 2);
- txn->mt_dbs[0] = meta->mm_dbs[0];
- txn->mt_dbs[1] = meta->mm_dbs[1];
- txn->mt_canary = meta->mm_canary;
-
- /* LY: Retry on a race, ITS#7970.
- * The barrier is not needed here since C11-atomics are used,
- * but it is reasonable paranoia to avoid compiler misoptimization
- * and makes clarity for code readers. */
- mdbx_compiler_barrier();
- if (likely(meta == meta_prefer_last(env) &&
- snap == meta_txnid(env, meta) &&
- snap >= atomic_load64(&env->me_lck->mti_oldest_reader,
- mo_AcquireRelease))) {
- /* workaround for todo4recovery://erased_by_github/libmdbx/issues/269
- */
- rc = meta_waittxnid(env, (const MDBX_meta *)meta, ×tamp);
- mdbx_jitter4testing(false);
- if (likely(rc == MDBX_SUCCESS))
- break;
- if (likely(rc == MDBX_RESULT_TRUE))
- continue;
- goto bailout;
- }
- }
- } else {
- /* r/o recovery mode */
- MDBX_meta *const meta = METAPAGE(env, env->me_stuck_meta);
- txn->mt_txnid = constmeta_txnid(env, meta);
- txn->mt_geo = meta->mm_geo;
- memcpy(txn->mt_dbs, meta->mm_dbs, CORE_DBS * sizeof(MDBX_db));
- txn->mt_canary = meta->mm_canary;
- if (likely(r)) {
- atomic_store32(&r->mr_snapshot_pages_used, meta->mm_geo.next,
- mo_Relaxed);
- atomic_store64(&r->mr_snapshot_pages_retired,
- unaligned_peek_u64(4, meta->mm_pages_retired),
- mo_Relaxed);
- atomic_store64(&r->mr_txnid, txn->mt_txnid, mo_Relaxed);
- mdbx_jitter4testing(false);
- mdbx_assert(env, r->mr_pid.weak == mdbx_getpid());
- mdbx_assert(
- env, r->mr_tid.weak ==
- ((env->me_flags & MDBX_NOTLS) ? 0 : mdbx_thread_self()));
- mdbx_assert(env, r->mr_txnid.weak == txn->mt_txnid);
- atomic_store32(&env->me_lck->mti_readers_refresh_flag, true,
- mo_Relaxed);
- }
- }
-
- if (unlikely(txn->mt_txnid < MIN_TXNID || txn->mt_txnid > MAX_TXNID)) {
- mdbx_error("%s", "environment corrupted by died writer, must shutdown!");
- rc = MDBX_CORRUPTED;
- goto bailout;
- }
- mdbx_assert(env, txn->mt_txnid >= env->me_lck->mti_oldest_reader.weak);
- txn->mt_dbxs = env->me_dbxs; /* mostly static anyway */
- mdbx_ensure(env, txn->mt_txnid >=
- /* paranoia is appropriate here */ env->me_lck
- ->mti_oldest_reader.weak);
- txn->mt_numdbs = env->me_numdbs;
- } else {
- mdbx_assert(env, (flags & ~(MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_SPILLS |
- MDBX_WRITEMAP)) == 0);
- if (unlikely(txn->mt_owner == tid ||
- /* not recovery mode */ env->me_stuck_meta >= 0))
- return MDBX_BUSY;
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (lck && (env->me_flags & MDBX_NOTLS) == 0 &&
- (mdbx_runtime_flags & MDBX_DBG_LEGACY_OVERLAP) == 0) {
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- for (unsigned i = 0; i < snap_nreaders; ++i) {
- if (atomic_load32(&lck->mti_readers[i].mr_pid, mo_Relaxed) ==
- env->me_pid &&
- unlikely(atomic_load64(&lck->mti_readers[i].mr_tid, mo_Relaxed) ==
- tid)) {
- const txnid_t txnid = safe64_read(&lck->mti_readers[i].mr_txnid);
- if (txnid >= MIN_TXNID && txnid <= MAX_TXNID)
- return MDBX_TXN_OVERLAPPING;
- }
- }
- }
-
- /* Not yet touching txn == env->me_txn0, it may be active */
- mdbx_jitter4testing(false);
- rc = mdbx_txn_lock(env, F_ISSET(flags, MDBX_TXN_TRY));
- if (unlikely(rc))
- return rc;
- if (unlikely(env->me_flags & MDBX_FATAL_ERROR)) {
- mdbx_txn_unlock(env);
- return MDBX_PANIC;
- }
-#if defined(_WIN32) || defined(_WIN64)
- if (unlikely(!env->me_map)) {
- mdbx_txn_unlock(env);
- return MDBX_EPERM;
- }
-#endif /* Windows */
-
- mdbx_jitter4testing(false);
- const MDBX_meta *meta = constmeta_prefer_last(env);
- uint64_t timestamp = 0;
- while (
- "workaround for todo4recovery://erased_by_github/libmdbx/issues/269") {
- rc = meta_waittxnid(env, (const MDBX_meta *)meta, ×tamp);
- if (likely(rc == MDBX_SUCCESS))
- break;
- if (unlikely(rc != MDBX_RESULT_TRUE))
- goto bailout;
- }
- mdbx_jitter4testing(false);
- txn->mt_canary = meta->mm_canary;
- const txnid_t snap = constmeta_txnid(env, meta);
- txn->mt_txnid = safe64_txnid_next(snap);
- if (unlikely(txn->mt_txnid > MAX_TXNID)) {
- rc = MDBX_TXN_FULL;
- mdbx_error("txnid overflow, raise %d", rc);
- goto bailout;
- }
-
- txn->mt_flags = flags;
- txn->mt_child = NULL;
- txn->tw.loose_pages = NULL;
- txn->tw.loose_count = 0;
-#if MDBX_ENABLE_REFUND
- txn->tw.loose_refund_wl = 0;
-#endif /* MDBX_ENABLE_REFUND */
- MDBX_PNL_SIZE(txn->tw.retired_pages) = 0;
- txn->tw.spill_pages = NULL;
- txn->tw.spill_least_removed = 0;
- txn->tw.last_reclaimed = 0;
- if (txn->tw.lifo_reclaimed)
- MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) = 0;
- env->me_txn = txn;
- txn->mt_numdbs = env->me_numdbs;
- memcpy(txn->mt_dbiseqs, env->me_dbiseqs, txn->mt_numdbs * sizeof(unsigned));
- /* Copy the DB info and flags */
- memcpy(txn->mt_dbs, meta->mm_dbs, CORE_DBS * sizeof(MDBX_db));
- /* Moved to here to avoid a data race in read TXNs */
- txn->mt_geo = meta->mm_geo;
-
- rc = mdbx_dpl_alloc(txn);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- txn->tw.dirtyroom = txn->mt_env->me_options.dp_limit;
- txn->tw.dirtylru = MDBX_DEBUG ? ~42u : 0;
- }
-
- /* Setup db info */
- mdbx_compiler_barrier();
- memset(txn->mt_cursors, 0, sizeof(MDBX_cursor *) * txn->mt_numdbs);
- for (unsigned i = CORE_DBS; i < txn->mt_numdbs; i++) {
- const unsigned db_flags = env->me_dbflags[i];
- txn->mt_dbs[i].md_flags = db_flags & DB_PERSISTENT_FLAGS;
- txn->mt_dbistate[i] =
- (db_flags & DB_VALID) ? DBI_VALID | DBI_USRVALID | DBI_STALE : 0;
- }
- txn->mt_dbistate[MAIN_DBI] = DBI_VALID | DBI_USRVALID;
- txn->mt_dbistate[FREE_DBI] = DBI_VALID;
- txn->mt_front =
- txn->mt_txnid + ((flags & (MDBX_WRITEMAP | MDBX_RDONLY)) == 0);
-
- if (unlikely(env->me_flags & MDBX_FATAL_ERROR)) {
- mdbx_warning("%s", "environment had fatal error, must shutdown!");
- rc = MDBX_PANIC;
- } else {
- const size_t size =
- pgno2bytes(env, (txn->mt_flags & MDBX_TXN_RDONLY) ? txn->mt_next_pgno
- : txn->mt_end_pgno);
- if (unlikely(size > env->me_dxb_mmap.limit)) {
- if (txn->mt_geo.upper > MAX_PAGENO + 1 ||
- bytes2pgno(env, pgno2bytes(env, txn->mt_geo.upper)) !=
- txn->mt_geo.upper) {
- rc = MDBX_UNABLE_EXTEND_MAPSIZE;
- goto bailout;
- }
- rc = mdbx_mapresize(env, txn->mt_next_pgno, txn->mt_end_pgno,
- txn->mt_geo.upper,
- (txn->mt_flags & MDBX_TXN_RDONLY) ? true : false);
- if (rc != MDBX_SUCCESS)
- goto bailout;
- }
- if (txn->mt_flags & MDBX_TXN_RDONLY) {
-#if defined(_WIN32) || defined(_WIN64)
- if (((size > env->me_dbgeo.lower && env->me_dbgeo.shrink) ||
- (mdbx_RunningUnderWine() &&
- /* under Wine acquisition of remap_guard is always required,
- * since Wine don't support section extending,
- * i.e. in both cases unmap+map are required. */
- size < env->me_dbgeo.upper && env->me_dbgeo.grow)) &&
- /* avoid recursive use SRW */ (txn->mt_flags & MDBX_NOTLS) == 0) {
- txn->mt_flags |= MDBX_SHRINK_ALLOWED;
- mdbx_srwlock_AcquireShared(&env->me_remap_guard);
- }
-#endif /* Windows */
- } else {
- env->me_dxb_mmap.current = size;
- env->me_dxb_mmap.filesize =
- (env->me_dxb_mmap.filesize < size) ? size : env->me_dxb_mmap.filesize;
- }
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- mdbx_txn_valgrind(env, txn);
-#endif
- txn->mt_owner = tid;
- return MDBX_SUCCESS;
- }
-bailout:
- mdbx_tassert(txn, rc != MDBX_SUCCESS);
- mdbx_txn_end(txn, MDBX_END_SLOT | MDBX_END_FAIL_BEGIN);
- return rc;
-}
-
-static __always_inline int check_txn(const MDBX_txn *txn, int bad_bits) {
- if (unlikely(!txn))
- return MDBX_EINVAL;
-
- if (unlikely(txn->mt_signature != MDBX_MT_SIGNATURE))
- return MDBX_EBADSIGN;
-
- if (unlikely(txn->mt_flags & bad_bits))
- return MDBX_BAD_TXN;
-
-#if MDBX_TXN_CHECKOWNER
- if ((txn->mt_flags & MDBX_NOTLS) == 0 &&
- unlikely(txn->mt_owner != mdbx_thread_self()))
- return txn->mt_owner ? MDBX_THREAD_MISMATCH : MDBX_BAD_TXN;
-#endif /* MDBX_TXN_CHECKOWNER */
-
- if (unlikely(!txn->mt_env->me_map))
- return MDBX_EPERM;
-
- return MDBX_SUCCESS;
-}
-
-static __always_inline int check_txn_rw(const MDBX_txn *txn, int bad_bits) {
- int err = check_txn(txn, bad_bits);
- if (unlikely(err))
- return err;
-
- if (unlikely(F_ISSET(txn->mt_flags, MDBX_TXN_RDONLY)))
- return MDBX_EACCESS;
-
- return MDBX_SUCCESS;
-}
-
-int mdbx_txn_renew(MDBX_txn *txn) {
- if (unlikely(!txn))
- return MDBX_EINVAL;
-
- if (unlikely(txn->mt_signature != MDBX_MT_SIGNATURE))
- return MDBX_EBADSIGN;
-
- if (unlikely((txn->mt_flags & MDBX_TXN_RDONLY) == 0))
- return MDBX_EINVAL;
-
- int rc;
- if (unlikely(txn->mt_owner != 0 || !(txn->mt_flags & MDBX_TXN_FINISHED))) {
- rc = mdbx_txn_reset(txn);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
-
- rc = mdbx_txn_renew0(txn, MDBX_TXN_RDONLY);
- if (rc == MDBX_SUCCESS) {
- txn->mt_owner = mdbx_thread_self();
- mdbx_debug("renew txn %" PRIaTXN "%c %p on env %p, root page %" PRIaPGNO
- "/%" PRIaPGNO,
- txn->mt_txnid, (txn->mt_flags & MDBX_TXN_RDONLY) ? 'r' : 'w',
- (void *)txn, (void *)txn->mt_env, txn->mt_dbs[MAIN_DBI].md_root,
- txn->mt_dbs[FREE_DBI].md_root);
- }
- return rc;
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-int mdbx_txn_begin(MDBX_env *env, MDBX_txn *parent, MDBX_txn_flags_t flags,
- MDBX_txn **ret) {
- return __inline_mdbx_txn_begin(env, parent, flags, ret);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-int mdbx_txn_set_userctx(MDBX_txn *txn, void *ctx) {
- int rc = check_txn(txn, MDBX_TXN_FINISHED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- txn->mt_userctx = ctx;
- return MDBX_SUCCESS;
-}
-
-void *mdbx_txn_get_userctx(const MDBX_txn *txn) {
- return check_txn(txn, MDBX_TXN_FINISHED) ? nullptr : txn->mt_userctx;
-}
-
-int mdbx_txn_begin_ex(MDBX_env *env, MDBX_txn *parent, MDBX_txn_flags_t flags,
- MDBX_txn **ret, void *context) {
- MDBX_txn *txn;
- unsigned size, tsize;
-
- if (unlikely(!ret))
- return MDBX_EINVAL;
- *ret = NULL;
-
- if (unlikely((flags & ~MDBX_TXN_RW_BEGIN_FLAGS) &&
- (flags & ~MDBX_TXN_RO_BEGIN_FLAGS)))
- return MDBX_EINVAL;
-
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(env->me_flags & MDBX_RDONLY &
- ~flags)) /* write txn in RDONLY env */
- return MDBX_EACCESS;
-
- flags |= env->me_flags & MDBX_WRITEMAP;
-
- if (parent) {
- /* Nested transactions: Max 1 child, write txns only, no writemap */
- rc = check_txn_rw(parent,
- MDBX_TXN_RDONLY | MDBX_WRITEMAP | MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (env->me_options.spill_parent4child_denominator) {
- /* Spill dirty-pages of parent to provide dirtyroom for child txn */
- rc = mdbx_txn_spill(parent, nullptr,
- parent->tw.dirtylist->length /
- env->me_options.spill_parent4child_denominator);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
- mdbx_tassert(parent, mdbx_audit_ex(parent, 0, false) == 0);
-
- flags |= parent->mt_flags & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_SPILLS);
- } else if (flags & MDBX_TXN_RDONLY) {
- if (env->me_txn0 &&
- unlikely(env->me_txn0->mt_owner == mdbx_thread_self()) &&
- (mdbx_runtime_flags & MDBX_DBG_LEGACY_OVERLAP) == 0)
- return MDBX_TXN_OVERLAPPING;
- } else {
- /* Reuse preallocated write txn. However, do not touch it until
- * mdbx_txn_renew0() succeeds, since it currently may be active. */
- txn = env->me_txn0;
- goto renew;
- }
-
- size = env->me_maxdbs * (sizeof(MDBX_db) + sizeof(MDBX_cursor *) + 1);
- size += tsize = sizeof(MDBX_txn);
- if (unlikely((txn = mdbx_malloc(size)) == NULL)) {
- mdbx_debug("calloc: %s", "failed");
- return MDBX_ENOMEM;
- }
-#if MDBX_DEBUG
- memset(txn, 0xCD, size);
- VALGRIND_MAKE_MEM_UNDEFINED(txn, size);
-#endif /* MDBX_DEBUG */
- memset(txn, 0, tsize);
- txn->mt_dbxs = env->me_dbxs; /* static */
- txn->mt_dbs = (MDBX_db *)((char *)txn + tsize);
- txn->mt_cursors = (MDBX_cursor **)(txn->mt_dbs + env->me_maxdbs);
- txn->mt_dbistate = (uint8_t *)txn + size - env->me_maxdbs;
- txn->mt_flags = flags;
- txn->mt_env = env;
-
- if (parent) {
- mdbx_tassert(parent, mdbx_dirtylist_check(parent));
- txn->mt_dbiseqs = parent->mt_dbiseqs;
- txn->mt_geo = parent->mt_geo;
- rc = mdbx_dpl_alloc(txn);
- if (likely(rc == MDBX_SUCCESS)) {
- const unsigned len =
- MDBX_PNL_SIZE(parent->tw.reclaimed_pglist) + parent->tw.loose_count;
- txn->tw.reclaimed_pglist =
- mdbx_pnl_alloc((len > MDBX_PNL_INITIAL) ? len : MDBX_PNL_INITIAL);
- if (unlikely(!txn->tw.reclaimed_pglist))
- rc = MDBX_ENOMEM;
- }
- if (unlikely(rc != MDBX_SUCCESS)) {
- nested_failed:
- mdbx_pnl_free(txn->tw.reclaimed_pglist);
- mdbx_dpl_free(txn);
- mdbx_free(txn);
- return rc;
- }
-
- /* Move loose pages to reclaimed list */
- if (parent->tw.loose_count) {
- do {
- MDBX_page *lp = parent->tw.loose_pages;
- const unsigned di = mdbx_dpl_exist(parent, lp->mp_pgno);
- mdbx_tassert(parent, di && parent->tw.dirtylist->items[di].ptr == lp);
- mdbx_tassert(parent, lp->mp_flags == P_LOOSE);
- rc =
- mdbx_pnl_insert_range(&parent->tw.reclaimed_pglist, lp->mp_pgno, 1);
- if (unlikely(rc != MDBX_SUCCESS))
- goto nested_failed;
- parent->tw.loose_pages = lp->mp_next;
- /* Remove from dirty list */
- mdbx_page_wash(parent, di, lp, 1);
- } while (parent->tw.loose_pages);
- parent->tw.loose_count = 0;
-#if MDBX_ENABLE_REFUND
- parent->tw.loose_refund_wl = 0;
-#endif /* MDBX_ENABLE_REFUND */
- mdbx_tassert(parent, mdbx_dirtylist_check(parent));
- }
- txn->tw.dirtyroom = parent->tw.dirtyroom;
- txn->tw.dirtylru = parent->tw.dirtylru;
-
- mdbx_dpl_sort(parent);
- if (parent->tw.spill_pages)
- mdbx_spill_purge(parent);
-
- mdbx_tassert(txn, MDBX_PNL_ALLOCLEN(txn->tw.reclaimed_pglist) >=
- MDBX_PNL_SIZE(parent->tw.reclaimed_pglist));
- memcpy(txn->tw.reclaimed_pglist, parent->tw.reclaimed_pglist,
- MDBX_PNL_SIZEOF(parent->tw.reclaimed_pglist));
- mdbx_assert(env, mdbx_pnl_check4assert(
- txn->tw.reclaimed_pglist,
- (txn->mt_next_pgno /* LY: intentional assignment here,
- only for assertion */
- = parent->mt_next_pgno) -
- MDBX_ENABLE_REFUND));
-
- txn->tw.last_reclaimed = parent->tw.last_reclaimed;
- if (parent->tw.lifo_reclaimed) {
- txn->tw.lifo_reclaimed = parent->tw.lifo_reclaimed;
- parent->tw.lifo_reclaimed =
- (void *)(intptr_t)MDBX_PNL_SIZE(parent->tw.lifo_reclaimed);
- }
-
- txn->tw.retired_pages = parent->tw.retired_pages;
- parent->tw.retired_pages =
- (void *)(intptr_t)MDBX_PNL_SIZE(parent->tw.retired_pages);
-
- txn->mt_txnid = parent->mt_txnid;
- txn->mt_front = parent->mt_front + 1;
-#if MDBX_ENABLE_REFUND
- txn->tw.loose_refund_wl = 0;
-#endif /* MDBX_ENABLE_REFUND */
- txn->mt_canary = parent->mt_canary;
- parent->mt_flags |= MDBX_TXN_HAS_CHILD;
- parent->mt_child = txn;
- txn->mt_parent = parent;
- txn->mt_numdbs = parent->mt_numdbs;
- txn->mt_owner = parent->mt_owner;
- memcpy(txn->mt_dbs, parent->mt_dbs, txn->mt_numdbs * sizeof(MDBX_db));
- /* Copy parent's mt_dbistate, but clear DB_NEW */
- for (unsigned i = 0; i < txn->mt_numdbs; i++)
- txn->mt_dbistate[i] =
- parent->mt_dbistate[i] & ~(DBI_FRESH | DBI_CREAT | DBI_DIRTY);
- mdbx_tassert(parent,
- parent->tw.dirtyroom + parent->tw.dirtylist->length ==
- (parent->mt_parent ? parent->mt_parent->tw.dirtyroom
- : parent->mt_env->me_options.dp_limit));
- mdbx_tassert(txn, txn->tw.dirtyroom + txn->tw.dirtylist->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
- env->me_txn = txn;
- rc = mdbx_cursor_shadow(parent, txn);
- if (mdbx_audit_enabled() && mdbx_assert_enabled()) {
- txn->mt_signature = MDBX_MT_SIGNATURE;
- mdbx_tassert(txn, mdbx_audit_ex(txn, 0, false) == 0);
- }
- if (unlikely(rc != MDBX_SUCCESS))
- mdbx_txn_end(txn, MDBX_END_FAIL_BEGINCHILD);
- } else { /* MDBX_TXN_RDONLY */
- txn->mt_dbiseqs = env->me_dbiseqs;
- renew:
- rc = mdbx_txn_renew0(txn, flags);
- }
-
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (txn != env->me_txn0)
- mdbx_free(txn);
- } else {
- if (flags & (MDBX_TXN_RDONLY_PREPARE - MDBX_TXN_RDONLY))
- mdbx_assert(env, txn->mt_flags == (MDBX_TXN_RDONLY | MDBX_TXN_FINISHED));
- else if (flags & MDBX_TXN_RDONLY)
- mdbx_assert(env, (txn->mt_flags &
- ~(MDBX_NOTLS | MDBX_TXN_RDONLY | MDBX_WRITEMAP |
- /* Win32: SRWL flag */ MDBX_SHRINK_ALLOWED)) == 0);
- else {
- mdbx_assert(env, (txn->mt_flags & ~(MDBX_WRITEMAP | MDBX_SHRINK_ALLOWED |
- MDBX_NOMETASYNC | MDBX_SAFE_NOSYNC |
- MDBX_TXN_SPILLS)) == 0);
- assert(!txn->tw.spill_pages && !txn->tw.spill_least_removed);
- }
- txn->mt_signature = MDBX_MT_SIGNATURE;
- txn->mt_userctx = context;
- *ret = txn;
- mdbx_debug("begin txn %" PRIaTXN "%c %p on env %p, root page %" PRIaPGNO
- "/%" PRIaPGNO,
- txn->mt_txnid, (flags & MDBX_TXN_RDONLY) ? 'r' : 'w',
- (void *)txn, (void *)env, txn->mt_dbs[MAIN_DBI].md_root,
- txn->mt_dbs[FREE_DBI].md_root);
- }
-
- return rc;
-}
-
-int mdbx_txn_info(const MDBX_txn *txn, MDBX_txn_info *info, bool scan_rlt) {
- int rc = check_txn(txn, MDBX_TXN_FINISHED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!info))
- return MDBX_EINVAL;
-
- MDBX_env *const env = txn->mt_env;
-#if MDBX_ENV_CHECKPID
- if (unlikely(env->me_pid != mdbx_getpid())) {
- env->me_flags |= MDBX_FATAL_ERROR;
- return MDBX_PANIC;
- }
-#endif /* MDBX_ENV_CHECKPID */
-
- info->txn_id = txn->mt_txnid;
- info->txn_space_used = pgno2bytes(env, txn->mt_geo.next);
-
- if (txn->mt_flags & MDBX_TXN_RDONLY) {
- volatile const MDBX_meta *head_meta;
- txnid_t head_txnid;
- uint64_t head_retired;
- do {
- /* fetch info from volatile head */
- head_meta = meta_prefer_last(env);
- head_txnid = meta_txnid(env, head_meta);
- head_retired =
- unaligned_peek_u64_volatile(4, head_meta->mm_pages_retired);
- info->txn_space_limit_soft = pgno2bytes(env, head_meta->mm_geo.now);
- info->txn_space_limit_hard = pgno2bytes(env, head_meta->mm_geo.upper);
- info->txn_space_leftover =
- pgno2bytes(env, head_meta->mm_geo.now - head_meta->mm_geo.next);
- mdbx_compiler_barrier();
- } while (unlikely(head_meta != meta_prefer_last(env) ||
- head_txnid != meta_txnid(env, head_meta)));
-
- info->txn_reader_lag = head_txnid - info->txn_id;
- info->txn_space_dirty = info->txn_space_retired = 0;
- uint64_t reader_snapshot_pages_retired;
- if (txn->to.reader &&
- head_retired >
- (reader_snapshot_pages_retired = atomic_load64(
- &txn->to.reader->mr_snapshot_pages_retired, mo_Relaxed))) {
- info->txn_space_dirty = info->txn_space_retired = pgno2bytes(
- env, (pgno_t)(head_retired - reader_snapshot_pages_retired));
-
- size_t retired_next_reader = 0;
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (scan_rlt && info->txn_reader_lag > 1 && lck) {
- /* find next more recent reader */
- txnid_t next_reader = head_txnid;
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- for (unsigned i = 0; i < snap_nreaders; ++i) {
- retry:
- if (atomic_load32(&lck->mti_readers[i].mr_pid, mo_AcquireRelease)) {
- mdbx_jitter4testing(true);
- const txnid_t snap_txnid =
- safe64_read(&lck->mti_readers[i].mr_txnid);
- const uint64_t snap_retired =
- atomic_load64(&lck->mti_readers[i].mr_snapshot_pages_retired,
- mo_AcquireRelease);
- if (unlikely(snap_retired !=
- atomic_load64(
- &lck->mti_readers[i].mr_snapshot_pages_retired,
- mo_Relaxed)) ||
- snap_txnid != safe64_read(&lck->mti_readers[i].mr_txnid))
- goto retry;
- if (snap_txnid <= txn->mt_txnid) {
- retired_next_reader = 0;
- break;
- }
- if (snap_txnid < next_reader) {
- next_reader = snap_txnid;
- retired_next_reader = pgno2bytes(
- env, (pgno_t)(snap_retired -
- atomic_load64(
- &txn->to.reader->mr_snapshot_pages_retired,
- mo_Relaxed)));
- }
- }
- }
- }
- info->txn_space_dirty = retired_next_reader;
- }
- } else {
- info->txn_space_limit_soft = pgno2bytes(env, txn->mt_geo.now);
- info->txn_space_limit_hard = pgno2bytes(env, txn->mt_geo.upper);
- info->txn_space_retired = pgno2bytes(
- env, txn->mt_child ? (unsigned)(uintptr_t)txn->tw.retired_pages
- : MDBX_PNL_SIZE(txn->tw.retired_pages));
- info->txn_space_leftover = pgno2bytes(env, txn->tw.dirtyroom);
- info->txn_space_dirty =
- pgno2bytes(env, txn->mt_env->me_options.dp_limit - txn->tw.dirtyroom);
- info->txn_reader_lag = INT64_MAX;
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (scan_rlt && lck) {
- txnid_t oldest_snapshot = txn->mt_txnid;
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- if (snap_nreaders) {
- oldest_snapshot = mdbx_find_oldest(txn);
- if (oldest_snapshot == txn->mt_txnid - 1) {
- /* check if there is at least one reader */
- bool exists = false;
- for (unsigned i = 0; i < snap_nreaders; ++i) {
- if (atomic_load32(&lck->mti_readers[i].mr_pid, mo_Relaxed) &&
- txn->mt_txnid > safe64_read(&lck->mti_readers[i].mr_txnid)) {
- exists = true;
- break;
- }
- }
- oldest_snapshot += !exists;
- }
- }
- info->txn_reader_lag = txn->mt_txnid - oldest_snapshot;
- }
- }
-
- return MDBX_SUCCESS;
-}
-
-MDBX_env *mdbx_txn_env(const MDBX_txn *txn) {
- if (unlikely(!txn || txn->mt_signature != MDBX_MT_SIGNATURE ||
- txn->mt_env->me_signature.weak != MDBX_ME_SIGNATURE))
- return NULL;
- return txn->mt_env;
-}
-
-uint64_t mdbx_txn_id(const MDBX_txn *txn) {
- if (unlikely(!txn || txn->mt_signature != MDBX_MT_SIGNATURE))
- return 0;
- return txn->mt_txnid;
-}
-
-int mdbx_txn_flags(const MDBX_txn *txn) {
- if (unlikely(!txn || txn->mt_signature != MDBX_MT_SIGNATURE))
- return -1;
- return txn->mt_flags;
-}
-
-/* Check for misused dbi handles */
-#define TXN_DBI_CHANGED(txn, dbi) \
- ((txn)->mt_dbiseqs[dbi] != (txn)->mt_env->me_dbiseqs[dbi])
-
-static void dbi_import_locked(MDBX_txn *txn) {
- MDBX_env *const env = txn->mt_env;
- const unsigned n = env->me_numdbs;
- for (unsigned i = CORE_DBS; i < n; ++i) {
- if (i >= txn->mt_numdbs) {
- txn->mt_dbistate[i] = 0;
- txn->mt_cursors[i] = NULL;
- }
- if ((env->me_dbflags[i] & DB_VALID) &&
- !(txn->mt_dbistate[i] & DBI_USRVALID)) {
- txn->mt_dbiseqs[i] = env->me_dbiseqs[i];
- txn->mt_dbs[i].md_flags = env->me_dbflags[i] & DB_PERSISTENT_FLAGS;
- txn->mt_dbistate[i] = DBI_VALID | DBI_USRVALID | DBI_STALE;
- mdbx_tassert(txn, txn->mt_dbxs[i].md_cmp != NULL);
- mdbx_tassert(txn, txn->mt_dbxs[i].md_name.iov_base != NULL);
- }
- }
- txn->mt_numdbs = n;
-}
-
-/* Import DBI which opened after txn started into context */
-__cold static bool dbi_import(MDBX_txn *txn, MDBX_dbi dbi) {
- if (dbi < CORE_DBS || dbi >= txn->mt_env->me_numdbs)
- return false;
-
- mdbx_ensure(txn->mt_env, mdbx_fastmutex_acquire(&txn->mt_env->me_dbi_lock) ==
- MDBX_SUCCESS);
- dbi_import_locked(txn);
- mdbx_ensure(txn->mt_env, mdbx_fastmutex_release(&txn->mt_env->me_dbi_lock) ==
- MDBX_SUCCESS);
- return txn->mt_dbistate[dbi] & DBI_USRVALID;
-}
-
-/* Export or close DBI handles opened in this txn. */
-static void dbi_update(MDBX_txn *txn, int keep) {
- mdbx_tassert(txn, !txn->mt_parent && txn == txn->mt_env->me_txn0);
- MDBX_dbi n = txn->mt_numdbs;
- if (n) {
- bool locked = false;
- MDBX_env *const env = txn->mt_env;
-
- for (unsigned i = n; --i >= CORE_DBS;) {
- if (likely((txn->mt_dbistate[i] & DBI_CREAT) == 0))
- continue;
- if (!locked) {
- mdbx_ensure(env,
- mdbx_fastmutex_acquire(&env->me_dbi_lock) == MDBX_SUCCESS);
- locked = true;
- }
- if (env->me_numdbs <= i || txn->mt_dbiseqs[i] != env->me_dbiseqs[i])
- continue /* dbi explicitly closed and/or then re-opened by other txn */;
- if (keep) {
- env->me_dbflags[i] = txn->mt_dbs[i].md_flags | DB_VALID;
- } else {
- char *ptr = env->me_dbxs[i].md_name.iov_base;
- if (ptr) {
- env->me_dbxs[i].md_name.iov_len = 0;
- mdbx_memory_fence(mo_AcquireRelease, true);
- mdbx_assert(env, env->me_dbflags[i] == 0);
- env->me_dbiseqs[i]++;
- env->me_dbxs[i].md_name.iov_base = NULL;
- mdbx_free(ptr);
- }
- }
- }
-
- n = env->me_numdbs;
- if (n > CORE_DBS && unlikely(!(env->me_dbflags[n - 1] & DB_VALID))) {
- if (!locked) {
- mdbx_ensure(env,
- mdbx_fastmutex_acquire(&env->me_dbi_lock) == MDBX_SUCCESS);
- locked = true;
- }
-
- n = env->me_numdbs;
- while (n > CORE_DBS && !(env->me_dbflags[n - 1] & DB_VALID))
- --n;
- env->me_numdbs = n;
- }
-
- if (unlikely(locked))
- mdbx_ensure(env,
- mdbx_fastmutex_release(&env->me_dbi_lock) == MDBX_SUCCESS);
- }
-}
-
-/* Filter-out pgno list from transaction's dirty-page list */
-static void mdbx_dpl_sift(MDBX_txn *const txn, MDBX_PNL pl,
- const bool spilled) {
- if (MDBX_PNL_SIZE(pl) && txn->tw.dirtylist->length) {
- mdbx_tassert(
- txn, mdbx_pnl_check4assert(pl, (size_t)txn->mt_next_pgno << spilled));
- MDBX_dpl *dl = mdbx_dpl_sort(txn);
-
- /* Scanning in ascend order */
- const int step = MDBX_PNL_ASCENDING ? 1 : -1;
- const int begin = MDBX_PNL_ASCENDING ? 1 : MDBX_PNL_SIZE(pl);
- const int end = MDBX_PNL_ASCENDING ? MDBX_PNL_SIZE(pl) + 1 : 0;
- mdbx_tassert(txn, pl[begin] <= pl[end - step]);
-
- unsigned r = mdbx_dpl_search(txn, pl[begin] >> spilled);
- mdbx_tassert(txn, dl->sorted == dl->length);
- for (int i = begin; r <= dl->length;) { /* scan loop */
- assert(i != end);
- mdbx_tassert(txn, !spilled || (pl[i] & 1) == 0);
- pgno_t pl_pgno = pl[i] >> spilled;
- pgno_t dp_pgno = dl->items[r].pgno;
- if (likely(dp_pgno != pl_pgno)) {
- const bool cmp = dp_pgno < pl_pgno;
- r += cmp;
- i += cmp ? 0 : step;
- if (likely(i != end))
- continue;
- return;
- }
-
- /* update loop */
- unsigned w = r;
- remove_dl:
- if ((txn->mt_env->me_flags & MDBX_WRITEMAP) == 0) {
- MDBX_page *dp = dl->items[r].ptr;
- mdbx_dpage_free(txn->mt_env, dp, dpl_npages(dl, r));
- }
- ++r;
- next_i:
- i += step;
- if (unlikely(i == end)) {
- while (r <= dl->length)
- dl->items[w++] = dl->items[r++];
- } else {
- while (r <= dl->length) {
- assert(i != end);
- mdbx_tassert(txn, !spilled || (pl[i] & 1) == 0);
- pl_pgno = pl[i] >> spilled;
- dp_pgno = dl->items[r].pgno;
- if (dp_pgno < pl_pgno)
- dl->items[w++] = dl->items[r++];
- else if (dp_pgno > pl_pgno)
- goto next_i;
- else
- goto remove_dl;
- }
- }
- dl->sorted = dpl_setlen(dl, w - 1);
- txn->tw.dirtyroom += r - w;
- mdbx_tassert(txn,
- txn->tw.dirtyroom + txn->tw.dirtylist->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
- return;
- }
- }
-}
-
-/* End a transaction, except successful commit of a nested transaction.
- * May be called twice for readonly txns: First reset it, then abort.
- * [in] txn the transaction handle to end
- * [in] mode why and how to end the transaction */
-static int mdbx_txn_end(MDBX_txn *txn, const unsigned mode) {
- MDBX_env *env = txn->mt_env;
- static const char *const names[] = MDBX_END_NAMES;
-
-#if MDBX_ENV_CHECKPID
- if (unlikely(txn->mt_env->me_pid != mdbx_getpid())) {
- env->me_flags |= MDBX_FATAL_ERROR;
- return MDBX_PANIC;
- }
-#endif /* MDBX_ENV_CHECKPID */
-
- mdbx_debug("%s txn %" PRIaTXN "%c %p on mdbenv %p, root page %" PRIaPGNO
- "/%" PRIaPGNO,
- names[mode & MDBX_END_OPMASK], txn->mt_txnid,
- (txn->mt_flags & MDBX_TXN_RDONLY) ? 'r' : 'w', (void *)txn,
- (void *)env, txn->mt_dbs[MAIN_DBI].md_root,
- txn->mt_dbs[FREE_DBI].md_root);
-
- mdbx_ensure(env, txn->mt_txnid >=
- /* paranoia is appropriate here */ env->me_lck
- ->mti_oldest_reader.weak);
-
- if (!(mode & MDBX_END_EOTDONE)) /* !(already closed cursors) */
- mdbx_cursors_eot(txn, false);
-
- int rc = MDBX_SUCCESS;
- if (F_ISSET(txn->mt_flags, MDBX_TXN_RDONLY)) {
- if (txn->to.reader) {
- MDBX_reader *slot = txn->to.reader;
- mdbx_assert(env, slot->mr_pid.weak == env->me_pid);
- if (likely(!F_ISSET(txn->mt_flags, MDBX_TXN_FINISHED))) {
- mdbx_assert(env, txn->mt_txnid == slot->mr_txnid.weak &&
- slot->mr_txnid.weak >=
- env->me_lck->mti_oldest_reader.weak);
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- mdbx_txn_valgrind(env, nullptr);
-#endif
- atomic_store32(&slot->mr_snapshot_pages_used, 0, mo_Relaxed);
- safe64_reset(&slot->mr_txnid, false);
- atomic_store32(&env->me_lck->mti_readers_refresh_flag, true,
- mo_Relaxed);
- } else {
- mdbx_assert(env, slot->mr_pid.weak == env->me_pid);
- mdbx_assert(env, slot->mr_txnid.weak >= SAFE64_INVALID_THRESHOLD);
- }
- if (mode & MDBX_END_SLOT) {
- if ((env->me_flags & MDBX_ENV_TXKEY) == 0)
- atomic_store32(&slot->mr_pid, 0, mo_Relaxed);
- txn->to.reader = NULL;
- }
- }
-#if defined(_WIN32) || defined(_WIN64)
- if (txn->mt_flags & MDBX_SHRINK_ALLOWED)
- mdbx_srwlock_ReleaseShared(&env->me_remap_guard);
-#endif
- txn->mt_numdbs = 0; /* prevent further DBI activity */
- txn->mt_flags = MDBX_TXN_RDONLY | MDBX_TXN_FINISHED;
- txn->mt_owner = 0;
- } else if (!F_ISSET(txn->mt_flags, MDBX_TXN_FINISHED)) {
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- if (txn == env->me_txn0)
- mdbx_txn_valgrind(env, nullptr);
-#endif
-
- txn->mt_flags = MDBX_TXN_FINISHED;
- txn->mt_owner = 0;
- env->me_txn = txn->mt_parent;
- mdbx_pnl_free(txn->tw.spill_pages);
- txn->tw.spill_pages = nullptr;
- if (txn == env->me_txn0) {
- mdbx_assert(env, txn->mt_parent == NULL);
- /* Export or close DBI handles created in this txn */
- dbi_update(txn, mode & MDBX_END_UPDATE);
- mdbx_pnl_shrink(&txn->tw.retired_pages);
- mdbx_pnl_shrink(&txn->tw.reclaimed_pglist);
- if (!(env->me_flags & MDBX_WRITEMAP))
- mdbx_dlist_free(txn);
- /* The writer mutex was locked in mdbx_txn_begin. */
- mdbx_txn_unlock(env);
- } else {
- mdbx_assert(env, txn->mt_parent != NULL);
- MDBX_txn *const parent = txn->mt_parent;
- mdbx_assert(env, parent->mt_signature == MDBX_MT_SIGNATURE);
- mdbx_assert(env, parent->mt_child == txn &&
- (parent->mt_flags & MDBX_TXN_HAS_CHILD) != 0);
- mdbx_assert(
- env, mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
-
- if (txn->tw.lifo_reclaimed) {
- mdbx_assert(env, MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) >=
- (unsigned)(uintptr_t)parent->tw.lifo_reclaimed);
- MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) =
- (unsigned)(uintptr_t)parent->tw.lifo_reclaimed;
- parent->tw.lifo_reclaimed = txn->tw.lifo_reclaimed;
- }
-
- if (txn->tw.retired_pages) {
- mdbx_assert(env, MDBX_PNL_SIZE(txn->tw.retired_pages) >=
- (unsigned)(uintptr_t)parent->tw.retired_pages);
- MDBX_PNL_SIZE(txn->tw.retired_pages) =
- (unsigned)(uintptr_t)parent->tw.retired_pages;
- parent->tw.retired_pages = txn->tw.retired_pages;
- }
-
- parent->mt_child = nullptr;
- parent->mt_flags &= ~MDBX_TXN_HAS_CHILD;
- parent->tw.dirtylru = txn->tw.dirtylru;
- mdbx_tassert(parent, mdbx_dirtylist_check(parent));
- mdbx_tassert(parent, mdbx_audit_ex(parent, 0, false) == 0);
- if (!(env->me_flags & MDBX_WRITEMAP))
- mdbx_dlist_free(txn);
- mdbx_dpl_free(txn);
- mdbx_pnl_free(txn->tw.reclaimed_pglist);
-
- if (parent->mt_geo.upper != txn->mt_geo.upper ||
- parent->mt_geo.now != txn->mt_geo.now) {
- /* undo resize performed by child txn */
- rc = mdbx_mapresize_implicit(env, parent->mt_next_pgno,
- parent->mt_geo.now, parent->mt_geo.upper);
- if (rc == MDBX_EPERM) {
- /* unable undo resize (it is regular for Windows),
- * therefore promote size changes from child to the parent txn */
- mdbx_warning("unable undo resize performed by child txn, promote to "
- "the parent (%u->%u, %u->%u)",
- txn->mt_geo.now, parent->mt_geo.now, txn->mt_geo.upper,
- parent->mt_geo.upper);
- parent->mt_geo.now = txn->mt_geo.now;
- parent->mt_geo.upper = txn->mt_geo.upper;
- parent->mt_flags |= MDBX_TXN_DIRTY;
- rc = MDBX_SUCCESS;
- } else if (unlikely(rc != MDBX_SUCCESS)) {
- mdbx_error("error %d while undo resize performed by child txn, fail "
- "the parent",
- rc);
- parent->mt_flags |= MDBX_TXN_ERROR;
- if (!env->me_dxb_mmap.address)
- env->me_flags |= MDBX_FATAL_ERROR;
- }
- }
- }
- }
-
- mdbx_assert(env, txn == env->me_txn0 || txn->mt_owner == 0);
- if ((mode & MDBX_END_FREE) != 0 && txn != env->me_txn0) {
- txn->mt_signature = 0;
- mdbx_free(txn);
- }
-
- return rc;
-}
-
-int mdbx_txn_reset(MDBX_txn *txn) {
- int rc = check_txn(txn, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- /* This call is only valid for read-only txns */
- if (unlikely((txn->mt_flags & MDBX_TXN_RDONLY) == 0))
- return MDBX_EINVAL;
-
- /* LY: don't close DBI-handles */
- rc = mdbx_txn_end(txn, MDBX_END_RESET | MDBX_END_UPDATE);
- if (rc == MDBX_SUCCESS) {
- mdbx_tassert(txn, txn->mt_signature == MDBX_MT_SIGNATURE);
- mdbx_tassert(txn, txn->mt_owner == 0);
- }
- return rc;
-}
-
-int mdbx_txn_break(MDBX_txn *txn) {
- do {
- int rc = check_txn(txn, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- txn->mt_flags |= MDBX_TXN_ERROR;
- if (txn->mt_flags & MDBX_TXN_RDONLY)
- break;
- txn = txn->mt_child;
- } while (txn);
- return MDBX_SUCCESS;
-}
-
-int mdbx_txn_abort(MDBX_txn *txn) {
- int rc = check_txn(txn, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (F_ISSET(txn->mt_flags, MDBX_TXN_RDONLY))
- /* LY: don't close DBI-handles */
- return mdbx_txn_end(txn, MDBX_END_ABORT | MDBX_END_UPDATE | MDBX_END_SLOT |
- MDBX_END_FREE);
-
- if (txn->mt_child)
- mdbx_txn_abort(txn->mt_child);
-
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- return mdbx_txn_end(txn, MDBX_END_ABORT | MDBX_END_SLOT | MDBX_END_FREE);
-}
-
-/* Count all the pages in each DB and in the GC and make sure
- * it matches the actual number of pages being used. */
-__cold static int mdbx_audit_ex(MDBX_txn *txn, unsigned retired_stored,
- bool dont_filter_gc) {
- pgno_t pending = 0;
- if ((txn->mt_flags & MDBX_TXN_RDONLY) == 0) {
- pending = txn->tw.loose_count + MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) +
- (MDBX_PNL_SIZE(txn->tw.retired_pages) - retired_stored);
- }
-
- MDBX_cursor_couple cx;
- int rc = mdbx_cursor_init(&cx.outer, txn, FREE_DBI);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- pgno_t gc = 0;
- MDBX_val key, data;
- while ((rc = mdbx_cursor_get(&cx.outer, &key, &data, MDBX_NEXT)) == 0) {
- if (!dont_filter_gc) {
- if (unlikely(key.iov_len != sizeof(txnid_t)))
- return MDBX_CORRUPTED;
- txnid_t id = unaligned_peek_u64(4, key.iov_base);
- if (txn->tw.lifo_reclaimed) {
- for (unsigned i = 1; i <= MDBX_PNL_SIZE(txn->tw.lifo_reclaimed); ++i)
- if (id == txn->tw.lifo_reclaimed[i])
- goto skip;
- } else if (id <= txn->tw.last_reclaimed)
- goto skip;
- }
-
- gc += *(pgno_t *)data.iov_base;
- skip:;
- }
- mdbx_tassert(txn, rc == MDBX_NOTFOUND);
-
- for (MDBX_dbi i = FREE_DBI; i < txn->mt_numdbs; i++)
- txn->mt_dbistate[i] &= ~DBI_AUDITED;
-
- pgno_t used = NUM_METAS;
- for (MDBX_dbi i = FREE_DBI; i <= MAIN_DBI; i++) {
- if (!(txn->mt_dbistate[i] & DBI_VALID))
- continue;
- rc = mdbx_cursor_init(&cx.outer, txn, i);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- txn->mt_dbistate[i] |= DBI_AUDITED;
- if (txn->mt_dbs[i].md_root == P_INVALID)
- continue;
- used += txn->mt_dbs[i].md_branch_pages + txn->mt_dbs[i].md_leaf_pages +
- txn->mt_dbs[i].md_overflow_pages;
-
- if (i != MAIN_DBI)
- continue;
- rc = mdbx_page_search(&cx.outer, NULL, MDBX_PS_FIRST);
- while (rc == MDBX_SUCCESS) {
- MDBX_page *mp = cx.outer.mc_pg[cx.outer.mc_top];
- for (unsigned j = 0; j < page_numkeys(mp); j++) {
- MDBX_node *node = page_node(mp, j);
- if (node_flags(node) == F_SUBDATA) {
- if (unlikely(node_ds(node) != sizeof(MDBX_db)))
- return MDBX_CORRUPTED;
- MDBX_db db_copy, *db;
- memcpy(db = &db_copy, node_data(node), sizeof(db_copy));
- if ((txn->mt_flags & MDBX_TXN_RDONLY) == 0) {
- for (MDBX_dbi k = txn->mt_numdbs; --k > MAIN_DBI;) {
- if ((txn->mt_dbistate[k] & DBI_VALID) &&
- /* txn->mt_dbxs[k].md_name.iov_len > 0 && */
- node_ks(node) == txn->mt_dbxs[k].md_name.iov_len &&
- memcmp(node_key(node), txn->mt_dbxs[k].md_name.iov_base,
- node_ks(node)) == 0) {
- txn->mt_dbistate[k] |= DBI_AUDITED;
- if (!(txn->mt_dbistate[k] & MDBX_DBI_STALE))
- db = txn->mt_dbs + k;
- break;
- }
- }
- }
- used +=
- db->md_branch_pages + db->md_leaf_pages + db->md_overflow_pages;
- }
- }
- rc = mdbx_cursor_sibling(&cx.outer, SIBLING_RIGHT);
- }
- mdbx_tassert(txn, rc == MDBX_NOTFOUND);
- }
-
- for (MDBX_dbi i = FREE_DBI; i < txn->mt_numdbs; i++) {
- if ((txn->mt_dbistate[i] & (DBI_VALID | DBI_AUDITED | DBI_STALE)) !=
- DBI_VALID)
- continue;
- for (MDBX_txn *t = txn; t; t = t->mt_parent)
- if (F_ISSET(t->mt_dbistate[i], DBI_DIRTY | DBI_CREAT)) {
- used += t->mt_dbs[i].md_branch_pages + t->mt_dbs[i].md_leaf_pages +
- t->mt_dbs[i].md_overflow_pages;
- txn->mt_dbistate[i] |= DBI_AUDITED;
- break;
- }
- if (!(txn->mt_dbistate[i] & DBI_AUDITED)) {
- mdbx_warning("audit %s@%" PRIaTXN
- ": unable account dbi %d / \"%*s\", state 0x%02x",
- txn->mt_parent ? "nested-" : "", txn->mt_txnid, i,
- (int)txn->mt_dbxs[i].md_name.iov_len,
- (const char *)txn->mt_dbxs[i].md_name.iov_base,
- txn->mt_dbistate[i]);
- }
- }
-
- if (pending + gc + used == txn->mt_next_pgno)
- return MDBX_SUCCESS;
-
- if ((txn->mt_flags & MDBX_TXN_RDONLY) == 0)
- mdbx_error("audit @%" PRIaTXN ": %u(pending) = %u(loose) + "
- "%u(reclaimed) + %u(retired-pending) - %u(retired-stored)",
- txn->mt_txnid, pending, txn->tw.loose_count,
- MDBX_PNL_SIZE(txn->tw.reclaimed_pglist),
- txn->tw.retired_pages ? MDBX_PNL_SIZE(txn->tw.retired_pages) : 0,
- retired_stored);
- mdbx_error("audit @%" PRIaTXN ": %" PRIaPGNO "(pending) + %" PRIaPGNO
- "(gc) + %" PRIaPGNO "(count) = %" PRIaPGNO "(total) <> %" PRIaPGNO
- "(allocated)",
- txn->mt_txnid, pending, gc, used, pending + gc + used,
- txn->mt_next_pgno);
- return MDBX_PROBLEM;
-}
-
-static __always_inline unsigned backlog_size(MDBX_txn *txn) {
- return MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) + txn->tw.loose_count;
-}
-
-/* LY: Prepare a backlog of pages to modify GC itself,
- * while reclaiming is prohibited. It should be enough to prevent search
- * in mdbx_page_alloc() during a deleting, when GC tree is unbalanced. */
-static int mdbx_prep_backlog(MDBX_txn *txn, MDBX_cursor *gc_cursor,
- const size_t pnl_bytes, unsigned *retired_stored) {
- const unsigned linear4list = number_of_ovpages(txn->mt_env, pnl_bytes);
- const unsigned backlog4cow = txn->mt_dbs[FREE_DBI].md_depth;
- const unsigned backlog4rebalance = backlog4cow + 1;
-
- if (likely(linear4list == 1 &&
- backlog_size(txn) > (pnl_bytes
- ? backlog4rebalance
- : (backlog4cow + backlog4rebalance))))
- return MDBX_SUCCESS;
-
- mdbx_trace(">> pnl_bytes %zu, backlog %u, 4list %u, 4cow %u, 4rebalance %u",
- pnl_bytes, backlog_size(txn), linear4list, backlog4cow,
- backlog4rebalance);
-
- MDBX_val gc_key, fake_val;
- int err;
- if (unlikely(linear4list > 2)) {
- gc_key.iov_base = fake_val.iov_base = nullptr;
- gc_key.iov_len = sizeof(txnid_t);
- fake_val.iov_len = pnl_bytes;
- err = mdbx_cursor_spill(gc_cursor, &gc_key, &fake_val);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
-
- gc_cursor->mc_flags &= ~C_RECLAIMING;
- err = mdbx_cursor_touch(gc_cursor);
- mdbx_trace("== after-touch, backlog %u, err %d", backlog_size(txn), err);
-
- if (unlikely(linear4list > 1) && err == MDBX_SUCCESS) {
- if (retired_stored) {
- gc_key.iov_base = &txn->mt_txnid;
- gc_key.iov_len = sizeof(txn->mt_txnid);
- const struct cursor_set_result csr =
- mdbx_cursor_set(gc_cursor, &gc_key, &fake_val, MDBX_SET);
- if (csr.err == MDBX_SUCCESS && csr.exact) {
- *retired_stored = 0;
- err = mdbx_cursor_del(gc_cursor, 0);
- mdbx_trace("== clear-4linear, backlog %u, err %d", backlog_size(txn),
- err);
- }
- }
- err =
- mdbx_page_alloc(gc_cursor, linear4list, MDBX_ALLOC_GC | MDBX_ALLOC_FAKE)
- .err;
- mdbx_trace("== after-4linear, backlog %u, err %d", backlog_size(txn), err);
- mdbx_cassert(gc_cursor,
- backlog_size(txn) >= linear4list || err != MDBX_SUCCESS);
- }
-
- while (backlog_size(txn) < backlog4cow + linear4list && err == MDBX_SUCCESS)
- err = mdbx_page_alloc(gc_cursor, 0,
- MDBX_ALLOC_GC | MDBX_ALLOC_SLOT | MDBX_ALLOC_FAKE)
- .err;
-
- gc_cursor->mc_flags |= C_RECLAIMING;
- mdbx_trace("<< backlog %u, err %d", backlog_size(txn), err);
- return (err != MDBX_NOTFOUND) ? err : MDBX_SUCCESS;
-}
-
-static __inline void clean_reserved_gc_pnl(MDBX_env *env, MDBX_val pnl) {
- /* PNL is initially empty, zero out at least the length */
- memset(pnl.iov_base, 0, sizeof(pgno_t));
- if ((env->me_flags & (MDBX_WRITEMAP | MDBX_NOMEMINIT)) == 0)
- /* zero out to avoid leaking values from uninitialized malloc'ed memory
- * to the file in non-writemap mode if length of the saving page-list
- * was changed during space reservation. */
- memset(pnl.iov_base, 0, pnl.iov_len);
-}
-
-/* Cleanups reclaimed GC (aka freeDB) records, saves the retired-list (aka
- * freelist) of current transaction to GC, puts back into GC leftover of the
- * reclaimed pages with chunking. This recursive changes the reclaimed-list,
- * loose-list and retired-list. Keep trying until it stabilizes.
- *
- * NOTE: This code is a consequence of many iterations of adding crutches (aka
- * "checks and balances") to partially bypass the fundamental design problems
- * inherited from LMDB. So do not try to understand it completely in order to
- * avoid your madness. */
-static int mdbx_update_gc(MDBX_txn *txn) {
- /* txn->tw.reclaimed_pglist[] can grow and shrink during this call.
- * txn->tw.last_reclaimed and txn->tw.retired_pages[] can only grow.
- * Page numbers cannot disappear from txn->tw.retired_pages[]. */
- MDBX_env *const env = txn->mt_env;
- const bool lifo = (env->me_flags & MDBX_LIFORECLAIM) != 0;
- const char *dbg_prefix_mode = lifo ? " lifo" : " fifo";
- (void)dbg_prefix_mode;
- mdbx_trace("\n>>> @%" PRIaTXN, txn->mt_txnid);
-
- unsigned retired_stored = 0, loop = 0;
- MDBX_cursor_couple couple;
- int rc = mdbx_cursor_init(&couple.outer, txn, FREE_DBI);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout_notracking;
-
- couple.outer.mc_flags |= C_RECLAIMING;
- couple.outer.mc_next = txn->mt_cursors[FREE_DBI];
- txn->mt_cursors[FREE_DBI] = &couple.outer;
- bool dense_gc = false;
-
-retry:
- ++loop;
- mdbx_trace("%s", " >> restart");
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- if (unlikely(/* paranoia */ loop > ((MDBX_DEBUG > 0) ? 12 : 42))) {
- mdbx_error("too more loops %u, bailout", loop);
- rc = MDBX_PROBLEM;
- goto bailout;
- }
-
- if (unlikely(dense_gc) && retired_stored) {
- rc = mdbx_prep_backlog(txn, &couple.outer,
- MDBX_PNL_SIZEOF(txn->tw.retired_pages),
- &retired_stored);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
-
- unsigned settled = 0, cleaned_gc_slot = 0, reused_gc_slot = 0,
- filled_gc_slot = ~0u;
- txnid_t cleaned_gc_id = 0, gc_rid = txn->tw.last_reclaimed;
- while (true) {
- /* Come back here after each Put() in case retired-list changed */
- MDBX_val key, data;
- mdbx_trace("%s", " >> continue");
-
- if (retired_stored != MDBX_PNL_SIZE(txn->tw.retired_pages) &&
- MDBX_PNL_SIZE(txn->tw.retired_pages) > env->me_maxgc_ov1page) {
- rc = mdbx_prep_backlog(txn, &couple.outer,
- MDBX_PNL_SIZEOF(txn->tw.retired_pages),
- &retired_stored);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
-
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- if (lifo) {
- if (cleaned_gc_slot < (txn->tw.lifo_reclaimed
- ? MDBX_PNL_SIZE(txn->tw.lifo_reclaimed)
- : 0)) {
- settled = 0;
- cleaned_gc_slot = 0;
- reused_gc_slot = 0;
- filled_gc_slot = ~0u;
- /* LY: cleanup reclaimed records. */
- do {
- cleaned_gc_id = txn->tw.lifo_reclaimed[++cleaned_gc_slot];
- mdbx_tassert(txn,
- cleaned_gc_slot > 0 &&
- cleaned_gc_id < env->me_lck->mti_oldest_reader.weak);
- key.iov_base = &cleaned_gc_id;
- key.iov_len = sizeof(cleaned_gc_id);
- rc = mdbx_cursor_get(&couple.outer, &key, NULL, MDBX_SET);
- if (rc == MDBX_NOTFOUND)
- continue;
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- if (likely(!dense_gc)) {
- rc = mdbx_prep_backlog(txn, &couple.outer, 0, nullptr);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- mdbx_tassert(txn,
- cleaned_gc_id < env->me_lck->mti_oldest_reader.weak);
- mdbx_trace("%s: cleanup-reclaimed-id [%u]%" PRIaTXN, dbg_prefix_mode,
- cleaned_gc_slot, cleaned_gc_id);
- mdbx_tassert(txn, *txn->mt_cursors == &couple.outer);
- rc = mdbx_cursor_del(&couple.outer, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- } while (cleaned_gc_slot < MDBX_PNL_SIZE(txn->tw.lifo_reclaimed));
- mdbx_txl_sort(txn->tw.lifo_reclaimed);
- }
- } else {
- /* If using records from GC which we have not yet deleted,
- * now delete them and any we reserved for tw.reclaimed_pglist. */
- while (cleaned_gc_id <= txn->tw.last_reclaimed) {
- rc = mdbx_cursor_first(&couple.outer, &key, NULL);
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc == MDBX_NOTFOUND)
- break;
- goto bailout;
- }
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(key.iov_len != sizeof(txnid_t))) {
- rc = MDBX_CORRUPTED;
- goto bailout;
- }
- gc_rid = cleaned_gc_id;
- settled = 0;
- reused_gc_slot = 0;
- cleaned_gc_id = unaligned_peek_u64(4, key.iov_base);
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(cleaned_gc_id < MIN_TXNID || cleaned_gc_id > MAX_TXNID)) {
- rc = MDBX_CORRUPTED;
- goto bailout;
- }
- if (cleaned_gc_id > txn->tw.last_reclaimed)
- break;
- if (likely(!dense_gc)) {
- rc = mdbx_prep_backlog(txn, &couple.outer, 0, nullptr);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- mdbx_tassert(txn, cleaned_gc_id <= txn->tw.last_reclaimed);
- mdbx_tassert(txn, cleaned_gc_id < env->me_lck->mti_oldest_reader.weak);
- mdbx_trace("%s: cleanup-reclaimed-id %" PRIaTXN, dbg_prefix_mode,
- cleaned_gc_id);
- mdbx_tassert(txn, *txn->mt_cursors == &couple.outer);
- rc = mdbx_cursor_del(&couple.outer, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- }
-
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- if (mdbx_audit_enabled()) {
- rc = mdbx_audit_ex(txn, retired_stored, false);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
-
- /* return suitable into unallocated space */
- if (mdbx_refund(txn)) {
- mdbx_tassert(
- txn, mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- if (mdbx_audit_enabled()) {
- rc = mdbx_audit_ex(txn, retired_stored, false);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- }
-
- /* handle loose pages - put ones into the reclaimed- or retired-list */
- if (txn->tw.loose_pages) {
- /* Return loose page numbers to tw.reclaimed_pglist,
- * though usually none are left at this point.
- * The pages themselves remain in dirtylist. */
- if (unlikely(!txn->tw.lifo_reclaimed && txn->tw.last_reclaimed < 1)) {
- if (txn->tw.loose_count > 0) {
- /* Put loose page numbers in tw.retired_pages,
- * since unable to return them to tw.reclaimed_pglist. */
- if (unlikely((rc = mdbx_pnl_need(&txn->tw.retired_pages,
- txn->tw.loose_count)) != 0))
- goto bailout;
- for (MDBX_page *mp = txn->tw.loose_pages; mp; mp = mp->mp_next)
- mdbx_pnl_xappend(txn->tw.retired_pages, mp->mp_pgno);
- mdbx_trace("%s: append %u loose-pages to retired-pages",
- dbg_prefix_mode, txn->tw.loose_count);
- }
- } else {
- /* Room for loose pages + temp PNL with same */
- rc = mdbx_pnl_need(&txn->tw.reclaimed_pglist,
- 2 * txn->tw.loose_count + 2);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- MDBX_PNL loose = txn->tw.reclaimed_pglist +
- MDBX_PNL_ALLOCLEN(txn->tw.reclaimed_pglist) -
- txn->tw.loose_count - 1;
- unsigned count = 0;
- for (MDBX_page *mp = txn->tw.loose_pages; mp; mp = mp->mp_next) {
- mdbx_tassert(txn, mp->mp_flags == P_LOOSE);
- loose[++count] = mp->mp_pgno;
- }
- mdbx_tassert(txn, count == txn->tw.loose_count);
- MDBX_PNL_SIZE(loose) = count;
- mdbx_pnl_sort(loose, txn->mt_next_pgno);
- mdbx_pnl_xmerge(txn->tw.reclaimed_pglist, loose);
- mdbx_trace("%s: append %u loose-pages to reclaimed-pages",
- dbg_prefix_mode, txn->tw.loose_count);
- }
-
- /* filter-out list of dirty-pages from loose-pages */
- MDBX_dpl *const dl = txn->tw.dirtylist;
- unsigned w = 0;
- for (unsigned r = w; ++r <= dl->length;) {
- MDBX_page *dp = dl->items[r].ptr;
- mdbx_tassert(txn, dp->mp_flags == P_LOOSE || IS_MODIFIABLE(txn, dp));
- mdbx_tassert(txn, dpl_endpgno(dl, r) <= txn->mt_next_pgno);
- if ((dp->mp_flags & P_LOOSE) == 0) {
- if (++w != r)
- dl->items[w] = dl->items[r];
- } else {
- mdbx_tassert(txn, dp->mp_flags == P_LOOSE);
- if ((env->me_flags & MDBX_WRITEMAP) == 0)
- mdbx_dpage_free(env, dp, 1);
- }
- }
- mdbx_trace("%s: filtered-out loose-pages from %u -> %u dirty-pages",
- dbg_prefix_mode, dl->length, w);
- mdbx_tassert(txn, txn->tw.loose_count == dl->length - w);
- dpl_setlen(dl, w);
- dl->sorted = 0;
- txn->tw.dirtyroom += txn->tw.loose_count;
- mdbx_tassert(txn,
- txn->tw.dirtyroom + txn->tw.dirtylist->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
- txn->tw.loose_pages = NULL;
- txn->tw.loose_count = 0;
-#if MDBX_ENABLE_REFUND
- txn->tw.loose_refund_wl = 0;
-#endif /* MDBX_ENABLE_REFUND */
- }
-
- const unsigned amount = (unsigned)MDBX_PNL_SIZE(txn->tw.reclaimed_pglist);
- /* handle retired-list - store ones into single gc-record */
- if (retired_stored < MDBX_PNL_SIZE(txn->tw.retired_pages)) {
- if (unlikely(!retired_stored)) {
- /* Make sure last page of GC is touched and on retired-list */
- couple.outer.mc_flags &= ~C_RECLAIMING;
- rc = mdbx_page_search(&couple.outer, NULL,
- MDBX_PS_LAST | MDBX_PS_MODIFY);
- couple.outer.mc_flags |= C_RECLAIMING;
- if (unlikely(rc != MDBX_SUCCESS) && rc != MDBX_NOTFOUND)
- goto bailout;
- }
- /* Write to last page of GC */
- key.iov_len = sizeof(txn->mt_txnid);
- key.iov_base = &txn->mt_txnid;
- do {
- data.iov_len = MDBX_PNL_SIZEOF(txn->tw.retired_pages);
- mdbx_prep_backlog(txn, &couple.outer, data.iov_len, &retired_stored);
- rc = mdbx_cursor_put(&couple.outer, &key, &data, MDBX_RESERVE);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- /* Retry if tw.retired_pages[] grew during the Put() */
- } while (data.iov_len < MDBX_PNL_SIZEOF(txn->tw.retired_pages));
-
- retired_stored = (unsigned)MDBX_PNL_SIZE(txn->tw.retired_pages);
- mdbx_pnl_sort(txn->tw.retired_pages, txn->mt_next_pgno);
- mdbx_assert(env, data.iov_len == MDBX_PNL_SIZEOF(txn->tw.retired_pages));
- memcpy(data.iov_base, txn->tw.retired_pages, data.iov_len);
-
- mdbx_trace("%s: put-retired #%u @ %" PRIaTXN, dbg_prefix_mode,
- retired_stored, txn->mt_txnid);
-
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) {
- unsigned i = retired_stored;
- mdbx_debug_extra("PNL write txn %" PRIaTXN " root %" PRIaPGNO
- " num %u, PNL",
- txn->mt_txnid, txn->mt_dbs[FREE_DBI].md_root, i);
- for (; i; i--)
- mdbx_debug_extra_print(" %" PRIaPGNO, txn->tw.retired_pages[i]);
- mdbx_debug_extra_print("%s\n", ".");
- }
- if (unlikely(amount != MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) &&
- settled)) {
- mdbx_trace("%s: reclaimed-list changed %u -> %u, retry",
- dbg_prefix_mode, amount,
- (unsigned)MDBX_PNL_SIZE(txn->tw.reclaimed_pglist));
- goto retry /* rare case, but avoids GC fragmentation
- and one cycle. */
- ;
- }
- continue;
- }
-
- /* handle reclaimed and lost pages - merge and store both into gc */
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- mdbx_tassert(txn, txn->tw.loose_count == 0);
-
- mdbx_trace("%s", " >> reserving");
- if (mdbx_audit_enabled()) {
- rc = mdbx_audit_ex(txn, retired_stored, false);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- const unsigned left = amount - settled;
- mdbx_trace("%s: amount %u, settled %d, left %d, lifo-reclaimed-slots %u, "
- "reused-gc-slots %u",
- dbg_prefix_mode, amount, settled, (int)left,
- txn->tw.lifo_reclaimed
- ? (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed)
- : 0,
- reused_gc_slot);
- if (0 >= (int)left)
- break;
-
- const unsigned prefer_max_scatter = 257;
- txnid_t reservation_gc_id;
- if (lifo) {
- if (txn->tw.lifo_reclaimed == nullptr) {
- txn->tw.lifo_reclaimed = mdbx_txl_alloc();
- if (unlikely(!txn->tw.lifo_reclaimed)) {
- rc = MDBX_ENOMEM;
- goto bailout;
- }
- }
- if ((unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) <
- prefer_max_scatter &&
- left > ((unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) -
- reused_gc_slot) *
- env->me_maxgc_ov1page &&
- !dense_gc) {
- /* LY: need just a txn-id for save page list. */
- bool need_cleanup = false;
- txnid_t snap_oldest;
- retry_rid:
- couple.outer.mc_flags &= ~C_RECLAIMING;
- do {
- snap_oldest = mdbx_find_oldest(txn);
- rc =
- mdbx_page_alloc(&couple.outer, 0,
- MDBX_ALLOC_GC | MDBX_ALLOC_SLOT | MDBX_ALLOC_FAKE)
- .err;
- if (likely(rc == MDBX_SUCCESS)) {
- mdbx_trace("%s: took @%" PRIaTXN " from GC", dbg_prefix_mode,
- MDBX_PNL_LAST(txn->tw.lifo_reclaimed));
- need_cleanup = true;
- }
- } while (rc == MDBX_SUCCESS &&
- (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) <
- prefer_max_scatter &&
- left > ((unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) -
- reused_gc_slot) *
- env->me_maxgc_ov1page);
- couple.outer.mc_flags |= C_RECLAIMING;
-
- if (likely(rc == MDBX_SUCCESS)) {
- mdbx_trace("%s: got enough from GC.", dbg_prefix_mode);
- continue;
- } else if (unlikely(rc != MDBX_NOTFOUND))
- /* LY: some troubles... */
- goto bailout;
-
- if (MDBX_PNL_SIZE(txn->tw.lifo_reclaimed)) {
- if (need_cleanup) {
- mdbx_txl_sort(txn->tw.lifo_reclaimed);
- cleaned_gc_slot = 0;
- }
- gc_rid = MDBX_PNL_LAST(txn->tw.lifo_reclaimed);
- } else {
- mdbx_tassert(txn, txn->tw.last_reclaimed == 0);
- if (unlikely(mdbx_find_oldest(txn) != snap_oldest))
- /* should retry mdbx_page_alloc(MDBX_ALLOC_GC)
- * if the oldest reader changes since the last attempt */
- goto retry_rid;
- /* no reclaimable GC entries,
- * therefore no entries with ID < mdbx_find_oldest(txn) */
- txn->tw.last_reclaimed = gc_rid = snap_oldest - 1;
- mdbx_trace("%s: none recycled yet, set rid to @%" PRIaTXN,
- dbg_prefix_mode, gc_rid);
- }
-
- /* LY: GC is empty, will look any free txn-id in high2low order. */
- while (MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) < prefer_max_scatter &&
- left > ((unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) -
- reused_gc_slot) *
- env->me_maxgc_ov1page) {
- if (unlikely(gc_rid <= MIN_TXNID)) {
- if (unlikely(MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) <=
- reused_gc_slot)) {
- mdbx_notice("** restart: reserve depleted (reused_gc_slot %u >= "
- "lifo_reclaimed %u" PRIaTXN,
- reused_gc_slot,
- (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed));
- goto retry;
- }
- break;
- }
-
- mdbx_tassert(txn, gc_rid >= MIN_TXNID && gc_rid <= MAX_TXNID);
- --gc_rid;
- key.iov_base = &gc_rid;
- key.iov_len = sizeof(gc_rid);
- rc = mdbx_cursor_get(&couple.outer, &key, &data, MDBX_SET_KEY);
- if (unlikely(rc == MDBX_SUCCESS)) {
- mdbx_debug("%s: GC's id %" PRIaTXN
- " is used, continue bottom-up search",
- dbg_prefix_mode, gc_rid);
- ++gc_rid;
- rc = mdbx_cursor_get(&couple.outer, &key, &data, MDBX_FIRST);
- if (rc == MDBX_NOTFOUND) {
- mdbx_debug("%s: GC is empty (going dense-mode)", dbg_prefix_mode);
- dense_gc = true;
- break;
- }
- if (unlikely(rc != MDBX_SUCCESS ||
- key.iov_len != sizeof(mdbx_tid_t))) {
- rc = MDBX_CORRUPTED;
- goto bailout;
- }
- txnid_t gc_first = unaligned_peek_u64(4, key.iov_base);
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(gc_first < MIN_TXNID || gc_first > MAX_TXNID)) {
- rc = MDBX_CORRUPTED;
- goto bailout;
- }
- if (gc_first <= MIN_TXNID) {
- mdbx_debug("%s: no free GC's id(s) less than %" PRIaTXN
- " (going dense-mode)",
- dbg_prefix_mode, gc_rid);
- dense_gc = true;
- break;
- }
- gc_rid = gc_first - 1;
- }
-
- mdbx_assert(env, !dense_gc);
- rc = mdbx_txl_append(&txn->tw.lifo_reclaimed, gc_rid);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
-
- if (reused_gc_slot)
- /* rare case, but it is better to clear and re-create GC entries
- * with less fragmentation. */
- need_cleanup = true;
- else
- cleaned_gc_slot +=
- 1 /* mark cleanup is not needed for added slot. */;
-
- mdbx_trace("%s: append @%" PRIaTXN
- " to lifo-reclaimed, cleaned-gc-slot = %u",
- dbg_prefix_mode, gc_rid, cleaned_gc_slot);
- }
-
- if (need_cleanup || dense_gc) {
- if (cleaned_gc_slot)
- mdbx_trace(
- "%s: restart inner-loop to clear and re-create GC entries",
- dbg_prefix_mode);
- cleaned_gc_slot = 0;
- continue;
- }
- }
-
- const unsigned i =
- (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) - reused_gc_slot;
- mdbx_tassert(txn, i > 0 && i <= MDBX_PNL_SIZE(txn->tw.lifo_reclaimed));
- reservation_gc_id = txn->tw.lifo_reclaimed[i];
- mdbx_trace("%s: take @%" PRIaTXN " from lifo-reclaimed[%u]",
- dbg_prefix_mode, reservation_gc_id, i);
- } else {
- mdbx_tassert(txn, txn->tw.lifo_reclaimed == NULL);
- if (unlikely(gc_rid == 0)) {
- gc_rid = mdbx_find_oldest(txn) - 1;
- rc = mdbx_cursor_get(&couple.outer, &key, NULL, MDBX_FIRST);
- if (rc == MDBX_SUCCESS) {
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(key.iov_len != sizeof(txnid_t))) {
- rc = MDBX_CORRUPTED;
- goto bailout;
- }
- txnid_t gc_first = unaligned_peek_u64(4, key.iov_base);
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(gc_first < MIN_TXNID || gc_first > MAX_TXNID)) {
- rc = MDBX_CORRUPTED;
- goto bailout;
- }
- if (gc_rid >= gc_first)
- gc_rid = gc_first - 1;
- if (unlikely(gc_rid == 0)) {
- mdbx_error("%s", "** no GC tail-space to store (going dense-mode)");
- dense_gc = true;
- goto retry;
- }
- } else if (rc != MDBX_NOTFOUND)
- goto bailout;
- txn->tw.last_reclaimed = gc_rid;
- cleaned_gc_id = gc_rid + 1;
- }
- reservation_gc_id = gc_rid--;
- mdbx_trace("%s: take @%" PRIaTXN " from head-gc-id", dbg_prefix_mode,
- reservation_gc_id);
- }
- ++reused_gc_slot;
-
- unsigned chunk = left;
- if (unlikely(chunk > env->me_maxgc_ov1page)) {
- const unsigned avail_gc_slots =
- txn->tw.lifo_reclaimed
- ? (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) -
- reused_gc_slot + 1
- : (gc_rid < INT16_MAX) ? (unsigned)gc_rid
- : INT16_MAX;
- if (avail_gc_slots > 1) {
- if (chunk < env->me_maxgc_ov1page * 2)
- chunk /= 2;
- else {
- const unsigned threshold =
- env->me_maxgc_ov1page * ((avail_gc_slots < prefer_max_scatter)
- ? avail_gc_slots
- : prefer_max_scatter);
- if (left < threshold)
- chunk = env->me_maxgc_ov1page;
- else {
- const unsigned tail = left - threshold + env->me_maxgc_ov1page + 1;
- unsigned span = 1;
- unsigned avail = (unsigned)((pgno2bytes(env, span) - PAGEHDRSZ) /
- sizeof(pgno_t)) /* - 1 + span */;
- if (tail > avail) {
- for (unsigned i = amount - span; i > 0; --i) {
- if (MDBX_PNL_ASCENDING
- ? (txn->tw.reclaimed_pglist[i] + span)
- : (txn->tw.reclaimed_pglist[i] - span) ==
- txn->tw.reclaimed_pglist[i + span]) {
- span += 1;
- avail = (unsigned)((pgno2bytes(env, span) - PAGEHDRSZ) /
- sizeof(pgno_t)) -
- 1 + span;
- if (avail >= tail)
- break;
- }
- }
- }
-
- chunk = (avail >= tail) ? tail - span
- : (avail_gc_slots > 3 &&
- reused_gc_slot < prefer_max_scatter - 3)
- ? avail - span
- : tail;
- }
- }
- }
- }
- mdbx_tassert(txn, chunk > 0);
-
- mdbx_trace("%s: gc_rid %" PRIaTXN ", reused_gc_slot %u, reservation-id "
- "%" PRIaTXN,
- dbg_prefix_mode, gc_rid, reused_gc_slot, reservation_gc_id);
-
- mdbx_trace("%s: chunk %u, gc-per-ovpage %u", dbg_prefix_mode, chunk,
- env->me_maxgc_ov1page);
-
- mdbx_tassert(txn, reservation_gc_id < env->me_lck->mti_oldest_reader.weak);
- if (unlikely(
- reservation_gc_id < MIN_TXNID ||
- reservation_gc_id >=
- atomic_load64(&env->me_lck->mti_oldest_reader, mo_Relaxed))) {
- mdbx_error("** internal error (reservation_gc_id %" PRIaTXN ")",
- reservation_gc_id);
- rc = MDBX_PROBLEM;
- goto bailout;
- }
-
- key.iov_len = sizeof(reservation_gc_id);
- key.iov_base = &reservation_gc_id;
- data.iov_len = (chunk + 1) * sizeof(pgno_t);
- mdbx_trace("%s: reserve %u [%u...%u) @%" PRIaTXN, dbg_prefix_mode, chunk,
- settled + 1, settled + chunk + 1, reservation_gc_id);
- mdbx_prep_backlog(txn, &couple.outer, data.iov_len, nullptr);
- rc = mdbx_cursor_put(&couple.outer, &key, &data,
- MDBX_RESERVE | MDBX_NOOVERWRITE);
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
-
- clean_reserved_gc_pnl(env, data);
- settled += chunk;
- mdbx_trace("%s: settled %u (+%u), continue", dbg_prefix_mode, settled,
- chunk);
-
- if (txn->tw.lifo_reclaimed &&
- unlikely(amount < MDBX_PNL_SIZE(txn->tw.reclaimed_pglist)) &&
- (loop < 5 || MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) - amount >
- env->me_maxgc_ov1page)) {
- mdbx_notice("** restart: reclaimed-list growth %u -> %u", amount,
- (unsigned)MDBX_PNL_SIZE(txn->tw.reclaimed_pglist));
- goto retry;
- }
-
- continue;
- }
-
- mdbx_tassert(
- txn,
- cleaned_gc_slot ==
- (txn->tw.lifo_reclaimed ? MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) : 0));
-
- mdbx_trace("%s", " >> filling");
- /* Fill in the reserved records */
- filled_gc_slot =
- txn->tw.lifo_reclaimed
- ? (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed) - reused_gc_slot
- : reused_gc_slot;
- rc = MDBX_SUCCESS;
- mdbx_tassert(txn,
- mdbx_pnl_check4assert(txn->tw.reclaimed_pglist,
- txn->mt_next_pgno - MDBX_ENABLE_REFUND));
- mdbx_tassert(txn, mdbx_dirtylist_check(txn));
- if (MDBX_PNL_SIZE(txn->tw.reclaimed_pglist)) {
- MDBX_val key, data;
- key.iov_len = data.iov_len = 0; /* avoid MSVC warning */
- key.iov_base = data.iov_base = NULL;
-
- const unsigned amount = MDBX_PNL_SIZE(txn->tw.reclaimed_pglist);
- unsigned left = amount;
- if (txn->tw.lifo_reclaimed == nullptr) {
- mdbx_tassert(txn, lifo == 0);
- rc = mdbx_cursor_first(&couple.outer, &key, &data);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- } else {
- mdbx_tassert(txn, lifo != 0);
- }
-
- while (true) {
- txnid_t fill_gc_id;
- mdbx_trace("%s: left %u of %u", dbg_prefix_mode, left,
- (unsigned)MDBX_PNL_SIZE(txn->tw.reclaimed_pglist));
- if (txn->tw.lifo_reclaimed == nullptr) {
- mdbx_tassert(txn, lifo == 0);
- fill_gc_id = unaligned_peek_u64(4, key.iov_base);
- if (filled_gc_slot-- == 0 || fill_gc_id > txn->tw.last_reclaimed) {
- mdbx_notice(
- "** restart: reserve depleted (filled_slot %u, fill_id %" PRIaTXN
- " > last_reclaimed %" PRIaTXN,
- filled_gc_slot, fill_gc_id, txn->tw.last_reclaimed);
- goto retry;
- }
- } else {
- mdbx_tassert(txn, lifo != 0);
- if (++filled_gc_slot >
- (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed)) {
- mdbx_notice("** restart: reserve depleted (filled_gc_slot %u > "
- "lifo_reclaimed %u" PRIaTXN,
- filled_gc_slot,
- (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed));
- goto retry;
- }
- fill_gc_id = txn->tw.lifo_reclaimed[filled_gc_slot];
- mdbx_trace("%s: seek-reservation @%" PRIaTXN " at lifo_reclaimed[%u]",
- dbg_prefix_mode, fill_gc_id, filled_gc_slot);
- key.iov_base = &fill_gc_id;
- key.iov_len = sizeof(fill_gc_id);
- rc = mdbx_cursor_get(&couple.outer, &key, &data, MDBX_SET_KEY);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- mdbx_tassert(txn, cleaned_gc_slot ==
- (txn->tw.lifo_reclaimed
- ? MDBX_PNL_SIZE(txn->tw.lifo_reclaimed)
- : 0));
- mdbx_tassert(txn, fill_gc_id > 0 &&
- fill_gc_id < env->me_lck->mti_oldest_reader.weak);
- key.iov_base = &fill_gc_id;
- key.iov_len = sizeof(fill_gc_id);
-
- mdbx_tassert(txn, data.iov_len >= sizeof(pgno_t) * 2);
- couple.outer.mc_flags |= C_GCFREEZE;
- unsigned chunk = (unsigned)(data.iov_len / sizeof(pgno_t)) - 1;
- if (unlikely(chunk > left)) {
- mdbx_trace("%s: chunk %u > left %u, @%" PRIaTXN, dbg_prefix_mode, chunk,
- left, fill_gc_id);
- if ((loop < 5 && chunk - left > loop / 2) ||
- chunk - left > env->me_maxgc_ov1page) {
- data.iov_len = (left + 1) * sizeof(pgno_t);
- if (loop < 7)
- couple.outer.mc_flags &= ~C_GCFREEZE;
- }
- chunk = left;
- }
- rc = mdbx_cursor_put(&couple.outer, &key, &data,
- MDBX_CURRENT | MDBX_RESERVE);
- couple.outer.mc_flags &= ~C_GCFREEZE;
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- clean_reserved_gc_pnl(env, data);
-
- if (unlikely(txn->tw.loose_count ||
- amount != MDBX_PNL_SIZE(txn->tw.reclaimed_pglist))) {
- mdbx_notice("** restart: reclaimed-list growth (%u -> %u, loose +%u)",
- amount, MDBX_PNL_SIZE(txn->tw.reclaimed_pglist),
- txn->tw.loose_count);
- goto retry;
- }
- if (unlikely(txn->tw.lifo_reclaimed
- ? cleaned_gc_slot < MDBX_PNL_SIZE(txn->tw.lifo_reclaimed)
- : cleaned_gc_id < txn->tw.last_reclaimed)) {
- mdbx_notice("%s", "** restart: reclaimed-slots changed");
- goto retry;
- }
- if (unlikely(retired_stored != MDBX_PNL_SIZE(txn->tw.retired_pages))) {
- mdbx_tassert(txn,
- retired_stored < MDBX_PNL_SIZE(txn->tw.retired_pages));
- mdbx_notice("** restart: retired-list growth (%u -> %u)",
- retired_stored, MDBX_PNL_SIZE(txn->tw.retired_pages));
- goto retry;
- }
-
- pgno_t *dst = data.iov_base;
- *dst++ = chunk;
- pgno_t *src = MDBX_PNL_BEGIN(txn->tw.reclaimed_pglist) + left - chunk;
- memcpy(dst, src, chunk * sizeof(pgno_t));
- pgno_t *from = src, *to = src + chunk;
- mdbx_trace("%s: fill %u [ %u:%" PRIaPGNO "...%u:%" PRIaPGNO
- "] @%" PRIaTXN,
- dbg_prefix_mode, chunk,
- (unsigned)(from - txn->tw.reclaimed_pglist), from[0],
- (unsigned)(to - txn->tw.reclaimed_pglist), to[-1], fill_gc_id);
-
- left -= chunk;
- if (mdbx_audit_enabled()) {
- rc = mdbx_audit_ex(txn, retired_stored + amount - left, true);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- if (left == 0) {
- rc = MDBX_SUCCESS;
- break;
- }
-
- if (txn->tw.lifo_reclaimed == nullptr) {
- mdbx_tassert(txn, lifo == 0);
- rc = mdbx_cursor_next(&couple.outer, &key, &data, MDBX_NEXT);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- } else {
- mdbx_tassert(txn, lifo != 0);
- }
- }
- }
-
- mdbx_tassert(txn, rc == MDBX_SUCCESS);
- if (unlikely(txn->tw.loose_count != 0)) {
- mdbx_notice("** restart: got %u loose pages", txn->tw.loose_count);
- goto retry;
- }
- if (unlikely(filled_gc_slot !=
- (txn->tw.lifo_reclaimed
- ? (unsigned)MDBX_PNL_SIZE(txn->tw.lifo_reclaimed)
- : 0))) {
-
- const bool will_retry = loop < 9;
- mdbx_notice("** %s: reserve excess (filled-slot %u, loop %u)",
- will_retry ? "restart" : "ignore", filled_gc_slot, loop);
- if (will_retry)
- goto retry;
- }
-
- mdbx_tassert(txn,
- txn->tw.lifo_reclaimed == NULL ||
- cleaned_gc_slot == MDBX_PNL_SIZE(txn->tw.lifo_reclaimed));
-
-bailout:
- txn->mt_cursors[FREE_DBI] = couple.outer.mc_next;
-
-bailout_notracking:
- MDBX_PNL_SIZE(txn->tw.reclaimed_pglist) = 0;
- mdbx_trace("<<< %u loops, rc = %d", loop, rc);
- return rc;
-}
-
-static int mdbx_txn_write(MDBX_txn *txn, struct mdbx_iov_ctx *ctx) {
- MDBX_dpl *const dl =
- (txn->mt_flags & MDBX_WRITEMAP) ? txn->tw.dirtylist : mdbx_dpl_sort(txn);
- int rc = MDBX_SUCCESS;
- unsigned r, w;
- for (w = 0, r = 1; r <= dl->length; ++r) {
- MDBX_page *dp = dl->items[r].ptr;
- if (dp->mp_flags & P_LOOSE) {
- dl->items[++w] = dl->items[r];
- continue;
- }
- unsigned npages = dpl_npages(dl, r);
- rc = iov_page(txn, ctx, dp, npages);
- if (unlikely(rc != MDBX_SUCCESS))
- break;
- }
-
- if (ctx->iov_items)
- rc = mdbx_iov_write(txn, ctx);
-
- while (r <= dl->length)
- dl->items[++w] = dl->items[r++];
-
- dl->sorted = dpl_setlen(dl, w);
- txn->tw.dirtyroom += r - 1 - w;
- mdbx_tassert(txn, txn->tw.dirtyroom + txn->tw.dirtylist->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
- return rc;
-}
-
-/* Check txn and dbi arguments to a function */
-static __always_inline bool check_dbi(MDBX_txn *txn, MDBX_dbi dbi,
- unsigned validity) {
- if (likely(dbi < txn->mt_numdbs))
- return likely((txn->mt_dbistate[dbi] & validity) &&
- !TXN_DBI_CHANGED(txn, dbi) &&
- (txn->mt_dbxs[dbi].md_name.iov_base || dbi < CORE_DBS));
-
- return dbi_import(txn, dbi);
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-int mdbx_txn_commit(MDBX_txn *txn) { return __inline_mdbx_txn_commit(txn); }
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-/* Merge child txn into parent */
-static __inline void mdbx_txn_merge(MDBX_txn *const parent, MDBX_txn *const txn,
- const unsigned parent_retired_len) {
- MDBX_dpl *const src = mdbx_dpl_sort(txn);
-
- /* Remove refunded pages from parent's dirty list */
- MDBX_dpl *const dst = mdbx_dpl_sort(parent);
- if (MDBX_ENABLE_REFUND) {
- unsigned n = dst->length;
- while (n && dst->items[n].pgno >= parent->mt_next_pgno) {
- if (!(txn->mt_env->me_flags & MDBX_WRITEMAP)) {
- MDBX_page *dp = dst->items[n].ptr;
- mdbx_dpage_free(txn->mt_env, dp, dpl_npages(dst, n));
- }
- --n;
- }
- parent->tw.dirtyroom += dst->sorted - n;
- dst->sorted = dpl_setlen(dst, n);
- mdbx_tassert(parent,
- parent->tw.dirtyroom + parent->tw.dirtylist->length ==
- (parent->mt_parent ? parent->mt_parent->tw.dirtyroom
- : parent->mt_env->me_options.dp_limit));
- }
-
- /* Remove reclaimed pages from parent's dirty list */
- const MDBX_PNL reclaimed_list = parent->tw.reclaimed_pglist;
- mdbx_dpl_sift(parent, reclaimed_list, false);
-
- /* Move retired pages from parent's dirty & spilled list to reclaimed */
- unsigned r, w, d, s, l;
- for (r = w = parent_retired_len;
- ++r <= MDBX_PNL_SIZE(parent->tw.retired_pages);) {
- const pgno_t pgno = parent->tw.retired_pages[r];
- const unsigned di = mdbx_dpl_exist(parent, pgno);
- const unsigned si = !di ? mdbx_search_spilled(parent, pgno) : 0;
- unsigned npages;
- const char *kind;
- if (di) {
- MDBX_page *dp = dst->items[di].ptr;
- mdbx_tassert(parent, (dp->mp_flags & ~(P_LEAF | P_LEAF2 | P_BRANCH |
- P_OVERFLOW | P_SPILLED)) == 0);
- npages = dpl_npages(dst, di);
- mdbx_page_wash(parent, di, dp, npages);
- kind = "dirty";
- l = 1;
- if (unlikely(npages > l)) {
- /* OVERFLOW-Ñтраница могла быть переиÑпользована по чаÑÑ‚Ñм. Тогда
- * в retired-ÑпиÑке может быть только начало поÑледовательноÑти,
- * а оÑтаток раÑтащен по dirty, spilled и reclaimed ÑпиÑкам. ПоÑтому
- * переноÑим в reclaimed Ñ Ð¿Ñ€Ð¾Ð²ÐµÑ€ÐºÐ¾Ð¹ на обрыв поÑледовательноÑти.
- * Ð’ любом Ñлучае, вÑе оÑколки будут учтены и отфильтрованы, Ñ‚.е. еÑли
- * Ñтраница была разбита на чаÑти, то важно удалить dirty-Ñлемент,
- * а вÑе оÑколки будут учтены отдельно. */
-
- /* СпиÑок retired Ñтраниц не Ñортирован, но Ð´Ð»Ñ ÑƒÑÐºÐ¾Ñ€ÐµÐ½Ð¸Ñ Ñортировки
- * дополнÑетÑÑ Ð² ÑоответÑтвии Ñ MDBX_PNL_ASCENDING */
-#if MDBX_PNL_ASCENDING
- const unsigned len = MDBX_PNL_SIZE(parent->tw.retired_pages);
- while (r < len && parent->tw.retired_pages[r + 1] == pgno + l) {
- ++r;
- if (++l == npages)
- break;
- }
-#else
- while (w > parent_retired_len &&
- parent->tw.retired_pages[w - 1] == pgno + l) {
- --w;
- if (++l == npages)
- break;
- }
-#endif
- }
- } else if (unlikely(si)) {
- l = npages = 1;
- mdbx_spill_remove(parent, si, 1);
- kind = "spilled";
- } else {
- parent->tw.retired_pages[++w] = pgno;
- continue;
- }
-
- mdbx_debug("reclaim retired parent's %u->%u %s page %" PRIaPGNO, npages, l,
- kind, pgno);
- int err = mdbx_pnl_insert_range(&parent->tw.reclaimed_pglist, pgno, l);
- mdbx_ensure(txn->mt_env, err == MDBX_SUCCESS);
- }
- MDBX_PNL_SIZE(parent->tw.retired_pages) = w;
-
- /* Filter-out parent spill list */
- if (parent->tw.spill_pages && MDBX_PNL_SIZE(parent->tw.spill_pages) > 0) {
- const MDBX_PNL sl = mdbx_spill_purge(parent);
- unsigned len = MDBX_PNL_SIZE(sl);
- if (len) {
- /* Remove refunded pages from parent's spill list */
- if (MDBX_ENABLE_REFUND &&
- MDBX_PNL_MOST(sl) >= (parent->mt_next_pgno << 1)) {
-#if MDBX_PNL_ASCENDING
- unsigned i = MDBX_PNL_SIZE(sl);
- assert(MDBX_PNL_MOST(sl) == MDBX_PNL_LAST(sl));
- do {
- if ((sl[i] & 1) == 0)
- mdbx_debug("refund parent's spilled page %" PRIaPGNO, sl[i] >> 1);
- i -= 1;
- } while (i && sl[i] >= (parent->mt_next_pgno << 1));
- MDBX_PNL_SIZE(sl) = i;
-#else
- assert(MDBX_PNL_MOST(sl) == MDBX_PNL_FIRST(sl));
- unsigned i = 0;
- do {
- ++i;
- if ((sl[i] & 1) == 0)
- mdbx_debug("refund parent's spilled page %" PRIaPGNO, sl[i] >> 1);
- } while (i < len && sl[i + 1] >= (parent->mt_next_pgno << 1));
- MDBX_PNL_SIZE(sl) = len -= i;
- memmove(sl + 1, sl + 1 + i, len * sizeof(sl[0]));
-#endif
- }
- mdbx_tassert(
- txn, mdbx_pnl_check4assert(sl, (size_t)parent->mt_next_pgno << 1));
-
- /* Remove reclaimed pages from parent's spill list */
- s = MDBX_PNL_SIZE(sl), r = MDBX_PNL_SIZE(reclaimed_list);
- /* Scanning from end to begin */
- while (s && r) {
- if (sl[s] & 1) {
- --s;
- continue;
- }
- const pgno_t spilled_pgno = sl[s] >> 1;
- const pgno_t reclaimed_pgno = reclaimed_list[r];
- if (reclaimed_pgno != spilled_pgno) {
- const bool cmp = MDBX_PNL_ORDERED(spilled_pgno, reclaimed_pgno);
- s -= !cmp;
- r -= cmp;
- } else {
- mdbx_debug("remove reclaimed parent's spilled page %" PRIaPGNO,
- reclaimed_pgno);
- mdbx_spill_remove(parent, s, 1);
- --s;
- --r;
- }
- }
-
- /* Remove anything in our dirty list from parent's spill list */
- /* Scanning spill list in descend order */
- const int step = MDBX_PNL_ASCENDING ? -1 : 1;
- s = MDBX_PNL_ASCENDING ? MDBX_PNL_SIZE(sl) : 1;
- d = src->length;
- while (d && (MDBX_PNL_ASCENDING ? s > 0 : s <= MDBX_PNL_SIZE(sl))) {
- if (sl[s] & 1) {
- s += step;
- continue;
- }
- const pgno_t spilled_pgno = sl[s] >> 1;
- const pgno_t dirty_pgno_form = src->items[d].pgno;
- const unsigned npages = dpl_npages(src, d);
- const pgno_t dirty_pgno_to = dirty_pgno_form + npages;
- if (dirty_pgno_form > spilled_pgno) {
- --d;
- continue;
- }
- if (dirty_pgno_to <= spilled_pgno) {
- s += step;
- continue;
- }
-
- mdbx_debug("remove dirtied parent's spilled %u page %" PRIaPGNO, npages,
- dirty_pgno_form);
- mdbx_spill_remove(parent, s, 1);
- s += step;
- }
-
- /* Squash deleted pagenums if we deleted any */
- mdbx_spill_purge(parent);
- }
- }
-
- /* Remove anything in our spill list from parent's dirty list */
- if (txn->tw.spill_pages) {
- mdbx_tassert(txn, mdbx_pnl_check4assert(txn->tw.spill_pages,
- (size_t)parent->mt_next_pgno << 1));
- mdbx_dpl_sift(parent, txn->tw.spill_pages, true);
- mdbx_tassert(parent,
- parent->tw.dirtyroom + parent->tw.dirtylist->length ==
- (parent->mt_parent ? parent->mt_parent->tw.dirtyroom
- : parent->mt_env->me_options.dp_limit));
- }
-
- /* Find length of merging our dirty list with parent's and release
- * filter-out pages */
- for (l = 0, d = dst->length, s = src->length; d > 0 && s > 0;) {
- MDBX_page *sp = src->items[s].ptr;
- mdbx_tassert(parent,
- (sp->mp_flags & ~(P_LEAF | P_LEAF2 | P_BRANCH | P_OVERFLOW |
- P_LOOSE | P_SPILLED)) == 0);
- const unsigned s_npages = dpl_npages(src, s);
- const pgno_t s_pgno = src->items[s].pgno;
-
- MDBX_page *dp = dst->items[d].ptr;
- mdbx_tassert(parent, (dp->mp_flags & ~(P_LEAF | P_LEAF2 | P_BRANCH |
- P_OVERFLOW | P_SPILLED)) == 0);
- const unsigned d_npages = dpl_npages(dst, d);
- const pgno_t d_pgno = dst->items[d].pgno;
-
- if (d_pgno >= s_pgno + s_npages) {
- --d;
- ++l;
- } else if (d_pgno + d_npages <= s_pgno) {
- if (sp->mp_flags != P_LOOSE) {
- sp->mp_txnid = parent->mt_front;
- sp->mp_flags &= ~P_SPILLED;
- }
- --s;
- ++l;
- } else {
- dst->items[d--].ptr = nullptr;
- if ((txn->mt_flags & MDBX_WRITEMAP) == 0)
- mdbx_dpage_free(txn->mt_env, dp, d_npages);
- }
- }
- assert(dst->sorted == dst->length);
- mdbx_tassert(parent, dst->detent >= l + d + s);
- dst->sorted = l + d + s; /* the merged length */
-
- while (s > 0) {
- MDBX_page *sp = src->items[s].ptr;
- mdbx_tassert(parent,
- (sp->mp_flags & ~(P_LEAF | P_LEAF2 | P_BRANCH | P_OVERFLOW |
- P_LOOSE | P_SPILLED)) == 0);
- if (sp->mp_flags != P_LOOSE) {
- sp->mp_txnid = parent->mt_front;
- sp->mp_flags &= ~P_SPILLED;
- }
- --s;
- }
-
- /* Merge our dirty list into parent's, i.e. merge(dst, src) -> dst */
- if (dst->sorted >= dst->length) {
- /* from end to begin with dst extending */
- for (l = dst->sorted, s = src->length, d = dst->length; s > 0 && d > 0;) {
- if (unlikely(l <= d)) {
- /* squash to get a gap of free space for merge */
- for (r = w = 1; r <= d; ++r)
- if (dst->items[r].ptr) {
- if (w != r) {
- dst->items[w] = dst->items[r];
- dst->items[r].ptr = nullptr;
- }
- ++w;
- }
- mdbx_notice("squash to begin for extending-merge %u -> %u", d, w - 1);
- d = w - 1;
- continue;
- }
- assert(l > d);
- if (dst->items[d].ptr) {
- dst->items[l--] = (dst->items[d].pgno > src->items[s].pgno)
- ? dst->items[d--]
- : src->items[s--];
- } else
- --d;
- }
- if (s > 0) {
- assert(l == s);
- while (d > 0) {
- assert(dst->items[d].ptr == nullptr);
- --d;
- }
- do {
- assert(l > 0);
- dst->items[l--] = src->items[s--];
- } while (s > 0);
- } else {
- assert(l == d);
- while (l > 0) {
- assert(dst->items[l].ptr != nullptr);
- --l;
- }
- }
- } else {
- /* from begin to end with dst shrinking (a lot of new overflow pages) */
- for (l = s = d = 1; s <= src->length && d <= dst->length;) {
- if (unlikely(l >= d)) {
- /* squash to get a gap of free space for merge */
- for (r = w = dst->length; r >= d; --r)
- if (dst->items[r].ptr) {
- if (w != r) {
- dst->items[w] = dst->items[r];
- dst->items[r].ptr = nullptr;
- }
- --w;
- }
- mdbx_notice("squash to end for shrinking-merge %u -> %u", d, w + 1);
- d = w + 1;
- continue;
- }
- assert(l < d);
- if (dst->items[d].ptr) {
- dst->items[l++] = (dst->items[d].pgno < src->items[s].pgno)
- ? dst->items[d++]
- : src->items[s++];
- } else
- ++d;
- }
- if (s <= src->length) {
- assert(dst->sorted - l == src->length - s);
- while (d <= dst->length) {
- assert(dst->items[d].ptr == nullptr);
- --d;
- }
- do {
- assert(l <= dst->sorted);
- dst->items[l++] = src->items[s++];
- } while (s <= src->length);
- } else {
- assert(dst->sorted - l == dst->length - d);
- while (l <= dst->sorted) {
- assert(l <= d && d <= dst->length && dst->items[d].ptr);
- dst->items[l++] = dst->items[d++];
- }
- }
- }
- parent->tw.dirtyroom -= dst->sorted - dst->length;
- assert(parent->tw.dirtyroom <= parent->mt_env->me_options.dp_limit);
- dpl_setlen(dst, dst->sorted);
- parent->tw.dirtylru = txn->tw.dirtylru;
- mdbx_tassert(parent, mdbx_dirtylist_check(parent));
- mdbx_dpl_free(txn);
-
- if (txn->tw.spill_pages) {
- if (parent->tw.spill_pages) {
- /* Must not fail since space was preserved above. */
- mdbx_pnl_xmerge(parent->tw.spill_pages, txn->tw.spill_pages);
- mdbx_pnl_free(txn->tw.spill_pages);
- } else {
- parent->tw.spill_pages = txn->tw.spill_pages;
- parent->tw.spill_least_removed = txn->tw.spill_least_removed;
- }
- mdbx_tassert(parent, mdbx_dirtylist_check(parent));
- }
-
- parent->mt_flags &= ~MDBX_TXN_HAS_CHILD;
- if (parent->tw.spill_pages) {
- assert(mdbx_pnl_check4assert(parent->tw.spill_pages,
- (size_t)parent->mt_next_pgno << 1));
- if (MDBX_PNL_SIZE(parent->tw.spill_pages))
- parent->mt_flags |= MDBX_TXN_SPILLS;
- }
-}
-
-int mdbx_txn_commit_ex(MDBX_txn *txn, MDBX_commit_latency *latency) {
- STATIC_ASSERT(MDBX_TXN_FINISHED ==
- MDBX_TXN_BLOCKED - MDBX_TXN_HAS_CHILD - MDBX_TXN_ERROR);
- const uint64_t ts_0 = latency ? mdbx_osal_monotime() : 0;
- uint64_t ts_1 = 0, ts_2 = 0, ts_3 = 0, ts_4 = 0;
- uint32_t audit_duration = 0;
-
- int rc = check_txn(txn, MDBX_TXN_FINISHED);
- if (unlikely(rc != MDBX_SUCCESS))
- goto provide_latency;
-
- if (unlikely(txn->mt_flags & MDBX_TXN_ERROR)) {
- rc = MDBX_RESULT_TRUE;
- goto fail;
- }
-
- MDBX_env *env = txn->mt_env;
-#if MDBX_ENV_CHECKPID
- if (unlikely(env->me_pid != mdbx_getpid())) {
- env->me_flags |= MDBX_FATAL_ERROR;
- rc = MDBX_PANIC;
- goto provide_latency;
- }
-#endif /* MDBX_ENV_CHECKPID */
-
- /* mdbx_txn_end() mode for a commit which writes nothing */
- unsigned end_mode =
- MDBX_END_PURE_COMMIT | MDBX_END_UPDATE | MDBX_END_SLOT | MDBX_END_FREE;
- if (unlikely(F_ISSET(txn->mt_flags, MDBX_TXN_RDONLY)))
- goto done;
-
- if (txn->mt_child) {
- rc = mdbx_txn_commit_ex(txn->mt_child, NULL);
- mdbx_tassert(txn, txn->mt_child == NULL);
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- }
-
- if (unlikely(txn != env->me_txn)) {
- mdbx_debug("%s", "attempt to commit unknown transaction");
- rc = MDBX_EINVAL;
- goto fail;
- }
-
- if (txn->mt_parent) {
- mdbx_tassert(txn, mdbx_audit_ex(txn, 0, false) == 0);
- mdbx_assert(env, txn != env->me_txn0);
- MDBX_txn *const parent = txn->mt_parent;
- mdbx_assert(env, parent->mt_signature == MDBX_MT_SIGNATURE);
- mdbx_assert(env, parent->mt_child == txn &&
- (parent->mt_flags & MDBX_TXN_HAS_CHILD) != 0);
- mdbx_assert(env, mdbx_dirtylist_check(txn));
-
- if (txn->tw.dirtylist->length == 0 && !(txn->mt_flags & MDBX_TXN_DIRTY) &&
- parent->mt_numdbs == txn->mt_numdbs) {
- for (int i = txn->mt_numdbs; --i >= 0;) {
- mdbx_tassert(txn, (txn->mt_dbistate[i] & DBI_DIRTY) == 0);
- if ((txn->mt_dbistate[i] & DBI_STALE) &&
- !(parent->mt_dbistate[i] & DBI_STALE))
- mdbx_tassert(txn, memcmp(&parent->mt_dbs[i], &txn->mt_dbs[i],
- sizeof(MDBX_db)) == 0);
- }
-
- mdbx_tassert(txn, memcmp(&parent->mt_geo, &txn->mt_geo,
- sizeof(parent->mt_geo)) == 0);
- mdbx_tassert(txn, memcmp(&parent->mt_canary, &txn->mt_canary,
- sizeof(parent->mt_canary)) == 0);
- mdbx_tassert(txn, !txn->tw.spill_pages ||
- MDBX_PNL_SIZE(txn->tw.spill_pages) == 0);
- mdbx_tassert(txn, txn->tw.loose_count == 0);
-
- /* fast completion of pure nested transaction */
- end_mode = MDBX_END_PURE_COMMIT | MDBX_END_SLOT | MDBX_END_FREE;
- goto done;
- }
-
- /* Preserve space for spill list to avoid parent's state corruption
- * if allocation fails. */
- const unsigned parent_retired_len =
- (unsigned)(uintptr_t)parent->tw.retired_pages;
- mdbx_tassert(txn,
- parent_retired_len <= MDBX_PNL_SIZE(txn->tw.retired_pages));
- const unsigned retired_delta =
- MDBX_PNL_SIZE(txn->tw.retired_pages) - parent_retired_len;
- if (retired_delta) {
- rc = mdbx_pnl_need(&txn->tw.reclaimed_pglist, retired_delta);
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- }
-
- if (txn->tw.spill_pages) {
- if (parent->tw.spill_pages) {
- rc = mdbx_pnl_need(&parent->tw.spill_pages,
- MDBX_PNL_SIZE(txn->tw.spill_pages));
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- }
- mdbx_spill_purge(txn);
- }
-
- if (unlikely(txn->tw.dirtylist->length + parent->tw.dirtylist->length >
- parent->tw.dirtylist->detent &&
- !mdbx_dpl_reserve(parent, txn->tw.dirtylist->length +
- parent->tw.dirtylist->length))) {
- rc = MDBX_ENOMEM;
- goto fail;
- }
-
- //-------------------------------------------------------------------------
-
- parent->tw.lifo_reclaimed = txn->tw.lifo_reclaimed;
- txn->tw.lifo_reclaimed = NULL;
-
- parent->tw.retired_pages = txn->tw.retired_pages;
- txn->tw.retired_pages = NULL;
-
- mdbx_pnl_free(parent->tw.reclaimed_pglist);
- parent->tw.reclaimed_pglist = txn->tw.reclaimed_pglist;
- txn->tw.reclaimed_pglist = NULL;
- parent->tw.last_reclaimed = txn->tw.last_reclaimed;
-
- parent->mt_geo = txn->mt_geo;
- parent->mt_canary = txn->mt_canary;
- parent->mt_flags |= txn->mt_flags & MDBX_TXN_DIRTY;
-
- /* Move loose pages to parent */
-#if MDBX_ENABLE_REFUND
- parent->tw.loose_refund_wl = txn->tw.loose_refund_wl;
-#endif /* MDBX_ENABLE_REFUND */
- parent->tw.loose_count = txn->tw.loose_count;
- parent->tw.loose_pages = txn->tw.loose_pages;
-
- /* Merge our cursors into parent's and close them */
- mdbx_cursors_eot(txn, true);
- end_mode |= MDBX_END_EOTDONE;
-
- /* Update parent's DBs array */
- memcpy(parent->mt_dbs, txn->mt_dbs, txn->mt_numdbs * sizeof(MDBX_db));
- parent->mt_numdbs = txn->mt_numdbs;
- for (unsigned i = 0; i < txn->mt_numdbs; i++) {
- /* preserve parent's status */
- const uint8_t state =
- txn->mt_dbistate[i] |
- (parent->mt_dbistate[i] & (DBI_CREAT | DBI_FRESH | DBI_DIRTY));
- mdbx_debug("db %u dbi-state %s 0x%02x -> 0x%02x", i,
- (parent->mt_dbistate[i] != state) ? "update" : "still",
- parent->mt_dbistate[i], state);
- parent->mt_dbistate[i] = state;
- }
-
- ts_1 = latency ? mdbx_osal_monotime() : 0;
- mdbx_txn_merge(parent, txn, parent_retired_len);
- ts_2 = latency ? mdbx_osal_monotime() : 0;
- env->me_txn = parent;
- parent->mt_child = NULL;
- mdbx_tassert(parent, mdbx_dirtylist_check(parent));
-
-#if MDBX_ENABLE_REFUND
- mdbx_refund(parent);
- if (mdbx_assert_enabled()) {
- /* Check parent's loose pages not suitable for refund */
- for (MDBX_page *lp = parent->tw.loose_pages; lp; lp = lp->mp_next)
- mdbx_tassert(parent, lp->mp_pgno < parent->tw.loose_refund_wl &&
- lp->mp_pgno + 1 < parent->mt_next_pgno);
- /* Check parent's reclaimed pages not suitable for refund */
- if (MDBX_PNL_SIZE(parent->tw.reclaimed_pglist))
- mdbx_tassert(parent, MDBX_PNL_MOST(parent->tw.reclaimed_pglist) + 1 <
- parent->mt_next_pgno);
- }
-#endif /* MDBX_ENABLE_REFUND */
-
- ts_4 = ts_3 = latency ? mdbx_osal_monotime() : 0;
- txn->mt_signature = 0;
- mdbx_free(txn);
- mdbx_tassert(parent, mdbx_audit_ex(parent, 0, false) == 0);
- rc = MDBX_SUCCESS;
- goto provide_latency;
- }
-
- mdbx_tassert(txn, txn->tw.dirtyroom + txn->tw.dirtylist->length ==
- (txn->mt_parent ? txn->mt_parent->tw.dirtyroom
- : txn->mt_env->me_options.dp_limit));
- mdbx_cursors_eot(txn, false);
- end_mode |= MDBX_END_EOTDONE;
-
- if (txn->tw.dirtylist->length == 0 &&
- (txn->mt_flags & (MDBX_TXN_DIRTY | MDBX_TXN_SPILLS)) == 0) {
- for (int i = txn->mt_numdbs; --i >= 0;)
- mdbx_tassert(txn, (txn->mt_dbistate[i] & DBI_DIRTY) == 0);
- rc = MDBX_SUCCESS;
- goto done;
- }
-
- mdbx_debug("committing txn %" PRIaTXN " %p on mdbenv %p, root page %" PRIaPGNO
- "/%" PRIaPGNO,
- txn->mt_txnid, (void *)txn, (void *)env,
- txn->mt_dbs[MAIN_DBI].md_root, txn->mt_dbs[FREE_DBI].md_root);
-
- /* Update DB root pointers */
- if (txn->mt_numdbs > CORE_DBS) {
- MDBX_cursor_couple couple;
- MDBX_val data;
- data.iov_len = sizeof(MDBX_db);
-
- rc = mdbx_cursor_init(&couple.outer, txn, MAIN_DBI);
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- for (MDBX_dbi i = CORE_DBS; i < txn->mt_numdbs; i++) {
- if (txn->mt_dbistate[i] & DBI_DIRTY) {
- if (unlikely(TXN_DBI_CHANGED(txn, i))) {
- rc = MDBX_BAD_DBI;
- goto fail;
- }
- MDBX_db *db = &txn->mt_dbs[i];
- mdbx_debug("update main's entry for sub-db %u, mod_txnid %" PRIaTXN
- " -> %" PRIaTXN,
- i, db->md_mod_txnid, txn->mt_txnid);
- db->md_mod_txnid = txn->mt_txnid;
- data.iov_base = db;
- WITH_CURSOR_TRACKING(couple.outer,
- rc = mdbx_cursor_put(&couple.outer,
- &txn->mt_dbxs[i].md_name,
- &data, F_SUBDATA));
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- }
- }
- }
-
- ts_1 = latency ? mdbx_osal_monotime() : 0;
- rc = mdbx_update_gc(txn);
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
-
- txn->mt_dbs[FREE_DBI].md_mod_txnid = (txn->mt_dbistate[FREE_DBI] & DBI_DIRTY)
- ? txn->mt_txnid
- : txn->mt_dbs[FREE_DBI].md_mod_txnid;
-
- txn->mt_dbs[MAIN_DBI].md_mod_txnid = (txn->mt_dbistate[MAIN_DBI] & DBI_DIRTY)
- ? txn->mt_txnid
- : txn->mt_dbs[MAIN_DBI].md_mod_txnid;
-
- ts_2 = latency ? mdbx_osal_monotime() : 0;
- if (mdbx_audit_enabled()) {
- rc = mdbx_audit_ex(txn, MDBX_PNL_SIZE(txn->tw.retired_pages), true);
- const uint64_t audit_end = mdbx_osal_monotime();
- audit_duration = mdbx_osal_monotime_to_16dot16(audit_end - ts_2);
- ts_2 = audit_end;
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- }
-
- struct mdbx_iov_ctx ctx;
- mdbx_iov_init(txn, &ctx);
- rc = mdbx_txn_write(txn, &ctx);
- if (likely(rc == MDBX_SUCCESS))
- mdbx_iov_done(txn, &ctx);
- /* TODO: use ctx.flush_begin & ctx.flush_end for range-sync */
- ts_3 = latency ? mdbx_osal_monotime() : 0;
-
- if (likely(rc == MDBX_SUCCESS)) {
- const MDBX_meta *head = constmeta_prefer_last(env);
- MDBX_meta meta;
- memcpy(meta.mm_magic_and_version, head->mm_magic_and_version, 8);
- meta.mm_extra_flags = head->mm_extra_flags;
- meta.mm_validator_id = head->mm_validator_id;
- meta.mm_extra_pagehdr = head->mm_extra_pagehdr;
- unaligned_poke_u64(4, meta.mm_pages_retired,
- unaligned_peek_u64(4, head->mm_pages_retired) +
- MDBX_PNL_SIZE(txn->tw.retired_pages));
- meta.mm_geo = txn->mt_geo;
- meta.mm_dbs[FREE_DBI] = txn->mt_dbs[FREE_DBI];
- meta.mm_dbs[MAIN_DBI] = txn->mt_dbs[MAIN_DBI];
- meta.mm_canary = txn->mt_canary;
- meta_set_txnid(env, &meta, txn->mt_txnid);
-
- rc = mdbx_sync_locked(
- env, env->me_flags | txn->mt_flags | MDBX_SHRINK_ALLOWED, &meta);
- }
- ts_4 = latency ? mdbx_osal_monotime() : 0;
- if (unlikely(rc != MDBX_SUCCESS)) {
- env->me_flags |= MDBX_FATAL_ERROR;
- goto fail;
- }
-
- end_mode = MDBX_END_COMMITTED | MDBX_END_UPDATE | MDBX_END_EOTDONE;
-
-done:
- rc = mdbx_txn_end(txn, end_mode);
-
-provide_latency:
- if (latency) {
- latency->audit = audit_duration;
- latency->preparation =
- ts_1 ? mdbx_osal_monotime_to_16dot16(ts_1 - ts_0) : 0;
- latency->gc =
- (ts_1 && ts_2) ? mdbx_osal_monotime_to_16dot16(ts_2 - ts_1) : 0;
- latency->write =
- (ts_2 && ts_3) ? mdbx_osal_monotime_to_16dot16(ts_3 - ts_2) : 0;
- latency->sync =
- (ts_3 && ts_4) ? mdbx_osal_monotime_to_16dot16(ts_4 - ts_3) : 0;
- const uint64_t ts_5 = mdbx_osal_monotime();
- latency->ending = ts_4 ? mdbx_osal_monotime_to_16dot16(ts_5 - ts_4) : 0;
- latency->whole = mdbx_osal_monotime_to_16dot16(ts_5 - ts_0);
- }
- return rc;
-
-fail:
- mdbx_txn_abort(txn);
- goto provide_latency;
-}
-
-static int mdbx_validate_meta(MDBX_env *env, MDBX_meta *const meta,
- const MDBX_page *const page,
- const unsigned meta_number,
- unsigned *guess_pagesize) {
- const uint64_t magic_and_version =
- unaligned_peek_u64(4, &meta->mm_magic_and_version);
- if (unlikely(magic_and_version != MDBX_DATA_MAGIC &&
- magic_and_version != MDBX_DATA_MAGIC_LEGACY_COMPAT &&
- magic_and_version != MDBX_DATA_MAGIC_LEGACY_DEVEL)) {
- mdbx_error("meta[%u] has invalid magic/version %" PRIx64, meta_number,
- magic_and_version);
- return ((magic_and_version >> 8) != MDBX_MAGIC) ? MDBX_INVALID
- : MDBX_VERSION_MISMATCH;
- }
-
- if (unlikely(page->mp_pgno != meta_number)) {
- mdbx_error("meta[%u] has invalid pageno %" PRIaPGNO, meta_number,
- page->mp_pgno);
- return MDBX_INVALID;
- }
-
- if (unlikely(page->mp_flags != P_META)) {
- mdbx_error("page #%u not a meta-page", meta_number);
- return MDBX_INVALID;
- }
-
- /* LY: check pagesize */
- if (unlikely(!is_powerof2(meta->mm_psize) || meta->mm_psize < MIN_PAGESIZE ||
- meta->mm_psize > MAX_PAGESIZE)) {
- mdbx_warning("meta[%u] has invalid pagesize (%u), skip it", meta_number,
- meta->mm_psize);
- return is_powerof2(meta->mm_psize) ? MDBX_VERSION_MISMATCH : MDBX_INVALID;
- }
-
- if (guess_pagesize && *guess_pagesize != meta->mm_psize) {
- *guess_pagesize = meta->mm_psize;
- mdbx_verbose("meta[%u] took pagesize %u", meta_number, meta->mm_psize);
- }
-
- const txnid_t txnid = unaligned_peek_u64(4, &meta->mm_txnid_a);
- if (unlikely(txnid != unaligned_peek_u64(4, &meta->mm_txnid_b))) {
- mdbx_warning("meta[%u] not completely updated, skip it", meta_number);
- return MDBX_RESULT_TRUE;
- }
-
- /* LY: check signature as a checksum */
- if (META_IS_STEADY(meta) &&
- unlikely(unaligned_peek_u64(4, &meta->mm_datasync_sign) !=
- meta_sign(meta))) {
- mdbx_warning("meta[%u] has invalid steady-checksum (0x%" PRIx64
- " != 0x%" PRIx64 "), skip it",
- meta_number, unaligned_peek_u64(4, &meta->mm_datasync_sign),
- meta_sign(meta));
- return MDBX_RESULT_TRUE;
- }
-
- mdbx_debug("checking meta%" PRIaPGNO " = root %" PRIaPGNO "/%" PRIaPGNO
- ", geo %" PRIaPGNO "/%" PRIaPGNO "-%" PRIaPGNO "/%" PRIaPGNO
- " +%u -%u, txn_id %" PRIaTXN ", %s",
- page->mp_pgno, meta->mm_dbs[MAIN_DBI].md_root,
- meta->mm_dbs[FREE_DBI].md_root, meta->mm_geo.lower,
- meta->mm_geo.next, meta->mm_geo.now, meta->mm_geo.upper,
- pv2pages(meta->mm_geo.grow_pv), pv2pages(meta->mm_geo.shrink_pv),
- txnid, mdbx_durable_str(meta));
-
- if (unlikely(txnid < MIN_TXNID || txnid > MAX_TXNID)) {
- mdbx_warning("meta[%u] has invalid txnid %" PRIaTXN ", skip it",
- meta_number, txnid);
- return MDBX_RESULT_TRUE;
- }
-
- /* LY: check min-pages value */
- if (unlikely(meta->mm_geo.lower < MIN_PAGENO ||
- meta->mm_geo.lower > MAX_PAGENO + 1)) {
- mdbx_warning("meta[%u] has invalid min-pages (%" PRIaPGNO "), skip it",
- meta_number, meta->mm_geo.lower);
- return MDBX_INVALID;
- }
-
- /* LY: check max-pages value */
- if (unlikely(meta->mm_geo.upper < MIN_PAGENO ||
- meta->mm_geo.upper > MAX_PAGENO + 1 ||
- meta->mm_geo.upper < meta->mm_geo.lower)) {
- mdbx_warning("meta[%u] has invalid max-pages (%" PRIaPGNO "), skip it",
- meta_number, meta->mm_geo.upper);
- return MDBX_INVALID;
- }
-
- /* LY: check last_pgno */
- if (unlikely(meta->mm_geo.next < MIN_PAGENO ||
- meta->mm_geo.next - 1 > MAX_PAGENO)) {
- mdbx_warning("meta[%u] has invalid next-pageno (%" PRIaPGNO "), skip it",
- meta_number, meta->mm_geo.next);
- return MDBX_CORRUPTED;
- }
-
- /* LY: check filesize & used_bytes */
- const uint64_t used_bytes = meta->mm_geo.next * (uint64_t)meta->mm_psize;
- if (unlikely(used_bytes > env->me_dxb_mmap.filesize)) {
- /* Here could be a race with DB-shrinking performed by other process */
- int err = mdbx_filesize(env->me_lazy_fd, &env->me_dxb_mmap.filesize);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- if (unlikely(used_bytes > env->me_dxb_mmap.filesize)) {
- mdbx_warning("meta[%u] used-bytes (%" PRIu64 ") beyond filesize (%" PRIu64
- "), skip it",
- meta_number, used_bytes, env->me_dxb_mmap.filesize);
- return MDBX_CORRUPTED;
- }
- }
- if (unlikely(meta->mm_geo.next - 1 > MAX_PAGENO ||
- used_bytes > MAX_MAPSIZE)) {
- mdbx_warning("meta[%u] has too large used-space (%" PRIu64 "), skip it",
- meta_number, used_bytes);
- return MDBX_TOO_LARGE;
- }
-
- /* LY: check mapsize limits */
- pgno_t geo_lower = meta->mm_geo.lower;
- uint64_t mapsize_min = geo_lower * (uint64_t)meta->mm_psize;
- STATIC_ASSERT(MAX_MAPSIZE < PTRDIFF_MAX - MAX_PAGESIZE);
- STATIC_ASSERT(MIN_MAPSIZE < MAX_MAPSIZE);
- STATIC_ASSERT((uint64_t)(MAX_PAGENO + 1) * MIN_PAGESIZE % (4ul << 20) == 0);
- if (unlikely(mapsize_min < MIN_MAPSIZE || mapsize_min > MAX_MAPSIZE)) {
- if (MAX_MAPSIZE != MAX_MAPSIZE64 && mapsize_min > MAX_MAPSIZE &&
- mapsize_min <= MAX_MAPSIZE64) {
- mdbx_assert(env, meta->mm_geo.next - 1 <= MAX_PAGENO &&
- used_bytes <= MAX_MAPSIZE);
- mdbx_warning("meta[%u] has too large min-mapsize (%" PRIu64 "), "
- "but size of used space still acceptable (%" PRIu64 ")",
- meta_number, mapsize_min, used_bytes);
- geo_lower = (pgno_t)((mapsize_min = MAX_MAPSIZE) / meta->mm_psize);
- if (geo_lower > MAX_PAGENO + 1) {
- geo_lower = MAX_PAGENO + 1;
- mapsize_min = geo_lower * (uint64_t)meta->mm_psize;
- }
- mdbx_warning("meta[%u] consider get-%s pageno is %" PRIaPGNO
- " instead of wrong %" PRIaPGNO
- ", will be corrected on next commit(s)",
- meta_number, "lower", geo_lower, meta->mm_geo.lower);
- meta->mm_geo.lower = geo_lower;
- } else {
- mdbx_warning("meta[%u] has invalid min-mapsize (%" PRIu64 "), skip it",
- meta_number, mapsize_min);
- return MDBX_VERSION_MISMATCH;
- }
- }
-
- pgno_t geo_upper = meta->mm_geo.upper;
- uint64_t mapsize_max = geo_upper * (uint64_t)meta->mm_psize;
- STATIC_ASSERT(MIN_MAPSIZE < MAX_MAPSIZE);
- if (unlikely(mapsize_max > MAX_MAPSIZE ||
- (MAX_PAGENO + 1) <
- ceil_powerof2((size_t)mapsize_max, env->me_os_psize) /
- (size_t)meta->mm_psize)) {
- if (mapsize_max > MAX_MAPSIZE64) {
- mdbx_warning("meta[%u] has invalid max-mapsize (%" PRIu64 "), skip it",
- meta_number, mapsize_max);
- return MDBX_VERSION_MISMATCH;
- }
- /* allow to open large DB from a 32-bit environment */
- mdbx_assert(env, meta->mm_geo.next - 1 <= MAX_PAGENO &&
- used_bytes <= MAX_MAPSIZE);
- mdbx_warning("meta[%u] has too large max-mapsize (%" PRIu64 "), "
- "but size of used space still acceptable (%" PRIu64 ")",
- meta_number, mapsize_max, used_bytes);
- geo_upper = (pgno_t)((mapsize_max = MAX_MAPSIZE) / meta->mm_psize);
- if (geo_upper > MAX_PAGENO + 1) {
- geo_upper = MAX_PAGENO + 1;
- mapsize_max = geo_upper * (uint64_t)meta->mm_psize;
- }
- mdbx_warning("meta[%u] consider get-%s pageno is %" PRIaPGNO
- " instead of wrong %" PRIaPGNO
- ", will be corrected on next commit(s)",
- meta_number, "upper", geo_upper, meta->mm_geo.upper);
- meta->mm_geo.upper = geo_upper;
- }
-
- /* LY: check and silently put mm_geo.now into [geo.lower...geo.upper].
- *
- * Copy-with-compaction by previous version of libmdbx could produce DB-file
- * less than meta.geo.lower bound, in case actual filling is low or no data
- * at all. This is not a problem as there is no damage or loss of data.
- * Therefore it is better not to consider such situation as an error, but
- * silently correct it. */
- pgno_t geo_now = meta->mm_geo.now;
- if (geo_now < geo_lower)
- geo_now = geo_lower;
- if (geo_now > geo_upper && meta->mm_geo.next <= geo_upper)
- geo_now = geo_upper;
-
- if (unlikely(meta->mm_geo.next > geo_now)) {
- mdbx_warning("meta[%u] next-pageno (%" PRIaPGNO
- ") is beyond end-pgno (%" PRIaPGNO "), skip it",
- meta_number, meta->mm_geo.next, geo_now);
- return MDBX_CORRUPTED;
- }
- if (meta->mm_geo.now != geo_now) {
- mdbx_warning("meta[%u] consider geo-%s pageno is %" PRIaPGNO
- " instead of wrong %" PRIaPGNO
- ", will be corrected on next commit(s)",
- meta_number, "now", geo_now, meta->mm_geo.now);
- meta->mm_geo.now = geo_now;
- }
-
- /* GC */
- if (meta->mm_dbs[FREE_DBI].md_root == P_INVALID) {
- if (unlikely(meta->mm_dbs[FREE_DBI].md_branch_pages ||
- meta->mm_dbs[FREE_DBI].md_depth ||
- meta->mm_dbs[FREE_DBI].md_entries ||
- meta->mm_dbs[FREE_DBI].md_leaf_pages ||
- meta->mm_dbs[FREE_DBI].md_overflow_pages)) {
- mdbx_warning("meta[%u] has false-empty %s, skip it", meta_number, "GC");
- return MDBX_CORRUPTED;
- }
- } else if (unlikely(meta->mm_dbs[FREE_DBI].md_root >= meta->mm_geo.next)) {
- mdbx_warning("meta[%u] has invalid %s-root %" PRIaPGNO ", skip it",
- meta_number, "GC", meta->mm_dbs[FREE_DBI].md_root);
- return MDBX_CORRUPTED;
- }
-
- /* MainDB */
- if (meta->mm_dbs[MAIN_DBI].md_root == P_INVALID) {
- if (unlikely(meta->mm_dbs[MAIN_DBI].md_branch_pages ||
- meta->mm_dbs[MAIN_DBI].md_depth ||
- meta->mm_dbs[MAIN_DBI].md_entries ||
- meta->mm_dbs[MAIN_DBI].md_leaf_pages ||
- meta->mm_dbs[MAIN_DBI].md_overflow_pages)) {
- mdbx_warning("meta[%u] has false-empty %s", meta_number, "MainDB");
- return MDBX_CORRUPTED;
- }
- } else if (unlikely(meta->mm_dbs[MAIN_DBI].md_root >= meta->mm_geo.next)) {
- mdbx_warning("meta[%u] has invalid %s-root %" PRIaPGNO ", skip it",
- meta_number, "MainDB", meta->mm_dbs[MAIN_DBI].md_root);
- return MDBX_CORRUPTED;
- }
-
- if (unlikely(meta->mm_dbs[FREE_DBI].md_mod_txnid > txnid)) {
- mdbx_warning("meta[%u] has wrong md_mod_txnid %" PRIaTXN " for %s, skip it",
- meta_number, meta->mm_dbs[FREE_DBI].md_mod_txnid, "GC");
- return MDBX_CORRUPTED;
- }
-
- if (unlikely(meta->mm_dbs[MAIN_DBI].md_mod_txnid > txnid)) {
- mdbx_warning("meta[%u] has wrong md_mod_txnid %" PRIaTXN " for %s, skip it",
- meta_number, meta->mm_dbs[MAIN_DBI].md_mod_txnid, "MainDB");
- return MDBX_CORRUPTED;
- }
-
- return MDBX_SUCCESS;
-}
-
-static int mdbx_validate_meta_copy(MDBX_env *env, const MDBX_meta *meta,
- MDBX_meta *dest) {
- *dest = *meta;
- return mdbx_validate_meta(env, dest, data_page(meta),
- bytes2pgno(env, (uint8_t *)meta - env->me_map),
- nullptr);
-}
-
-/* Read the environment parameters of a DB environment
- * before mapping it into memory. */
-__cold static int mdbx_read_header(MDBX_env *env, MDBX_meta *dest,
- const int lck_exclusive,
- const mdbx_mode_t mode_bits) {
- int rc = mdbx_filesize(env->me_lazy_fd, &env->me_dxb_mmap.filesize);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- memset(dest, 0, sizeof(MDBX_meta));
- unaligned_poke_u64(4, dest->mm_datasync_sign, MDBX_DATASIGN_WEAK);
- rc = MDBX_CORRUPTED;
-
- /* Read twice all meta pages so we can find the latest one. */
- unsigned loop_limit = NUM_METAS * 2;
- /* We don't know the page size on first time. So, just guess it. */
- unsigned guess_pagesize = 0;
- for (unsigned loop_count = 0; loop_count < loop_limit; ++loop_count) {
- const unsigned meta_number = loop_count % NUM_METAS;
- const unsigned offset = (guess_pagesize ? guess_pagesize
- : (loop_count > NUM_METAS) ? env->me_psize
- : env->me_os_psize) *
- meta_number;
-
- char buffer[MIN_PAGESIZE];
- unsigned retryleft = 42;
- while (1) {
- mdbx_trace("reading meta[%d]: offset %u, bytes %u, retry-left %u",
- meta_number, offset, MIN_PAGESIZE, retryleft);
- int err = mdbx_pread(env->me_lazy_fd, buffer, MIN_PAGESIZE, offset);
- if (err != MDBX_SUCCESS) {
- if (err == MDBX_ENODATA && offset == 0 && loop_count == 0 &&
- env->me_dxb_mmap.filesize == 0 &&
- mode_bits /* non-zero for DB creation */ != 0)
- mdbx_notice("read meta: empty file (%d, %s)", err,
- mdbx_strerror(err));
- else
- mdbx_error("read meta[%u,%u]: %i, %s", offset, MIN_PAGESIZE, err,
- mdbx_strerror(err));
- return err;
- }
-
- char again[MIN_PAGESIZE];
- err = mdbx_pread(env->me_lazy_fd, again, MIN_PAGESIZE, offset);
- if (err != MDBX_SUCCESS) {
- mdbx_error("read meta[%u,%u]: %i, %s", offset, MIN_PAGESIZE, err,
- mdbx_strerror(err));
- return err;
- }
-
- if (memcmp(buffer, again, MIN_PAGESIZE) == 0 || --retryleft == 0)
- break;
-
- mdbx_verbose("meta[%u] was updated, re-read it", meta_number);
- }
-
- if (!retryleft) {
- mdbx_error("meta[%u] is too volatile, skip it", meta_number);
- continue;
- }
-
- MDBX_page *const page = (MDBX_page *)buffer;
- MDBX_meta *const meta = page_meta(page);
- rc = mdbx_validate_meta(env, meta, page, meta_number, &guess_pagesize);
- if (rc != MDBX_SUCCESS)
- continue;
-
- if ((env->me_stuck_meta < 0)
- ? meta_ot(meta_bootid_match(meta) ? prefer_last : prefer_steady,
- env, dest, meta)
- : (meta_number == (unsigned)env->me_stuck_meta)) {
- *dest = *meta;
- if (!lck_exclusive && !META_IS_STEADY(dest))
- loop_limit += 1; /* LY: should re-read to hush race with update */
- mdbx_verbose("latch meta[%u]", meta_number);
- }
- }
-
- if (dest->mm_psize == 0 ||
- (env->me_stuck_meta < 0 &&
- !(META_IS_STEADY(dest) ||
- meta_weak_acceptable(env, dest, lck_exclusive)))) {
- mdbx_error("%s", "no usable meta-pages, database is corrupted");
- if (rc == MDBX_SUCCESS) {
- /* TODO: try to restore the database by fully checking b-tree structure
- * for the each meta page, if the corresponding option was given */
- return MDBX_CORRUPTED;
- }
- return rc;
- }
-
- return MDBX_SUCCESS;
-}
-
-__cold static MDBX_page *mdbx_meta_model(const MDBX_env *env, MDBX_page *model,
- unsigned num) {
- mdbx_ensure(env, is_powerof2(env->me_psize));
- mdbx_ensure(env, env->me_psize >= MIN_PAGESIZE);
- mdbx_ensure(env, env->me_psize <= MAX_PAGESIZE);
- mdbx_ensure(env, env->me_dbgeo.lower >= MIN_MAPSIZE);
- mdbx_ensure(env, env->me_dbgeo.upper <= MAX_MAPSIZE);
- mdbx_ensure(env, env->me_dbgeo.now >= env->me_dbgeo.lower);
- mdbx_ensure(env, env->me_dbgeo.now <= env->me_dbgeo.upper);
-
- memset(model, 0, env->me_psize);
- model->mp_pgno = num;
- model->mp_flags = P_META;
- MDBX_meta *const model_meta = page_meta(model);
- unaligned_poke_u64(4, model_meta->mm_magic_and_version, MDBX_DATA_MAGIC);
-
- model_meta->mm_geo.lower = bytes2pgno(env, env->me_dbgeo.lower);
- model_meta->mm_geo.upper = bytes2pgno(env, env->me_dbgeo.upper);
- model_meta->mm_geo.grow_pv = pages2pv(bytes2pgno(env, env->me_dbgeo.grow));
- model_meta->mm_geo.shrink_pv =
- pages2pv(bytes2pgno(env, env->me_dbgeo.shrink));
- model_meta->mm_geo.now = bytes2pgno(env, env->me_dbgeo.now);
- model_meta->mm_geo.next = NUM_METAS;
-
- mdbx_ensure(env, model_meta->mm_geo.lower >= MIN_PAGENO);
- mdbx_ensure(env, model_meta->mm_geo.upper <= MAX_PAGENO + 1);
- mdbx_ensure(env, model_meta->mm_geo.now >= model_meta->mm_geo.lower);
- mdbx_ensure(env, model_meta->mm_geo.now <= model_meta->mm_geo.upper);
- mdbx_ensure(env, model_meta->mm_geo.next >= MIN_PAGENO);
- mdbx_ensure(env, model_meta->mm_geo.next <= model_meta->mm_geo.now);
- mdbx_ensure(env, model_meta->mm_geo.grow_pv ==
- pages2pv(pv2pages(model_meta->mm_geo.grow_pv)));
- mdbx_ensure(env, model_meta->mm_geo.shrink_pv ==
- pages2pv(pv2pages(model_meta->mm_geo.shrink_pv)));
-
- model_meta->mm_psize = env->me_psize;
- model_meta->mm_dbs[FREE_DBI].md_flags = MDBX_INTEGERKEY;
- model_meta->mm_dbs[FREE_DBI].md_root = P_INVALID;
- model_meta->mm_dbs[MAIN_DBI].md_root = P_INVALID;
- meta_set_txnid(env, model_meta, MIN_TXNID + num);
- unaligned_poke_u64(4, model_meta->mm_datasync_sign, meta_sign(model_meta));
- mdbx_assert(env, meta_checktxnid(env, model_meta, true));
- return (MDBX_page *)((uint8_t *)model + env->me_psize);
-}
-
-/* Fill in most of the zeroed meta-pages for an empty database environment.
- * Return pointer to recently (head) meta-page. */
-__cold static MDBX_meta *mdbx_init_metas(const MDBX_env *env, void *buffer) {
- MDBX_page *page0 = (MDBX_page *)buffer;
- MDBX_page *page1 = mdbx_meta_model(env, page0, 0);
- MDBX_page *page2 = mdbx_meta_model(env, page1, 1);
- mdbx_meta_model(env, page2, 2);
- mdbx_assert(env, !meta_eq(env, page_meta(page0), page_meta(page1)));
- mdbx_assert(env, !meta_eq(env, page_meta(page1), page_meta(page2)));
- mdbx_assert(env, !meta_eq(env, page_meta(page2), page_meta(page0)));
- return page_meta(page2);
-}
-
-#if MDBX_ENABLE_MADVISE && !(defined(_WIN32) || defined(_WIN64))
-static size_t mdbx_madvise_threshold(const MDBX_env *env,
- const size_t largest_bytes) {
- /* TODO: use options */
- const unsigned factor = 9;
- const size_t threshold = (largest_bytes < (65536ul << factor))
- ? 65536 /* minimal threshold */
- : (largest_bytes > (MEGABYTE * 4 << factor))
- ? MEGABYTE * 4 /* maximal threshold */
- : largest_bytes >> factor;
- return bytes_align2os_bytes(env, threshold);
-}
-#endif /* MDBX_ENABLE_MADVISE */
-
-static int mdbx_sync_locked(MDBX_env *env, unsigned flags,
- MDBX_meta *const pending) {
- mdbx_assert(env, ((env->me_flags ^ flags) & MDBX_WRITEMAP) == 0);
- const MDBX_meta *const meta0 = METAPAGE(env, 0);
- const MDBX_meta *const meta1 = METAPAGE(env, 1);
- const MDBX_meta *const meta2 = METAPAGE(env, 2);
- const MDBX_meta *const head = constmeta_prefer_last(env);
- int rc;
-
- mdbx_assert(env, meta_eq_mask(env) == 0);
- mdbx_assert(env,
- pending < METAPAGE(env, 0) || pending > METAPAGE(env, NUM_METAS));
- mdbx_assert(env, (env->me_flags & (MDBX_RDONLY | MDBX_FATAL_ERROR)) == 0);
- mdbx_assert(env, pending->mm_geo.next <= pending->mm_geo.now);
-
- if (flags & MDBX_SAFE_NOSYNC) {
- /* Check auto-sync conditions */
- const pgno_t autosync_threshold =
- atomic_load32(&env->me_lck->mti_autosync_threshold, mo_Relaxed);
- const uint64_t autosync_period =
- atomic_load64(&env->me_lck->mti_autosync_period, mo_Relaxed);
- if ((autosync_threshold &&
- atomic_load32(&env->me_lck->mti_unsynced_pages, mo_Relaxed) >=
- autosync_threshold) ||
- (autosync_period &&
- mdbx_osal_monotime() -
- atomic_load64(&env->me_lck->mti_sync_timestamp, mo_Relaxed) >=
- autosync_period))
- flags &= MDBX_WRITEMAP | MDBX_SHRINK_ALLOWED; /* force steady */
- }
-
- pgno_t shrink = 0;
- if (flags & MDBX_SHRINK_ALLOWED) {
- /* LY: check conditions to discard unused pages */
- const pgno_t largest_pgno = mdbx_find_largest(
- env, (head->mm_geo.next > pending->mm_geo.next) ? head->mm_geo.next
- : pending->mm_geo.next);
- mdbx_assert(env, largest_pgno >= NUM_METAS);
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- const pgno_t edge = env->me_poison_edge;
- if (edge > largest_pgno) {
- env->me_poison_edge = largest_pgno;
- VALGRIND_MAKE_MEM_NOACCESS(env->me_map + pgno2bytes(env, largest_pgno),
- pgno2bytes(env, edge - largest_pgno));
- MDBX_ASAN_POISON_MEMORY_REGION(env->me_map +
- pgno2bytes(env, largest_pgno),
- pgno2bytes(env, edge - largest_pgno));
- }
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-#if MDBX_ENABLE_MADVISE && \
- (defined(MADV_DONTNEED) || defined(POSIX_MADV_DONTNEED))
- const size_t largest_bytes = pgno2bytes(env, largest_pgno);
- /* threshold to avoid unreasonable frequent madvise() calls */
- const size_t madvise_threshold = mdbx_madvise_threshold(env, largest_bytes);
- const size_t discard_edge_bytes = bytes_align2os_bytes(
- env, ((MDBX_RDONLY &
- (env->me_lck_mmap.lck ? env->me_lck_mmap.lck->mti_envmode.weak
- : env->me_flags))
- ? largest_bytes
- : largest_bytes + madvise_threshold));
- const pgno_t discard_edge_pgno = bytes2pgno(env, discard_edge_bytes);
- const pgno_t prev_discarded_pgno =
- atomic_load32(&env->me_lck->mti_discarded_tail, mo_Relaxed);
- if (prev_discarded_pgno >=
- discard_edge_pgno + bytes2pgno(env, madvise_threshold)) {
- mdbx_notice("open-MADV_%s %u..%u", "DONTNEED", largest_pgno,
- prev_discarded_pgno);
- atomic_store32(&env->me_lck->mti_discarded_tail, discard_edge_pgno,
- mo_Relaxed);
- const size_t prev_discarded_bytes =
- ceil_powerof2(pgno2bytes(env, prev_discarded_pgno), env->me_os_psize);
- mdbx_ensure(env, prev_discarded_bytes > discard_edge_bytes);
-#if defined(MADV_DONTNEED)
- int advise = MADV_DONTNEED;
-#if defined(MADV_FREE) && \
- 0 /* MADV_FREE works for only anonymous vma at the moment */
- if ((env->me_flags & MDBX_WRITEMAP) &&
- mdbx_linux_kernel_version > 0x04050000)
- advise = MADV_FREE;
-#endif /* MADV_FREE */
- int err = madvise(env->me_map + discard_edge_bytes,
- prev_discarded_bytes - discard_edge_bytes, advise)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
-#else
- int err = ignore_enosys(posix_madvise(
- env->me_map + discard_edge_bytes,
- prev_discarded_bytes - discard_edge_bytes, POSIX_MADV_DONTNEED));
-#endif
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
- }
-#endif /* MDBX_ENABLE_MADVISE && (MADV_DONTNEED || POSIX_MADV_DONTNEED) */
-
- /* LY: check conditions to shrink datafile */
- const pgno_t backlog_gap = 3 + pending->mm_dbs[FREE_DBI].md_depth * 3;
- pgno_t shrink_step = 0;
- if (pending->mm_geo.shrink_pv &&
- pending->mm_geo.now - pending->mm_geo.next >
- (shrink_step = pv2pages(pending->mm_geo.shrink_pv)) + backlog_gap) {
- if (pending->mm_geo.now > largest_pgno &&
- pending->mm_geo.now - largest_pgno > shrink_step + backlog_gap) {
- pgno_t grow_step = 0;
- const pgno_t aligner =
- pending->mm_geo.grow_pv
- ? (grow_step = pv2pages(pending->mm_geo.grow_pv))
- : shrink_step;
- const pgno_t with_backlog_gap = largest_pgno + backlog_gap;
- const pgno_t aligned = pgno_align2os_pgno(
- env, with_backlog_gap + aligner - with_backlog_gap % aligner);
- const pgno_t bottom =
- (aligned > pending->mm_geo.lower) ? aligned : pending->mm_geo.lower;
- if (pending->mm_geo.now > bottom) {
- if (META_IS_STEADY(meta_prefer_steady(env)))
- /* force steady, but only if steady-checkpoint is present */
- flags &= MDBX_WRITEMAP | MDBX_SHRINK_ALLOWED;
- shrink = pending->mm_geo.now - bottom;
- pending->mm_geo.now = bottom;
- if (unlikely(constmeta_txnid(env, head) ==
- unaligned_peek_u64(4, pending->mm_txnid_a))) {
- const txnid_t txnid =
- safe64_txnid_next(unaligned_peek_u64(4, pending->mm_txnid_a));
- mdbx_notice("force-forward pending-txn %" PRIaTXN " -> %" PRIaTXN,
- unaligned_peek_u64(4, pending->mm_txnid_a), txnid);
- mdbx_ensure(env, env->me_txn0->mt_owner != mdbx_thread_self() &&
- !env->me_txn);
- if (unlikely(txnid > MAX_TXNID)) {
- rc = MDBX_TXN_FULL;
- mdbx_error("txnid overflow, raise %d", rc);
- goto fail;
- }
- meta_set_txnid(env, pending, txnid);
- mdbx_assert(env, meta_checktxnid(env, pending, true));
- }
- }
- }
- }
- }
-
- /* LY: step#1 - sync previously written/updated data-pages */
- rc = MDBX_RESULT_FALSE /* carry steady */;
- if (atomic_load32(&env->me_lck->mti_unsynced_pages, mo_Relaxed)) {
- mdbx_assert(env, ((flags ^ env->me_flags) & MDBX_WRITEMAP) == 0);
- enum mdbx_syncmode_bits mode_bits = MDBX_SYNC_NONE;
- if ((flags & MDBX_SAFE_NOSYNC) == 0) {
- mode_bits = MDBX_SYNC_DATA;
- if (pending->mm_geo.next > meta_prefer_steady(env)->mm_geo.now)
- mode_bits |= MDBX_SYNC_SIZE;
- if (flags & MDBX_NOMETASYNC)
- mode_bits |= MDBX_SYNC_IODQ;
- }
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- if (flags & MDBX_WRITEMAP)
- rc =
- mdbx_msync(&env->me_dxb_mmap, 0,
- pgno_align2os_bytes(env, pending->mm_geo.next), mode_bits);
- else
- rc = mdbx_fsync(env->me_lazy_fd, mode_bits);
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- rc = (flags & MDBX_SAFE_NOSYNC) ? MDBX_RESULT_TRUE /* carry non-steady */
- : MDBX_RESULT_FALSE /* carry steady */;
- }
- mdbx_assert(env, meta_checktxnid(env, pending, true));
-
- /* Steady or Weak */
- if (rc == MDBX_RESULT_FALSE /* carry steady */) {
- atomic_store64(&env->me_lck->mti_sync_timestamp, mdbx_osal_monotime(),
- mo_Relaxed);
- unaligned_poke_u64(4, pending->mm_datasync_sign, meta_sign(pending));
- atomic_store32(&env->me_lck->mti_unsynced_pages, 0, mo_Relaxed);
- } else {
- assert(rc == MDBX_RESULT_TRUE /* carry non-steady */);
- unaligned_poke_u64(4, pending->mm_datasync_sign, MDBX_DATASIGN_WEAK);
- }
-
- MDBX_meta *target = nullptr;
- if (constmeta_txnid(env, head) ==
- unaligned_peek_u64(4, pending->mm_txnid_a)) {
- mdbx_assert(env, memcmp(&head->mm_dbs, &pending->mm_dbs,
- sizeof(head->mm_dbs)) == 0);
- mdbx_assert(env, memcmp(&head->mm_canary, &pending->mm_canary,
- sizeof(head->mm_canary)) == 0);
- mdbx_assert(env, memcmp(&head->mm_geo, &pending->mm_geo,
- sizeof(pending->mm_geo)) == 0);
- if (!META_IS_STEADY(head) && META_IS_STEADY(pending))
- target = (MDBX_meta *)head;
- else {
- mdbx_ensure(env, meta_eq(env, head, pending));
- mdbx_debug("%s", "skip update meta");
- return MDBX_SUCCESS;
- }
- } else if (head == meta0)
- target = (MDBX_meta *)meta_ancient_prefer_weak(env, meta1, meta2);
- else if (head == meta1)
- target = (MDBX_meta *)meta_ancient_prefer_weak(env, meta0, meta2);
- else {
- mdbx_assert(env, head == meta2);
- target = (MDBX_meta *)meta_ancient_prefer_weak(env, meta0, meta1);
- }
-
- /* LY: step#2 - update meta-page. */
- mdbx_debug(
- "writing meta%" PRIaPGNO " = root %" PRIaPGNO "/%" PRIaPGNO
- ", geo %" PRIaPGNO "/%" PRIaPGNO "-%" PRIaPGNO "/%" PRIaPGNO
- " +%u -%u, txn_id %" PRIaTXN ", %s",
- data_page(target)->mp_pgno, pending->mm_dbs[MAIN_DBI].md_root,
- pending->mm_dbs[FREE_DBI].md_root, pending->mm_geo.lower,
- pending->mm_geo.next, pending->mm_geo.now, pending->mm_geo.upper,
- pv2pages(pending->mm_geo.grow_pv), pv2pages(pending->mm_geo.shrink_pv),
- unaligned_peek_u64(4, pending->mm_txnid_a), mdbx_durable_str(pending));
-
- mdbx_debug("meta0: %s, %s, txn_id %" PRIaTXN ", root %" PRIaPGNO
- "/%" PRIaPGNO,
- (meta0 == head) ? "head"
- : (meta0 == target) ? "tail"
- : "stay",
- mdbx_durable_str(meta0), meta_txnid(env, meta0),
- meta0->mm_dbs[MAIN_DBI].md_root, meta0->mm_dbs[FREE_DBI].md_root);
- mdbx_debug("meta1: %s, %s, txn_id %" PRIaTXN ", root %" PRIaPGNO
- "/%" PRIaPGNO,
- (meta1 == head) ? "head"
- : (meta1 == target) ? "tail"
- : "stay",
- mdbx_durable_str(meta1), meta_txnid(env, meta1),
- meta1->mm_dbs[MAIN_DBI].md_root, meta1->mm_dbs[FREE_DBI].md_root);
- mdbx_debug("meta2: %s, %s, txn_id %" PRIaTXN ", root %" PRIaPGNO
- "/%" PRIaPGNO,
- (meta2 == head) ? "head"
- : (meta2 == target) ? "tail"
- : "stay",
- mdbx_durable_str(meta2), meta_txnid(env, meta2),
- meta2->mm_dbs[MAIN_DBI].md_root, meta2->mm_dbs[FREE_DBI].md_root);
-
- mdbx_assert(env, !meta_eq(env, pending, meta0));
- mdbx_assert(env, !meta_eq(env, pending, meta1));
- mdbx_assert(env, !meta_eq(env, pending, meta2));
-
- mdbx_assert(env, ((env->me_flags ^ flags) & MDBX_WRITEMAP) == 0);
- mdbx_ensure(env,
- target == head || constmeta_txnid(env, target) <
- unaligned_peek_u64(4, pending->mm_txnid_a));
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- if (flags & MDBX_WRITEMAP) {
- mdbx_jitter4testing(true);
- if (likely(target != head)) {
- /* LY: 'invalidate' the meta. */
- meta_update_begin(env, target,
- unaligned_peek_u64(4, pending->mm_txnid_a));
- unaligned_poke_u64(4, target->mm_datasync_sign, MDBX_DATASIGN_WEAK);
-#ifndef NDEBUG
- /* debug: provoke failure to catch a violators, but don't touch mm_psize
- * to allow readers catch actual pagesize. */
- uint8_t *provoke_begin = (uint8_t *)&target->mm_dbs[FREE_DBI].md_root;
- uint8_t *provoke_end = (uint8_t *)&target->mm_datasync_sign;
- memset(provoke_begin, 0xCC, provoke_end - provoke_begin);
- mdbx_jitter4testing(false);
-#endif
-
- /* LY: update info */
- target->mm_geo = pending->mm_geo;
- target->mm_dbs[FREE_DBI] = pending->mm_dbs[FREE_DBI];
- target->mm_dbs[MAIN_DBI] = pending->mm_dbs[MAIN_DBI];
- target->mm_canary = pending->mm_canary;
- memcpy(target->mm_pages_retired, pending->mm_pages_retired, 8);
- mdbx_jitter4testing(true);
-
- /* LY: 'commit' the meta */
- meta_update_end(env, target, unaligned_peek_u64(4, pending->mm_txnid_b));
- mdbx_jitter4testing(true);
- mdbx_assert(env, meta_checktxnid(env, target, true));
- } else {
- /* dangerous case (target == head), only mm_datasync_sign could
- * me updated, check assertions once again */
- mdbx_ensure(env, constmeta_txnid(env, head) ==
- unaligned_peek_u64(4, pending->mm_txnid_a) &&
- !META_IS_STEADY(head) && META_IS_STEADY(pending));
- mdbx_ensure(env, memcmp(&head->mm_geo, &pending->mm_geo,
- sizeof(head->mm_geo)) == 0);
- mdbx_ensure(env, memcmp(&head->mm_dbs, &pending->mm_dbs,
- sizeof(head->mm_dbs)) == 0);
- mdbx_ensure(env, memcmp(&head->mm_canary, &pending->mm_canary,
- sizeof(head->mm_canary)) == 0);
- }
- memcpy(target->mm_datasync_sign, pending->mm_datasync_sign, 8);
- mdbx_flush_incoherent_cpu_writeback();
- mdbx_jitter4testing(true);
- /* sync meta-pages */
- rc =
- mdbx_msync(&env->me_dxb_mmap, 0, pgno_align2os_bytes(env, NUM_METAS),
- (flags & MDBX_NOMETASYNC) ? MDBX_SYNC_NONE
- : MDBX_SYNC_DATA | MDBX_SYNC_IODQ);
- if (unlikely(rc != MDBX_SUCCESS))
- goto fail;
- } else {
- const MDBX_meta undo_meta = *target;
- const mdbx_filehandle_t fd = (env->me_dsync_fd != INVALID_HANDLE_VALUE)
- ? env->me_dsync_fd
- : env->me_lazy_fd;
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- rc = mdbx_pwrite(fd, pending, sizeof(MDBX_meta),
- (uint8_t *)target - env->me_map);
- if (unlikely(rc != MDBX_SUCCESS)) {
- undo:
- mdbx_debug("%s", "write failed, disk error?");
- /* On a failure, the pagecache still contains the new data.
- * Try write some old data back, to prevent it from being used. */
- mdbx_pwrite(fd, &undo_meta, sizeof(MDBX_meta),
- (uint8_t *)target - env->me_map);
- goto fail;
- }
- mdbx_flush_incoherent_mmap(target, sizeof(MDBX_meta), env->me_os_psize);
- /* sync meta-pages */
- if ((flags & MDBX_NOMETASYNC) == 0 && fd == env->me_lazy_fd) {
- rc = mdbx_fsync(env->me_lazy_fd, MDBX_SYNC_DATA | MDBX_SYNC_IODQ);
- if (rc != MDBX_SUCCESS)
- goto undo;
- }
- mdbx_assert(env, meta_checktxnid(env, target, true));
- }
- env->me_lck->mti_meta_sync_txnid.weak =
- (uint32_t)unaligned_peek_u64(4, pending->mm_txnid_a) -
- ((flags & MDBX_NOMETASYNC) ? UINT32_MAX / 3 : 0);
-
- /* LY: shrink datafile if needed */
- if (unlikely(shrink)) {
- mdbx_verbose("shrink to %" PRIaPGNO " pages (-%" PRIaPGNO ")",
- pending->mm_geo.now, shrink);
- rc = mdbx_mapresize_implicit(env, pending->mm_geo.next, pending->mm_geo.now,
- pending->mm_geo.upper);
- if (rc != MDBX_SUCCESS && rc != MDBX_EPERM)
- goto fail;
- mdbx_assert(env, meta_checktxnid(env, target, true));
- }
-
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (likely(lck))
- /* toggle oldest refresh */
- atomic_store32(&lck->mti_readers_refresh_flag, false, mo_Relaxed);
-
- return MDBX_SUCCESS;
-
-fail:
- env->me_flags |= MDBX_FATAL_ERROR;
- return rc;
-}
-
-static void recalculate_merge_threshold(MDBX_env *env) {
- const unsigned bytes = page_space(env);
- env->me_merge_threshold =
- (uint16_t)(bytes -
- (bytes * env->me_options.merge_threshold_16dot16_percent >>
- 16));
- env->me_merge_threshold_gc =
- (uint16_t)(bytes -
- ((env->me_options.merge_threshold_16dot16_percent > 19005)
- ? bytes / 3 /* 33 % */
- : bytes / 4 /* 25 % */));
-}
-
-__cold static void mdbx_setup_pagesize(MDBX_env *env, const size_t pagesize) {
- STATIC_ASSERT(PTRDIFF_MAX > MAX_MAPSIZE);
- STATIC_ASSERT(MIN_PAGESIZE > sizeof(MDBX_page) + sizeof(MDBX_meta));
- mdbx_ensure(env, is_powerof2(pagesize));
- mdbx_ensure(env, pagesize >= MIN_PAGESIZE);
- mdbx_ensure(env, pagesize <= MAX_PAGESIZE);
- env->me_psize = (unsigned)pagesize;
- if (env->me_pbuf) {
- mdbx_memalign_free(env->me_pbuf);
- env->me_pbuf = nullptr;
- }
-
- STATIC_ASSERT(MAX_GC1OVPAGE(MIN_PAGESIZE) > 4);
- STATIC_ASSERT(MAX_GC1OVPAGE(MAX_PAGESIZE) < MDBX_PGL_LIMIT);
- const intptr_t maxgc_ov1page = (pagesize - PAGEHDRSZ) / sizeof(pgno_t) - 1;
- mdbx_ensure(env, maxgc_ov1page > 42 &&
- maxgc_ov1page < (intptr_t)MDBX_PGL_LIMIT / 4);
- env->me_maxgc_ov1page = (unsigned)maxgc_ov1page;
-
- STATIC_ASSERT(LEAF_NODE_MAX(MIN_PAGESIZE) > sizeof(MDBX_db) + NODESIZE + 42);
- STATIC_ASSERT(LEAF_NODE_MAX(MAX_PAGESIZE) < UINT16_MAX);
- STATIC_ASSERT(LEAF_NODE_MAX(MIN_PAGESIZE) >= BRANCH_NODE_MAX(MIN_PAGESIZE));
- STATIC_ASSERT(BRANCH_NODE_MAX(MAX_PAGESIZE) > NODESIZE + 42);
- STATIC_ASSERT(BRANCH_NODE_MAX(MAX_PAGESIZE) < UINT16_MAX);
- const intptr_t branch_nodemax = BRANCH_NODE_MAX(pagesize);
- const intptr_t leaf_nodemax = LEAF_NODE_MAX(pagesize);
- mdbx_ensure(env,
- branch_nodemax > (intptr_t)(NODESIZE + 42) &&
- branch_nodemax % 2 == 0 &&
- leaf_nodemax > (intptr_t)(sizeof(MDBX_db) + NODESIZE + 42) &&
- leaf_nodemax >= branch_nodemax &&
- leaf_nodemax < (int)UINT16_MAX && leaf_nodemax % 2 == 0);
- env->me_leaf_nodemax = (unsigned)leaf_nodemax;
- env->me_psize2log = (uint8_t)log2n_powerof2(pagesize);
- mdbx_assert(env, pgno2bytes(env, 1) == pagesize);
- mdbx_assert(env, bytes2pgno(env, pagesize + pagesize) == 2);
- recalculate_merge_threshold(env);
-
- const pgno_t max_pgno = bytes2pgno(env, MAX_MAPSIZE);
- if (!env->me_options.flags.non_auto.dp_limit) {
- /* auto-setup dp_limit by "The42" ;-) */
- intptr_t total_ram_pages, avail_ram_pages;
- int err = mdbx_get_sysraminfo(nullptr, &total_ram_pages, &avail_ram_pages);
- if (unlikely(err != MDBX_SUCCESS))
- mdbx_error("mdbx_get_sysraminfo(), rc %d", err);
- else {
- size_t reasonable_dpl_limit =
- (size_t)(total_ram_pages + avail_ram_pages) / 42;
- if (pagesize > env->me_os_psize)
- reasonable_dpl_limit /= pagesize / env->me_os_psize;
- else if (pagesize < env->me_os_psize)
- reasonable_dpl_limit *= env->me_os_psize / pagesize;
- reasonable_dpl_limit = (reasonable_dpl_limit < MDBX_PGL_LIMIT)
- ? reasonable_dpl_limit
- : MDBX_PGL_LIMIT;
- reasonable_dpl_limit = (reasonable_dpl_limit > CURSOR_STACK * 4)
- ? reasonable_dpl_limit
- : CURSOR_STACK * 4;
- env->me_options.dp_limit = (unsigned)reasonable_dpl_limit;
- }
- }
- if (env->me_options.dp_limit > max_pgno - NUM_METAS)
- env->me_options.dp_limit = max_pgno - NUM_METAS;
- if (env->me_options.dp_initial > env->me_options.dp_limit)
- env->me_options.dp_initial = env->me_options.dp_limit;
-}
-
-static __inline MDBX_CONST_FUNCTION MDBX_lockinfo *
-lckless_stub(const MDBX_env *env) {
- uintptr_t stub = (uintptr_t)&env->x_lckless_stub;
- /* align to avoid false-positive alarm from UndefinedBehaviorSanitizer */
- stub = (stub + MDBX_CACHELINE_SIZE - 1) & ~(MDBX_CACHELINE_SIZE - 1);
- return (MDBX_lockinfo *)stub;
-}
-
-__cold int mdbx_env_create(MDBX_env **penv) {
- MDBX_env *env = mdbx_calloc(1, sizeof(MDBX_env));
- if (unlikely(!env))
- return MDBX_ENOMEM;
-
- env->me_maxreaders = DEFAULT_READERS;
- env->me_maxdbs = env->me_numdbs = CORE_DBS;
- env->me_lazy_fd = INVALID_HANDLE_VALUE;
- env->me_dsync_fd = INVALID_HANDLE_VALUE;
- env->me_lfd = INVALID_HANDLE_VALUE;
- env->me_pid = mdbx_getpid();
- env->me_stuck_meta = -1;
-
- env->me_options.dp_reserve_limit = 1024;
- env->me_options.rp_augment_limit = 256 * 1024;
- env->me_options.dp_limit = 64 * 1024;
- if (env->me_options.dp_limit > MAX_PAGENO + 1 - NUM_METAS)
- env->me_options.dp_limit = MAX_PAGENO + 1 - NUM_METAS;
- env->me_options.dp_initial = MDBX_PNL_INITIAL;
- if (env->me_options.dp_initial > env->me_options.dp_limit)
- env->me_options.dp_initial = env->me_options.dp_limit;
- env->me_options.spill_max_denominator = 8;
- env->me_options.spill_min_denominator = 8;
- env->me_options.spill_parent4child_denominator = 0;
- env->me_options.dp_loose_limit = 64;
- env->me_options.merge_threshold_16dot16_percent = 65536 / 4 /* 25% */;
-
- int rc;
- const size_t os_psize = mdbx_syspagesize();
- if (unlikely(!is_powerof2(os_psize) || os_psize < MIN_PAGESIZE)) {
- mdbx_error("unsuitable system pagesize %" PRIuPTR, os_psize);
- rc = MDBX_INCOMPATIBLE;
- goto bailout;
- }
- env->me_os_psize = (unsigned)os_psize;
- mdbx_setup_pagesize(env, (env->me_os_psize < MAX_PAGESIZE) ? env->me_os_psize
- : MAX_PAGESIZE);
-
- rc = mdbx_fastmutex_init(&env->me_dbi_lock);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
-
-#if defined(_WIN32) || defined(_WIN64)
- mdbx_srwlock_Init(&env->me_remap_guard);
- InitializeCriticalSection(&env->me_windowsbug_lock);
-#else
- rc = mdbx_fastmutex_init(&env->me_remap_guard);
- if (unlikely(rc != MDBX_SUCCESS)) {
- mdbx_fastmutex_destroy(&env->me_dbi_lock);
- goto bailout;
- }
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV
- MDBX_lockinfo *const stub = lckless_stub(env);
- rc = mdbx_ipclock_stub(&stub->mti_wlock);
-#endif /* MDBX_LOCKING */
- if (unlikely(rc != MDBX_SUCCESS)) {
- mdbx_fastmutex_destroy(&env->me_remap_guard);
- mdbx_fastmutex_destroy(&env->me_dbi_lock);
- goto bailout;
- }
-#endif /* Windows */
-
- VALGRIND_CREATE_MEMPOOL(env, 0, 0);
- env->me_signature.weak = MDBX_ME_SIGNATURE;
- *penv = env;
- return MDBX_SUCCESS;
-
-bailout:
- mdbx_free(env);
- *penv = nullptr;
- return rc;
-}
-
-__cold static intptr_t get_reasonable_db_maxsize(intptr_t *cached_result) {
- if (*cached_result == 0) {
- intptr_t pagesize, total_ram_pages;
- if (unlikely(mdbx_get_sysraminfo(&pagesize, &total_ram_pages, nullptr) !=
- MDBX_SUCCESS))
- return *cached_result = MAX_MAPSIZE32 /* the 32-bit limit is good enough
- for fallback */
- ;
-
- if (unlikely((size_t)total_ram_pages * 2 > MAX_MAPSIZE / (size_t)pagesize))
- return *cached_result = MAX_MAPSIZE;
- assert(MAX_MAPSIZE >= (size_t)(total_ram_pages * pagesize * 2));
-
- /* Suggesting should not be more than golden ratio of the size of RAM. */
- *cached_result = (intptr_t)((size_t)total_ram_pages * 207 >> 7) * pagesize;
-
- /* Round to the nearest human-readable granulation. */
- for (size_t unit = MEGABYTE; unit; unit <<= 5) {
- const size_t floor = floor_powerof2(*cached_result, unit);
- const size_t ceil = ceil_powerof2(*cached_result, unit);
- const size_t threshold = (size_t)*cached_result >> 4;
- const bool down =
- *cached_result - floor < ceil - *cached_result || ceil > MAX_MAPSIZE;
- if (threshold < (down ? *cached_result - floor : ceil - *cached_result))
- break;
- *cached_result = down ? floor : ceil;
- }
- }
- return *cached_result;
-}
-
-__cold LIBMDBX_API int
-mdbx_env_set_geometry(MDBX_env *env, intptr_t size_lower, intptr_t size_now,
- intptr_t size_upper, intptr_t growth_step,
- intptr_t shrink_threshold, intptr_t pagesize) {
- int rc = check_env(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- const bool inside_txn =
- (env->me_txn0 && env->me_txn0->mt_owner == mdbx_thread_self());
-
-#if MDBX_DEBUG
- if (growth_step < 0) {
- growth_step = 1;
- if (shrink_threshold < 0)
- shrink_threshold = 1;
- }
-#endif /* MDBX_DEBUG */
-
- intptr_t reasonable_maxsize = 0;
- bool need_unlock = false;
- if (env->me_map) {
- /* env already mapped */
- if (unlikely(env->me_flags & MDBX_RDONLY))
- return MDBX_EACCESS;
-
- if (!inside_txn) {
- int err = mdbx_txn_lock(env, false);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- need_unlock = true;
- }
- const MDBX_meta *head = constmeta_prefer_last(env);
- if (!inside_txn) {
- env->me_txn0->mt_txnid = constmeta_txnid(env, head);
- mdbx_find_oldest(env->me_txn0);
- }
-
- /* get untouched params from DB */
- if (pagesize <= 0 || pagesize >= INT_MAX)
- pagesize = env->me_psize;
- if (size_lower < 0)
- size_lower = pgno2bytes(env, head->mm_geo.lower);
- if (size_now < 0)
- size_now = pgno2bytes(env, head->mm_geo.now);
- if (size_upper < 0)
- size_upper = pgno2bytes(env, head->mm_geo.upper);
- if (growth_step < 0)
- growth_step = pgno2bytes(env, pv2pages(head->mm_geo.grow_pv));
- if (shrink_threshold < 0)
- shrink_threshold = pgno2bytes(env, pv2pages(head->mm_geo.shrink_pv));
-
- if (pagesize != (intptr_t)env->me_psize) {
- rc = MDBX_EINVAL;
- goto bailout;
- }
- const size_t usedbytes =
- pgno2bytes(env, mdbx_find_largest(env, head->mm_geo.next));
- if ((size_t)size_upper < usedbytes) {
- rc = MDBX_MAP_FULL;
- goto bailout;
- }
- if ((size_t)size_now < usedbytes)
- size_now = usedbytes;
- } else {
- /* env NOT yet mapped */
- if (unlikely(inside_txn))
- return MDBX_PANIC;
-
- /* is requested some auto-value for pagesize ? */
- if (pagesize >= INT_MAX /* maximal */)
- pagesize = MAX_PAGESIZE;
- else if (pagesize <= 0) {
- if (pagesize < 0 /* default */) {
- pagesize = env->me_os_psize;
- if ((uintptr_t)pagesize > MAX_PAGESIZE)
- pagesize = MAX_PAGESIZE;
- mdbx_assert(env, (uintptr_t)pagesize >= MIN_PAGESIZE);
- } else if (pagesize == 0 /* minimal */)
- pagesize = MIN_PAGESIZE;
-
- /* choose pagesize */
- intptr_t max_size = (size_now > size_lower) ? size_now : size_lower;
- max_size = (size_upper > max_size) ? size_upper : max_size;
- if (max_size < 0 /* default */)
- max_size = DEFAULT_MAPSIZE;
- else if (max_size == 0 /* minimal */)
- max_size = MIN_MAPSIZE;
- else if (max_size >= (intptr_t)MAX_MAPSIZE /* maximal */)
- max_size = get_reasonable_db_maxsize(&reasonable_maxsize);
-
- while (max_size > pagesize * (int64_t)(MAX_PAGENO + 1) &&
- pagesize < MAX_PAGESIZE)
- pagesize <<= 1;
- }
- }
-
- if (pagesize < (intptr_t)MIN_PAGESIZE || pagesize > (intptr_t)MAX_PAGESIZE ||
- !is_powerof2(pagesize)) {
- rc = MDBX_EINVAL;
- goto bailout;
- }
-
- if (size_lower <= 0) {
- size_lower = MIN_MAPSIZE;
- if (MIN_MAPSIZE / pagesize < MIN_PAGENO)
- size_lower = MIN_PAGENO * pagesize;
- }
- if (size_lower >= INTPTR_MAX) {
- size_lower = get_reasonable_db_maxsize(&reasonable_maxsize);
- if ((size_t)size_lower / pagesize > MAX_PAGENO + 1)
- size_lower = pagesize * (MAX_PAGENO + 1);
- }
-
- if (size_now <= 0) {
- size_now = size_lower;
- if (size_upper >= size_lower && size_now > size_upper)
- size_now = size_upper;
- }
- if (size_now >= INTPTR_MAX) {
- size_now = get_reasonable_db_maxsize(&reasonable_maxsize);
- if ((size_t)size_now / pagesize > MAX_PAGENO + 1)
- size_now = pagesize * (MAX_PAGENO + 1);
- }
-
- if (size_upper <= 0) {
- if (size_now >= get_reasonable_db_maxsize(&reasonable_maxsize) / 2)
- size_upper = get_reasonable_db_maxsize(&reasonable_maxsize);
- else if (MAX_MAPSIZE != MAX_MAPSIZE32 &&
- (size_t)size_now >= MAX_MAPSIZE32 / 2 &&
- (size_t)size_now <= MAX_MAPSIZE32 / 4 * 3)
- size_upper = MAX_MAPSIZE32;
- else {
- size_upper = size_now + size_now;
- if ((size_t)size_upper < DEFAULT_MAPSIZE * 2)
- size_upper = DEFAULT_MAPSIZE * 2;
- }
- if ((size_t)size_upper / pagesize > (MAX_PAGENO + 1))
- size_upper = pagesize * (MAX_PAGENO + 1);
- } else if (size_upper >= INTPTR_MAX) {
- size_upper = get_reasonable_db_maxsize(&reasonable_maxsize);
- if ((size_t)size_upper / pagesize > MAX_PAGENO + 1)
- size_upper = pagesize * (MAX_PAGENO + 1);
- }
-
- if (unlikely(size_lower < (intptr_t)MIN_MAPSIZE || size_lower > size_upper)) {
- rc = MDBX_EINVAL;
- goto bailout;
- }
-
- if ((uint64_t)size_lower / pagesize < MIN_PAGENO) {
- rc = MDBX_EINVAL;
- goto bailout;
- }
-
- if (unlikely((size_t)size_upper > MAX_MAPSIZE ||
- (uint64_t)size_upper / pagesize > MAX_PAGENO + 1)) {
- rc = MDBX_TOO_LARGE;
- goto bailout;
- }
-
- const size_t unit = (env->me_os_psize > (size_t)pagesize) ? env->me_os_psize
- : (size_t)pagesize;
- size_lower = ceil_powerof2(size_lower, unit);
- size_upper = ceil_powerof2(size_upper, unit);
- size_now = ceil_powerof2(size_now, unit);
-
- /* LY: подбираем значение size_upper:
- * - кратное размеру Ñтраницы
- * - без Ð½Ð°Ñ€ÑƒÑˆÐµÐ½Ð¸Ñ MAX_MAPSIZE и MAX_PAGENO */
- while (unlikely((size_t)size_upper > MAX_MAPSIZE ||
- (uint64_t)size_upper / pagesize > MAX_PAGENO + 1)) {
- if ((size_t)size_upper < unit + MIN_MAPSIZE ||
- (size_t)size_upper < (size_t)pagesize * (MIN_PAGENO + 1)) {
- /* паранойа на Ñлучай Ð¿ÐµÑ€ÐµÐ¿Ð¾Ð»Ð½ÐµÐ½Ð¸Ñ Ð¿Ñ€Ð¸ невероÑтных значениÑÑ… */
- rc = MDBX_EINVAL;
- goto bailout;
- }
- size_upper -= unit;
- if ((size_t)size_upper < (size_t)size_lower)
- size_lower = size_upper;
- }
- mdbx_assert(env, (size_upper - size_lower) % env->me_os_psize == 0);
-
- if (size_now < size_lower)
- size_now = size_lower;
- if (size_now > size_upper)
- size_now = size_upper;
-
- if (growth_step < 0) {
- growth_step = ((size_t)(size_upper - size_lower)) / 42;
- if (growth_step > size_lower && size_lower < (intptr_t)MEGABYTE)
- growth_step = size_lower;
- if (growth_step < 65536)
- growth_step = 65536;
- if ((size_t)growth_step > MAX_MAPSIZE / 64)
- growth_step = MAX_MAPSIZE / 64;
- }
- if (growth_step == 0 && shrink_threshold > 0)
- growth_step = 1;
- growth_step = ceil_powerof2(growth_step, unit);
-
- if (shrink_threshold < 0)
- shrink_threshold = growth_step + growth_step;
- shrink_threshold = ceil_powerof2(shrink_threshold, unit);
-
- //----------------------------------------------------------------------------
-
- if (!env->me_map) {
- /* save user's geo-params for future open/create */
- if (pagesize != (intptr_t)env->me_psize)
- mdbx_setup_pagesize(env, pagesize);
- env->me_dbgeo.lower = size_lower;
- env->me_dbgeo.now = size_now;
- env->me_dbgeo.upper = size_upper;
- env->me_dbgeo.grow =
- pgno2bytes(env, pv2pages(pages2pv(bytes2pgno(env, growth_step))));
- env->me_dbgeo.shrink =
- pgno2bytes(env, pv2pages(pages2pv(bytes2pgno(env, shrink_threshold))));
-
- mdbx_ensure(env, env->me_dbgeo.lower >= MIN_MAPSIZE);
- mdbx_ensure(env, env->me_dbgeo.lower / (unsigned)pagesize >= MIN_PAGENO);
- mdbx_ensure(env, env->me_dbgeo.lower % (unsigned)pagesize == 0);
- mdbx_ensure(env, env->me_dbgeo.lower % env->me_os_psize == 0);
-
- mdbx_ensure(env, env->me_dbgeo.upper <= MAX_MAPSIZE);
- mdbx_ensure(env,
- env->me_dbgeo.upper / (unsigned)pagesize <= MAX_PAGENO + 1);
- mdbx_ensure(env, env->me_dbgeo.upper % (unsigned)pagesize == 0);
- mdbx_ensure(env, env->me_dbgeo.upper % env->me_os_psize == 0);
-
- mdbx_ensure(env, env->me_dbgeo.now >= env->me_dbgeo.lower);
- mdbx_ensure(env, env->me_dbgeo.now <= env->me_dbgeo.upper);
- mdbx_ensure(env, env->me_dbgeo.now % (unsigned)pagesize == 0);
- mdbx_ensure(env, env->me_dbgeo.now % env->me_os_psize == 0);
-
- mdbx_ensure(env, env->me_dbgeo.grow % (unsigned)pagesize == 0);
- mdbx_ensure(env, env->me_dbgeo.grow % env->me_os_psize == 0);
- mdbx_ensure(env, env->me_dbgeo.shrink % (unsigned)pagesize == 0);
- mdbx_ensure(env, env->me_dbgeo.shrink % env->me_os_psize == 0);
-
- rc = MDBX_SUCCESS;
- } else {
- /* apply new params to opened environment */
- mdbx_ensure(env, pagesize == (intptr_t)env->me_psize);
- MDBX_meta meta;
- memset(&meta, 0, sizeof(meta));
- const MDBX_geo *current_geo;
- if (!inside_txn) {
- mdbx_assert(env, need_unlock);
- const MDBX_meta *head = constmeta_prefer_last(env);
-
- uint64_t timestamp = 0;
- while ("workaround for "
- "todo4recovery://erased_by_github/libmdbx/issues/269") {
- meta = *head;
- rc = meta_waittxnid(env, &meta, ×tamp);
- if (likely(rc == MDBX_SUCCESS))
- break;
- if (unlikely(rc != MDBX_RESULT_TRUE))
- goto bailout;
- }
- const txnid_t txnid = safe64_txnid_next(constmeta_txnid(env, &meta));
- if (unlikely(txnid > MAX_TXNID)) {
- rc = MDBX_TXN_FULL;
- mdbx_error("txnid overflow, raise %d", rc);
- goto bailout;
- }
- meta_set_txnid(env, &meta, txnid);
- current_geo = &meta.mm_geo;
- } else {
- current_geo = &env->me_txn->mt_geo;
- }
-
- MDBX_geo new_geo;
- new_geo.lower = bytes2pgno(env, size_lower);
- new_geo.now = bytes2pgno(env, size_now);
- new_geo.upper = bytes2pgno(env, size_upper);
- new_geo.grow_pv = pages2pv(bytes2pgno(env, growth_step));
- new_geo.shrink_pv = pages2pv(bytes2pgno(env, shrink_threshold));
- new_geo.next = current_geo->next;
-
- mdbx_ensure(env,
- pgno_align2os_bytes(env, new_geo.lower) == (size_t)size_lower);
- mdbx_ensure(env,
- pgno_align2os_bytes(env, new_geo.upper) == (size_t)size_upper);
- mdbx_ensure(env, pgno_align2os_bytes(env, new_geo.now) == (size_t)size_now);
- mdbx_ensure(env, new_geo.grow_pv == pages2pv(pv2pages(new_geo.grow_pv)));
- mdbx_ensure(env,
- new_geo.shrink_pv == pages2pv(pv2pages(new_geo.shrink_pv)));
-
- mdbx_ensure(env, (size_t)size_lower >= MIN_MAPSIZE);
- mdbx_ensure(env, new_geo.lower >= MIN_PAGENO);
- mdbx_ensure(env, (size_t)size_upper <= MAX_MAPSIZE);
- mdbx_ensure(env, new_geo.upper <= MAX_PAGENO + 1);
- mdbx_ensure(env, new_geo.now >= new_geo.next);
- mdbx_ensure(env, new_geo.upper >= new_geo.now);
- mdbx_ensure(env, new_geo.now >= new_geo.lower);
-
- if (memcmp(current_geo, &new_geo, sizeof(MDBX_geo)) != 0) {
-#if defined(_WIN32) || defined(_WIN64)
- /* Was DB shrinking disabled before and now it will be enabled? */
- if (new_geo.lower < new_geo.upper && new_geo.shrink_pv &&
- !(current_geo->lower < current_geo->upper &&
- current_geo->shrink_pv)) {
- if (!env->me_lck_mmap.lck) {
- rc = MDBX_EPERM;
- goto bailout;
- }
- int err = mdbx_rdt_lock(env);
- if (unlikely(MDBX_IS_ERROR(err))) {
- rc = err;
- goto bailout;
- }
-
- /* Check if there are any reading threads that do not use the SRWL */
- const size_t CurrentTid = GetCurrentThreadId();
- const MDBX_reader *const begin = env->me_lck_mmap.lck->mti_readers;
- const MDBX_reader *const end =
- begin + atomic_load32(&env->me_lck_mmap.lck->mti_numreaders,
- mo_AcquireRelease);
- for (const MDBX_reader *reader = begin; reader < end; ++reader) {
- if (reader->mr_pid.weak == env->me_pid && reader->mr_tid.weak &&
- reader->mr_tid.weak != CurrentTid) {
- /* At least one thread may don't use SRWL */
- rc = MDBX_EPERM;
- break;
- }
- }
-
- mdbx_rdt_unlock(env);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
-#endif
-
- if (new_geo.now != current_geo->now ||
- new_geo.upper != current_geo->upper) {
- rc = mdbx_mapresize(env, current_geo->next, new_geo.now, new_geo.upper,
- false);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- if (inside_txn) {
- env->me_txn->mt_geo = new_geo;
- env->me_txn->mt_flags |= MDBX_TXN_DIRTY;
- } else {
- meta.mm_geo = new_geo;
- rc = mdbx_sync_locked(env, env->me_flags, &meta);
- }
-
- if (likely(rc == MDBX_SUCCESS)) {
- /* store new geo to env to avoid influences */
- env->me_dbgeo.now = pgno2bytes(env, new_geo.now);
- env->me_dbgeo.lower = pgno2bytes(env, new_geo.lower);
- env->me_dbgeo.upper = pgno2bytes(env, new_geo.upper);
- env->me_dbgeo.grow = pgno2bytes(env, pv2pages(new_geo.grow_pv));
- env->me_dbgeo.shrink = pgno2bytes(env, pv2pages(new_geo.shrink_pv));
- }
- }
- }
-
-bailout:
- if (need_unlock)
- mdbx_txn_unlock(env);
- return rc;
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-__cold int mdbx_env_set_mapsize(MDBX_env *env, size_t size) {
- return __inline_mdbx_env_set_mapsize(env, size);
-}
-
-__cold int mdbx_env_set_maxdbs(MDBX_env *env, MDBX_dbi dbs) {
- return __inline_mdbx_env_set_maxdbs(env, dbs);
-}
-
-__cold int mdbx_env_get_maxdbs(const MDBX_env *env, MDBX_dbi *dbs) {
- return __inline_mdbx_env_get_maxdbs(env, dbs);
-}
-
-__cold int mdbx_env_set_maxreaders(MDBX_env *env, unsigned readers) {
- return __inline_mdbx_env_set_maxreaders(env, readers);
-}
-
-__cold int mdbx_env_get_maxreaders(const MDBX_env *env, unsigned *readers) {
- return __inline_mdbx_env_get_maxreaders(env, readers);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-__cold static int alloc_page_buf(MDBX_env *env) {
- return env->me_pbuf
- ? MDBX_SUCCESS
- : mdbx_memalign_alloc(env->me_os_psize, env->me_psize * NUM_METAS,
- &env->me_pbuf);
-}
-
-/* Further setup required for opening an MDBX environment */
-__cold static int mdbx_setup_dxb(MDBX_env *env, const int lck_rc,
- const mdbx_mode_t mode_bits) {
- MDBX_meta meta;
- int rc = MDBX_RESULT_FALSE;
- int err = mdbx_read_header(env, &meta, lck_rc, mode_bits);
- if (unlikely(err != MDBX_SUCCESS)) {
- if (lck_rc != /* lck exclusive */ MDBX_RESULT_TRUE || err != MDBX_ENODATA ||
- (env->me_flags & MDBX_RDONLY) != 0 ||
- /* recovery mode */ env->me_stuck_meta >= 0)
- return err;
-
- mdbx_debug("%s", "create new database");
- rc = /* new database */ MDBX_RESULT_TRUE;
-
- if (!env->me_dbgeo.now) {
- /* set defaults if not configured */
- err = mdbx_env_set_geometry(env, 0, -1, DEFAULT_MAPSIZE, -1, -1, -1);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
-
- err = alloc_page_buf(env);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- meta = *mdbx_init_metas(env, env->me_pbuf);
- err = mdbx_pwrite(env->me_lazy_fd, env->me_pbuf, env->me_psize * NUM_METAS,
- 0);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- err = mdbx_ftruncate(env->me_lazy_fd,
- env->me_dxb_mmap.filesize = env->me_dbgeo.now);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
-#ifndef NDEBUG /* just for checking */
- err = mdbx_read_header(env, &meta, lck_rc, mode_bits);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-#endif
- }
-
- mdbx_verbose(
- "header: root %" PRIaPGNO "/%" PRIaPGNO ", geo %" PRIaPGNO "/%" PRIaPGNO
- "-%" PRIaPGNO "/%" PRIaPGNO " +%u -%u, txn_id %" PRIaTXN ", %s",
- meta.mm_dbs[MAIN_DBI].md_root, meta.mm_dbs[FREE_DBI].md_root,
- meta.mm_geo.lower, meta.mm_geo.next, meta.mm_geo.now, meta.mm_geo.upper,
- pv2pages(meta.mm_geo.grow_pv), pv2pages(meta.mm_geo.shrink_pv),
- unaligned_peek_u64(4, meta.mm_txnid_a), mdbx_durable_str(&meta));
-
- if (env->me_psize != meta.mm_psize)
- mdbx_setup_pagesize(env, meta.mm_psize);
- const size_t used_bytes = pgno2bytes(env, meta.mm_geo.next);
- const size_t used_aligned2os_bytes =
- ceil_powerof2(used_bytes, env->me_os_psize);
- if ((env->me_flags & MDBX_RDONLY) /* readonly */
- || lck_rc != MDBX_RESULT_TRUE /* not exclusive */
- || /* recovery mode */ env->me_stuck_meta >= 0) {
- /* use present params from db */
- const size_t pagesize = meta.mm_psize;
- err = mdbx_env_set_geometry(
- env, meta.mm_geo.lower * pagesize, meta.mm_geo.now * pagesize,
- meta.mm_geo.upper * pagesize, pv2pages(meta.mm_geo.grow_pv) * pagesize,
- pv2pages(meta.mm_geo.shrink_pv) * pagesize, meta.mm_psize);
- if (unlikely(err != MDBX_SUCCESS)) {
- mdbx_error("%s: err %d", "could not apply preconfigured geometry from db",
- err);
- return (err == MDBX_EINVAL) ? MDBX_INCOMPATIBLE : err;
- }
- } else if (env->me_dbgeo.now) {
- /* silently growth to last used page */
- if (env->me_dbgeo.now < used_aligned2os_bytes)
- env->me_dbgeo.now = used_aligned2os_bytes;
- if (env->me_dbgeo.upper < used_aligned2os_bytes)
- env->me_dbgeo.upper = used_aligned2os_bytes;
-
- /* apply preconfigured params, but only if substantial changes:
- * - upper or lower limit changes
- * - shrink threshold or growth step
- * But ignore change just a 'now/current' size. */
- if (bytes_align2os_bytes(env, env->me_dbgeo.upper) !=
- pgno2bytes(env, meta.mm_geo.upper) ||
- bytes_align2os_bytes(env, env->me_dbgeo.lower) !=
- pgno2bytes(env, meta.mm_geo.lower) ||
- bytes_align2os_bytes(env, env->me_dbgeo.shrink) !=
- pgno2bytes(env, pv2pages(meta.mm_geo.shrink_pv)) ||
- bytes_align2os_bytes(env, env->me_dbgeo.grow) !=
- pgno2bytes(env, pv2pages(meta.mm_geo.grow_pv))) {
-
- if (env->me_dbgeo.shrink && env->me_dbgeo.now > used_bytes)
- /* pre-shrink if enabled */
- env->me_dbgeo.now = used_bytes + env->me_dbgeo.shrink -
- used_bytes % env->me_dbgeo.shrink;
-
- err = mdbx_env_set_geometry(env, env->me_dbgeo.lower, env->me_dbgeo.now,
- env->me_dbgeo.upper, env->me_dbgeo.grow,
- env->me_dbgeo.shrink, meta.mm_psize);
- if (unlikely(err != MDBX_SUCCESS)) {
- mdbx_error("%s: err %d", "could not apply preconfigured db-geometry",
- err);
- return (err == MDBX_EINVAL) ? MDBX_INCOMPATIBLE : err;
- }
-
- /* update meta fields */
- meta.mm_geo.now = bytes2pgno(env, env->me_dbgeo.now);
- meta.mm_geo.lower = bytes2pgno(env, env->me_dbgeo.lower);
- meta.mm_geo.upper = bytes2pgno(env, env->me_dbgeo.upper);
- meta.mm_geo.grow_pv = pages2pv(bytes2pgno(env, env->me_dbgeo.grow));
- meta.mm_geo.shrink_pv = pages2pv(bytes2pgno(env, env->me_dbgeo.shrink));
-
- mdbx_verbose("amended: root %" PRIaPGNO "/%" PRIaPGNO ", geo %" PRIaPGNO
- "/%" PRIaPGNO "-%" PRIaPGNO "/%" PRIaPGNO
- " +%u -%u, txn_id %" PRIaTXN ", %s",
- meta.mm_dbs[MAIN_DBI].md_root, meta.mm_dbs[FREE_DBI].md_root,
- meta.mm_geo.lower, meta.mm_geo.next, meta.mm_geo.now,
- meta.mm_geo.upper, pv2pages(meta.mm_geo.grow_pv),
- pv2pages(meta.mm_geo.shrink_pv),
- unaligned_peek_u64(4, meta.mm_txnid_a),
- mdbx_durable_str(&meta));
- } else {
- /* fetch back 'now/current' size, since it was ignored during comparison
- * and may differ. */
- env->me_dbgeo.now = pgno_align2os_bytes(env, meta.mm_geo.now);
- }
- mdbx_ensure(env, meta.mm_geo.now >= meta.mm_geo.next);
- } else {
- /* geo-params are not pre-configured by user,
- * get current values from the meta. */
- env->me_dbgeo.now = pgno2bytes(env, meta.mm_geo.now);
- env->me_dbgeo.lower = pgno2bytes(env, meta.mm_geo.lower);
- env->me_dbgeo.upper = pgno2bytes(env, meta.mm_geo.upper);
- env->me_dbgeo.grow = pgno2bytes(env, pv2pages(meta.mm_geo.grow_pv));
- env->me_dbgeo.shrink = pgno2bytes(env, pv2pages(meta.mm_geo.shrink_pv));
- }
-
- mdbx_ensure(env,
- pgno_align2os_bytes(env, meta.mm_geo.now) == env->me_dbgeo.now);
- mdbx_ensure(env, env->me_dbgeo.now >= used_bytes);
- const uint64_t filesize_before = env->me_dxb_mmap.filesize;
- if (unlikely(filesize_before != env->me_dbgeo.now)) {
- if (lck_rc != /* lck exclusive */ MDBX_RESULT_TRUE) {
- mdbx_verbose("filesize mismatch (expect %" PRIuPTR "b/%" PRIaPGNO
- "p, have %" PRIu64 "b/%" PRIaPGNO "p), "
- "assume other process working",
- env->me_dbgeo.now, bytes2pgno(env, env->me_dbgeo.now),
- filesize_before, bytes2pgno(env, (size_t)filesize_before));
- } else {
- mdbx_warning("filesize mismatch (expect %" PRIuSIZE "b/%" PRIaPGNO
- "p, have %" PRIu64 "b/%" PRIaPGNO "p)",
- env->me_dbgeo.now, bytes2pgno(env, env->me_dbgeo.now),
- filesize_before, bytes2pgno(env, (size_t)filesize_before));
- if (filesize_before < used_bytes) {
- mdbx_error("last-page beyond end-of-file (last %" PRIaPGNO
- ", have %" PRIaPGNO ")",
- meta.mm_geo.next, bytes2pgno(env, (size_t)filesize_before));
- return MDBX_CORRUPTED;
- }
-
- if (env->me_flags & MDBX_RDONLY) {
- if (filesize_before & (env->me_os_psize - 1)) {
- mdbx_error("%s", "filesize should be rounded-up to system page");
- return MDBX_WANNA_RECOVERY;
- }
- mdbx_warning("%s", "ignore filesize mismatch in readonly-mode");
- } else {
- mdbx_verbose("will resize datafile to %" PRIuSIZE " bytes, %" PRIaPGNO
- " pages",
- env->me_dbgeo.now, bytes2pgno(env, env->me_dbgeo.now));
- }
- }
- }
-
- mdbx_verbose("current boot-id %" PRIx64 "-%" PRIx64 " (%savailable)",
- bootid.x, bootid.y, (bootid.x | bootid.y) ? "" : "not-");
-
-#if MDBX_ENABLE_MADVISE
- /* calculate readahead hint before mmap with zero redundant pages */
- const bool readahead =
- !(env->me_flags & MDBX_NORDAHEAD) &&
- mdbx_is_readahead_reasonable(used_bytes, 0) == MDBX_RESULT_TRUE;
-#endif /* MDBX_ENABLE_MADVISE */
-
- err = mdbx_mmap(env->me_flags, &env->me_dxb_mmap, env->me_dbgeo.now,
- env->me_dbgeo.upper, lck_rc ? MMAP_OPTION_TRUNCATE : 0);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
-#if MDBX_ENABLE_MADVISE
-#if defined(MADV_DONTDUMP)
- err = madvise(env->me_map, env->me_dxb_mmap.limit, MADV_DONTDUMP)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#endif /* MADV_DONTDUMP */
-#if defined(MADV_DODUMP)
- if (mdbx_runtime_flags & MDBX_DBG_DUMP) {
- const size_t meta_length_aligned2os = pgno_align2os_bytes(env, NUM_METAS);
- err = madvise(env->me_map, meta_length_aligned2os, MADV_DODUMP)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
- }
-#endif /* MADV_DODUMP */
-#endif /* MDBX_ENABLE_MADVISE */
-
-#ifdef MDBX_USE_VALGRIND
- env->me_valgrind_handle =
- VALGRIND_CREATE_BLOCK(env->me_map, env->me_dxb_mmap.limit, "mdbx");
-#endif /* MDBX_USE_VALGRIND */
-
- mdbx_assert(env, used_bytes >= pgno2bytes(env, NUM_METAS) &&
- used_bytes <= env->me_dxb_mmap.limit);
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- if (env->me_dxb_mmap.filesize > used_bytes &&
- env->me_dxb_mmap.filesize < env->me_dxb_mmap.limit) {
- VALGRIND_MAKE_MEM_NOACCESS(env->me_map + used_bytes,
- env->me_dxb_mmap.filesize - used_bytes);
- MDBX_ASAN_POISON_MEMORY_REGION(env->me_map + used_bytes,
- env->me_dxb_mmap.filesize - used_bytes);
- }
- env->me_poison_edge =
- bytes2pgno(env, (env->me_dxb_mmap.filesize < env->me_dxb_mmap.limit)
- ? env->me_dxb_mmap.filesize
- : env->me_dxb_mmap.limit);
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
- //-------------------------------- validate/rollback head & steady meta-pages
- if (unlikely(env->me_stuck_meta >= 0)) {
- /* recovery mode */
- MDBX_meta clone;
- MDBX_meta const *const target = METAPAGE(env, env->me_stuck_meta);
- err = mdbx_validate_meta_copy(env, target, &clone);
- if (unlikely(err != MDBX_SUCCESS)) {
- mdbx_error("target meta[%u] is corrupted",
- bytes2pgno(env, (uint8_t *)data_page(target) - env->me_map));
- return MDBX_CORRUPTED;
- }
- } else /* not recovery mode */
- while (1) {
- const unsigned meta_clash_mask = meta_eq_mask(env);
- if (unlikely(meta_clash_mask)) {
- mdbx_error("meta-pages are clashed: mask 0x%d", meta_clash_mask);
- return MDBX_CORRUPTED;
- }
-
- if (lck_rc != /* lck exclusive */ MDBX_RESULT_TRUE) {
- /* non-exclusive mode,
- * meta-pages should be validated by a first process opened the DB */
- volatile const MDBX_meta *const head = meta_prefer_last(env);
- volatile const MDBX_meta *const steady = meta_prefer_steady(env);
- const txnid_t head_txnid = meta_txnid(env, head);
- const txnid_t steady_txnid = meta_txnid(env, steady);
- if (head_txnid == steady_txnid)
- break;
-
- if (!env->me_lck_mmap.lck) {
- /* LY: without-lck (read-only) mode, so it is impossible that other
- * process made weak checkpoint. */
- mdbx_error("%s", "without-lck, unable recovery/rollback");
- return MDBX_WANNA_RECOVERY;
- }
-
- /* LY: assume just have a collision with other running process,
- * or someone make a weak checkpoint */
- mdbx_verbose("%s", "assume collision or online weak checkpoint");
- break;
- }
- mdbx_assert(env, lck_rc == MDBX_RESULT_TRUE);
- /* exclusive mode */
-
- MDBX_meta clone;
- const MDBX_meta *const steady = constmeta_prefer_steady(env);
- const MDBX_meta *const head = constmeta_prefer_last(env);
- const txnid_t steady_txnid = meta_txnid(env, steady);
- if (META_IS_STEADY(steady)) {
- err = mdbx_validate_meta_copy(env, steady, &clone);
- if (unlikely(err != MDBX_SUCCESS)) {
- mdbx_error("meta[%u] with %s txnid %" PRIaTXN
- " is corrupted, %s needed",
- bytes2pgno(env, (uint8_t *)steady - env->me_map), "steady",
- steady_txnid, "manual recovery");
- return MDBX_CORRUPTED;
- }
- if (steady == head)
- break;
- }
-
- const pgno_t pgno = bytes2pgno(env, (uint8_t *)head - env->me_map);
- const txnid_t head_txnid = meta_txnid(env, head);
- const bool head_valid =
- mdbx_validate_meta_copy(env, head, &clone) == MDBX_SUCCESS;
- mdbx_assert(env, !META_IS_STEADY(steady) || head_txnid != steady_txnid);
- if (unlikely(!head_valid)) {
- if (unlikely(!META_IS_STEADY(steady))) {
- mdbx_error("%s for open or automatic rollback, %s",
- "there are no suitable meta-pages",
- "manual recovery is required");
- return MDBX_CORRUPTED;
- }
- mdbx_warning("meta[%u] with last txnid %" PRIaTXN
- " is corrupted, rollback needed",
- pgno, head_txnid);
- goto purge_meta_head;
- }
-
- if (meta_bootid_match(head)) {
- if (env->me_flags & MDBX_RDONLY) {
- mdbx_error("%s, but boot-id(%016" PRIx64 "-%016" PRIx64 ") is MATCH: "
- "rollback NOT needed, steady-sync NEEDED%s",
- "opening after an unclean shutdown", bootid.x, bootid.y,
- ", but unable in read-only mode");
- return MDBX_WANNA_RECOVERY;
- }
- mdbx_warning("%s, but boot-id(%016" PRIx64 "-%016" PRIx64 ") is MATCH: "
- "rollback NOT needed, steady-sync NEEDED%s",
- "opening after an unclean shutdown", bootid.x, bootid.y,
- "");
- meta = clone;
- atomic_store32(&env->me_lck->mti_unsynced_pages, meta.mm_geo.next,
- mo_Relaxed);
- break;
- }
- if (unlikely(!META_IS_STEADY(steady))) {
- mdbx_error("%s, but %s for automatic rollback: %s",
- "opening after an unclean shutdown",
- "there are no suitable meta-pages",
- "manual recovery is required");
- return MDBX_CORRUPTED;
- }
- if (env->me_flags & MDBX_RDONLY) {
- mdbx_error("%s and rollback needed: (from head %" PRIaTXN
- " to steady %" PRIaTXN ")%s",
- "opening after an unclean shutdown", head_txnid,
- steady_txnid, ", but unable in read-only mode");
- return MDBX_WANNA_RECOVERY;
- }
-
- purge_meta_head:
- mdbx_notice("%s and doing automatic rollback: "
- "purge%s meta[%u] with%s txnid %" PRIaTXN,
- "opening after an unclean shutdown",
- head_valid ? "" : " invalid", pgno, head_valid ? " weak" : "",
- head_txnid);
- mdbx_ensure(env, META_IS_STEADY(steady));
- err = mdbx_override_meta(env, pgno, 0, head_valid ? head : steady);
- if (err) {
- mdbx_error("rollback: overwrite meta[%u] with txnid %" PRIaTXN
- ", error %d",
- pgno, head_txnid, err);
- return err;
- }
- mdbx_ensure(env, 0 == meta_txnid(env, head));
- mdbx_ensure(env, 0 == meta_eq_mask(env));
- }
-
- if (lck_rc == /* lck exclusive */ MDBX_RESULT_TRUE) {
- //-------------------------------------------------- shrink DB & update geo
- const MDBX_meta *head = constmeta_prefer_last(env);
- /* re-check size after mmap */
- if ((env->me_dxb_mmap.current & (env->me_os_psize - 1)) != 0 ||
- env->me_dxb_mmap.current < used_bytes) {
- mdbx_error("unacceptable/unexpected datafile size %" PRIuPTR,
- env->me_dxb_mmap.current);
- return MDBX_PROBLEM;
- }
- if (env->me_dxb_mmap.current != env->me_dbgeo.now) {
- meta.mm_geo.now = bytes2pgno(env, env->me_dxb_mmap.current);
- mdbx_notice("need update meta-geo to filesize %" PRIuPTR
- " bytes, %" PRIaPGNO " pages",
- env->me_dxb_mmap.current, meta.mm_geo.now);
- }
-
- if (memcmp(&meta.mm_geo, &head->mm_geo, sizeof(meta.mm_geo))) {
- if ((env->me_flags & MDBX_RDONLY) != 0 ||
- /* recovery mode */ env->me_stuck_meta >= 0) {
- mdbx_warning(
- "skipped update meta.geo in %s mode: from l%" PRIaPGNO
- "-n%" PRIaPGNO "-u%" PRIaPGNO "/s%u-g%u, to l%" PRIaPGNO
- "-n%" PRIaPGNO "-u%" PRIaPGNO "/s%u-g%u",
- (env->me_stuck_meta < 0) ? "read-only" : "recovery",
- head->mm_geo.lower, head->mm_geo.now, head->mm_geo.upper,
- pv2pages(head->mm_geo.shrink_pv), pv2pages(head->mm_geo.grow_pv),
- meta.mm_geo.lower, meta.mm_geo.now, meta.mm_geo.upper,
- pv2pages(meta.mm_geo.shrink_pv), pv2pages(meta.mm_geo.grow_pv));
- } else {
- const txnid_t txnid = constmeta_txnid(env, head);
- const txnid_t next_txnid = safe64_txnid_next(txnid);
- if (unlikely(txnid > MAX_TXNID)) {
- mdbx_error("txnid overflow, raise %d", MDBX_TXN_FULL);
- return MDBX_TXN_FULL;
- }
- mdbx_notice("updating meta.geo: "
- "from l%" PRIaPGNO "-n%" PRIaPGNO "-u%" PRIaPGNO
- "/s%u-g%u (txn#%" PRIaTXN "), "
- "to l%" PRIaPGNO "-n%" PRIaPGNO "-u%" PRIaPGNO
- "/s%u-g%u (txn#%" PRIaTXN ")",
- head->mm_geo.lower, head->mm_geo.now, head->mm_geo.upper,
- pv2pages(head->mm_geo.shrink_pv),
- pv2pages(head->mm_geo.grow_pv), txnid, meta.mm_geo.lower,
- meta.mm_geo.now, meta.mm_geo.upper,
- pv2pages(meta.mm_geo.shrink_pv),
- pv2pages(meta.mm_geo.grow_pv), next_txnid);
-
- mdbx_ensure(env, meta_eq(env, &meta, head));
- meta_set_txnid(env, &meta, next_txnid);
- err = mdbx_sync_locked(env, env->me_flags | MDBX_SHRINK_ALLOWED, &meta);
- if (err) {
- mdbx_error("error %d, while updating meta.geo: "
- "from l%" PRIaPGNO "-n%" PRIaPGNO "-u%" PRIaPGNO
- "/s%u-g%u (txn#%" PRIaTXN "), "
- "to l%" PRIaPGNO "-n%" PRIaPGNO "-u%" PRIaPGNO
- "/s%u-g%u (txn#%" PRIaTXN ")",
- err, head->mm_geo.lower, head->mm_geo.now,
- head->mm_geo.upper, pv2pages(head->mm_geo.shrink_pv),
- pv2pages(head->mm_geo.grow_pv), txnid, meta.mm_geo.lower,
- meta.mm_geo.now, meta.mm_geo.upper,
- pv2pages(meta.mm_geo.shrink_pv),
- pv2pages(meta.mm_geo.grow_pv), next_txnid);
- return err;
- }
- }
- }
-
- atomic_store32(&env->me_lck->mti_discarded_tail,
- bytes2pgno(env, used_aligned2os_bytes), mo_Relaxed);
-
- if ((env->me_flags & MDBX_RDONLY) == 0 && env->me_stuck_meta < 0 &&
- (mdbx_runtime_flags & MDBX_DBG_DONT_UPGRADE) == 0) {
- for (int n = 0; n < NUM_METAS; ++n) {
- MDBX_meta *const pmeta = METAPAGE(env, n);
- if (unlikely(unaligned_peek_u64(4, &pmeta->mm_magic_and_version) !=
- MDBX_DATA_MAGIC)) {
- const txnid_t txnid = meta_txnid(env, pmeta);
- mdbx_notice("%s %s"
- "meta[%u], txnid %" PRIaTXN,
- "updating db-format signature for",
- META_IS_STEADY(pmeta) ? "stead-" : "weak-", n, txnid);
- err = mdbx_override_meta(env, n, txnid, pmeta);
- if (unlikely(err != MDBX_SUCCESS) &&
- /* Just ignore the MDBX_PROBLEM error, since here it is
- * returned only in case of the attempt to upgrade an obsolete
- * meta-page that is invalid for current state of a DB,
- * e.g. after shrinking DB file */
- err != MDBX_PROBLEM) {
- mdbx_error("%s meta[%u], txnid %" PRIaTXN ", error %d",
- "updating db-format signature for", n, txnid, err);
- return err;
- }
- }
- }
- }
- } /* lck exclusive, lck_rc == MDBX_RESULT_TRUE */
-
- //---------------------------------------------------- setup madvise/readahead
-#if MDBX_ENABLE_MADVISE
- if (used_aligned2os_bytes < env->me_dxb_mmap.current) {
-#if defined(MADV_REMOVE)
- if (lck_rc && (env->me_flags & MDBX_WRITEMAP) != 0 &&
- /* not recovery mode */ env->me_stuck_meta < 0) {
- mdbx_notice("open-MADV_%s %u..%u", "REMOVE (deallocate file space)",
- env->me_lck->mti_discarded_tail.weak,
- bytes2pgno(env, env->me_dxb_mmap.current));
- err =
- madvise(env->me_map + used_aligned2os_bytes,
- env->me_dxb_mmap.current - used_aligned2os_bytes, MADV_REMOVE)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
- }
-#endif /* MADV_REMOVE */
-#if defined(MADV_DONTNEED)
- mdbx_notice("open-MADV_%s %u..%u", "DONTNEED",
- env->me_lck->mti_discarded_tail.weak,
- bytes2pgno(env, env->me_dxb_mmap.current));
- err =
- madvise(env->me_map + used_aligned2os_bytes,
- env->me_dxb_mmap.current - used_aligned2os_bytes, MADV_DONTNEED)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(POSIX_MADV_DONTNEED)
- err = ignore_enosys(posix_madvise(
- env->me_map + used_aligned2os_bytes,
- env->me_dxb_mmap.current - used_aligned2os_bytes, POSIX_MADV_DONTNEED));
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#elif defined(POSIX_FADV_DONTNEED)
- err = ignore_enosys(posix_fadvise(
- env->me_lazy_fd, used_aligned2os_bytes,
- env->me_dxb_mmap.current - used_aligned2os_bytes, POSIX_FADV_DONTNEED));
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
-#endif /* MADV_DONTNEED */
- }
-
- err = mdbx_set_readahead(env, bytes2pgno(env, used_bytes), readahead, true);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-#endif /* MDBX_ENABLE_MADVISE */
-
- return rc;
-}
-
-/******************************************************************************/
-
-/* Open and/or initialize the lock region for the environment. */
-__cold static int mdbx_setup_lck(MDBX_env *env, char *lck_pathname,
- mdbx_mode_t mode) {
- mdbx_assert(env, env->me_lazy_fd != INVALID_HANDLE_VALUE);
- mdbx_assert(env, env->me_lfd == INVALID_HANDLE_VALUE);
-
- int err = mdbx_openfile(MDBX_OPEN_LCK, env, lck_pathname, &env->me_lfd, mode);
- if (err != MDBX_SUCCESS) {
- if (!(err == MDBX_ENOFILE && (env->me_flags & MDBX_EXCLUSIVE)) &&
- !((err == MDBX_EROFS || err == MDBX_EACCESS || err == MDBX_EPERM) &&
- (env->me_flags & MDBX_RDONLY)))
- return err;
-
- /* ensure the file system is read-only */
- err = mdbx_check_fs_rdonly(env->me_lazy_fd, lck_pathname, err);
- if (err != MDBX_SUCCESS &&
- /* ignore ERROR_NOT_SUPPORTED for exclusive mode */
- !(err == MDBX_ENOSYS && (env->me_flags & MDBX_EXCLUSIVE)))
- return err;
-
- /* LY: without-lck mode (e.g. exclusive or on read-only filesystem) */
- /* beginning of a locked section ---------------------------------------- */
- lcklist_lock();
- mdbx_assert(env, env->me_lcklist_next == nullptr);
- env->me_lfd = INVALID_HANDLE_VALUE;
- const int rc = mdbx_lck_seize(env);
- if (MDBX_IS_ERROR(rc)) {
- /* Calling lcklist_detach_locked() is required to restore POSIX-filelock
- * and this job will be done by mdbx_env_close0(). */
- lcklist_unlock();
- return rc;
- }
- /* insert into inprocess lck-list */
- env->me_lcklist_next = inprocess_lcklist_head;
- inprocess_lcklist_head = env;
- lcklist_unlock();
- /* end of a locked section ---------------------------------------------- */
-
- env->me_lck = lckless_stub(env);
- env->me_maxreaders = UINT_MAX;
- mdbx_debug("lck-setup:%s%s%s", " lck-less",
- (env->me_flags & MDBX_RDONLY) ? " readonly" : "",
- (rc == MDBX_RESULT_TRUE) ? " exclusive" : " cooperative");
- return rc;
- }
-
- /* beginning of a locked section ------------------------------------------ */
- lcklist_lock();
- mdbx_assert(env, env->me_lcklist_next == nullptr);
-
- /* Try to get exclusive lock. If we succeed, then
- * nobody is using the lock region and we should initialize it. */
- err = mdbx_lck_seize(env);
- if (MDBX_IS_ERROR(err)) {
- bailout:
- /* Calling lcklist_detach_locked() is required to restore POSIX-filelock
- * and this job will be done by mdbx_env_close0(). */
- lcklist_unlock();
- return err;
- }
-
- MDBX_env *inprocess_neighbor = nullptr;
- if (err == MDBX_RESULT_TRUE) {
- err = uniq_check(&env->me_lck_mmap, &inprocess_neighbor);
- if (MDBX_IS_ERROR(err))
- goto bailout;
- if (inprocess_neighbor &&
- ((mdbx_runtime_flags & MDBX_DBG_LEGACY_MULTIOPEN) == 0 ||
- (inprocess_neighbor->me_flags & MDBX_EXCLUSIVE) != 0)) {
- err = MDBX_BUSY;
- goto bailout;
- }
- }
- const int lck_seize_rc = err;
-
- mdbx_debug("lck-setup:%s%s%s", " with-lck",
- (env->me_flags & MDBX_RDONLY) ? " readonly" : "",
- (lck_seize_rc == MDBX_RESULT_TRUE) ? " exclusive"
- : " cooperative");
-
- uint64_t size = 0;
- err = mdbx_filesize(env->me_lfd, &size);
- if (unlikely(err != MDBX_SUCCESS))
- goto bailout;
-
- if (lck_seize_rc == MDBX_RESULT_TRUE) {
- size = ceil_powerof2(env->me_maxreaders * sizeof(MDBX_reader) +
- sizeof(MDBX_lockinfo),
- env->me_os_psize);
- mdbx_jitter4testing(false);
- } else {
- if (env->me_flags & MDBX_EXCLUSIVE) {
- err = MDBX_BUSY;
- goto bailout;
- }
- if (size > INT_MAX || (size & (env->me_os_psize - 1)) != 0 ||
- size < env->me_os_psize) {
- mdbx_error("lck-file has invalid size %" PRIu64 " bytes", size);
- err = MDBX_PROBLEM;
- goto bailout;
- }
- }
-
- const size_t maxreaders =
- ((size_t)size - sizeof(MDBX_lockinfo)) / sizeof(MDBX_reader);
- if (maxreaders < 4) {
- mdbx_error("lck-size too small (up to %" PRIuPTR " readers)", maxreaders);
- err = MDBX_PROBLEM;
- goto bailout;
- }
- env->me_maxreaders = (maxreaders <= MDBX_READERS_LIMIT)
- ? (unsigned)maxreaders
- : (unsigned)MDBX_READERS_LIMIT;
-
- err = mdbx_mmap((env->me_flags & MDBX_EXCLUSIVE) | MDBX_WRITEMAP,
- &env->me_lck_mmap, (size_t)size, (size_t)size,
- lck_seize_rc ? MMAP_OPTION_TRUNCATE | MMAP_OPTION_SEMAPHORE
- : MMAP_OPTION_SEMAPHORE);
- if (unlikely(err != MDBX_SUCCESS))
- goto bailout;
-
-#if MDBX_ENABLE_MADVISE
-#ifdef MADV_DODUMP
- err = madvise(env->me_lck_mmap.lck, size, MADV_DODUMP) ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- goto bailout;
-#endif /* MADV_DODUMP */
-
-#ifdef MADV_WILLNEED
- err = madvise(env->me_lck_mmap.lck, size, MADV_WILLNEED)
- ? ignore_enosys(errno)
- : MDBX_SUCCESS;
- if (unlikely(MDBX_IS_ERROR(err)))
- goto bailout;
-#elif defined(POSIX_MADV_WILLNEED)
- err = ignore_enosys(
- posix_madvise(env->me_lck_mmap.lck, size, POSIX_MADV_WILLNEED));
- if (unlikely(MDBX_IS_ERROR(err)))
- goto bailout;
-#endif /* MADV_WILLNEED */
-#endif /* MDBX_ENABLE_MADVISE */
-
- struct MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (lck_seize_rc == MDBX_RESULT_TRUE) {
- /* LY: exclusive mode, check and reset lck content */
- memset(lck, 0, (size_t)size);
- mdbx_jitter4testing(false);
- lck->mti_magic_and_version = MDBX_LOCK_MAGIC;
- lck->mti_os_and_format = MDBX_LOCK_FORMAT;
-#if MDBX_ENABLE_PGOP_STAT
- lck->mti_pgop_stat.wops.weak = 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- err = mdbx_msync(&env->me_lck_mmap, 0, (size_t)size, MDBX_SYNC_NONE);
- if (unlikely(err != MDBX_SUCCESS)) {
- mdbx_error("initial-%s for lck-file failed", "msync");
- goto bailout;
- }
- err = mdbx_fsync(env->me_lck_mmap.fd, MDBX_SYNC_SIZE);
- if (unlikely(err != MDBX_SUCCESS)) {
- mdbx_error("initial-%s for lck-file failed", "fsync");
- goto bailout;
- }
- } else {
- if (lck->mti_magic_and_version != MDBX_LOCK_MAGIC) {
- const bool invalid = (lck->mti_magic_and_version >> 8) != MDBX_MAGIC;
- mdbx_error(
- "lock region has %s",
- invalid
- ? "invalid magic"
- : "incompatible version (only applications with nearly or the "
- "same versions of libmdbx can share the same database)");
- err = invalid ? MDBX_INVALID : MDBX_VERSION_MISMATCH;
- goto bailout;
- }
- if (lck->mti_os_and_format != MDBX_LOCK_FORMAT) {
- mdbx_error("lock region has os/format signature 0x%" PRIx32
- ", expected 0x%" PRIx32,
- lck->mti_os_and_format, MDBX_LOCK_FORMAT);
- err = MDBX_VERSION_MISMATCH;
- goto bailout;
- }
- }
-
- err = mdbx_lck_init(env, inprocess_neighbor, lck_seize_rc);
- if (MDBX_IS_ERROR(err))
- goto bailout;
-
- mdbx_ensure(env, env->me_lcklist_next == nullptr);
- /* insert into inprocess lck-list */
- env->me_lcklist_next = inprocess_lcklist_head;
- inprocess_lcklist_head = env;
- lcklist_unlock();
- /* end of a locked section ------------------------------------------------ */
-
- mdbx_assert(env, !MDBX_IS_ERROR(lck_seize_rc));
- env->me_lck = lck;
- return lck_seize_rc;
-}
-
-__cold int mdbx_is_readahead_reasonable(size_t volume, intptr_t redundancy) {
- if (volume <= 1024 * 1024 * 4ul)
- return MDBX_RESULT_TRUE;
-
- intptr_t pagesize, total_ram_pages;
- int err = mdbx_get_sysraminfo(&pagesize, &total_ram_pages, nullptr);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- const int log2page = log2n_powerof2(pagesize);
- const intptr_t volume_pages = (volume + pagesize - 1) >> log2page;
- const intptr_t redundancy_pages =
- (redundancy < 0) ? -(intptr_t)((-redundancy + pagesize - 1) >> log2page)
- : (intptr_t)(redundancy + pagesize - 1) >> log2page;
- if (volume_pages >= total_ram_pages ||
- volume_pages + redundancy_pages >= total_ram_pages)
- return MDBX_RESULT_FALSE;
-
- intptr_t avail_ram_pages;
- err = mdbx_get_sysraminfo(nullptr, nullptr, &avail_ram_pages);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- return (volume_pages + redundancy_pages >= avail_ram_pages)
- ? MDBX_RESULT_FALSE
- : MDBX_RESULT_TRUE;
-}
-
-/* Merge sync flags */
-static uint32_t merge_sync_flags(const uint32_t a, const uint32_t b) {
- uint32_t r = a | b;
-
- /* avoid false MDBX_UTTERLY_NOSYNC */
- if (F_ISSET(r, MDBX_UTTERLY_NOSYNC) && !F_ISSET(a, MDBX_UTTERLY_NOSYNC) &&
- !F_ISSET(b, MDBX_UTTERLY_NOSYNC))
- r = (r - MDBX_UTTERLY_NOSYNC) | MDBX_SAFE_NOSYNC;
-
- /* convert MDBX_DEPRECATED_MAPASYNC to MDBX_SAFE_NOSYNC */
- if ((r & (MDBX_WRITEMAP | MDBX_DEPRECATED_MAPASYNC)) ==
- (MDBX_WRITEMAP | MDBX_DEPRECATED_MAPASYNC) &&
- !F_ISSET(r, MDBX_UTTERLY_NOSYNC))
- r = (r - MDBX_DEPRECATED_MAPASYNC) | MDBX_SAFE_NOSYNC;
-
- /* force MDBX_NOMETASYNC if MDBX_SAFE_NOSYNC enabled */
- if (r & MDBX_SAFE_NOSYNC)
- r |= MDBX_NOMETASYNC;
-
- assert(!(F_ISSET(r, MDBX_UTTERLY_NOSYNC) &&
- !F_ISSET(a, MDBX_UTTERLY_NOSYNC) &&
- !F_ISSET(b, MDBX_UTTERLY_NOSYNC)));
- return r;
-}
-
-__cold static int __must_check_result mdbx_override_meta(
- MDBX_env *env, unsigned target, txnid_t txnid, const MDBX_meta *shape) {
- int rc = alloc_page_buf(env);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- MDBX_page *const page = env->me_pbuf;
- mdbx_meta_model(env, page, target);
- MDBX_meta *const model = page_meta(page);
- meta_set_txnid(env, model, txnid);
- mdbx_assert(env, meta_checktxnid(env, model, true));
- if (shape) {
- if (txnid && unlikely(!meta_checktxnid(env, shape, false))) {
- mdbx_error("bailout overriding meta-%u since model failed "
- "freedb/maindb %s-check for txnid #%" PRIaTXN,
- target, "pre", constmeta_txnid(env, shape));
- return MDBX_PROBLEM;
- }
- if (mdbx_runtime_flags & MDBX_DBG_DONT_UPGRADE)
- memcpy(&model->mm_magic_and_version, &shape->mm_magic_and_version,
- sizeof(model->mm_magic_and_version));
- model->mm_extra_flags = shape->mm_extra_flags;
- model->mm_validator_id = shape->mm_validator_id;
- model->mm_extra_pagehdr = shape->mm_extra_pagehdr;
- memcpy(&model->mm_geo, &shape->mm_geo, sizeof(model->mm_geo));
- memcpy(&model->mm_dbs, &shape->mm_dbs, sizeof(model->mm_dbs));
- memcpy(&model->mm_canary, &shape->mm_canary, sizeof(model->mm_canary));
- memcpy(&model->mm_pages_retired, &shape->mm_pages_retired,
- sizeof(model->mm_pages_retired));
- if (txnid) {
- if ((!model->mm_dbs[FREE_DBI].md_mod_txnid &&
- model->mm_dbs[FREE_DBI].md_root != P_INVALID) ||
- (!model->mm_dbs[MAIN_DBI].md_mod_txnid &&
- model->mm_dbs[MAIN_DBI].md_root != P_INVALID))
- memcpy(&model->mm_magic_and_version, &shape->mm_magic_and_version,
- sizeof(model->mm_magic_and_version));
- if (unlikely(!meta_checktxnid(env, model, false))) {
- mdbx_error("bailout overriding meta-%u since model failed "
- "freedb/maindb %s-check for txnid #%" PRIaTXN,
- target, "post", txnid);
- return MDBX_PROBLEM;
- }
- }
- }
- unaligned_poke_u64(4, model->mm_datasync_sign, meta_sign(model));
- rc = mdbx_validate_meta(env, model, page, target, nullptr);
- if (unlikely(MDBX_IS_ERROR(rc)))
- return MDBX_PROBLEM;
-
- if (shape && memcmp(model, shape, sizeof(MDBX_meta)) == 0)
- return MDBX_SUCCESS;
-
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.wops.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- if (env->me_flags & MDBX_WRITEMAP) {
- rc = mdbx_msync(&env->me_dxb_mmap, 0,
- pgno_align2os_bytes(env, model->mm_geo.next),
- MDBX_SYNC_DATA | MDBX_SYNC_IODQ);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- /* mdbx_override_meta() called only while current process have exclusive
- * lock of a DB file. So meta-page could be updated directly without
- * clearing consistency flag by mdbx_meta_update_begin() */
- memcpy(pgno2page(env, target), page, env->me_psize);
- mdbx_flush_incoherent_cpu_writeback();
- rc = mdbx_msync(&env->me_dxb_mmap, 0, pgno_align2os_bytes(env, target + 1),
- MDBX_SYNC_DATA | MDBX_SYNC_IODQ);
- } else {
- const mdbx_filehandle_t fd = (env->me_dsync_fd != INVALID_HANDLE_VALUE)
- ? env->me_dsync_fd
- : env->me_lazy_fd;
- rc = mdbx_pwrite(fd, page, env->me_psize, pgno2bytes(env, target));
- if (rc == MDBX_SUCCESS && fd == env->me_lazy_fd)
- rc = mdbx_fsync(env->me_lazy_fd, MDBX_SYNC_DATA | MDBX_SYNC_IODQ);
- }
- mdbx_flush_incoherent_mmap(env->me_map, pgno2bytes(env, NUM_METAS),
- env->me_os_psize);
- return rc;
-}
-
-__cold int mdbx_env_turn_for_recovery(MDBX_env *env, unsigned target) {
- if (unlikely(target >= NUM_METAS))
- return MDBX_EINVAL;
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely((env->me_flags & (MDBX_EXCLUSIVE | MDBX_RDONLY)) !=
- MDBX_EXCLUSIVE))
- return MDBX_EPERM;
-
- const MDBX_meta *target_meta = METAPAGE(env, target);
- txnid_t new_txnid = safe64_txnid_next(constmeta_txnid(env, target_meta));
- for (unsigned n = 0; n < NUM_METAS; ++n) {
- MDBX_page *page = pgno2page(env, n);
- MDBX_meta meta = *page_meta(page);
- if (n == target)
- continue;
- if (mdbx_validate_meta(env, &meta, page, n, nullptr) != MDBX_SUCCESS) {
- int err = mdbx_override_meta(env, n, 0, nullptr);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- } else {
- txnid_t txnid = constmeta_txnid(env, &meta);
- if (new_txnid <= txnid)
- new_txnid = safe64_txnid_next(txnid);
- }
- }
-
- if (unlikely(new_txnid > MAX_TXNID)) {
- mdbx_error("txnid overflow, raise %d", MDBX_TXN_FULL);
- return MDBX_TXN_FULL;
- }
- return mdbx_override_meta(env, target, new_txnid, target_meta);
-}
-
-__cold int mdbx_env_open_for_recovery(MDBX_env *env, const char *pathname,
- unsigned target_meta, bool writeable) {
- if (unlikely(target_meta >= NUM_METAS))
- return MDBX_EINVAL;
- int rc = check_env(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- if (unlikely(env->me_map))
- return MDBX_EPERM;
-
- env->me_stuck_meta = (int8_t)target_meta;
- return mdbx_env_open(
- env, pathname, writeable ? MDBX_EXCLUSIVE : MDBX_EXCLUSIVE | MDBX_RDONLY,
- 0);
-}
-
-typedef struct {
- void *buffer_for_free;
- char *lck, *dxb;
- size_t ent_len;
-} MDBX_handle_env_pathname;
-
-__cold static int mdbx_handle_env_pathname(MDBX_handle_env_pathname *ctx,
- const char *pathname,
- MDBX_env_flags_t *flags,
- const mdbx_mode_t mode) {
- int rc;
- memset(ctx, 0, sizeof(*ctx));
- if (unlikely(!pathname))
- return MDBX_EINVAL;
-
-#if defined(_WIN32) || defined(_WIN64)
- const size_t wlen = mbstowcs(nullptr, pathname, INT_MAX);
- if (wlen < 1 || wlen > /* MAX_PATH */ INT16_MAX)
- return ERROR_INVALID_NAME;
- wchar_t *const pathnameW = _alloca((wlen + 1) * sizeof(wchar_t));
- if (wlen != mbstowcs(pathnameW, pathname, wlen + 1))
- return ERROR_INVALID_NAME;
-
- const DWORD dwAttrib = GetFileAttributesW(pathnameW);
- if (dwAttrib == INVALID_FILE_ATTRIBUTES) {
- rc = GetLastError();
- if (rc != MDBX_ENOFILE)
- return rc;
- if (mode == 0 || (*flags & MDBX_RDONLY) != 0)
- /* can't open existing */
- return rc;
-
- /* auto-create directory if requested */
- if ((*flags & MDBX_NOSUBDIR) == 0 &&
- !CreateDirectoryW(pathnameW, nullptr)) {
- rc = GetLastError();
- if (rc != ERROR_ALREADY_EXISTS)
- return rc;
- }
- } else {
- /* ignore passed MDBX_NOSUBDIR flag and set it automatically */
- *flags |= MDBX_NOSUBDIR;
- if (dwAttrib & FILE_ATTRIBUTE_DIRECTORY)
- *flags -= MDBX_NOSUBDIR;
- }
-#else
- struct stat st;
- if (stat(pathname, &st)) {
- rc = errno;
- if (rc != MDBX_ENOFILE)
- return rc;
- if (mode == 0 || (*flags & MDBX_RDONLY) != 0)
- /* can't open existing */
- return rc;
-
- /* auto-create directory if requested */
- const mdbx_mode_t dir_mode =
- (/* inherit read/write permissions for group and others */ mode &
- (S_IRGRP | S_IWGRP | S_IROTH | S_IWOTH)) |
- /* always add read/write/search for owner */ S_IRWXU |
- ((mode & S_IRGRP) ? /* +search if readable by group */ S_IXGRP : 0) |
- ((mode & S_IROTH) ? /* +search if readable by others */ S_IXOTH : 0);
- if ((*flags & MDBX_NOSUBDIR) == 0 && mkdir(pathname, dir_mode)) {
- rc = errno;
- if (rc != EEXIST)
- return rc;
- }
- } else {
- /* ignore passed MDBX_NOSUBDIR flag and set it automatically */
- *flags |= MDBX_NOSUBDIR;
- if (S_ISDIR(st.st_mode))
- *flags -= MDBX_NOSUBDIR;
- }
-#endif
-
- static const char dxb_name[] = MDBX_DATANAME;
- static const size_t dxb_name_len = sizeof(dxb_name) - 1;
- static const char lck_name[] = MDBX_LOCKNAME;
- static const char lock_suffix[] = MDBX_LOCK_SUFFIX;
-
- ctx->ent_len = strlen(pathname);
- if ((*flags & MDBX_NOSUBDIR) && ctx->ent_len >= dxb_name_len &&
- !memcmp(dxb_name, pathname + ctx->ent_len - dxb_name_len, dxb_name_len)) {
- *flags -= MDBX_NOSUBDIR;
- ctx->ent_len -= dxb_name_len;
- }
-
- const size_t bytes_needed =
- ctx->ent_len * 2 + ((*flags & MDBX_NOSUBDIR)
- ? sizeof(lock_suffix) + 1
- : sizeof(lck_name) + sizeof(dxb_name));
- ctx->buffer_for_free = mdbx_malloc(bytes_needed);
- if (!ctx->buffer_for_free)
- return MDBX_ENOMEM;
-
- ctx->lck = ctx->buffer_for_free;
- if (*flags & MDBX_NOSUBDIR) {
- ctx->dxb = ctx->lck + ctx->ent_len + sizeof(lock_suffix);
- sprintf(ctx->lck, "%s%s", pathname, lock_suffix);
- strcpy(ctx->dxb, pathname);
- } else {
- ctx->dxb = ctx->lck + ctx->ent_len + sizeof(lck_name);
- sprintf(ctx->lck, "%.*s%s", (int)ctx->ent_len, pathname, lck_name);
- sprintf(ctx->dxb, "%.*s%s", (int)ctx->ent_len, pathname, dxb_name);
- }
-
- return MDBX_SUCCESS;
-}
-
-__cold int mdbx_env_delete(const char *pathname, MDBX_env_delete_mode_t mode) {
- switch (mode) {
- default:
- return MDBX_EINVAL;
- case MDBX_ENV_JUST_DELETE:
- case MDBX_ENV_ENSURE_UNUSED:
- case MDBX_ENV_WAIT_FOR_UNUSED:
- break;
- }
-
-#ifdef __e2k__ /* https://bugs.mcst.ru/bugzilla/show_bug.cgi?id=6011 */
- MDBX_env *const dummy_env = alloca(sizeof(MDBX_env));
-#else
- MDBX_env dummy_env_silo, *const dummy_env = &dummy_env_silo;
-#endif
- memset(dummy_env, 0, sizeof(*dummy_env));
- dummy_env->me_flags =
- (mode == MDBX_ENV_ENSURE_UNUSED) ? MDBX_EXCLUSIVE : MDBX_ENV_DEFAULTS;
- dummy_env->me_os_psize = (unsigned)mdbx_syspagesize();
- dummy_env->me_psize = (unsigned)mdbx_default_pagesize();
- dummy_env->me_pathname = (char *)pathname;
-
- MDBX_handle_env_pathname env_pathname;
- STATIC_ASSERT(sizeof(dummy_env->me_flags) == sizeof(MDBX_env_flags_t));
- int rc = MDBX_RESULT_TRUE,
- err = mdbx_handle_env_pathname(
- &env_pathname, pathname, (MDBX_env_flags_t *)&dummy_env->me_flags, 0);
- if (likely(err == MDBX_SUCCESS)) {
- mdbx_filehandle_t clk_handle = INVALID_HANDLE_VALUE,
- dxb_handle = INVALID_HANDLE_VALUE;
- if (mode > MDBX_ENV_JUST_DELETE) {
- err = mdbx_openfile(MDBX_OPEN_DELETE, dummy_env, env_pathname.dxb,
- &dxb_handle, 0);
- err = (err == MDBX_ENOFILE) ? MDBX_SUCCESS : err;
- if (err == MDBX_SUCCESS) {
- err = mdbx_openfile(MDBX_OPEN_DELETE, dummy_env, env_pathname.lck,
- &clk_handle, 0);
- err = (err == MDBX_ENOFILE) ? MDBX_SUCCESS : err;
- }
- if (err == MDBX_SUCCESS && clk_handle != INVALID_HANDLE_VALUE)
- err = mdbx_lockfile(clk_handle, mode == MDBX_ENV_WAIT_FOR_UNUSED);
- if (err == MDBX_SUCCESS && dxb_handle != INVALID_HANDLE_VALUE)
- err = mdbx_lockfile(dxb_handle, mode == MDBX_ENV_WAIT_FOR_UNUSED);
- }
-
- if (err == MDBX_SUCCESS) {
- err = mdbx_removefile(env_pathname.dxb);
- if (err == MDBX_SUCCESS)
- rc = MDBX_SUCCESS;
- else if (err == MDBX_ENOFILE)
- err = MDBX_SUCCESS;
- }
-
- if (err == MDBX_SUCCESS) {
- err = mdbx_removefile(env_pathname.lck);
- if (err == MDBX_SUCCESS)
- rc = MDBX_SUCCESS;
- else if (err == MDBX_ENOFILE)
- err = MDBX_SUCCESS;
- }
-
- if (err == MDBX_SUCCESS && !(dummy_env->me_flags & MDBX_NOSUBDIR)) {
- err = mdbx_removedirectory(pathname);
- if (err == MDBX_SUCCESS)
- rc = MDBX_SUCCESS;
- else if (err == MDBX_ENOFILE)
- err = MDBX_SUCCESS;
- }
-
- if (dxb_handle != INVALID_HANDLE_VALUE)
- mdbx_closefile(dxb_handle);
- if (clk_handle != INVALID_HANDLE_VALUE)
- mdbx_closefile(clk_handle);
- } else if (err == MDBX_ENOFILE)
- err = MDBX_SUCCESS;
-
- mdbx_free(env_pathname.buffer_for_free);
- return (err == MDBX_SUCCESS) ? rc : err;
-}
-
-__cold int mdbx_env_open(MDBX_env *env, const char *pathname,
- MDBX_env_flags_t flags, mdbx_mode_t mode) {
- int rc = check_env(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(flags & ~ENV_USABLE_FLAGS))
- return MDBX_EINVAL;
-
- if (flags & MDBX_RDONLY)
- mode = 0;
-
- if (env->me_lazy_fd != INVALID_HANDLE_VALUE ||
- (env->me_flags & MDBX_ENV_ACTIVE) != 0 || env->me_map)
- return MDBX_EPERM;
-
- /* pickup previously mdbx_env_set_flags(),
- * but avoid MDBX_UTTERLY_NOSYNC by disjunction */
- const uint32_t saved_me_flags = env->me_flags;
- flags = merge_sync_flags(flags, env->me_flags);
-
- MDBX_handle_env_pathname env_pathname;
- rc = mdbx_handle_env_pathname(&env_pathname, pathname, &flags, mode);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
-
- if (flags & MDBX_RDONLY) {
- /* LY: silently ignore irrelevant flags when
- * we're only getting read access */
- flags &= ~(MDBX_WRITEMAP | MDBX_DEPRECATED_MAPASYNC | MDBX_SAFE_NOSYNC |
- MDBX_NOMETASYNC | MDBX_COALESCE | MDBX_LIFORECLAIM |
- MDBX_NOMEMINIT | MDBX_ACCEDE);
- } else {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- /* Temporary `workaround` for OpenBSD kernel's flaw.
- * See todo4recovery://erased_by_github/libmdbx/issues/67 */
- if ((flags & MDBX_WRITEMAP) == 0) {
- if (flags & MDBX_ACCEDE)
- flags |= MDBX_WRITEMAP;
- else {
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__,
- "System (i.e. OpenBSD) requires MDBX_WRITEMAP because "
- "of an internal flaw(s) in a file/buffer/page cache.\n");
- rc = 42 /* ENOPROTOOPT */;
- goto bailout;
- }
- }
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
- }
-
- env->me_flags = (flags & ~MDBX_FATAL_ERROR) | MDBX_ENV_ACTIVE;
- env->me_pathname = mdbx_calloc(env_pathname.ent_len + 1, 1);
- env->me_dbxs = mdbx_calloc(env->me_maxdbs, sizeof(MDBX_dbx));
- env->me_dbflags = mdbx_calloc(env->me_maxdbs, sizeof(env->me_dbflags[0]));
- env->me_dbiseqs = mdbx_calloc(env->me_maxdbs, sizeof(env->me_dbiseqs[0]));
- if (!(env->me_dbxs && env->me_pathname && env->me_dbflags &&
- env->me_dbiseqs)) {
- rc = MDBX_ENOMEM;
- goto bailout;
- }
- memcpy(env->me_pathname, env_pathname.dxb, env_pathname.ent_len);
- env->me_dbxs[FREE_DBI].md_cmp = cmp_int_align4; /* aligned MDBX_INTEGERKEY */
- env->me_dbxs[FREE_DBI].md_dcmp = cmp_lenfast;
-
- rc = mdbx_openfile(F_ISSET(flags, MDBX_RDONLY) ? MDBX_OPEN_DXB_READ
- : MDBX_OPEN_DXB_LAZY,
- env, env_pathname.dxb, &env->me_lazy_fd, mode);
- if (rc != MDBX_SUCCESS)
- goto bailout;
-
- mdbx_assert(env, env->me_dsync_fd == INVALID_HANDLE_VALUE);
- if ((flags & (MDBX_RDONLY | MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC)) == 0) {
- rc = mdbx_openfile(MDBX_OPEN_DXB_DSYNC, env, env_pathname.dxb,
- &env->me_dsync_fd, 0);
- mdbx_ensure(env, (rc != MDBX_SUCCESS) ==
- (env->me_dsync_fd == INVALID_HANDLE_VALUE));
- }
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- env->me_sysv_ipc.key = ftok(env_pathname.dxb, 42);
- if (env->me_sysv_ipc.key == -1) {
- rc = errno;
- goto bailout;
- }
-#endif /* MDBX_LOCKING */
-
-#if !(defined(_WIN32) || defined(_WIN64))
- if (mode == 0) {
- /* pickup mode for lck-file */
- struct stat st;
- if (fstat(env->me_lazy_fd, &st)) {
- rc = errno;
- goto bailout;
- }
- mode = st.st_mode;
- }
- mode = (/* inherit read permissions for group and others */ mode &
- (S_IRGRP | S_IWGRP | S_IROTH | S_IWOTH)) |
- /* always add read/write/search for owner */ S_IRUSR | S_IWUSR |
- ((mode & S_IRGRP) ? /* +write if readable by group */ S_IWGRP : 0) |
- ((mode & S_IROTH) ? /* +write if readable by others */ S_IWOTH : 0);
-#endif /* !Windows */
- const int lck_rc = mdbx_setup_lck(env, env_pathname.lck, mode);
- if (MDBX_IS_ERROR(lck_rc)) {
- rc = lck_rc;
- goto bailout;
- }
-
- /* Set the position in files outside of the data to avoid corruption
- * due to erroneous use of file descriptors in the application code. */
- mdbx_fseek(env->me_lfd, UINT64_C(1) << 63);
- mdbx_fseek(env->me_lazy_fd, UINT64_C(1) << 63);
- if (env->me_dsync_fd != INVALID_HANDLE_VALUE)
- mdbx_fseek(env->me_dsync_fd, UINT64_C(1) << 63);
-
- const MDBX_env_flags_t rigorous_flags =
- MDBX_SAFE_NOSYNC | MDBX_DEPRECATED_MAPASYNC;
- const MDBX_env_flags_t mode_flags = rigorous_flags | MDBX_NOMETASYNC |
- MDBX_LIFORECLAIM | MDBX_COALESCE |
- MDBX_NORDAHEAD;
-
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (lck && lck_rc != MDBX_RESULT_TRUE && (env->me_flags & MDBX_RDONLY) == 0) {
- while (atomic_load32(&lck->mti_envmode, mo_AcquireRelease) == MDBX_RDONLY) {
- if (atomic_cas32(&lck->mti_envmode, MDBX_RDONLY,
- env->me_flags & mode_flags)) {
- /* The case:
- * - let's assume that for some reason the DB file is smaller
- * than it should be according to the geometry,
- * but not smaller than the last page used;
- * - the first process that opens the database (lc_rc = true)
- * does this in readonly mode and therefore cannot bring
- * the file size back to normal;
- * - some next process (lc_rc = false) opens the DB in read-write
- * mode and now is here.
- *
- * FIXME: Should we re-check and set the size of DB-file right here? */
- break;
- }
- atomic_yield();
- }
-
- if (env->me_flags & MDBX_ACCEDE) {
- /* pickup current mode-flags, including MDBX_LIFORECLAIM |
- * MDBX_COALESCE | MDBX_NORDAHEAD */
- const unsigned diff =
- (lck->mti_envmode.weak ^ env->me_flags) & mode_flags;
- mdbx_notice("accede mode-flags: 0x%X, 0x%X -> 0x%X", diff, env->me_flags,
- env->me_flags ^ diff);
- env->me_flags ^= diff;
- }
-
- if ((lck->mti_envmode.weak ^ env->me_flags) & rigorous_flags) {
- mdbx_error("%s", "current mode/flags incompatible with requested");
- rc = MDBX_INCOMPATIBLE;
- goto bailout;
- }
- }
-
- const int dxb_rc = mdbx_setup_dxb(env, lck_rc, mode);
- if (MDBX_IS_ERROR(dxb_rc)) {
- rc = dxb_rc;
- goto bailout;
- }
-
- if (unlikely(/* recovery mode */ env->me_stuck_meta >= 0) &&
- (lck_rc != /* exclusive */ MDBX_RESULT_TRUE ||
- (flags & MDBX_EXCLUSIVE) == 0)) {
- mdbx_error("%s", "recovery requires exclusive mode");
- rc = MDBX_BUSY;
- goto bailout;
- }
-
- mdbx_debug("opened dbenv %p", (void *)env);
- if (lck) {
- if (lck_rc == MDBX_RESULT_TRUE) {
- lck->mti_envmode.weak = env->me_flags & (mode_flags | MDBX_RDONLY);
- rc = mdbx_lck_downgrade(env);
- mdbx_debug("lck-downgrade-%s: rc %i",
- (env->me_flags & MDBX_EXCLUSIVE) ? "partial" : "full", rc);
- if (rc != MDBX_SUCCESS)
- goto bailout;
- } else {
- rc = mdbx_cleanup_dead_readers(env, false, NULL);
- if (MDBX_IS_ERROR(rc))
- goto bailout;
- }
-
- if ((env->me_flags & MDBX_NOTLS) == 0) {
- rc = mdbx_rthc_alloc(&env->me_txkey, &lck->mti_readers[0],
- &lck->mti_readers[env->me_maxreaders]);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- env->me_flags |= MDBX_ENV_TXKEY;
- }
- }
-
- if ((flags & MDBX_RDONLY) == 0) {
- const size_t tsize = sizeof(MDBX_txn),
- size = tsize + env->me_maxdbs *
- (sizeof(MDBX_db) + sizeof(MDBX_cursor *) +
- sizeof(unsigned) + 1);
- rc = alloc_page_buf(env);
- if (rc == MDBX_SUCCESS) {
- memset(env->me_pbuf, -1, env->me_psize * 2);
- MDBX_txn *txn = mdbx_calloc(1, size);
- if (txn) {
- txn->mt_dbs = (MDBX_db *)((char *)txn + tsize);
- txn->mt_cursors = (MDBX_cursor **)(txn->mt_dbs + env->me_maxdbs);
- txn->mt_dbiseqs = (unsigned *)(txn->mt_cursors + env->me_maxdbs);
- txn->mt_dbistate = (uint8_t *)(txn->mt_dbiseqs + env->me_maxdbs);
- txn->mt_env = env;
- txn->mt_dbxs = env->me_dbxs;
- txn->mt_flags = MDBX_TXN_FINISHED;
- env->me_txn0 = txn;
- txn->tw.retired_pages = mdbx_pnl_alloc(MDBX_PNL_INITIAL);
- txn->tw.reclaimed_pglist = mdbx_pnl_alloc(MDBX_PNL_INITIAL);
- if (unlikely(!txn->tw.retired_pages || !txn->tw.reclaimed_pglist))
- rc = MDBX_ENOMEM;
- } else
- rc = MDBX_ENOMEM;
- }
- }
-
-#if MDBX_DEBUG
- if (rc == MDBX_SUCCESS) {
- const MDBX_meta *meta = constmeta_prefer_last(env);
- const MDBX_db *db = &meta->mm_dbs[MAIN_DBI];
-
- mdbx_debug("opened database version %u, pagesize %u",
- (uint8_t)unaligned_peek_u64(4, meta->mm_magic_and_version),
- env->me_psize);
- mdbx_debug("using meta page %" PRIaPGNO ", txn %" PRIaTXN,
- data_page(meta)->mp_pgno, meta_txnid(env, meta));
- mdbx_debug("depth: %u", db->md_depth);
- mdbx_debug("entries: %" PRIu64, db->md_entries);
- mdbx_debug("branch pages: %" PRIaPGNO, db->md_branch_pages);
- mdbx_debug("leaf pages: %" PRIaPGNO, db->md_leaf_pages);
- mdbx_debug("overflow pages: %" PRIaPGNO, db->md_overflow_pages);
- mdbx_debug("root: %" PRIaPGNO, db->md_root);
- mdbx_debug("schema_altered: %" PRIaTXN, db->md_mod_txnid);
- }
-#endif
-
-bailout:
- if (rc != MDBX_SUCCESS) {
- rc = mdbx_env_close0(env) ? MDBX_PANIC : rc;
- env->me_flags =
- saved_me_flags | ((rc != MDBX_PANIC) ? 0 : MDBX_FATAL_ERROR);
- } else {
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- mdbx_txn_valgrind(env, nullptr);
-#endif
- }
- mdbx_free(env_pathname.buffer_for_free);
- return rc;
-}
-
-/* Destroy resources from mdbx_env_open(), clear our readers & DBIs */
-__cold static int mdbx_env_close0(MDBX_env *env) {
- const unsigned flags = env->me_flags;
- if (!(flags & MDBX_ENV_ACTIVE)) {
- mdbx_ensure(env, env->me_lcklist_next == nullptr);
- return MDBX_SUCCESS;
- }
-
- env->me_flags &= ~ENV_INTERNAL_FLAGS;
- env->me_lck = nullptr;
- if (flags & MDBX_ENV_TXKEY) {
- mdbx_rthc_remove(env->me_txkey);
- env->me_txkey = (mdbx_thread_key_t)0;
- }
-
- lcklist_lock();
- const int rc = lcklist_detach_locked(env);
- lcklist_unlock();
-
- if (env->me_map) {
- mdbx_munmap(&env->me_dxb_mmap);
-#ifdef MDBX_USE_VALGRIND
- VALGRIND_DISCARD(env->me_valgrind_handle);
- env->me_valgrind_handle = -1;
-#endif
- }
-
- if (env->me_dsync_fd != INVALID_HANDLE_VALUE) {
- (void)mdbx_closefile(env->me_dsync_fd);
- env->me_dsync_fd = INVALID_HANDLE_VALUE;
- }
-
- if (env->me_lazy_fd != INVALID_HANDLE_VALUE) {
- (void)mdbx_closefile(env->me_lazy_fd);
- env->me_lazy_fd = INVALID_HANDLE_VALUE;
- }
-
- if (env->me_lck_mmap.lck)
- mdbx_munmap(&env->me_lck_mmap);
-
- if (env->me_lfd != INVALID_HANDLE_VALUE) {
- (void)mdbx_closefile(env->me_lfd);
- env->me_lfd = INVALID_HANDLE_VALUE;
- }
-
- if (env->me_dbxs) {
- for (unsigned i = env->me_numdbs; --i >= CORE_DBS;)
- mdbx_free(env->me_dbxs[i].md_name.iov_base);
- mdbx_free(env->me_dbxs);
- env->me_dbxs = nullptr;
- }
- if (env->me_pbuf) {
- mdbx_memalign_free(env->me_pbuf);
- env->me_pbuf = nullptr;
- }
- if (env->me_dbiseqs) {
- mdbx_free(env->me_dbiseqs);
- env->me_dbiseqs = nullptr;
- }
- if (env->me_dbflags) {
- mdbx_free(env->me_dbflags);
- env->me_dbflags = nullptr;
- }
- if (env->me_pathname) {
- mdbx_free(env->me_pathname);
- env->me_pathname = nullptr;
- }
- if (env->me_txn0) {
- mdbx_dpl_free(env->me_txn0);
- mdbx_txl_free(env->me_txn0->tw.lifo_reclaimed);
- mdbx_pnl_free(env->me_txn0->tw.retired_pages);
- mdbx_pnl_free(env->me_txn0->tw.spill_pages);
- mdbx_pnl_free(env->me_txn0->tw.reclaimed_pglist);
- mdbx_free(env->me_txn0);
- env->me_txn0 = nullptr;
- }
- env->me_stuck_meta = -1;
- return rc;
-}
-
-__cold int mdbx_env_close_ex(MDBX_env *env, bool dont_sync) {
- MDBX_page *dp;
- int rc = MDBX_SUCCESS;
-
- if (unlikely(!env))
- return MDBX_EINVAL;
-
- if (unlikely(env->me_signature.weak != MDBX_ME_SIGNATURE))
- return MDBX_EBADSIGN;
-
-#if MDBX_ENV_CHECKPID || !(defined(_WIN32) || defined(_WIN64))
- /* Check the PID even if MDBX_ENV_CHECKPID=0 on non-Windows
- * platforms (i.e. where fork() is available).
- * This is required to legitimize a call after fork()
- * from a child process, that should be allowed to free resources. */
- if (unlikely(env->me_pid != mdbx_getpid()))
- env->me_flags |= MDBX_FATAL_ERROR;
-#endif /* MDBX_ENV_CHECKPID */
-
- if (env->me_map && (env->me_flags & (MDBX_RDONLY | MDBX_FATAL_ERROR)) == 0 &&
- env->me_txn0) {
- if (env->me_txn0->mt_owner && env->me_txn0->mt_owner != mdbx_thread_self())
- return MDBX_BUSY;
- } else
- dont_sync = true;
-
- if (!atomic_cas32(&env->me_signature, MDBX_ME_SIGNATURE, 0))
- return MDBX_EBADSIGN;
-
- if (!dont_sync) {
-#if defined(_WIN32) || defined(_WIN64)
- /* On windows, without blocking is impossible to determine whether another
- * process is running a writing transaction or not.
- * Because in the "owner died" condition kernel don't release
- * file lock immediately. */
- rc = mdbx_env_sync_internal(env, true, false);
- rc = (rc == MDBX_RESULT_TRUE) ? MDBX_SUCCESS : rc;
-#else
- struct stat st;
- if (unlikely(fstat(env->me_lazy_fd, &st)))
- rc = errno;
- else if (st.st_nlink > 0 /* don't sync deleted files */) {
- rc = mdbx_env_sync_internal(env, true, true);
- rc = (rc == MDBX_BUSY || rc == EAGAIN || rc == EACCES || rc == EBUSY ||
- rc == EWOULDBLOCK || rc == MDBX_RESULT_TRUE)
- ? MDBX_SUCCESS
- : rc;
- }
-#endif
- }
-
- mdbx_assert(env, env->me_signature.weak == 0);
- rc = mdbx_env_close0(env) ? MDBX_PANIC : rc;
- mdbx_ensure(env, mdbx_fastmutex_destroy(&env->me_dbi_lock) == MDBX_SUCCESS);
-#if defined(_WIN32) || defined(_WIN64)
- /* me_remap_guard don't have destructor (Slim Reader/Writer Lock) */
- DeleteCriticalSection(&env->me_windowsbug_lock);
-#else
- mdbx_ensure(env,
- mdbx_fastmutex_destroy(&env->me_remap_guard) == MDBX_SUCCESS);
-#endif /* Windows */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV
- MDBX_lockinfo *const stub = lckless_stub(env);
- mdbx_ensure(env, mdbx_ipclock_destroy(&stub->mti_wlock) == 0);
-#endif /* MDBX_LOCKING */
-
- while ((dp = env->me_dp_reserve) != NULL) {
- MDBX_ASAN_UNPOISON_MEMORY_REGION(dp, env->me_psize);
- VALGRIND_MAKE_MEM_DEFINED(&dp->mp_next, sizeof(dp->mp_next));
- env->me_dp_reserve = dp->mp_next;
- mdbx_free(dp);
- }
- VALGRIND_DESTROY_MEMPOOL(env);
- mdbx_ensure(env, env->me_lcklist_next == nullptr);
- env->me_pid = 0;
- mdbx_free(env);
-
- return rc;
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-__cold int mdbx_env_close(MDBX_env *env) {
- return __inline_mdbx_env_close(env);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-/* Compare two items pointing at aligned unsigned int's. */
-static int __hot cmp_int_align4(const MDBX_val *a, const MDBX_val *b) {
- mdbx_assert(NULL, a->iov_len == b->iov_len);
- switch (a->iov_len) {
- case 4:
- return CMP2INT(unaligned_peek_u32(4, a->iov_base),
- unaligned_peek_u32(4, b->iov_base));
- case 8:
- return CMP2INT(unaligned_peek_u64(4, a->iov_base),
- unaligned_peek_u64(4, b->iov_base));
- default:
- mdbx_assert_fail(NULL, "invalid size for INTEGERKEY/INTEGERDUP", __func__,
- __LINE__);
- return 0;
- }
-}
-
-/* Compare two items pointing at 2-byte aligned unsigned int's. */
-static int __hot cmp_int_align2(const MDBX_val *a, const MDBX_val *b) {
- mdbx_assert(NULL, a->iov_len == b->iov_len);
- switch (a->iov_len) {
- case 4:
- return CMP2INT(unaligned_peek_u32(2, a->iov_base),
- unaligned_peek_u32(2, b->iov_base));
- case 8:
- return CMP2INT(unaligned_peek_u64(2, a->iov_base),
- unaligned_peek_u64(2, b->iov_base));
- default:
- mdbx_assert_fail(NULL, "invalid size for INTEGERKEY/INTEGERDUP", __func__,
- __LINE__);
- return 0;
- }
-}
-
-/* Compare two items pointing at unsigned values with unknown alignment.
- *
- * This is also set as MDBX_INTEGERDUP|MDBX_DUPFIXED's MDBX_dbx.md_dcmp. */
-static int __hot cmp_int_unaligned(const MDBX_val *a, const MDBX_val *b) {
- mdbx_assert(NULL, a->iov_len == b->iov_len);
- switch (a->iov_len) {
- case 4:
- return CMP2INT(unaligned_peek_u32(1, a->iov_base),
- unaligned_peek_u32(1, b->iov_base));
- case 8:
- return CMP2INT(unaligned_peek_u64(1, a->iov_base),
- unaligned_peek_u64(1, b->iov_base));
- default:
- mdbx_assert_fail(NULL, "invalid size for INTEGERKEY/INTEGERDUP", __func__,
- __LINE__);
- return 0;
- }
-}
-
-/* Compare two items lexically */
-static int __hot cmp_lexical(const MDBX_val *a, const MDBX_val *b) {
- if (a->iov_len == b->iov_len)
- return a->iov_len ? memcmp(a->iov_base, b->iov_base, a->iov_len) : 0;
-
- const int diff_len = (a->iov_len < b->iov_len) ? -1 : 1;
- const size_t shortest = (a->iov_len < b->iov_len) ? a->iov_len : b->iov_len;
- int diff_data = shortest ? memcmp(a->iov_base, b->iov_base, shortest) : 0;
- return likely(diff_data) ? diff_data : diff_len;
-}
-
-/* Compare two items in reverse byte order */
-static int __hot cmp_reverse(const MDBX_val *a, const MDBX_val *b) {
- const size_t shortest = (a->iov_len < b->iov_len) ? a->iov_len : b->iov_len;
- if (likely(shortest)) {
- const uint8_t *pa = (const uint8_t *)a->iov_base + a->iov_len;
- const uint8_t *pb = (const uint8_t *)b->iov_base + b->iov_len;
- const uint8_t *const end = pa - shortest;
- do {
- int diff = *--pa - *--pb;
- if (likely(diff))
- return diff;
- } while (pa != end);
- }
- return CMP2INT(a->iov_len, b->iov_len);
-}
-
-/* Fast non-lexically comparator */
-static int __hot cmp_lenfast(const MDBX_val *a, const MDBX_val *b) {
- int diff = CMP2INT(a->iov_len, b->iov_len);
- return likely(diff || a->iov_len == 0)
- ? diff
- : memcmp(a->iov_base, b->iov_base, a->iov_len);
-}
-
-static bool unsure_equal(MDBX_cmp_func cmp, const MDBX_val *a,
- const MDBX_val *b) {
- /* checking for the use of a known good comparator
- * or/otherwise for a full byte-to-byte match */
- return cmp == cmp_lenfast || cmp == cmp_lexical || cmp == cmp_reverse ||
- cmp == cmp_int_unaligned || cmp_lenfast(a, b) == 0;
-}
-
-/* Search for key within a page, using binary search.
- * Returns the smallest entry larger or equal to the key.
- * Updates the cursor index with the index of the found entry.
- * If no entry larger or equal to the key is found, returns NULL. */
-static struct node_result __hot mdbx_node_search(MDBX_cursor *mc,
- const MDBX_val *key) {
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- const int nkeys = page_numkeys(mp);
- DKBUF_DEBUG;
-
- mdbx_debug("searching %u keys in %s %spage %" PRIaPGNO, nkeys,
- IS_LEAF(mp) ? "leaf" : "branch", IS_SUBP(mp) ? "sub-" : "",
- mp->mp_pgno);
-
- struct node_result ret;
- ret.exact = false;
- STATIC_ASSERT(P_BRANCH == 1);
- int low = mp->mp_flags & P_BRANCH;
- int high = nkeys - 1;
- if (unlikely(high < low)) {
- mc->mc_ki[mc->mc_top] = 0;
- ret.node = NULL;
- return ret;
- }
-
- int cr = 0, i = 0;
- MDBX_cmp_func *cmp = mc->mc_dbx->md_cmp;
- MDBX_val nodekey;
- if (unlikely(IS_LEAF2(mp))) {
- mdbx_cassert(mc, mp->mp_leaf2_ksize == mc->mc_db->md_xsize);
- nodekey.iov_len = mp->mp_leaf2_ksize;
- do {
- i = (low + high) >> 1;
- nodekey.iov_base = page_leaf2key(mp, i, nodekey.iov_len);
- mdbx_cassert(mc, (char *)mp + mc->mc_txn->mt_env->me_psize >=
- (char *)nodekey.iov_base + nodekey.iov_len);
- cr = cmp(key, &nodekey);
- mdbx_debug("found leaf index %u [%s], rc = %i", i, DKEY_DEBUG(&nodekey),
- cr);
- if (unlikely(cr == 0)) {
- ret.exact = true;
- break;
- }
- low = (cr < 0) ? low : i + 1;
- high = (cr < 0) ? i - 1 : high;
- } while (likely(low <= high));
-
- /* Found entry is less than the key. */
- /* Skip to get the smallest entry larger than key. */
- i += cr > 0;
-
- /* store the key index */
- mc->mc_ki[mc->mc_top] = (indx_t)i;
- ret.node = (i < nkeys)
- ? /* fake for LEAF2 */ (MDBX_node *)(intptr_t)-1
- : /* There is no entry larger or equal to the key. */ NULL;
- return ret;
- }
-
- if (IS_BRANCH(mp) && cmp == cmp_int_align2)
- /* Branch pages have no data, so if using integer keys,
- * alignment is guaranteed. Use faster cmp_int_align4(). */
- cmp = cmp_int_align4;
-
- MDBX_node *node;
- do {
- i = (low + high) >> 1;
-
- node = page_node(mp, i);
- nodekey.iov_len = node_ks(node);
- nodekey.iov_base = node_key(node);
- mdbx_cassert(mc, (char *)mp + mc->mc_txn->mt_env->me_psize >=
- (char *)nodekey.iov_base + nodekey.iov_len);
-
- cr = cmp(key, &nodekey);
- if (IS_LEAF(mp))
- mdbx_debug("found leaf index %u [%s], rc = %i", i, DKEY_DEBUG(&nodekey),
- cr);
- else
- mdbx_debug("found branch index %u [%s -> %" PRIaPGNO "], rc = %i", i,
- DKEY_DEBUG(&nodekey), node_pgno(node), cr);
- if (unlikely(cr == 0)) {
- ret.exact = true;
- break;
- }
- low = (cr < 0) ? low : i + 1;
- high = (cr < 0) ? i - 1 : high;
- } while (likely(low <= high));
-
- /* Found entry is less than the key. */
- /* Skip to get the smallest entry larger than key. */
- i += cr > 0;
-
- /* store the key index */
- mc->mc_ki[mc->mc_top] = (indx_t)i;
- ret.node = (i < nkeys)
- ? page_node(mp, i)
- : /* There is no entry larger or equal to the key. */ NULL;
- return ret;
-}
-
-/* Pop a page off the top of the cursor's stack. */
-static __inline void mdbx_cursor_pop(MDBX_cursor *mc) {
- if (mc->mc_snum) {
- mdbx_debug("popped page %" PRIaPGNO " off db %d cursor %p",
- mc->mc_pg[mc->mc_top]->mp_pgno, DDBI(mc), (void *)mc);
- if (--mc->mc_snum) {
- mc->mc_top--;
- } else {
- mc->mc_flags &= ~C_INITIALIZED;
- }
- }
-}
-
-/* Push a page onto the top of the cursor's stack.
- * Set MDBX_TXN_ERROR on failure. */
-static __inline int mdbx_cursor_push(MDBX_cursor *mc, MDBX_page *mp) {
- mdbx_debug("pushing page %" PRIaPGNO " on db %d cursor %p", mp->mp_pgno,
- DDBI(mc), (void *)mc);
-
- if (unlikely(mc->mc_snum >= CURSOR_STACK)) {
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return MDBX_CURSOR_FULL;
- }
-
- mdbx_cassert(mc, mc->mc_snum < UINT16_MAX);
- mc->mc_top = mc->mc_snum++;
- mc->mc_pg[mc->mc_top] = mp;
- mc->mc_ki[mc->mc_top] = 0;
-
- return MDBX_SUCCESS;
-}
-
-__hot static struct page_result
-mdbx_page_get_ex(MDBX_cursor *const mc, const pgno_t pgno,
- /* TODO: use parent-page ptr */ txnid_t front) {
- struct page_result ret;
- MDBX_txn *const txn = mc->mc_txn;
- mdbx_tassert(txn, front <= txn->mt_front);
- if (unlikely(pgno >= txn->mt_next_pgno)) {
- mdbx_error("page #%" PRIaPGNO " beyond next-pgno", pgno);
- notfound:
- ret.page = nullptr;
- ret.err = MDBX_PAGE_NOTFOUND;
- bailout:
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return ret;
- }
-
- MDBX_env *const env = txn->mt_env;
- mdbx_assert(env, ((txn->mt_flags ^ env->me_flags) & MDBX_WRITEMAP) == 0);
- if (unlikely((txn->mt_flags & (MDBX_TXN_RDONLY | MDBX_WRITEMAP)) == 0)) {
- const MDBX_txn *spiller = txn;
- do {
- /* Spilled pages were dirtied in this txn and flushed
- * because the dirty list got full. Bring this page
- * back in from the map (but don't unspill it here,
- * leave that unless page_touch happens again). */
- if (unlikely(spiller->mt_flags & MDBX_TXN_SPILLS) &&
- mdbx_search_spilled(spiller, pgno)) {
- goto spilled;
- }
-
- const unsigned i = mdbx_dpl_search(spiller, pgno);
- assert((int)i > 0);
- if (spiller->tw.dirtylist->items[i].pgno == pgno) {
- ret.page = spiller->tw.dirtylist->items[i].ptr;
- spiller->tw.dirtylist->items[i].lru = txn->tw.dirtylru++;
- goto dirty;
- }
-
- spiller = spiller->mt_parent;
- } while (spiller != NULL);
- }
-
-spilled:
- ret.page = pgno2page(env, pgno);
-
-dirty:
- if (unlikely(ret.page->mp_pgno != pgno)) {
- bad_page(ret.page,
- "mismatch actual pgno (%" PRIaPGNO ") != expected (%" PRIaPGNO
- ")\n",
- ret.page->mp_pgno, pgno);
- goto notfound;
- }
-
-#if !MDBX_DISABLE_PAGECHECKS
- if (unlikely(ret.page->mp_flags & P_ILL_BITS)) {
- ret.err =
- bad_page(ret.page, "invalid page's flags (%u)\n", ret.page->mp_flags);
- goto bailout;
- }
-
- if (unlikely(ret.page->mp_txnid > front) &&
- unlikely(ret.page->mp_txnid > txn->mt_front || front < txn->mt_txnid)) {
- ret.err = bad_page(
- ret.page,
- "invalid page txnid (%" PRIaTXN ") for %s' txnid (%" PRIaTXN ")\n",
- ret.page->mp_txnid,
- (front == txn->mt_front && front != txn->mt_txnid) ? "front-txn"
- : "parent-page",
- front);
- goto bailout;
- }
-
- if (unlikely((ret.page->mp_upper < ret.page->mp_lower ||
- ((ret.page->mp_lower | ret.page->mp_upper) & 1) ||
- PAGEHDRSZ + ret.page->mp_upper > env->me_psize) &&
- !IS_OVERFLOW(ret.page))) {
- ret.err =
- bad_page(ret.page, "invalid page lower(%u)/upper(%u) with limit (%u)\n",
- ret.page->mp_lower, ret.page->mp_upper, page_space(env));
- goto bailout;
- }
-#endif /* !MDBX_DISABLE_PAGECHECKS */
-
- ret.err = MDBX_SUCCESS;
- if (mdbx_audit_enabled())
- ret.err = mdbx_page_check(mc, ret.page, C_UPDATING);
- return ret;
-}
-
-/* Finish mdbx_page_search() / mdbx_page_search_lowest().
- * The cursor is at the root page, set up the rest of it. */
-__hot static int mdbx_page_search_root(MDBX_cursor *mc, const MDBX_val *key,
- int flags) {
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- int rc;
- DKBUF_DEBUG;
-
- while (IS_BRANCH(mp)) {
- MDBX_node *node;
- int i;
-
- mdbx_debug("branch page %" PRIaPGNO " has %u keys", mp->mp_pgno,
- page_numkeys(mp));
- /* Don't assert on branch pages in the GC. We can get here
- * while in the process of rebalancing a GC branch page; we must
- * let that proceed. ITS#8336 */
- mdbx_cassert(mc, !mc->mc_dbi || page_numkeys(mp) > 1);
- mdbx_debug("found index 0 to page %" PRIaPGNO, node_pgno(page_node(mp, 0)));
-
- if (flags & (MDBX_PS_FIRST | MDBX_PS_LAST)) {
- i = 0;
- if (flags & MDBX_PS_LAST) {
- i = page_numkeys(mp) - 1;
- /* if already init'd, see if we're already in right place */
- if (mc->mc_flags & C_INITIALIZED) {
- if (mc->mc_ki[mc->mc_top] == i) {
- mc->mc_top = mc->mc_snum++;
- mp = mc->mc_pg[mc->mc_top];
- goto ready;
- }
- }
- }
- } else {
- const struct node_result nsr = mdbx_node_search(mc, key);
- if (nsr.node)
- i = mc->mc_ki[mc->mc_top] + nsr.exact - 1;
- else
- i = page_numkeys(mp) - 1;
- mdbx_debug("following index %u for key [%s]", i, DKEY_DEBUG(key));
- }
-
- mdbx_cassert(mc, i >= 0 && i < (int)page_numkeys(mp));
- node = page_node(mp, i);
-
- if (unlikely((rc = mdbx_page_get(mc, node_pgno(node), &mp,
- pp_txnid4chk(mp, mc->mc_txn))) != 0))
- return rc;
-
- mc->mc_ki[mc->mc_top] = (indx_t)i;
- if (unlikely(rc = mdbx_cursor_push(mc, mp)))
- return rc;
-
- ready:
- if (flags & MDBX_PS_MODIFY) {
- if (unlikely((rc = mdbx_page_touch(mc)) != 0))
- return rc;
- mp = mc->mc_pg[mc->mc_top];
- }
- }
-
-#if !MDBX_DISABLE_PAGECHECKS
- if (unlikely(!IS_LEAF(mp))) {
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return bad_page(mp, "index points to a page with 0x%02x flags\n",
- mp->mp_flags);
- }
-#endif /* !MDBX_DISABLE_PAGECHECKS */
-
- mdbx_debug("found leaf page %" PRIaPGNO " for key [%s]", mp->mp_pgno,
- DKEY_DEBUG(key));
- mc->mc_flags |= C_INITIALIZED;
- mc->mc_flags &= ~C_EOF;
-
- return MDBX_SUCCESS;
-}
-
-static int mdbx_setup_dbx(MDBX_dbx *const dbx, const MDBX_db *const db,
- const unsigned pagesize) {
- if (unlikely(!dbx->md_cmp)) {
- dbx->md_cmp = get_default_keycmp(db->md_flags);
- dbx->md_dcmp = get_default_datacmp(db->md_flags);
- }
-
- dbx->md_klen_min =
- (db->md_flags & MDBX_INTEGERKEY) ? 4 /* sizeof(uint32_t) */ : 0;
- dbx->md_klen_max = keysize_max(pagesize, db->md_flags);
- assert(dbx->md_klen_max != (unsigned)-1);
-
- dbx->md_vlen_min = (db->md_flags & MDBX_INTEGERDUP)
- ? 4 /* sizeof(uint32_t) */
- : ((db->md_flags & MDBX_DUPFIXED) ? 1 : 0);
- dbx->md_vlen_max = valsize_max(pagesize, db->md_flags);
- assert(dbx->md_vlen_max != (unsigned)-1);
-
- if ((db->md_flags & (MDBX_DUPFIXED | MDBX_INTEGERDUP)) != 0 && db->md_xsize) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(db->md_xsize < dbx->md_vlen_min ||
- db->md_xsize > dbx->md_vlen_max)) {
- mdbx_error("db.md_xsize (%u) <> min/max value-length (%zu/%zu)",
- db->md_xsize, dbx->md_vlen_min, dbx->md_vlen_max);
- return MDBX_CORRUPTED;
- }
- dbx->md_vlen_min = dbx->md_vlen_max = db->md_xsize;
- }
- return MDBX_SUCCESS;
-}
-
-static int mdbx_fetch_sdb(MDBX_txn *txn, MDBX_dbi dbi) {
- MDBX_cursor_couple couple;
- if (unlikely(TXN_DBI_CHANGED(txn, dbi)))
- return MDBX_BAD_DBI;
- int rc = mdbx_cursor_init(&couple.outer, txn, MAIN_DBI);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- MDBX_dbx *const dbx = &txn->mt_dbxs[dbi];
- rc = mdbx_page_search(&couple.outer, &dbx->md_name, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- return (rc == MDBX_NOTFOUND) ? MDBX_BAD_DBI : rc;
-
- MDBX_val data;
- struct node_result nsr = mdbx_node_search(&couple.outer, &dbx->md_name);
- if (unlikely(!nsr.exact))
- return MDBX_BAD_DBI;
- if (unlikely((node_flags(nsr.node) & (F_DUPDATA | F_SUBDATA)) != F_SUBDATA))
- return MDBX_INCOMPATIBLE; /* not a named DB */
-
- const txnid_t pp_txnid =
- pp_txnid4chk(couple.outer.mc_pg[couple.outer.mc_top], txn);
- rc = mdbx_node_read(&couple.outer, nsr.node, &data, pp_txnid);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(data.iov_len != sizeof(MDBX_db)))
- return MDBX_INCOMPATIBLE; /* not a named DB */
-
- uint16_t md_flags = UNALIGNED_PEEK_16(data.iov_base, MDBX_db, md_flags);
- /* The txn may not know this DBI, or another process may
- * have dropped and recreated the DB with other flags. */
- MDBX_db *const db = &txn->mt_dbs[dbi];
- if (unlikely((db->md_flags & DB_PERSISTENT_FLAGS) != md_flags))
- return MDBX_INCOMPATIBLE;
-
- memcpy(db, data.iov_base, sizeof(MDBX_db));
-#if !MDBX_DISABLE_PAGECHECKS
- mdbx_tassert(txn, txn->mt_front >= pp_txnid);
- if (unlikely(db->md_mod_txnid > pp_txnid)) {
- mdbx_error("db.md_mod_txnid (%" PRIaTXN ") > page-txnid (%" PRIaTXN ")",
- db->md_mod_txnid, pp_txnid);
- return MDBX_CORRUPTED;
- }
-#endif /* !MDBX_DISABLE_PAGECHECKS */
- rc = mdbx_setup_dbx(dbx, db, txn->mt_env->me_psize);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- txn->mt_dbistate[dbi] &= ~DBI_STALE;
- return MDBX_SUCCESS;
-}
-
-/* Search for the lowest key under the current branch page.
- * This just bypasses a numkeys check in the current page
- * before calling mdbx_page_search_root(), because the callers
- * are all in situations where the current page is known to
- * be underfilled. */
-__hot static int mdbx_page_search_lowest(MDBX_cursor *mc) {
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- mdbx_cassert(mc, IS_BRANCH(mp));
- MDBX_node *node = page_node(mp, 0);
- int rc;
-
- if (unlikely((rc = mdbx_page_get(mc, node_pgno(node), &mp,
- pp_txnid4chk(mp, mc->mc_txn))) != 0))
- return rc;
-
- mc->mc_ki[mc->mc_top] = 0;
- if (unlikely(rc = mdbx_cursor_push(mc, mp)))
- return rc;
- return mdbx_page_search_root(mc, NULL, MDBX_PS_FIRST);
-}
-
-/* Search for the page a given key should be in.
- * Push it and its parent pages on the cursor stack.
- *
- * [in,out] mc the cursor for this operation.
- * [in] key the key to search for, or NULL for first/last page.
- * [in] flags If MDBX_PS_MODIFY is set, visited pages in the DB
- * are touched (updated with new page numbers).
- * If MDBX_PS_FIRST or MDBX_PS_LAST is set, find first or last
- * leaf.
- * This is used by mdbx_cursor_first() and mdbx_cursor_last().
- * If MDBX_PS_ROOTONLY set, just fetch root node, no further
- * lookups.
- *
- * Returns 0 on success, non-zero on failure. */
-__hot static int mdbx_page_search(MDBX_cursor *mc, const MDBX_val *key,
- int flags) {
- int rc;
- pgno_t root;
-
- /* Make sure the txn is still viable, then find the root from
- * the txn's db table and set it as the root of the cursor's stack. */
- if (unlikely(mc->mc_txn->mt_flags & MDBX_TXN_BLOCKED)) {
- mdbx_debug("%s", "transaction has failed, must abort");
- return MDBX_BAD_TXN;
- }
-
- /* Make sure we're using an up-to-date root */
- if (unlikely(*mc->mc_dbistate & DBI_STALE)) {
- rc = mdbx_fetch_sdb(mc->mc_txn, mc->mc_dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
- root = mc->mc_db->md_root;
-
- if (unlikely(root == P_INVALID)) { /* Tree is empty. */
- mdbx_debug("%s", "tree is empty");
- return MDBX_NOTFOUND;
- }
-
- mdbx_cassert(mc, root >= NUM_METAS);
- if (!mc->mc_pg[0] || mc->mc_pg[0]->mp_pgno != root) {
- txnid_t pp_txnid = mc->mc_db->md_mod_txnid;
- pp_txnid = /* mc->mc_db->md_mod_txnid maybe zero in a legacy DB */ pp_txnid
- ? pp_txnid
- : mc->mc_txn->mt_txnid;
- if ((mc->mc_txn->mt_flags & MDBX_TXN_RDONLY) == 0) {
- MDBX_txn *scan = mc->mc_txn;
- do
- if ((scan->mt_flags & MDBX_TXN_DIRTY) &&
- (mc->mc_dbi == MAIN_DBI ||
- (scan->mt_dbistate[mc->mc_dbi] & DBI_DIRTY))) {
- pp_txnid = scan->mt_front;
- break;
- }
- while (unlikely((scan = scan->mt_parent) != nullptr));
- }
- if (unlikely((rc = mdbx_page_get(mc, root, &mc->mc_pg[0], pp_txnid)) != 0))
- return rc;
- }
-
- mc->mc_snum = 1;
- mc->mc_top = 0;
-
- mdbx_debug("db %d root page %" PRIaPGNO " has flags 0x%X", DDBI(mc), root,
- mc->mc_pg[0]->mp_flags);
-
- if (flags & MDBX_PS_MODIFY) {
- if (!(*mc->mc_dbistate & DBI_DIRTY) && unlikely(rc = mdbx_touch_dbi(mc)))
- return rc;
- if (unlikely(rc = mdbx_page_touch(mc)))
- return rc;
- }
-
- if (flags & MDBX_PS_ROOTONLY)
- return MDBX_SUCCESS;
-
- return mdbx_page_search_root(mc, key, flags);
-}
-
-/* Return the data associated with a given node.
- *
- * [in] mc The cursor for this operation.
- * [in] leaf The node being read.
- * [out] data Updated to point to the node's data.
- *
- * Returns 0 on success, non-zero on failure. */
-static __always_inline int mdbx_node_read(MDBX_cursor *mc,
- const MDBX_node *node, MDBX_val *data,
- const txnid_t front) {
- data->iov_len = node_ds(node);
- data->iov_base = node_data(node);
- if (unlikely(F_ISSET(node_flags(node), F_BIGDATA))) {
- /* Read overflow data. */
- MDBX_page *omp; /* overflow page */
- int rc = mdbx_page_get(mc, node_largedata_pgno(node), &omp, front);
- if (unlikely((rc != MDBX_SUCCESS))) {
- mdbx_debug("read overflow page %" PRIaPGNO " failed",
- node_largedata_pgno(node));
- return rc;
- }
- data->iov_base = page_data(omp);
- }
- return MDBX_SUCCESS;
-}
-
-int mdbx_get(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key, MDBX_val *data) {
- DKBUF_DEBUG;
- mdbx_debug("===> get db %u key [%s]", dbi, DKEY_DEBUG(key));
-
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!key || !data))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- MDBX_cursor_couple cx;
- rc = mdbx_cursor_init(&cx.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- return mdbx_cursor_set(&cx.outer, (MDBX_val *)key, data, MDBX_SET).err;
-}
-
-int mdbx_get_equal_or_great(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key,
- MDBX_val *data) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!key || !data))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- if (unlikely(txn->mt_flags & MDBX_TXN_BLOCKED))
- return MDBX_BAD_TXN;
-
- MDBX_cursor_couple cx;
- rc = mdbx_cursor_init(&cx.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- return mdbx_cursor_get(&cx.outer, key, data, MDBX_SET_LOWERBOUND);
-}
-
-int mdbx_get_ex(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key, MDBX_val *data,
- size_t *values_count) {
- DKBUF_DEBUG;
- mdbx_debug("===> get db %u key [%s]", dbi, DKEY_DEBUG(key));
-
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!key || !data))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- MDBX_cursor_couple cx;
- rc = mdbx_cursor_init(&cx.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- rc = mdbx_cursor_set(&cx.outer, key, data, MDBX_SET_KEY).err;
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc == MDBX_NOTFOUND && values_count)
- *values_count = 0;
- return rc;
- }
-
- if (values_count) {
- *values_count = 1;
- if (cx.outer.mc_xcursor != NULL) {
- MDBX_node *node = page_node(cx.outer.mc_pg[cx.outer.mc_top],
- cx.outer.mc_ki[cx.outer.mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- // coverity[uninit_use : FALSE]
- mdbx_tassert(txn, cx.outer.mc_xcursor == &cx.inner &&
- (cx.inner.mx_cursor.mc_flags & C_INITIALIZED));
- // coverity[uninit_use : FALSE]
- *values_count =
- (sizeof(*values_count) >= sizeof(cx.inner.mx_db.md_entries) ||
- cx.inner.mx_db.md_entries <= PTRDIFF_MAX)
- ? (size_t)cx.inner.mx_db.md_entries
- : PTRDIFF_MAX;
- }
- }
- }
- return MDBX_SUCCESS;
-}
-
-/* Find a sibling for a page.
- * Replaces the page at the top of the cursor's stack with the specified
- * sibling, if one exists.
- *
- * [in] mc The cursor for this operation.
- * [in] dir SIBLING_LEFT or SIBLING_RIGHT.
- *
- * Returns 0 on success, non-zero on failure. */
-static int mdbx_cursor_sibling(MDBX_cursor *mc, int dir) {
- int rc;
- MDBX_node *node;
- MDBX_page *mp;
- assert(dir == SIBLING_LEFT || dir == SIBLING_RIGHT);
-
- if (unlikely(mc->mc_snum < 2))
- return MDBX_NOTFOUND; /* root has no siblings */
-
- mdbx_cursor_pop(mc);
- mdbx_debug("parent page is page %" PRIaPGNO ", index %u",
- mc->mc_pg[mc->mc_top]->mp_pgno, mc->mc_ki[mc->mc_top]);
-
- if ((dir == SIBLING_RIGHT)
- ? (mc->mc_ki[mc->mc_top] + 1u >= page_numkeys(mc->mc_pg[mc->mc_top]))
- : (mc->mc_ki[mc->mc_top] == 0)) {
- mdbx_debug("no more keys aside, moving to next %s sibling",
- dir ? "right" : "left");
- if (unlikely((rc = mdbx_cursor_sibling(mc, dir)) != MDBX_SUCCESS)) {
- /* undo cursor_pop before returning */
- mc->mc_top++;
- mc->mc_snum++;
- return rc;
- }
- } else {
- assert((dir - 1) == -1 || (dir - 1) == 1);
- mc->mc_ki[mc->mc_top] += (indx_t)(dir - 1);
- mdbx_debug("just moving to %s index key %u",
- (dir == SIBLING_RIGHT) ? "right" : "left",
- mc->mc_ki[mc->mc_top]);
- }
- mdbx_cassert(mc, IS_BRANCH(mc->mc_pg[mc->mc_top]));
-
- node = page_node(mp = mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
- if (unlikely((rc = mdbx_page_get(mc, node_pgno(node), &mp,
- pp_txnid4chk(mp, mc->mc_txn))) != 0)) {
- /* mc will be inconsistent if caller does mc_snum++ as above */
- mc->mc_flags &= ~(C_INITIALIZED | C_EOF);
- return rc;
- }
-
- rc = mdbx_cursor_push(mc, mp);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- mc->mc_ki[mc->mc_top] =
- (indx_t)((dir == SIBLING_LEFT) ? page_numkeys(mp) - 1 : 0);
- return MDBX_SUCCESS;
-}
-
-/* Move the cursor to the next data item. */
-static int mdbx_cursor_next(MDBX_cursor *mc, MDBX_val *key, MDBX_val *data,
- MDBX_cursor_op op) {
- MDBX_page *mp;
- MDBX_node *node;
- int rc;
-
- if (unlikely(mc->mc_flags & C_DEL) && op == MDBX_NEXT_DUP)
- return MDBX_NOTFOUND;
-
- if (unlikely(!(mc->mc_flags & C_INITIALIZED)))
- return mdbx_cursor_first(mc, key, data);
-
- mp = mc->mc_pg[mc->mc_top];
- if (unlikely(mc->mc_flags & C_EOF)) {
- if (mc->mc_ki[mc->mc_top] + 1u >= page_numkeys(mp))
- return MDBX_NOTFOUND;
- mc->mc_flags ^= C_EOF;
- }
-
- if (mc->mc_db->md_flags & MDBX_DUPSORT) {
- node = page_node(mp, mc->mc_ki[mc->mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- if (op == MDBX_NEXT || op == MDBX_NEXT_DUP) {
- rc =
- mdbx_cursor_next(&mc->mc_xcursor->mx_cursor, data, NULL, MDBX_NEXT);
- if (op != MDBX_NEXT || rc != MDBX_NOTFOUND) {
- if (likely(rc == MDBX_SUCCESS))
- get_key_optional(node, key);
- return rc;
- }
- }
- } else {
- mc->mc_xcursor->mx_cursor.mc_flags &= ~(C_INITIALIZED | C_EOF);
- if (op == MDBX_NEXT_DUP)
- return MDBX_NOTFOUND;
- }
- }
-
- mdbx_debug("cursor_next: top page is %" PRIaPGNO " in cursor %p", mp->mp_pgno,
- (void *)mc);
- if (mc->mc_flags & C_DEL) {
- mc->mc_flags ^= C_DEL;
- goto skip;
- }
-
- int ki = mc->mc_ki[mc->mc_top];
- mc->mc_ki[mc->mc_top] = (indx_t)++ki;
- const int numkeys = page_numkeys(mp);
- if (unlikely(ki >= numkeys)) {
- mdbx_debug("%s", "=====> move to next sibling page");
- mc->mc_ki[mc->mc_top] = (indx_t)(numkeys - 1);
- if (unlikely((rc = mdbx_cursor_sibling(mc, SIBLING_RIGHT)) !=
- MDBX_SUCCESS)) {
- mc->mc_flags |= C_EOF;
- return rc;
- }
- mp = mc->mc_pg[mc->mc_top];
- mdbx_debug("next page is %" PRIaPGNO ", key index %u", mp->mp_pgno,
- mc->mc_ki[mc->mc_top]);
- }
-
-skip:
- mdbx_debug("==> cursor points to page %" PRIaPGNO
- " with %u keys, key index %u",
- mp->mp_pgno, page_numkeys(mp), mc->mc_ki[mc->mc_top]);
-
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(!IS_LEAF(mp)))
- return MDBX_CORRUPTED;
-
- if (IS_LEAF2(mp)) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely((mc->mc_flags & C_SUB) == 0)) {
- mdbx_error("unexpected LEAF2-page %" PRIaPGNO "for non-dupsort cursor",
- mp->mp_pgno);
- return MDBX_CORRUPTED;
- } else if (likely(key)) {
- key->iov_len = mc->mc_db->md_xsize;
- key->iov_base = page_leaf2key(mp, mc->mc_ki[mc->mc_top], key->iov_len);
- }
- return MDBX_SUCCESS;
- }
-
- node = page_node(mp, mc->mc_ki[mc->mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- rc = mdbx_xcursor_init1(mc, node, mp);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- rc = mdbx_cursor_first(&mc->mc_xcursor->mx_cursor, data, NULL);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- } else if (likely(data)) {
- if (unlikely((rc = mdbx_node_read(mc, node, data,
- pp_txnid4chk(mp, mc->mc_txn))) !=
- MDBX_SUCCESS))
- return rc;
- }
-
- get_key_optional(node, key);
- return MDBX_SUCCESS;
-}
-
-/* Move the cursor to the previous data item. */
-static int mdbx_cursor_prev(MDBX_cursor *mc, MDBX_val *key, MDBX_val *data,
- MDBX_cursor_op op) {
- MDBX_page *mp;
- MDBX_node *node;
- int rc;
-
- if (unlikely(mc->mc_flags & C_DEL) && op == MDBX_PREV_DUP)
- return MDBX_NOTFOUND;
-
- if (unlikely(!(mc->mc_flags & C_INITIALIZED))) {
- rc = mdbx_cursor_last(mc, key, data);
- if (unlikely(rc))
- return rc;
- mc->mc_ki[mc->mc_top]++;
- }
-
- mp = mc->mc_pg[mc->mc_top];
- if ((mc->mc_db->md_flags & MDBX_DUPSORT) &&
- mc->mc_ki[mc->mc_top] < page_numkeys(mp)) {
- node = page_node(mp, mc->mc_ki[mc->mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- if (op == MDBX_PREV || op == MDBX_PREV_DUP) {
- rc =
- mdbx_cursor_prev(&mc->mc_xcursor->mx_cursor, data, NULL, MDBX_PREV);
- if (op != MDBX_PREV || rc != MDBX_NOTFOUND) {
- if (likely(rc == MDBX_SUCCESS)) {
- get_key_optional(node, key);
- mc->mc_flags &= ~C_EOF;
- }
- return rc;
- }
- }
- } else {
- mc->mc_xcursor->mx_cursor.mc_flags &= ~(C_INITIALIZED | C_EOF);
- if (op == MDBX_PREV_DUP)
- return MDBX_NOTFOUND;
- }
- }
-
- mdbx_debug("cursor_prev: top page is %" PRIaPGNO " in cursor %p", mp->mp_pgno,
- (void *)mc);
-
- mc->mc_flags &= ~(C_EOF | C_DEL);
-
- int ki = mc->mc_ki[mc->mc_top];
- mc->mc_ki[mc->mc_top] = (indx_t)--ki;
- if (unlikely(ki < 0)) {
- mc->mc_ki[mc->mc_top] = 0;
- mdbx_debug("%s", "=====> move to prev sibling page");
- if ((rc = mdbx_cursor_sibling(mc, SIBLING_LEFT)) != MDBX_SUCCESS)
- return rc;
- mp = mc->mc_pg[mc->mc_top];
- mdbx_debug("prev page is %" PRIaPGNO ", key index %u", mp->mp_pgno,
- mc->mc_ki[mc->mc_top]);
- }
- mdbx_debug("==> cursor points to page %" PRIaPGNO
- " with %u keys, key index %u",
- mp->mp_pgno, page_numkeys(mp), mc->mc_ki[mc->mc_top]);
-
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(!IS_LEAF(mp)))
- return MDBX_CORRUPTED;
-
- if (IS_LEAF2(mp)) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely((mc->mc_flags & C_SUB) == 0)) {
- mdbx_error("unexpected LEAF2-page %" PRIaPGNO "for non-dupsort cursor",
- mp->mp_pgno);
- return MDBX_CORRUPTED;
- } else if (likely(key)) {
- key->iov_len = mc->mc_db->md_xsize;
- key->iov_base = page_leaf2key(mp, mc->mc_ki[mc->mc_top], key->iov_len);
- }
- return MDBX_SUCCESS;
- }
-
- node = page_node(mp, mc->mc_ki[mc->mc_top]);
-
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- rc = mdbx_xcursor_init1(mc, node, mp);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- rc = mdbx_cursor_last(&mc->mc_xcursor->mx_cursor, data, NULL);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- } else if (likely(data)) {
- if (unlikely((rc = mdbx_node_read(mc, node, data,
- pp_txnid4chk(mp, mc->mc_txn))) !=
- MDBX_SUCCESS))
- return rc;
- }
-
- get_key_optional(node, key);
- return MDBX_SUCCESS;
-}
-
-/* Set the cursor on a specific data item. */
-static struct cursor_set_result mdbx_cursor_set(MDBX_cursor *mc, MDBX_val *key,
- MDBX_val *data,
- MDBX_cursor_op op) {
- MDBX_page *mp;
- MDBX_node *node = NULL;
- DKBUF_DEBUG;
-
- struct cursor_set_result ret;
- ret.exact = false;
- if (unlikely(key->iov_len < mc->mc_dbx->md_klen_min ||
- key->iov_len > mc->mc_dbx->md_klen_max)) {
- mdbx_cassert(mc, !"Invalid key-size");
- ret.err = MDBX_BAD_VALSIZE;
- return ret;
- }
-
- MDBX_val aligned_key = *key;
- uint64_t aligned_keybytes;
- if (mc->mc_db->md_flags & MDBX_INTEGERKEY) {
- switch (aligned_key.iov_len) {
- default:
- mdbx_cassert(mc, !"key-size is invalid for MDBX_INTEGERKEY");
- ret.err = MDBX_BAD_VALSIZE;
- return ret;
- case 4:
- if (unlikely(3 & (uintptr_t)aligned_key.iov_base))
- /* copy instead of return error to avoid break compatibility */
- aligned_key.iov_base =
- memcpy(&aligned_keybytes, aligned_key.iov_base, 4);
- break;
- case 8:
- if (unlikely(7 & (uintptr_t)aligned_key.iov_base))
- /* copy instead of return error to avoid break compatibility */
- aligned_key.iov_base =
- memcpy(&aligned_keybytes, aligned_key.iov_base, 8);
- break;
- }
- }
-
- if (mc->mc_xcursor)
- mc->mc_xcursor->mx_cursor.mc_flags &= ~(C_INITIALIZED | C_EOF);
-
- /* See if we're already on the right page */
- if (mc->mc_flags & C_INITIALIZED) {
- MDBX_val nodekey;
-
- mdbx_cassert(mc, IS_LEAF(mc->mc_pg[mc->mc_top]));
- mp = mc->mc_pg[mc->mc_top];
- if (unlikely(!page_numkeys(mp))) {
- mc->mc_ki[mc->mc_top] = 0;
- mc->mc_flags |= C_EOF;
- ret.err = MDBX_NOTFOUND;
- return ret;
- }
- if (IS_LEAF2(mp)) {
- nodekey.iov_len = mc->mc_db->md_xsize;
- nodekey.iov_base = page_leaf2key(mp, 0, nodekey.iov_len);
- } else {
- node = page_node(mp, 0);
- get_key(node, &nodekey);
- }
- int cmp = mc->mc_dbx->md_cmp(&aligned_key, &nodekey);
- if (unlikely(cmp == 0)) {
- /* Probably happens rarely, but first node on the page
- * was the one we wanted. */
- mc->mc_ki[mc->mc_top] = 0;
- ret.exact = true;
- mdbx_cassert(mc, mc->mc_ki[mc->mc_top] <
- page_numkeys(mc->mc_pg[mc->mc_top]) ||
- (mc->mc_flags & C_EOF));
- goto got_node;
- }
- if (cmp > 0) {
- const unsigned nkeys = page_numkeys(mp);
- if (nkeys > 1) {
- if (IS_LEAF2(mp)) {
- nodekey.iov_base = page_leaf2key(mp, nkeys - 1, nodekey.iov_len);
- } else {
- node = page_node(mp, nkeys - 1);
- get_key(node, &nodekey);
- }
- cmp = mc->mc_dbx->md_cmp(&aligned_key, &nodekey);
- if (cmp == 0) {
- /* last node was the one we wanted */
- mdbx_cassert(mc, nkeys >= 1 && nkeys <= UINT16_MAX + 1);
- mc->mc_ki[mc->mc_top] = (indx_t)(nkeys - 1);
- ret.exact = true;
- mdbx_cassert(mc, mc->mc_ki[mc->mc_top] <
- page_numkeys(mc->mc_pg[mc->mc_top]) ||
- (mc->mc_flags & C_EOF));
- goto got_node;
- }
- if (cmp < 0) {
- if (mc->mc_ki[mc->mc_top] < page_numkeys(mp)) {
- /* This is definitely the right page, skip search_page */
- if (IS_LEAF2(mp)) {
- nodekey.iov_base =
- page_leaf2key(mp, mc->mc_ki[mc->mc_top], nodekey.iov_len);
- } else {
- node = page_node(mp, mc->mc_ki[mc->mc_top]);
- get_key(node, &nodekey);
- }
- cmp = mc->mc_dbx->md_cmp(&aligned_key, &nodekey);
- if (cmp == 0) {
- /* current node was the one we wanted */
- ret.exact = true;
- mdbx_cassert(mc, mc->mc_ki[mc->mc_top] <
- page_numkeys(mc->mc_pg[mc->mc_top]) ||
- (mc->mc_flags & C_EOF));
- goto got_node;
- }
- }
- mc->mc_flags &= ~C_EOF;
- goto search_node;
- }
- }
- /* If any parents have right-sibs, search.
- * Otherwise, there's nothing further. */
- unsigned i;
- for (i = 0; i < mc->mc_top; i++)
- if (mc->mc_ki[i] < page_numkeys(mc->mc_pg[i]) - 1)
- break;
- if (i == mc->mc_top) {
- /* There are no other pages */
- mdbx_cassert(mc, nkeys <= UINT16_MAX);
- mc->mc_ki[mc->mc_top] = (uint16_t)nkeys;
- mc->mc_flags |= C_EOF;
- ret.err = MDBX_NOTFOUND;
- return ret;
- }
- }
- if (!mc->mc_top) {
- /* There are no other pages */
- mc->mc_ki[mc->mc_top] = 0;
- if (op == MDBX_SET_RANGE)
- goto got_node;
-
- mdbx_cassert(mc, mc->mc_ki[mc->mc_top] <
- page_numkeys(mc->mc_pg[mc->mc_top]) ||
- (mc->mc_flags & C_EOF));
- ret.err = MDBX_NOTFOUND;
- return ret;
- }
- } else {
- mc->mc_pg[0] = 0;
- }
-
- ret.err = mdbx_page_search(mc, &aligned_key, 0);
- if (unlikely(ret.err != MDBX_SUCCESS))
- return ret;
-
- mp = mc->mc_pg[mc->mc_top];
- mdbx_cassert(mc, IS_LEAF(mp));
-
-search_node:;
- struct node_result nsr = mdbx_node_search(mc, &aligned_key);
- node = nsr.node;
- ret.exact = nsr.exact;
- if (!ret.exact) {
- if (op != MDBX_SET_RANGE) {
- /* MDBX_SET specified and not an exact match. */
- if (unlikely(mc->mc_ki[mc->mc_top] >=
- page_numkeys(mc->mc_pg[mc->mc_top])))
- mc->mc_flags |= C_EOF;
- ret.err = MDBX_NOTFOUND;
- return ret;
- }
-
- if (node == NULL) {
- mdbx_debug("%s", "===> inexact leaf not found, goto sibling");
- ret.err = mdbx_cursor_sibling(mc, SIBLING_RIGHT);
- if (unlikely(ret.err != MDBX_SUCCESS)) {
- mc->mc_flags |= C_EOF;
- return ret; /* no entries matched */
- }
- mp = mc->mc_pg[mc->mc_top];
- mdbx_cassert(mc, IS_LEAF(mp));
- if (!IS_LEAF2(mp))
- node = page_node(mp, 0);
- }
- }
- mdbx_cassert(mc,
- mc->mc_ki[mc->mc_top] < page_numkeys(mc->mc_pg[mc->mc_top]) ||
- (mc->mc_flags & C_EOF));
-
-got_node:
- mc->mc_flags |= C_INITIALIZED;
- mc->mc_flags &= ~C_EOF;
-
- if (IS_LEAF2(mp)) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely((mc->mc_flags & C_SUB) == 0)) {
- mdbx_error("unexpected LEAF2-page %" PRIaPGNO "for non-dupsort cursor",
- mp->mp_pgno);
- ret.err = MDBX_CORRUPTED;
- } else {
- if (op == MDBX_SET_RANGE || op == MDBX_SET_KEY) {
- key->iov_len = mc->mc_db->md_xsize;
- key->iov_base = page_leaf2key(mp, mc->mc_ki[mc->mc_top], key->iov_len);
- }
- ret.err = MDBX_SUCCESS;
- }
- return ret;
- }
-
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- ret.err = mdbx_xcursor_init1(mc, node, mp);
- if (unlikely(ret.err != MDBX_SUCCESS))
- return ret;
- if (op == MDBX_SET || op == MDBX_SET_KEY || op == MDBX_SET_RANGE) {
- ret.err = mdbx_cursor_first(&mc->mc_xcursor->mx_cursor, data, NULL);
- if (unlikely(ret.err != MDBX_SUCCESS))
- return ret;
- } else {
- ret = mdbx_cursor_set(&mc->mc_xcursor->mx_cursor, data, NULL,
- MDBX_SET_RANGE);
- if (unlikely(ret.err != MDBX_SUCCESS))
- return ret;
- if (op == MDBX_GET_BOTH && !ret.exact) {
- ret.err = MDBX_NOTFOUND;
- return ret;
- }
- }
- } else if (likely(data)) {
- if (op == MDBX_GET_BOTH || op == MDBX_GET_BOTH_RANGE) {
- if (unlikely(data->iov_len < mc->mc_dbx->md_vlen_min ||
- data->iov_len > mc->mc_dbx->md_vlen_max)) {
- mdbx_cassert(mc, !"Invalid data-size");
- ret.err = MDBX_BAD_VALSIZE;
- return ret;
- }
- MDBX_val aligned_data = *data;
- uint64_t aligned_databytes;
- if (mc->mc_db->md_flags & MDBX_INTEGERDUP) {
- switch (aligned_data.iov_len) {
- default:
- mdbx_cassert(mc, !"data-size is invalid for MDBX_INTEGERDUP");
- ret.err = MDBX_BAD_VALSIZE;
- return ret;
- case 4:
- if (unlikely(3 & (uintptr_t)aligned_data.iov_base))
- /* copy instead of return error to avoid break compatibility */
- aligned_data.iov_base =
- memcpy(&aligned_databytes, aligned_data.iov_base, 4);
- break;
- case 8:
- if (unlikely(7 & (uintptr_t)aligned_data.iov_base))
- /* copy instead of return error to avoid break compatibility */
- aligned_data.iov_base =
- memcpy(&aligned_databytes, aligned_data.iov_base, 8);
- break;
- }
- }
- MDBX_val actual_data;
- ret.err = mdbx_node_read(mc, node, &actual_data,
- pp_txnid4chk(mc->mc_pg[mc->mc_top], mc->mc_txn));
- if (unlikely(ret.err != MDBX_SUCCESS))
- return ret;
- const int cmp = mc->mc_dbx->md_dcmp(&aligned_data, &actual_data);
- if (cmp) {
- mdbx_cassert(mc, mc->mc_ki[mc->mc_top] <
- page_numkeys(mc->mc_pg[mc->mc_top]) ||
- (mc->mc_flags & C_EOF));
- if (op != MDBX_GET_BOTH_RANGE || cmp > 0) {
- ret.err = MDBX_NOTFOUND;
- return ret;
- }
- }
- *data = actual_data;
- } else {
- ret.err = mdbx_node_read(mc, node, data,
- pp_txnid4chk(mc->mc_pg[mc->mc_top], mc->mc_txn));
- if (unlikely(ret.err != MDBX_SUCCESS))
- return ret;
- }
- }
-
- /* The key already matches in all other cases */
- if (op == MDBX_SET_RANGE || op == MDBX_SET_KEY)
- get_key_optional(node, key);
-
- mdbx_debug("==> cursor placed on key [%s], data [%s]", DKEY_DEBUG(key),
- DVAL_DEBUG(data));
- ret.err = MDBX_SUCCESS;
- return ret;
-}
-
-/* Move the cursor to the first item in the database. */
-static int mdbx_cursor_first(MDBX_cursor *mc, MDBX_val *key, MDBX_val *data) {
- int rc;
-
- if (mc->mc_xcursor)
- mc->mc_xcursor->mx_cursor.mc_flags &= ~(C_INITIALIZED | C_EOF);
-
- if (!(mc->mc_flags & C_INITIALIZED) || mc->mc_top) {
- rc = mdbx_page_search(mc, NULL, MDBX_PS_FIRST);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
-
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(!IS_LEAF(mc->mc_pg[mc->mc_top])))
- return MDBX_CORRUPTED;
-
- mc->mc_flags |= C_INITIALIZED;
- mc->mc_flags &= ~C_EOF;
- mc->mc_ki[mc->mc_top] = 0;
-
- if (IS_LEAF2(mc->mc_pg[mc->mc_top])) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely((mc->mc_flags & C_SUB) == 0)) {
- mdbx_error("unexpected LEAF2-page %" PRIaPGNO "for non-dupsort cursor",
- mc->mc_pg[mc->mc_top]->mp_pgno);
- return MDBX_CORRUPTED;
- } else if (likely(key)) {
- key->iov_len = mc->mc_db->md_xsize;
- key->iov_base = page_leaf2key(mc->mc_pg[mc->mc_top], 0, key->iov_len);
- }
- return MDBX_SUCCESS;
- }
-
- MDBX_node *node = page_node(mc->mc_pg[mc->mc_top], 0);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- rc = mdbx_xcursor_init1(mc, node, mc->mc_pg[mc->mc_top]);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- rc = mdbx_cursor_first(&mc->mc_xcursor->mx_cursor, data, NULL);
- if (unlikely(rc))
- return rc;
- } else if (likely(data)) {
- if (unlikely((rc = mdbx_node_read(
- mc, node, data,
- pp_txnid4chk(mc->mc_pg[mc->mc_top], mc->mc_txn))) !=
- MDBX_SUCCESS))
- return rc;
- }
-
- get_key_optional(node, key);
- return MDBX_SUCCESS;
-}
-
-/* Move the cursor to the last item in the database. */
-static int mdbx_cursor_last(MDBX_cursor *mc, MDBX_val *key, MDBX_val *data) {
- int rc;
-
- if (mc->mc_xcursor)
- mc->mc_xcursor->mx_cursor.mc_flags &= ~(C_INITIALIZED | C_EOF);
-
- if (!(mc->mc_flags & C_INITIALIZED) || mc->mc_top) {
- rc = mdbx_page_search(mc, NULL, MDBX_PS_LAST);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
-
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(!IS_LEAF(mc->mc_pg[mc->mc_top])))
- return MDBX_CORRUPTED;
-
- mc->mc_ki[mc->mc_top] = (indx_t)page_numkeys(mc->mc_pg[mc->mc_top]) - 1;
- mc->mc_flags |= C_INITIALIZED | C_EOF;
-
- if (IS_LEAF2(mc->mc_pg[mc->mc_top])) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely((mc->mc_flags & C_SUB) == 0)) {
- mdbx_error("unexpected LEAF2-page %" PRIaPGNO "for non-dupsort cursor",
- mc->mc_pg[mc->mc_top]->mp_pgno);
- return MDBX_CORRUPTED;
- } else if (likely(key)) {
- key->iov_len = mc->mc_db->md_xsize;
- key->iov_base = page_leaf2key(mc->mc_pg[mc->mc_top],
- mc->mc_ki[mc->mc_top], key->iov_len);
- }
- return MDBX_SUCCESS;
- }
-
- MDBX_node *node = page_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- rc = mdbx_xcursor_init1(mc, node, mc->mc_pg[mc->mc_top]);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- rc = mdbx_cursor_last(&mc->mc_xcursor->mx_cursor, data, NULL);
- if (unlikely(rc))
- return rc;
- } else if (likely(data)) {
- if (unlikely((rc = mdbx_node_read(
- mc, node, data,
- pp_txnid4chk(mc->mc_pg[mc->mc_top], mc->mc_txn))) !=
- MDBX_SUCCESS))
- return rc;
- }
-
- get_key_optional(node, key);
- return MDBX_SUCCESS;
-}
-
-int mdbx_cursor_get(MDBX_cursor *mc, MDBX_val *key, MDBX_val *data,
- MDBX_cursor_op op) {
- if (unlikely(mc == NULL))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_LIVE))
- return (mc->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- int rc = check_txn(mc->mc_txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- int (*mfunc)(MDBX_cursor * mc, MDBX_val * key, MDBX_val * data);
- switch (op) {
- case MDBX_GET_CURRENT: {
- if (unlikely(!(mc->mc_flags & C_INITIALIZED)))
- return MDBX_ENODATA;
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- const unsigned nkeys = page_numkeys(mp);
- if (unlikely(mc->mc_ki[mc->mc_top] >= nkeys)) {
- mdbx_cassert(mc, nkeys <= UINT16_MAX);
- if (mc->mc_flags & C_EOF)
- return MDBX_ENODATA;
- mc->mc_ki[mc->mc_top] = (uint16_t)nkeys;
- mc->mc_flags |= C_EOF;
- return MDBX_NOTFOUND;
- }
- mdbx_cassert(mc, nkeys > 0);
-
- rc = MDBX_SUCCESS;
- if (IS_LEAF2(mp)) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely((mc->mc_flags & C_SUB) == 0)) {
- mdbx_error("unexpected LEAF2-page %" PRIaPGNO "for non-dupsort cursor",
- mp->mp_pgno);
- return MDBX_CORRUPTED;
- }
- key->iov_len = mc->mc_db->md_xsize;
- key->iov_base = page_leaf2key(mp, mc->mc_ki[mc->mc_top], key->iov_len);
- } else {
- MDBX_node *node = page_node(mp, mc->mc_ki[mc->mc_top]);
- get_key_optional(node, key);
- if (data) {
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- if (unlikely(!(mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED))) {
- rc = mdbx_xcursor_init1(mc, node, mp);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- rc = mdbx_cursor_first(&mc->mc_xcursor->mx_cursor, data, NULL);
- if (unlikely(rc))
- return rc;
- } else {
- rc = mdbx_cursor_get(&mc->mc_xcursor->mx_cursor, data, NULL,
- MDBX_GET_CURRENT);
- if (unlikely(rc))
- return rc;
- }
- } else {
- rc = mdbx_node_read(mc, node, data, pp_txnid4chk(mp, mc->mc_txn));
- if (unlikely(rc))
- return rc;
- }
- }
- }
- break;
- }
- case MDBX_GET_BOTH:
- case MDBX_GET_BOTH_RANGE:
- if (unlikely(data == NULL))
- return MDBX_EINVAL;
- if (unlikely(mc->mc_xcursor == NULL))
- return MDBX_INCOMPATIBLE;
- /* fall through */
- __fallthrough;
- case MDBX_SET:
- case MDBX_SET_KEY:
- case MDBX_SET_RANGE:
- if (unlikely(key == NULL))
- return MDBX_EINVAL;
- rc = mdbx_cursor_set(mc, key, data, op).err;
- if (mc->mc_flags & C_INITIALIZED) {
- mdbx_cassert(mc, mc->mc_snum > 0 && mc->mc_top < mc->mc_snum);
- mdbx_cassert(mc, mc->mc_ki[mc->mc_top] <
- page_numkeys(mc->mc_pg[mc->mc_top]) ||
- (mc->mc_flags & C_EOF));
- }
- break;
- case MDBX_GET_MULTIPLE:
- if (unlikely(data == NULL || !(mc->mc_flags & C_INITIALIZED)))
- return MDBX_EINVAL;
- if (unlikely(!(mc->mc_db->md_flags & MDBX_DUPFIXED)))
- return MDBX_INCOMPATIBLE;
- rc = MDBX_SUCCESS;
- if ((mc->mc_xcursor->mx_cursor.mc_flags & (C_INITIALIZED | C_EOF)) !=
- C_INITIALIZED)
- break;
- goto fetchm;
- case MDBX_NEXT_MULTIPLE:
- if (unlikely(data == NULL))
- return MDBX_EINVAL;
- if (unlikely(!(mc->mc_db->md_flags & MDBX_DUPFIXED)))
- return MDBX_INCOMPATIBLE;
- rc = mdbx_cursor_next(mc, key, data, MDBX_NEXT_DUP);
- if (rc == MDBX_SUCCESS) {
- if (mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED) {
- MDBX_cursor *mx;
- fetchm:
- mx = &mc->mc_xcursor->mx_cursor;
- data->iov_len =
- page_numkeys(mx->mc_pg[mx->mc_top]) * mx->mc_db->md_xsize;
- data->iov_base = page_data(mx->mc_pg[mx->mc_top]);
- mx->mc_ki[mx->mc_top] = (indx_t)page_numkeys(mx->mc_pg[mx->mc_top]) - 1;
- } else {
- rc = MDBX_NOTFOUND;
- }
- }
- break;
- case MDBX_PREV_MULTIPLE:
- if (data == NULL)
- return MDBX_EINVAL;
- if (!(mc->mc_db->md_flags & MDBX_DUPFIXED))
- return MDBX_INCOMPATIBLE;
- rc = MDBX_SUCCESS;
- if (!(mc->mc_flags & C_INITIALIZED))
- rc = mdbx_cursor_last(mc, key, data);
- if (rc == MDBX_SUCCESS) {
- MDBX_cursor *mx = &mc->mc_xcursor->mx_cursor;
- if (mx->mc_flags & C_INITIALIZED) {
- rc = mdbx_cursor_sibling(mx, SIBLING_LEFT);
- if (rc == MDBX_SUCCESS)
- goto fetchm;
- } else {
- rc = MDBX_NOTFOUND;
- }
- }
- break;
- case MDBX_NEXT:
- case MDBX_NEXT_DUP:
- case MDBX_NEXT_NODUP:
- rc = mdbx_cursor_next(mc, key, data, op);
- break;
- case MDBX_PREV:
- case MDBX_PREV_DUP:
- case MDBX_PREV_NODUP:
- rc = mdbx_cursor_prev(mc, key, data, op);
- break;
- case MDBX_FIRST:
- rc = mdbx_cursor_first(mc, key, data);
- break;
- case MDBX_FIRST_DUP:
- mfunc = mdbx_cursor_first;
- move:
- if (unlikely(data == NULL || !(mc->mc_flags & C_INITIALIZED)))
- return MDBX_EINVAL;
- if (unlikely(mc->mc_xcursor == NULL))
- return MDBX_INCOMPATIBLE;
- if (mc->mc_ki[mc->mc_top] >= page_numkeys(mc->mc_pg[mc->mc_top])) {
- mc->mc_ki[mc->mc_top] = (indx_t)page_numkeys(mc->mc_pg[mc->mc_top]);
- mc->mc_flags |= C_EOF;
- return MDBX_NOTFOUND;
- }
- {
- MDBX_node *node = page_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
- if (!F_ISSET(node_flags(node), F_DUPDATA)) {
- get_key_optional(node, key);
- rc = mdbx_node_read(mc, node, data,
- pp_txnid4chk(mc->mc_pg[mc->mc_top], mc->mc_txn));
- break;
- }
- }
- if (unlikely(!(mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED)))
- return MDBX_EINVAL;
- rc = mfunc(&mc->mc_xcursor->mx_cursor, data, NULL);
- break;
- case MDBX_LAST:
- rc = mdbx_cursor_last(mc, key, data);
- break;
- case MDBX_LAST_DUP:
- mfunc = mdbx_cursor_last;
- goto move;
- case MDBX_SET_UPPERBOUND: /* mostly same as MDBX_SET_LOWERBOUND */
- case MDBX_SET_LOWERBOUND: {
- if (unlikely(key == NULL || data == NULL))
- return MDBX_EINVAL;
- MDBX_val save_data = *data;
- struct cursor_set_result csr =
- mdbx_cursor_set(mc, key, data, MDBX_SET_RANGE);
- rc = csr.err;
- if (rc == MDBX_SUCCESS && csr.exact && mc->mc_xcursor) {
- mc->mc_flags &= ~C_DEL;
- csr.exact = false;
- if (!save_data.iov_base && (mc->mc_db->md_flags & MDBX_DUPFIXED)) {
- /* Avoiding search nested dupfixed hive if no data provided.
- * This is changes the semantic of MDBX_SET_LOWERBOUND but avoid
- * returning MDBX_BAD_VALSIZE. */
- } else if (mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED) {
- *data = save_data;
- csr = mdbx_cursor_set(&mc->mc_xcursor->mx_cursor, data, NULL,
- MDBX_SET_RANGE);
- rc = csr.err;
- if (rc == MDBX_NOTFOUND) {
- mdbx_cassert(mc, !csr.exact);
- rc = mdbx_cursor_next(mc, key, data, MDBX_NEXT_NODUP);
- }
- } else {
- int cmp = mc->mc_dbx->md_dcmp(&save_data, data);
- csr.exact = (cmp == 0);
- if (cmp > 0)
- rc = mdbx_cursor_next(mc, key, data, MDBX_NEXT_NODUP);
- }
- }
- if (rc == MDBX_SUCCESS && !csr.exact)
- rc = MDBX_RESULT_TRUE;
- if (unlikely(op == MDBX_SET_UPPERBOUND)) {
- /* minor fixups for MDBX_SET_UPPERBOUND */
- if (rc == MDBX_RESULT_TRUE)
- /* already at great-than by MDBX_SET_LOWERBOUND */
- rc = MDBX_SUCCESS;
- else if (rc == MDBX_SUCCESS)
- /* exactly match, going next */
- rc = mdbx_cursor_next(mc, key, data, MDBX_NEXT);
- }
- break;
- }
- default:
- mdbx_debug("unhandled/unimplemented cursor operation %u", op);
- return MDBX_EINVAL;
- }
-
- mc->mc_flags &= ~C_DEL;
- return rc;
-}
-
-static int cursor_first_batch(MDBX_cursor *mc) {
- if (!(mc->mc_flags & C_INITIALIZED) || mc->mc_top) {
- int err = mdbx_page_search(mc, NULL, MDBX_PS_FIRST);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- mdbx_cassert(mc, IS_LEAF(mc->mc_pg[mc->mc_top]));
-
- mc->mc_flags |= C_INITIALIZED;
- mc->mc_flags &= ~C_EOF;
- mc->mc_ki[mc->mc_top] = 0;
- return MDBX_SUCCESS;
-}
-
-static int cursor_next_batch(MDBX_cursor *mc) {
- if (unlikely(!(mc->mc_flags & C_INITIALIZED)))
- return cursor_first_batch(mc);
-
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- if (unlikely(mc->mc_flags & C_EOF)) {
- if ((unsigned)mc->mc_ki[mc->mc_top] + 1 >= page_numkeys(mp))
- return MDBX_NOTFOUND;
- mc->mc_flags ^= C_EOF;
- }
-
- int ki = mc->mc_ki[mc->mc_top];
- mc->mc_ki[mc->mc_top] = (indx_t)++ki;
- const int numkeys = page_numkeys(mp);
- if (likely(ki >= numkeys)) {
- mdbx_debug("%s", "=====> move to next sibling page");
- mc->mc_ki[mc->mc_top] = (indx_t)(numkeys - 1);
- int err = mdbx_cursor_sibling(mc, SIBLING_RIGHT);
- if (unlikely(err != MDBX_SUCCESS)) {
- mc->mc_flags |= C_EOF;
- return err;
- }
- mp = mc->mc_pg[mc->mc_top];
- mdbx_debug("next page is %" PRIaPGNO ", key index %u", mp->mp_pgno,
- mc->mc_ki[mc->mc_top]);
- }
- return MDBX_SUCCESS;
-}
-
-int mdbx_cursor_get_batch(MDBX_cursor *mc, size_t *count, MDBX_val *pairs,
- size_t limit, MDBX_cursor_op op) {
- if (unlikely(mc == NULL || count == NULL || limit < 4))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_LIVE))
- return (mc->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- int rc = check_txn(mc->mc_txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(mc->mc_db->md_flags & MDBX_DUPSORT))
- return MDBX_INCOMPATIBLE /* must be a non-dupsort subDB */;
-
- switch (op) {
- case MDBX_FIRST:
- rc = cursor_first_batch(mc);
- break;
- case MDBX_NEXT:
- rc = cursor_next_batch(mc);
- break;
- case MDBX_GET_CURRENT:
- rc = likely(mc->mc_flags & C_INITIALIZED) ? MDBX_SUCCESS : MDBX_ENODATA;
- break;
- default:
- mdbx_debug("unhandled/unimplemented cursor operation %u", op);
- rc = EINVAL;
- break;
- }
-
- if (unlikely(rc != MDBX_SUCCESS)) {
- *count = 0;
- return rc;
- }
-
- const MDBX_page *const page = mc->mc_pg[mc->mc_top];
- const unsigned nkeys = page_numkeys(page);
- unsigned i = mc->mc_ki[mc->mc_top], n = 0;
- if (unlikely(i >= nkeys)) {
- mdbx_cassert(mc, op == MDBX_GET_CURRENT);
- mdbx_cassert(mc, mdbx_cursor_on_last(mc) == MDBX_RESULT_TRUE);
- *count = 0;
- if (mc->mc_flags & C_EOF) {
- mdbx_cassert(mc, mdbx_cursor_on_last(mc) == MDBX_RESULT_TRUE);
- return MDBX_ENODATA;
- }
- if (mdbx_cursor_on_last(mc) != MDBX_RESULT_TRUE)
- return MDBX_EINVAL /* again MDBX_GET_CURRENT after MDBX_GET_CURRENT */;
- mc->mc_flags |= C_EOF;
- return MDBX_NOTFOUND;
- }
-
- const txnid_t pp_txnid = pp_txnid4chk(page, mc->mc_txn);
- do {
- if (unlikely(n + 2 > limit)) {
- rc = MDBX_RESULT_TRUE;
- break;
- }
- const MDBX_node *leaf = page_node(page, i);
- get_key(leaf, &pairs[n]);
- rc = mdbx_node_read(mc, leaf, &pairs[n + 1], pp_txnid);
- if (unlikely(rc != MDBX_SUCCESS))
- break;
- n += 2;
- } while (++i < nkeys);
-
- mc->mc_ki[mc->mc_top] = (indx_t)i;
- *count = n;
- return rc;
-}
-
-static int mdbx_touch_dbi(MDBX_cursor *mc) {
- mdbx_cassert(mc, (*mc->mc_dbistate & DBI_DIRTY) == 0);
- *mc->mc_dbistate |= DBI_DIRTY;
- mc->mc_txn->mt_flags |= MDBX_TXN_DIRTY;
- if (mc->mc_dbi >= CORE_DBS) {
- mdbx_cassert(mc, (mc->mc_flags & C_RECLAIMING) == 0);
- /* Touch DB record of named DB */
- MDBX_cursor_couple cx;
- int rc = mdbx_cursor_init(&cx.outer, mc->mc_txn, MAIN_DBI);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- mc->mc_txn->mt_dbistate[MAIN_DBI] |= DBI_DIRTY;
- rc = mdbx_page_search(&cx.outer, &mc->mc_dbx->md_name, MDBX_PS_MODIFY);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
- return MDBX_SUCCESS;
-}
-
-/* Touch all the pages in the cursor stack. Set mc_top.
- * Makes sure all the pages are writable, before attempting a write operation.
- * [in] mc The cursor to operate on. */
-static int mdbx_cursor_touch(MDBX_cursor *mc) {
- int rc = MDBX_SUCCESS;
- if (unlikely((*mc->mc_dbistate & DBI_DIRTY) == 0)) {
- rc = mdbx_touch_dbi(mc);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
- if (likely(mc->mc_snum)) {
- mc->mc_top = 0;
- do {
- rc = mdbx_page_touch(mc);
- } while (!rc && ++(mc->mc_top) < mc->mc_snum);
- mc->mc_top = mc->mc_snum - 1;
- }
- return rc;
-}
-
-int mdbx_cursor_put(MDBX_cursor *mc, const MDBX_val *key, MDBX_val *data,
- unsigned flags) {
- MDBX_env *env;
- MDBX_page *sub_root = NULL;
- MDBX_val xdata, *rdata, dkey, olddata;
- MDBX_db nested_dupdb;
- int err;
- DKBUF_DEBUG;
-
- if (unlikely(mc == NULL || key == NULL || data == NULL))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_LIVE))
- return (mc->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- int rc = check_txn_rw(mc->mc_txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(TXN_DBI_CHANGED(mc->mc_txn, mc->mc_dbi)))
- return MDBX_BAD_DBI;
-
- mdbx_cassert(mc, cursor_is_tracked(mc));
- env = mc->mc_txn->mt_env;
-
- /* Check this first so counter will always be zero on any early failures. */
- size_t mcount = 0, dcount = 0;
- if (unlikely(flags & MDBX_MULTIPLE)) {
- if (unlikely(flags & MDBX_RESERVE))
- return MDBX_EINVAL;
- if (unlikely(!F_ISSET(mc->mc_db->md_flags, MDBX_DUPFIXED)))
- return MDBX_INCOMPATIBLE;
- dcount = data[1].iov_len;
- if (unlikely(dcount < 2 || data->iov_len == 0))
- return MDBX_BAD_VALSIZE;
- if (unlikely(mc->mc_db->md_xsize != data->iov_len) && mc->mc_db->md_xsize)
- return MDBX_BAD_VALSIZE;
- if (unlikely(dcount > MAX_MAPSIZE / 2 /
- (BRANCH_NODE_MAX(MAX_PAGESIZE) - NODESIZE))) {
- /* checking for multiplication overflow */
- if (unlikely(dcount > MAX_MAPSIZE / 2 / data->iov_len))
- return MDBX_TOO_LARGE;
- }
- data[1].iov_len = 0 /* reset done item counter */;
- }
-
- if (flags & MDBX_RESERVE) {
- if (unlikely(mc->mc_db->md_flags & (MDBX_DUPSORT | MDBX_REVERSEDUP |
- MDBX_INTEGERDUP | MDBX_DUPFIXED)))
- return MDBX_INCOMPATIBLE;
- data->iov_base = nullptr;
- }
-
- const unsigned nospill = flags & MDBX_NOSPILL;
- flags -= nospill;
-
- if (unlikely(mc->mc_txn->mt_flags & (MDBX_TXN_RDONLY | MDBX_TXN_BLOCKED)))
- return (mc->mc_txn->mt_flags & MDBX_TXN_RDONLY) ? MDBX_EACCESS
- : MDBX_BAD_TXN;
-
- uint64_t aligned_keybytes, aligned_databytes;
- MDBX_val aligned_key, aligned_data;
- if (likely((mc->mc_flags & C_SUB) == 0)) {
- if (unlikely(key->iov_len < mc->mc_dbx->md_klen_min ||
- key->iov_len > mc->mc_dbx->md_klen_max)) {
- mdbx_cassert(mc, !"Invalid key-size");
- return MDBX_BAD_VALSIZE;
- }
- if (unlikely(data->iov_len < mc->mc_dbx->md_vlen_min ||
- data->iov_len > mc->mc_dbx->md_vlen_max)) {
- mdbx_cassert(mc, !"Invalid data-size");
- return MDBX_BAD_VALSIZE;
- }
-
- if (mc->mc_db->md_flags & MDBX_INTEGERKEY) {
- switch (key->iov_len) {
- default:
- mdbx_cassert(mc, !"key-size is invalid for MDBX_INTEGERKEY");
- return MDBX_BAD_VALSIZE;
- case 4:
- if (unlikely(3 & (uintptr_t)key->iov_base)) {
- /* copy instead of return error to avoid break compatibility */
- aligned_key.iov_base =
- memcpy(&aligned_keybytes, key->iov_base, aligned_key.iov_len = 4);
- key = &aligned_key;
- }
- break;
- case 8:
- if (unlikely(7 & (uintptr_t)key->iov_base)) {
- /* copy instead of return error to avoid break compatibility */
- aligned_key.iov_base =
- memcpy(&aligned_keybytes, key->iov_base, aligned_key.iov_len = 8);
- key = &aligned_key;
- }
- break;
- }
- }
- if (mc->mc_db->md_flags & MDBX_INTEGERDUP) {
- switch (data->iov_len) {
- default:
- mdbx_cassert(mc, !"data-size is invalid for MDBX_INTEGERKEY");
- return MDBX_BAD_VALSIZE;
- case 4:
- if (unlikely(3 & (uintptr_t)data->iov_base)) {
- if (unlikely(flags & MDBX_MULTIPLE))
- return MDBX_BAD_VALSIZE;
- /* copy instead of return error to avoid break compatibility */
- aligned_data.iov_base = memcpy(&aligned_databytes, data->iov_base,
- aligned_data.iov_len = 4);
- data = &aligned_data;
- }
- break;
- case 8:
- if (unlikely(7 & (uintptr_t)data->iov_base)) {
- if (unlikely(flags & MDBX_MULTIPLE))
- return MDBX_BAD_VALSIZE;
- /* copy instead of return error to avoid break compatibility */
- aligned_data.iov_base = memcpy(&aligned_databytes, data->iov_base,
- aligned_data.iov_len = 8);
- data = &aligned_data;
- }
- break;
- }
- }
- }
-
- mdbx_debug(
- "==> put db %d key [%s], size %" PRIuPTR ", data [%s] size %" PRIuPTR,
- DDBI(mc), DKEY_DEBUG(key), key->iov_len,
- DVAL_DEBUG((flags & MDBX_RESERVE) ? nullptr : data), data->iov_len);
-
- int dupdata_flag = 0;
- if ((flags & MDBX_CURRENT) != 0 && (mc->mc_flags & C_SUB) == 0) {
- if (unlikely(flags & (MDBX_APPEND | MDBX_NOOVERWRITE)))
- return MDBX_EINVAL;
- /* ÐžÐ¿Ñ†Ð¸Ñ MDBX_CURRENT означает, что запрошено обновление текущей запиÑи,
- * на которой ÑÐµÐ¹Ñ‡Ð°Ñ Ñтоит курÑор. ПроверÑем что переданный ключ Ñовпадает
- * Ñо значением в текущей позиции курÑора.
- * ЗдеÑÑŒ проще вызвать mdbx_cursor_get(), так как Ð´Ð»Ñ Ð¾Ð±ÑÐ»ÑƒÐ¶Ð¸Ð²Ð°Ð½Ð¸Ñ Ñ‚Ð°Ð±Ð»Ð¸Ñ†
- * Ñ MDBX_DUPSORT также требуетÑÑ Ñ‚ÐµÐºÑƒÑ‰Ð¸Ð¹ размер данных. */
- MDBX_val current_key, current_data;
- rc = mdbx_cursor_get(mc, ¤t_key, ¤t_data, MDBX_GET_CURRENT);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- if (mc->mc_dbx->md_cmp(key, ¤t_key) != 0)
- return MDBX_EKEYMISMATCH;
-
- if (unlikely((flags & MDBX_MULTIPLE)))
- goto drop_current;
-
- if (F_ISSET(mc->mc_db->md_flags, MDBX_DUPSORT)) {
- MDBX_node *node = page_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- mdbx_cassert(mc,
- mc->mc_xcursor != NULL &&
- (mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED));
- /* ЕÑли за ключом более одного значениÑ, либо еÑли размер данных
- * отличаетÑÑ, то вмеÑто Ð¾Ð±Ð½Ð¾Ð²Ð»ÐµÐ½Ð¸Ñ Ñ‚Ñ€ÐµÐ±ÑƒÐµÑ‚ÑÑ ÑƒÐ´Ð°Ð»ÐµÐ½Ð¸Ðµ и
- * поÑÐ»ÐµÐ´ÑƒÑŽÑ‰Ð°Ñ Ð²Ñтавка. */
- if (mc->mc_xcursor->mx_db.md_entries > 1 ||
- current_data.iov_len != data->iov_len) {
- drop_current:
- rc = mdbx_cursor_del(mc, flags & MDBX_ALLDUPS);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- flags -= MDBX_CURRENT;
- goto skip_check_samedata;
- }
- } else if (unlikely(node_size(key, data) > env->me_leaf_nodemax)) {
- rc = mdbx_cursor_del(mc, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- flags -= MDBX_CURRENT;
- goto skip_check_samedata;
- }
- }
- if (!(flags & MDBX_RESERVE) &&
- unlikely(cmp_lenfast(¤t_data, data) == 0))
- return MDBX_SUCCESS /* the same data, nothing to update */;
- skip_check_samedata:;
- }
-
- if (mc->mc_db->md_root == P_INVALID) {
- /* new database, cursor has nothing to point to */
- mc->mc_snum = 0;
- mc->mc_top = 0;
- mc->mc_flags &= ~C_INITIALIZED;
- rc = MDBX_NO_ROOT;
- } else if ((flags & MDBX_CURRENT) == 0) {
- bool exact = false;
- if ((flags & MDBX_APPEND) && mc->mc_db->md_entries > 0) {
- rc = mdbx_cursor_last(mc, &dkey, &olddata);
- if (likely(rc == MDBX_SUCCESS)) {
- rc = mc->mc_dbx->md_cmp(key, &dkey);
- if (likely(rc > 0)) {
- mc->mc_ki[mc->mc_top]++; /* step forward for appending */
- rc = MDBX_NOTFOUND;
- } else {
- if (unlikely(rc != MDBX_SUCCESS || !(flags & MDBX_APPENDDUP)))
- /* new-key < last-key
- * or new-key == last-key without MDBX_APPENDDUP */
- return MDBX_EKEYMISMATCH;
- exact = true;
- }
- }
- } else {
- struct cursor_set_result csr =
- /* olddata may not be updated in case LEAF2-page of dupfixed-subDB */
- mdbx_cursor_set(mc, (MDBX_val *)key, &olddata, MDBX_SET);
- rc = csr.err;
- exact = csr.exact;
- }
- if (likely(rc == MDBX_SUCCESS)) {
- if (exact) {
- if (unlikely(flags & MDBX_NOOVERWRITE)) {
- mdbx_debug("duplicate key [%s]", DKEY_DEBUG(key));
- *data = olddata;
- return MDBX_KEYEXIST;
- }
- if (unlikely(mc->mc_flags & C_SUB)) {
- /* nested subtree of DUPSORT-database with the same key,
- * nothing to update */
- mdbx_assert(env, data->iov_len == 0 &&
- (olddata.iov_len == 0 ||
- /* olddata may not be updated in case LEAF2-page
- of dupfixed-subDB */
- (mc->mc_db->md_flags & MDBX_DUPFIXED)));
- return MDBX_SUCCESS;
- }
- if (unlikely(flags & MDBX_ALLDUPS) && mc->mc_xcursor &&
- (mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED)) {
- rc = mdbx_cursor_del(mc, MDBX_ALLDUPS);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- flags -= MDBX_ALLDUPS;
- rc = MDBX_NOTFOUND;
- exact = false;
- } else /* checking for early exit without dirtying pages */
- if (!(flags & (MDBX_RESERVE | MDBX_MULTIPLE)) &&
- unlikely(mc->mc_dbx->md_dcmp(data, &olddata) == 0)) {
- if (!mc->mc_xcursor)
- /* the same data, nothing to update */
- return MDBX_SUCCESS;
- if (flags & MDBX_NODUPDATA)
- return MDBX_KEYEXIST;
- if (flags & MDBX_APPENDDUP)
- return MDBX_EKEYMISMATCH;
- if (likely(unsure_equal(mc->mc_dbx->md_dcmp, data, &olddata)))
- /* data is match exactly byte-to-byte, nothing to update */
- return MDBX_SUCCESS;
- else {
- /* The data has differences, but the user-provided comparator
- * considers them equal. So continue update since called without.
- * Continue to update since was called without MDBX_NODUPDATA. */
- }
- }
- }
- } else if (unlikely(rc != MDBX_NOTFOUND))
- return rc;
- }
-
- mc->mc_flags &= ~C_DEL;
-
- /* Cursor is positioned, check for room in the dirty list */
- if (!nospill) {
- rdata = data;
- if (unlikely(flags & MDBX_MULTIPLE)) {
- rdata = &xdata;
- xdata.iov_len = data->iov_len * dcount;
- }
- if (unlikely(err = mdbx_cursor_spill(mc, key, rdata)))
- return err;
- }
-
- if (unlikely(rc == MDBX_NO_ROOT)) {
- /* new database, write a root leaf page */
- mdbx_debug("%s", "allocating new root leaf page");
- if (unlikely((*mc->mc_dbistate & DBI_DIRTY) == 0)) {
- err = mdbx_touch_dbi(mc);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- struct page_result npr = mdbx_page_new(mc, P_LEAF, 1);
- if (unlikely(npr.err != MDBX_SUCCESS))
- return npr.err;
- npr.err = mdbx_cursor_push(mc, npr.page);
- if (unlikely(npr.err != MDBX_SUCCESS))
- return npr.err;
- mc->mc_db->md_root = npr.page->mp_pgno;
- mc->mc_db->md_depth++;
- if (mc->mc_db->md_flags & MDBX_INTEGERKEY) {
- assert(key->iov_len >= mc->mc_dbx->md_klen_min &&
- key->iov_len <= mc->mc_dbx->md_klen_max);
- mc->mc_dbx->md_klen_min = mc->mc_dbx->md_klen_max = key->iov_len;
- }
- if (mc->mc_db->md_flags & (MDBX_INTEGERDUP | MDBX_DUPFIXED)) {
- assert(data->iov_len >= mc->mc_dbx->md_vlen_min &&
- data->iov_len <= mc->mc_dbx->md_vlen_max);
- assert(mc->mc_xcursor != NULL);
- mc->mc_db->md_xsize = mc->mc_xcursor->mx_db.md_xsize =
- (unsigned)(mc->mc_dbx->md_vlen_min = mc->mc_dbx->md_vlen_max =
- mc->mc_xcursor->mx_dbx.md_klen_min =
- mc->mc_xcursor->mx_dbx.md_klen_max =
- data->iov_len);
- }
- if ((mc->mc_db->md_flags & (MDBX_DUPSORT | MDBX_DUPFIXED)) == MDBX_DUPFIXED)
- npr.page->mp_flags |= P_LEAF2;
- mc->mc_flags |= C_INITIALIZED;
- } else {
- /* make sure all cursor pages are writable */
- err = mdbx_cursor_touch(mc);
- if (unlikely(err))
- return err;
- }
-
- bool insert_key, insert_data, do_sub = false;
- insert_key = insert_data = (rc != MDBX_SUCCESS);
- uint16_t fp_flags = P_LEAF;
- MDBX_page *fp = env->me_pbuf;
- fp->mp_txnid = mc->mc_txn->mt_front;
- if (insert_key) {
- /* The key does not exist */
- mdbx_debug("inserting key at index %i", mc->mc_ki[mc->mc_top]);
- if ((mc->mc_db->md_flags & MDBX_DUPSORT) &&
- node_size(key, data) > env->me_leaf_nodemax) {
- /* Too big for a node, insert in sub-DB. Set up an empty
- * "old sub-page" for prep_subDB to expand to a full page. */
- fp->mp_leaf2_ksize =
- (mc->mc_db->md_flags & MDBX_DUPFIXED) ? (uint16_t)data->iov_len : 0;
- fp->mp_lower = fp->mp_upper = 0;
- olddata.iov_len = PAGEHDRSZ;
- goto prep_subDB;
- }
- } else {
- /* there's only a key anyway, so this is a no-op */
- if (IS_LEAF2(mc->mc_pg[mc->mc_top])) {
- char *ptr;
- unsigned ksize = mc->mc_db->md_xsize;
- if (unlikely(key->iov_len != ksize))
- return MDBX_BAD_VALSIZE;
- ptr = page_leaf2key(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top], ksize);
- memcpy(ptr, key->iov_base, ksize);
- fix_parent:
- /* if overwriting slot 0 of leaf, need to
- * update branch key if there is a parent page */
- if (mc->mc_top && !mc->mc_ki[mc->mc_top]) {
- unsigned dtop = 1;
- mc->mc_top--;
- /* slot 0 is always an empty key, find real slot */
- while (mc->mc_top && !mc->mc_ki[mc->mc_top]) {
- mc->mc_top--;
- dtop++;
- }
- err = MDBX_SUCCESS;
- if (mc->mc_ki[mc->mc_top])
- err = mdbx_update_key(mc, key);
- mdbx_cassert(mc, mc->mc_top + dtop < UINT16_MAX);
- mc->mc_top += (uint16_t)dtop;
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
-
- if (mdbx_audit_enabled()) {
- err = mdbx_cursor_check(mc, 0);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- return MDBX_SUCCESS;
- }
-
- more:;
- if (mdbx_audit_enabled()) {
- err = mdbx_cursor_check(mc, 0);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- MDBX_node *node = page_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
-
- /* Large/Overflow page overwrites need special handling */
- if (unlikely(F_ISSET(node_flags(node), F_BIGDATA))) {
- int dpages = (node_size(key, data) > env->me_leaf_nodemax)
- ? number_of_ovpages(env, data->iov_len)
- : 0;
-
- const pgno_t pgno = node_largedata_pgno(node);
- struct page_result pgr = mdbx_page_get_ex(
- mc, pgno, pp_txnid4chk(mc->mc_pg[mc->mc_top], mc->mc_txn));
- if (unlikely(pgr.err != MDBX_SUCCESS))
- return pgr.err;
- if (unlikely(!IS_OVERFLOW(pgr.page)))
- return MDBX_CORRUPTED;
-
- /* Is the ov page from this txn (or a parent) and big enough? */
- int ovpages = pgr.page->mp_pages;
- if (!IS_FROZEN(mc->mc_txn, pgr.page) &&
- (unlikely(mc->mc_flags & C_GCFREEZE)
- ? (ovpages >= dpages)
- : (ovpages ==
- /* LY: add configurable threshold to keep reserve space */
- dpages))) {
- /* yes, overwrite it. */
- if (!IS_MODIFIABLE(mc->mc_txn, pgr.page)) {
- if (IS_SPILLED(mc->mc_txn, pgr.page)) {
- pgr = /* TODO: avoid search and get txn & spill-index from
- page_result */
- mdbx_page_unspill(mc->mc_txn, pgr.page);
- if (unlikely(pgr.err))
- return pgr.err;
- } else {
- if (unlikely(!mc->mc_txn->mt_parent)) {
- mdbx_error(
- "Unexpected not frozen/modifiable/spilled but shadowed %s "
- "page %" PRIaPGNO " mod-txnid %" PRIaTXN ","
- " without parent transaction, current txn %" PRIaTXN
- " front %" PRIaTXN,
- "overflow/large", pgno, pgr.page->mp_txnid,
- mc->mc_txn->mt_txnid, mc->mc_txn->mt_front);
- return MDBX_PROBLEM;
- }
-
- /* It is writable only in a parent txn */
- MDBX_page *np = mdbx_page_malloc(mc->mc_txn, ovpages);
- if (unlikely(!np))
- return MDBX_ENOMEM;
-
- memcpy(np, pgr.page, PAGEHDRSZ); /* Copy header of page */
- err = mdbx_page_dirty(mc->mc_txn, pgr.page = np, ovpages);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
-#if MDBX_ENABLE_PGOP_STAT
- mc->mc_txn->mt_env->me_lck->mti_pgop_stat.clone.weak += ovpages;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- mdbx_cassert(mc, mdbx_dirtylist_check(mc->mc_txn));
- }
- }
- node_set_ds(node, data->iov_len);
- if (F_ISSET(flags, MDBX_RESERVE))
- data->iov_base = page_data(pgr.page);
- else
- memcpy(page_data(pgr.page), data->iov_base, data->iov_len);
-
- if (mdbx_audit_enabled()) {
- err = mdbx_cursor_check(mc, 0);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- return MDBX_SUCCESS;
- }
-
- if ((err = mdbx_page_retire(mc, pgr.page)) != MDBX_SUCCESS)
- return err;
- } else {
- olddata.iov_len = node_ds(node);
- olddata.iov_base = node_data(node);
- mdbx_cassert(mc, (char *)olddata.iov_base + olddata.iov_len <=
- (char *)(mc->mc_pg[mc->mc_top]) + env->me_psize);
-
- /* DB has dups? */
- if (F_ISSET(mc->mc_db->md_flags, MDBX_DUPSORT)) {
- /* Prepare (sub-)page/sub-DB to accept the new item, if needed.
- * fp: old sub-page or a header faking it.
- * mp: new (sub-)page. offset: growth in page size.
- * xdata: node data with new page or DB. */
- unsigned i;
- size_t offset = 0;
- MDBX_page *mp = fp = xdata.iov_base = env->me_pbuf;
- mp->mp_pgno = mc->mc_pg[mc->mc_top]->mp_pgno;
-
- /* Was a single item before, must convert now */
- if (!F_ISSET(node_flags(node), F_DUPDATA)) {
-
- /* does data match? */
- const int cmp = mc->mc_dbx->md_dcmp(data, &olddata);
- if ((flags & MDBX_APPENDDUP) && unlikely(cmp <= 0))
- return MDBX_EKEYMISMATCH;
- if (cmp == 0) {
- if (flags & MDBX_NODUPDATA)
- return MDBX_KEYEXIST;
- if (likely(unsure_equal(mc->mc_dbx->md_dcmp, data, &olddata))) {
- /* data is match exactly byte-to-byte, nothing to update */
- if (unlikely(flags & MDBX_MULTIPLE)) {
- rc = MDBX_SUCCESS;
- goto continue_multiple;
- }
- return MDBX_SUCCESS;
- } else {
- /* The data has differences, but the user-provided comparator
- * considers them equal. So continue update since called without.
- * Continue to update since was called without MDBX_NODUPDATA. */
- }
- mdbx_cassert(mc, node_size(key, data) <= env->me_leaf_nodemax);
- goto current;
- }
-
- /* Just overwrite the current item */
- if (flags & MDBX_CURRENT) {
- mdbx_cassert(mc, node_size(key, data) <= env->me_leaf_nodemax);
- goto current;
- }
-
- /* Back up original data item */
- memcpy(dkey.iov_base = fp + 1, olddata.iov_base,
- dkey.iov_len = olddata.iov_len);
- dupdata_flag = 1;
-
- /* Make sub-page header for the dup items, with dummy body */
- fp->mp_flags = P_LEAF | P_SUBP;
- fp->mp_lower = 0;
- xdata.iov_len = PAGEHDRSZ + dkey.iov_len + data->iov_len;
- if (mc->mc_db->md_flags & MDBX_DUPFIXED) {
- fp->mp_flags |= P_LEAF2;
- fp->mp_leaf2_ksize = (uint16_t)data->iov_len;
- xdata.iov_len += 2 * data->iov_len; /* leave space for 2 more */
- mdbx_cassert(mc, xdata.iov_len <= env->me_psize);
- } else {
- xdata.iov_len += 2 * (sizeof(indx_t) + NODESIZE) +
- (dkey.iov_len & 1) + (data->iov_len & 1);
- mdbx_cassert(mc, xdata.iov_len <= env->me_psize);
- }
- fp->mp_upper = (uint16_t)(xdata.iov_len - PAGEHDRSZ);
- olddata.iov_len = xdata.iov_len; /* pretend olddata is fp */
- } else if (node_flags(node) & F_SUBDATA) {
- /* Data is on sub-DB, just store it */
- flags |= F_DUPDATA | F_SUBDATA;
- goto put_sub;
- } else {
- /* Data is on sub-page */
- fp = olddata.iov_base;
- switch (flags) {
- default:
- if (!(mc->mc_db->md_flags & MDBX_DUPFIXED)) {
- offset = node_size(data, nullptr) + sizeof(indx_t);
- break;
- }
- offset = fp->mp_leaf2_ksize;
- if (page_room(fp) < offset) {
- offset *= 4; /* space for 4 more */
- break;
- }
- /* FALLTHRU: Big enough MDBX_DUPFIXED sub-page */
- __fallthrough;
- case MDBX_CURRENT | MDBX_NODUPDATA:
- case MDBX_CURRENT:
- fp->mp_txnid = mc->mc_txn->mt_front;
- fp->mp_pgno = mp->mp_pgno;
- mc->mc_xcursor->mx_cursor.mc_pg[0] = fp;
- flags |= F_DUPDATA;
- goto put_sub;
- }
- xdata.iov_len = olddata.iov_len + offset;
- }
-
- fp_flags = fp->mp_flags;
- if (node_size_len(node_ks(node), xdata.iov_len) >
- env->me_leaf_nodemax) {
- /* Too big for a sub-page, convert to sub-DB */
- fp_flags &= ~P_SUBP;
- prep_subDB:
- nested_dupdb.md_xsize = 0;
- nested_dupdb.md_flags = flags_db2sub(mc->mc_db->md_flags);
- if (mc->mc_db->md_flags & MDBX_DUPFIXED) {
- fp_flags |= P_LEAF2;
- nested_dupdb.md_xsize = fp->mp_leaf2_ksize;
- }
- nested_dupdb.md_depth = 1;
- nested_dupdb.md_branch_pages = 0;
- nested_dupdb.md_leaf_pages = 1;
- nested_dupdb.md_overflow_pages = 0;
- nested_dupdb.md_entries = page_numkeys(fp);
- xdata.iov_len = sizeof(nested_dupdb);
- xdata.iov_base = &nested_dupdb;
- const struct page_result par = mdbx_page_alloc(mc, 1, MDBX_ALLOC_ALL);
- mp = par.page;
- if (unlikely(par.err != MDBX_SUCCESS))
- return par.err;
- mc->mc_db->md_leaf_pages += 1;
- mdbx_cassert(mc, env->me_psize > olddata.iov_len);
- offset = env->me_psize - (unsigned)olddata.iov_len;
- flags |= F_DUPDATA | F_SUBDATA;
- nested_dupdb.md_root = mp->mp_pgno;
- nested_dupdb.md_seq = 0;
- nested_dupdb.md_mod_txnid = mc->mc_txn->mt_txnid;
- sub_root = mp;
- }
- if (mp != fp) {
- mp->mp_flags = fp_flags;
- mp->mp_txnid = mc->mc_txn->mt_front;
- mp->mp_leaf2_ksize = fp->mp_leaf2_ksize;
- mp->mp_lower = fp->mp_lower;
- mdbx_cassert(mc, fp->mp_upper + offset <= UINT16_MAX);
- mp->mp_upper = (indx_t)(fp->mp_upper + offset);
- if (unlikely(fp_flags & P_LEAF2)) {
- memcpy(page_data(mp), page_data(fp),
- page_numkeys(fp) * fp->mp_leaf2_ksize);
- } else {
- memcpy((char *)mp + mp->mp_upper + PAGEHDRSZ,
- (char *)fp + fp->mp_upper + PAGEHDRSZ,
- olddata.iov_len - fp->mp_upper - PAGEHDRSZ);
- memcpy((char *)(&mp->mp_ptrs), (char *)(&fp->mp_ptrs),
- page_numkeys(fp) * sizeof(mp->mp_ptrs[0]));
- for (i = 0; i < page_numkeys(fp); i++) {
- mdbx_cassert(mc, mp->mp_ptrs[i] + offset <= UINT16_MAX);
- mp->mp_ptrs[i] += (indx_t)offset;
- }
- }
- }
-
- rdata = &xdata;
- flags |= F_DUPDATA;
- do_sub = true;
- if (!insert_key)
- mdbx_node_del(mc, 0);
- goto new_sub;
- }
-
- /* MDBX passes F_SUBDATA in 'flags' to write a DB record */
- if (unlikely((node_flags(node) ^ flags) & F_SUBDATA))
- return MDBX_INCOMPATIBLE;
-
- current:
- if (data->iov_len == olddata.iov_len) {
- mdbx_cassert(mc, EVEN(key->iov_len) == EVEN(node_ks(node)));
- /* same size, just replace it. Note that we could
- * also reuse this node if the new data is smaller,
- * but instead we opt to shrink the node in that case. */
- if (F_ISSET(flags, MDBX_RESERVE))
- data->iov_base = olddata.iov_base;
- else if (!(mc->mc_flags & C_SUB))
- memcpy(olddata.iov_base, data->iov_base, data->iov_len);
- else {
- mdbx_cassert(mc, page_numkeys(mc->mc_pg[mc->mc_top]) == 1);
- mdbx_cassert(mc, PAGETYPE(mc->mc_pg[mc->mc_top]) == P_LEAF);
- mdbx_cassert(mc, node_ds(node) == 0);
- mdbx_cassert(mc, node_flags(node) == 0);
- mdbx_cassert(mc, key->iov_len < UINT16_MAX);
- node_set_ks(node, key->iov_len);
- memcpy(node_key(node), key->iov_base, key->iov_len);
- mdbx_cassert(mc, (char *)node_key(node) + node_ds(node) <
- (char *)(mc->mc_pg[mc->mc_top]) + env->me_psize);
- goto fix_parent;
- }
-
- if (mdbx_audit_enabled()) {
- err = mdbx_cursor_check(mc, 0);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- return MDBX_SUCCESS;
- }
- }
- mdbx_node_del(mc, 0);
- }
-
- rdata = data;
-
-new_sub:;
- unsigned nflags = flags & NODE_ADD_FLAGS;
- size_t nsize = IS_LEAF2(mc->mc_pg[mc->mc_top]) ? key->iov_len
- : leaf_size(env, key, rdata);
- if (page_room(mc->mc_pg[mc->mc_top]) < nsize) {
- if (!insert_key)
- nflags |= MDBX_SPLIT_REPLACE;
- rc = mdbx_page_split(mc, key, rdata, P_INVALID, nflags);
- if (rc == MDBX_SUCCESS && mdbx_audit_enabled())
- rc = mdbx_cursor_check(mc, 0);
- } else {
- /* There is room already in this leaf page. */
- if (IS_LEAF2(mc->mc_pg[mc->mc_top])) {
- mdbx_cassert(mc, (nflags & (F_BIGDATA | F_SUBDATA | F_DUPDATA)) == 0 &&
- rdata->iov_len == 0);
- rc = mdbx_node_add_leaf2(mc, mc->mc_ki[mc->mc_top], key);
- } else
- rc = mdbx_node_add_leaf(mc, mc->mc_ki[mc->mc_top], key, rdata, nflags);
- if (likely(rc == 0)) {
- /* Adjust other cursors pointing to mp */
- const MDBX_dbi dbi = mc->mc_dbi;
- const unsigned i = mc->mc_top;
- MDBX_page *const mp = mc->mc_pg[i];
- for (MDBX_cursor *m2 = mc->mc_txn->mt_cursors[dbi]; m2;
- m2 = m2->mc_next) {
- MDBX_cursor *m3 =
- (mc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (m3 == mc || m3->mc_snum < mc->mc_snum || m3->mc_pg[i] != mp)
- continue;
- if (m3->mc_ki[i] >= mc->mc_ki[i])
- m3->mc_ki[i] += insert_key;
- if (XCURSOR_INITED(m3))
- XCURSOR_REFRESH(m3, mp, m3->mc_ki[i]);
- }
- }
- }
-
- if (likely(rc == MDBX_SUCCESS)) {
- /* Now store the actual data in the child DB. Note that we're
- * storing the user data in the keys field, so there are strict
- * size limits on dupdata. The actual data fields of the child
- * DB are all zero size. */
- if (do_sub) {
- int xflags;
- size_t ecount;
- put_sub:
- xdata.iov_len = 0;
- xdata.iov_base = nullptr;
- MDBX_node *node = page_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
-#define SHIFT_MDBX_NODUPDATA_TO_MDBX_NOOVERWRITE 1
- STATIC_ASSERT(
- (MDBX_NODUPDATA >> SHIFT_MDBX_NODUPDATA_TO_MDBX_NOOVERWRITE) ==
- MDBX_NOOVERWRITE);
- xflags = MDBX_CURRENT | MDBX_NOSPILL |
- ((flags & MDBX_NODUPDATA) >>
- SHIFT_MDBX_NODUPDATA_TO_MDBX_NOOVERWRITE);
- if ((flags & MDBX_CURRENT) == 0) {
- xflags -= MDBX_CURRENT;
- err = mdbx_xcursor_init1(mc, node, mc->mc_pg[mc->mc_top]);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- if (sub_root)
- mc->mc_xcursor->mx_cursor.mc_pg[0] = sub_root;
- /* converted, write the original data first */
- if (dupdata_flag) {
- rc = mdbx_cursor_put(&mc->mc_xcursor->mx_cursor, &dkey, &xdata, xflags);
- if (unlikely(rc))
- goto bad_sub;
- /* we've done our job */
- dkey.iov_len = 0;
- }
- if (!(node_flags(node) & F_SUBDATA) || sub_root) {
- /* Adjust other cursors pointing to mp */
- MDBX_cursor *m2;
- MDBX_xcursor *mx = mc->mc_xcursor;
- unsigned i = mc->mc_top;
- MDBX_page *mp = mc->mc_pg[i];
- const int nkeys = page_numkeys(mp);
-
- for (m2 = mc->mc_txn->mt_cursors[mc->mc_dbi]; m2; m2 = m2->mc_next) {
- if (m2 == mc || m2->mc_snum < mc->mc_snum)
- continue;
- if (!(m2->mc_flags & C_INITIALIZED))
- continue;
- if (m2->mc_pg[i] == mp) {
- if (m2->mc_ki[i] == mc->mc_ki[i]) {
- err = mdbx_xcursor_init2(m2, mx, dupdata_flag);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- } else if (!insert_key && m2->mc_ki[i] < nkeys) {
- XCURSOR_REFRESH(m2, mp, m2->mc_ki[i]);
- }
- }
- }
- }
- mdbx_cassert(mc, mc->mc_xcursor->mx_db.md_entries < PTRDIFF_MAX);
- ecount = (size_t)mc->mc_xcursor->mx_db.md_entries;
-#define SHIFT_MDBX_APPENDDUP_TO_MDBX_APPEND 1
- STATIC_ASSERT((MDBX_APPENDDUP >> SHIFT_MDBX_APPENDDUP_TO_MDBX_APPEND) ==
- MDBX_APPEND);
- xflags |= (flags & MDBX_APPENDDUP) >> SHIFT_MDBX_APPENDDUP_TO_MDBX_APPEND;
- rc = mdbx_cursor_put(&mc->mc_xcursor->mx_cursor, data, &xdata, xflags);
- if (flags & F_SUBDATA) {
- void *db = node_data(node);
- mc->mc_xcursor->mx_db.md_mod_txnid = mc->mc_txn->mt_txnid;
- memcpy(db, &mc->mc_xcursor->mx_db, sizeof(MDBX_db));
- }
- insert_data = (ecount != (size_t)mc->mc_xcursor->mx_db.md_entries);
- }
- /* Increment count unless we just replaced an existing item. */
- if (insert_data)
- mc->mc_db->md_entries++;
- if (insert_key) {
- /* Invalidate txn if we created an empty sub-DB */
- if (unlikely(rc))
- goto bad_sub;
- /* If we succeeded and the key didn't exist before,
- * make sure the cursor is marked valid. */
- mc->mc_flags |= C_INITIALIZED;
- }
- if (unlikely(flags & MDBX_MULTIPLE)) {
- if (likely(rc == MDBX_SUCCESS)) {
- continue_multiple:
- mcount++;
- /* let caller know how many succeeded, if any */
- data[1].iov_len = mcount;
- if (mcount < dcount) {
- data[0].iov_base = (char *)data[0].iov_base + data[0].iov_len;
- insert_key = insert_data = false;
- goto more;
- }
- }
- }
- if (rc == MDBX_SUCCESS && mdbx_audit_enabled())
- rc = mdbx_cursor_check(mc, 0);
- return rc;
- bad_sub:
- if (unlikely(rc == MDBX_KEYEXIST)) {
- /* should not happen, we deleted that item */
- mdbx_error("Unexpected %i error while put to nested dupsort's hive", rc);
- rc = MDBX_PROBLEM;
- }
- }
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return rc;
-}
-
-int mdbx_cursor_del(MDBX_cursor *mc, MDBX_put_flags_t flags) {
- if (unlikely(!mc))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_LIVE))
- return (mc->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- int rc = check_txn_rw(mc->mc_txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(TXN_DBI_CHANGED(mc->mc_txn, mc->mc_dbi)))
- return MDBX_BAD_DBI;
-
- if (unlikely(!(mc->mc_flags & C_INITIALIZED)))
- return MDBX_ENODATA;
-
- if (unlikely(mc->mc_ki[mc->mc_top] >= page_numkeys(mc->mc_pg[mc->mc_top])))
- return MDBX_NOTFOUND;
-
- if (likely((flags & MDBX_NOSPILL) == 0) &&
- unlikely(rc = mdbx_cursor_spill(mc, NULL, NULL)))
- return rc;
-
- rc = mdbx_cursor_touch(mc);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(!IS_LEAF(mp)))
- return MDBX_CORRUPTED;
- if (IS_LEAF2(mp)) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely((mc->mc_flags & C_SUB) == 0)) {
- mdbx_error("unexpected LEAF2-page %" PRIaPGNO "for non-dupsort cursor",
- mp->mp_pgno);
- return MDBX_CORRUPTED;
- }
- goto del_key;
- }
-
- MDBX_node *node = page_node(mp, mc->mc_ki[mc->mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- if (flags & (MDBX_ALLDUPS | /* for compatibility */ MDBX_NODUPDATA)) {
- /* mdbx_cursor_del0() will subtract the final entry */
- mc->mc_db->md_entries -= mc->mc_xcursor->mx_db.md_entries - 1;
- mc->mc_xcursor->mx_cursor.mc_flags &= ~C_INITIALIZED;
- } else {
- if (!F_ISSET(node_flags(node), F_SUBDATA))
- mc->mc_xcursor->mx_cursor.mc_pg[0] = node_data(node);
- rc = mdbx_cursor_del(&mc->mc_xcursor->mx_cursor, MDBX_NOSPILL);
- if (unlikely(rc))
- return rc;
- /* If sub-DB still has entries, we're done */
- if (mc->mc_xcursor->mx_db.md_entries) {
- if (node_flags(node) & F_SUBDATA) {
- /* update subDB info */
- void *db = node_data(node);
- mc->mc_xcursor->mx_db.md_mod_txnid = mc->mc_txn->mt_txnid;
- memcpy(db, &mc->mc_xcursor->mx_db, sizeof(MDBX_db));
- } else {
- MDBX_cursor *m2;
- /* shrink fake page */
- mdbx_node_shrink(mp, mc->mc_ki[mc->mc_top]);
- node = page_node(mp, mc->mc_ki[mc->mc_top]);
- mc->mc_xcursor->mx_cursor.mc_pg[0] = node_data(node);
- /* fix other sub-DB cursors pointed at fake pages on this page */
- for (m2 = mc->mc_txn->mt_cursors[mc->mc_dbi]; m2; m2 = m2->mc_next) {
- if (m2 == mc || m2->mc_snum < mc->mc_snum)
- continue;
- if (!(m2->mc_flags & C_INITIALIZED))
- continue;
- if (m2->mc_pg[mc->mc_top] == mp) {
- MDBX_node *inner = node;
- if (m2->mc_ki[mc->mc_top] >= page_numkeys(mp))
- continue;
- if (m2->mc_ki[mc->mc_top] != mc->mc_ki[mc->mc_top]) {
- inner = page_node(mp, m2->mc_ki[mc->mc_top]);
- if (node_flags(inner) & F_SUBDATA)
- continue;
- }
- m2->mc_xcursor->mx_cursor.mc_pg[0] = node_data(inner);
- }
- }
- }
- mc->mc_db->md_entries--;
- mdbx_cassert(mc, mc->mc_db->md_entries > 0 && mc->mc_db->md_depth > 0 &&
- mc->mc_db->md_root != P_INVALID);
- return rc;
- } else {
- mc->mc_xcursor->mx_cursor.mc_flags &= ~C_INITIALIZED;
- }
- /* otherwise fall thru and delete the sub-DB */
- }
-
- if (node_flags(node) & F_SUBDATA) {
- /* add all the child DB's pages to the free list */
- rc = mdbx_drop_tree(&mc->mc_xcursor->mx_cursor, false);
- if (unlikely(rc))
- goto fail;
- }
- }
- /* MDBX passes F_SUBDATA in 'flags' to delete a DB record */
- else if (unlikely((node_flags(node) ^ flags) & F_SUBDATA))
- return MDBX_INCOMPATIBLE;
-
- /* add overflow pages to free list */
- if (F_ISSET(node_flags(node), F_BIGDATA)) {
- MDBX_page *omp;
- if (unlikely((rc = mdbx_page_get(mc, node_largedata_pgno(node), &omp,
- pp_txnid4chk(mp, mc->mc_txn))) ||
- (rc = mdbx_page_retire(mc, omp))))
- goto fail;
- }
-
-del_key:
- return mdbx_cursor_del0(mc);
-
-fail:
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return rc;
-}
-
-/* Allocate and initialize new pages for a database.
- * Set MDBX_TXN_ERROR on failure.
- *
- * [in] mc a cursor on the database being added to.
- * [in] flags flags defining what type of page is being allocated.
- * [in] num the number of pages to allocate. This is usually 1,
- * unless allocating overflow pages for a large record.
- * [out] mp Address of a page, or NULL on failure.
- *
- * Returns 0 on success, non-zero on failure. */
-static struct page_result mdbx_page_new(MDBX_cursor *mc, const unsigned flags,
- const unsigned npages) {
- struct page_result ret = mdbx_page_alloc(mc, npages, MDBX_ALLOC_ALL);
- if (unlikely(ret.err != MDBX_SUCCESS))
- return ret;
-
- mdbx_debug("db %u allocated new page %" PRIaPGNO ", num %u", mc->mc_dbi,
- ret.page->mp_pgno, npages);
- ret.page->mp_flags = (uint16_t)flags;
- ret.page->mp_txnid = mc->mc_txn->mt_front;
- mdbx_cassert(mc, *mc->mc_dbistate & DBI_DIRTY);
- mdbx_cassert(mc, mc->mc_txn->mt_flags & MDBX_TXN_DIRTY);
-#if MDBX_ENABLE_PGOP_STAT
- mc->mc_txn->mt_env->me_lck->mti_pgop_stat.newly.weak += npages;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
- if (likely((flags & P_OVERFLOW) == 0)) {
- STATIC_ASSERT(P_BRANCH == 1);
- const bool is_branch = flags & P_BRANCH;
- ret.page->mp_lower = 0;
- ret.page->mp_upper = (indx_t)(mc->mc_txn->mt_env->me_psize - PAGEHDRSZ);
- mc->mc_db->md_branch_pages += is_branch;
- mc->mc_db->md_leaf_pages += 1 - is_branch;
- if (unlikely(mc->mc_flags & C_SUB)) {
- MDBX_db *outer = mdbx_outer_db(mc);
- outer->md_branch_pages += is_branch;
- outer->md_leaf_pages += 1 - is_branch;
- }
- } else {
- mc->mc_db->md_overflow_pages += npages;
- ret.page->mp_pages = npages;
- mdbx_cassert(mc, !(mc->mc_flags & C_SUB));
- }
-
- return ret;
-}
-
-static int __must_check_result mdbx_node_add_leaf2(MDBX_cursor *mc,
- unsigned indx,
- const MDBX_val *key) {
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- DKBUF_DEBUG;
- mdbx_debug("add to leaf2-%spage %" PRIaPGNO " index %i, "
- " key size %" PRIuPTR " [%s]",
- IS_SUBP(mp) ? "sub-" : "", mp->mp_pgno, indx,
- key ? key->iov_len : 0, DKEY_DEBUG(key));
-
- mdbx_cassert(mc, key);
- mdbx_cassert(mc, PAGETYPE(mp) == (P_LEAF | P_LEAF2));
- const unsigned ksize = mc->mc_db->md_xsize;
- mdbx_cassert(mc, ksize == key->iov_len);
- const unsigned nkeys = page_numkeys(mp);
-
- /* Just using these for counting */
- const intptr_t lower = mp->mp_lower + sizeof(indx_t);
- const intptr_t upper = mp->mp_upper - (ksize - sizeof(indx_t));
- if (unlikely(lower > upper)) {
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return MDBX_PAGE_FULL;
- }
- mp->mp_lower = (indx_t)lower;
- mp->mp_upper = (indx_t)upper;
-
- char *const ptr = page_leaf2key(mp, indx, ksize);
- mdbx_cassert(mc, nkeys >= indx);
- const unsigned diff = nkeys - indx;
- if (likely(diff > 0))
- /* Move higher keys up one slot. */
- memmove(ptr + ksize, ptr, diff * ksize);
- /* insert new key */
- memcpy(ptr, key->iov_base, ksize);
- return MDBX_SUCCESS;
-}
-
-static int __must_check_result mdbx_node_add_branch(MDBX_cursor *mc,
- unsigned indx,
- const MDBX_val *key,
- pgno_t pgno) {
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- DKBUF_DEBUG;
- mdbx_debug("add to branch-%spage %" PRIaPGNO " index %i, node-pgno %" PRIaPGNO
- " key size %" PRIuPTR " [%s]",
- IS_SUBP(mp) ? "sub-" : "", mp->mp_pgno, indx, pgno,
- key ? key->iov_len : 0, DKEY_DEBUG(key));
-
- mdbx_cassert(mc, PAGETYPE(mp) == P_BRANCH);
- STATIC_ASSERT(NODESIZE % 2 == 0);
-
- /* Move higher pointers up one slot. */
- const unsigned nkeys = page_numkeys(mp);
- mdbx_cassert(mc, nkeys >= indx);
- for (unsigned i = nkeys; i > indx; --i)
- mp->mp_ptrs[i] = mp->mp_ptrs[i - 1];
-
- /* Adjust free space offsets. */
- const size_t branch_bytes = branch_size(mc->mc_txn->mt_env, key);
- const intptr_t lower = mp->mp_lower + sizeof(indx_t);
- const intptr_t upper = mp->mp_upper - (branch_bytes - sizeof(indx_t));
- if (unlikely(lower > upper)) {
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return MDBX_PAGE_FULL;
- }
- mp->mp_lower = (indx_t)lower;
- mp->mp_ptrs[indx] = mp->mp_upper = (indx_t)upper;
-
- /* Write the node data. */
- MDBX_node *node = page_node(mp, indx);
- node_set_pgno(node, pgno);
- node_set_flags(node, 0);
- UNALIGNED_POKE_8(node, MDBX_node, mn_extra, 0);
- node_set_ks(node, 0);
- if (likely(key != NULL)) {
- node_set_ks(node, key->iov_len);
- memcpy(node_key(node), key->iov_base, key->iov_len);
- }
- return MDBX_SUCCESS;
-}
-
-static int __must_check_result mdbx_node_add_leaf(MDBX_cursor *mc,
- unsigned indx,
- const MDBX_val *key,
- MDBX_val *data,
- unsigned flags) {
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- DKBUF_DEBUG;
- mdbx_debug("add to leaf-%spage %" PRIaPGNO " index %i, data size %" PRIuPTR
- " key size %" PRIuPTR " [%s]",
- IS_SUBP(mp) ? "sub-" : "", mp->mp_pgno, indx,
- data ? data->iov_len : 0, key ? key->iov_len : 0, DKEY_DEBUG(key));
- mdbx_cassert(mc, key != NULL && data != NULL);
- mdbx_cassert(mc, PAGETYPE(mp) == P_LEAF);
- mdbx_cassert(mc, page_room(mp) >= leaf_size(mc->mc_txn->mt_env, key, data));
- MDBX_page *largepage = NULL;
-
- size_t node_bytes;
- if (unlikely(flags & F_BIGDATA)) {
- /* Data already on overflow page. */
- STATIC_ASSERT(sizeof(pgno_t) % 2 == 0);
- node_bytes =
- node_size_len(key->iov_len, 0) + sizeof(pgno_t) + sizeof(indx_t);
- } else if (unlikely(node_size(key, data) >
- mc->mc_txn->mt_env->me_leaf_nodemax)) {
- /* Put data on overflow page. */
- if (unlikely(mc->mc_db->md_flags & MDBX_DUPSORT)) {
- mdbx_error("Unexpected target %s flags 0x%x for large data-item",
- "dupsort-db", mc->mc_db->md_flags);
- return MDBX_PROBLEM;
- }
- if (unlikely(flags & (F_DUPDATA | F_SUBDATA))) {
- mdbx_error("Unexpected target %s flags 0x%x for large data-item", "node",
- flags);
- return MDBX_PROBLEM;
- }
- const pgno_t ovpages = number_of_ovpages(mc->mc_txn->mt_env, data->iov_len);
- const struct page_result npr = mdbx_page_new(mc, P_OVERFLOW, ovpages);
- if (unlikely(npr.err != MDBX_SUCCESS))
- return npr.err;
- largepage = npr.page;
- mdbx_debug("allocated %u overflow page(s) %" PRIaPGNO "for %" PRIuPTR
- " data bytes",
- largepage->mp_pages, largepage->mp_pgno, data->iov_len);
- flags |= F_BIGDATA;
- node_bytes =
- node_size_len(key->iov_len, 0) + sizeof(pgno_t) + sizeof(indx_t);
- } else {
- node_bytes = node_size(key, data) + sizeof(indx_t);
- }
- mdbx_cassert(mc, node_bytes == leaf_size(mc->mc_txn->mt_env, key, data));
-
- /* Move higher pointers up one slot. */
- const unsigned nkeys = page_numkeys(mp);
- mdbx_cassert(mc, nkeys >= indx);
- for (unsigned i = nkeys; i > indx; --i)
- mp->mp_ptrs[i] = mp->mp_ptrs[i - 1];
-
- /* Adjust free space offsets. */
- const intptr_t lower = mp->mp_lower + sizeof(indx_t);
- const intptr_t upper = mp->mp_upper - (node_bytes - sizeof(indx_t));
- if (unlikely(lower > upper)) {
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return MDBX_PAGE_FULL;
- }
- mp->mp_lower = (indx_t)lower;
- mp->mp_ptrs[indx] = mp->mp_upper = (indx_t)upper;
-
- /* Write the node data. */
- MDBX_node *node = page_node(mp, indx);
- node_set_ks(node, key->iov_len);
- node_set_flags(node, (uint8_t)flags);
- UNALIGNED_POKE_8(node, MDBX_node, mn_extra, 0);
- node_set_ds(node, data->iov_len);
- memcpy(node_key(node), key->iov_base, key->iov_len);
-
- void *nodedata = node_data(node);
- if (likely(largepage == NULL)) {
- if (unlikely(flags & F_BIGDATA))
- memcpy(nodedata, data->iov_base, sizeof(pgno_t));
- else if (unlikely(flags & MDBX_RESERVE))
- data->iov_base = nodedata;
- else if (likely(nodedata != data->iov_base &&
- data->iov_len /* to avoid UBSAN traps*/ != 0))
- memcpy(nodedata, data->iov_base, data->iov_len);
- } else {
- poke_pgno(nodedata, largepage->mp_pgno);
- nodedata = page_data(largepage);
- if (unlikely(flags & MDBX_RESERVE))
- data->iov_base = nodedata;
- else if (likely(nodedata != data->iov_base &&
- data->iov_len /* to avoid UBSAN traps*/ != 0))
- memcpy(nodedata, data->iov_base, data->iov_len);
- }
- return MDBX_SUCCESS;
-}
-
-/* Delete the specified node from a page.
- * [in] mc Cursor pointing to the node to delete.
- * [in] ksize The size of a node. Only used if the page is
- * part of a MDBX_DUPFIXED database. */
-static void mdbx_node_del(MDBX_cursor *mc, size_t ksize) {
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- int indx = mc->mc_ki[mc->mc_top];
- int i, j, nkeys, ptr;
- MDBX_node *node;
- char *base;
-
- mdbx_debug("delete node %u on %s page %" PRIaPGNO, indx,
- IS_LEAF(mp) ? "leaf" : "branch", mp->mp_pgno);
- nkeys = page_numkeys(mp);
- mdbx_cassert(mc, indx < nkeys);
-
- if (IS_LEAF2(mp)) {
- mdbx_cassert(mc, ksize >= sizeof(indx_t));
- unsigned diff = nkeys - 1 - indx;
- base = page_leaf2key(mp, indx, ksize);
- if (diff)
- memmove(base, base + ksize, diff * ksize);
- mdbx_cassert(mc, mp->mp_lower >= sizeof(indx_t));
- mp->mp_lower -= sizeof(indx_t);
- mdbx_cassert(mc,
- (size_t)UINT16_MAX - mp->mp_upper >= ksize - sizeof(indx_t));
- mp->mp_upper += (indx_t)(ksize - sizeof(indx_t));
- return;
- }
-
- node = page_node(mp, indx);
- mdbx_cassert(mc, !IS_BRANCH(mp) || indx || node_ks(node) == 0);
- size_t sz = NODESIZE + node_ks(node);
- if (IS_LEAF(mp)) {
- if (F_ISSET(node_flags(node), F_BIGDATA))
- sz += sizeof(pgno_t);
- else
- sz += node_ds(node);
- }
- sz = EVEN(sz);
-
- ptr = mp->mp_ptrs[indx];
- for (i = j = 0; i < nkeys; i++) {
- if (i != indx) {
- mp->mp_ptrs[j] = mp->mp_ptrs[i];
- if (mp->mp_ptrs[i] < ptr) {
- mdbx_cassert(mc, (size_t)UINT16_MAX - mp->mp_ptrs[j] >= sz);
- mp->mp_ptrs[j] += (indx_t)sz;
- }
- j++;
- }
- }
-
- base = (char *)mp + mp->mp_upper + PAGEHDRSZ;
- memmove(base + sz, base, ptr - mp->mp_upper);
-
- mdbx_cassert(mc, mp->mp_lower >= sizeof(indx_t));
- mp->mp_lower -= sizeof(indx_t);
- mdbx_cassert(mc, (size_t)UINT16_MAX - mp->mp_upper >= sz);
- mp->mp_upper += (indx_t)sz;
-
-#if MDBX_DEBUG > 0
- if (mdbx_audit_enabled()) {
- int page_check_err = mdbx_page_check(mc, mp, C_UPDATING);
- mdbx_cassert(mc, page_check_err == MDBX_SUCCESS);
- }
-#endif
-}
-
-/* Compact the main page after deleting a node on a subpage.
- * [in] mp The main page to operate on.
- * [in] indx The index of the subpage on the main page. */
-static void mdbx_node_shrink(MDBX_page *mp, unsigned indx) {
- MDBX_node *node;
- MDBX_page *sp, *xp;
- char *base;
- size_t nsize, delta, len, ptr;
- int i;
-
- node = page_node(mp, indx);
- sp = (MDBX_page *)node_data(node);
- delta = page_room(sp);
- assert(delta > 0);
-
- /* Prepare to shift upward, set len = length(subpage part to shift) */
- if (IS_LEAF2(sp)) {
- delta &= /* do not make the node uneven-sized */ ~(size_t)1;
- if (unlikely(delta) == 0)
- return;
- nsize = node_ds(node) - delta;
- assert(nsize % 1 == 0);
- len = nsize;
- } else {
- xp = (MDBX_page *)((char *)sp + delta); /* destination subpage */
- for (i = page_numkeys(sp); --i >= 0;) {
- assert(sp->mp_ptrs[i] >= delta);
- xp->mp_ptrs[i] = (indx_t)(sp->mp_ptrs[i] - delta);
- }
- nsize = node_ds(node) - delta;
- len = PAGEHDRSZ;
- }
- sp->mp_upper = sp->mp_lower;
- sp->mp_pgno = mp->mp_pgno;
- node_set_ds(node, nsize);
-
- /* Shift <lower nodes...initial part of subpage> upward */
- base = (char *)mp + mp->mp_upper + PAGEHDRSZ;
- memmove(base + delta, base, (char *)sp + len - base);
-
- ptr = mp->mp_ptrs[indx];
- for (i = page_numkeys(mp); --i >= 0;) {
- if (mp->mp_ptrs[i] <= ptr) {
- assert((size_t)UINT16_MAX - mp->mp_ptrs[i] >= delta);
- mp->mp_ptrs[i] += (indx_t)delta;
- }
- }
- assert((size_t)UINT16_MAX - mp->mp_upper >= delta);
- mp->mp_upper += (indx_t)delta;
-}
-
-/* Initial setup of a sorted-dups cursor.
- *
- * Sorted duplicates are implemented as a sub-database for the given key.
- * The duplicate data items are actually keys of the sub-database.
- * Operations on the duplicate data items are performed using a sub-cursor
- * initialized when the sub-database is first accessed. This function does
- * the preliminary setup of the sub-cursor, filling in the fields that
- * depend only on the parent DB.
- *
- * [in] mc The main cursor whose sorted-dups cursor is to be initialized. */
-static int mdbx_xcursor_init0(MDBX_cursor *mc) {
- MDBX_xcursor *mx = mc->mc_xcursor;
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(mx == nullptr)) {
- mdbx_error("unexpected dupsort-page for non-dupsort db/cursor (dbi %u)",
- mc->mc_dbi);
- return MDBX_CORRUPTED;
- }
-
- mx->mx_cursor.mc_xcursor = NULL;
- mx->mx_cursor.mc_next = NULL;
- mx->mx_cursor.mc_txn = mc->mc_txn;
- mx->mx_cursor.mc_db = &mx->mx_db;
- mx->mx_cursor.mc_dbx = &mx->mx_dbx;
- mx->mx_cursor.mc_dbi = mc->mc_dbi;
- mx->mx_cursor.mc_dbistate = mc->mc_dbistate;
- mx->mx_cursor.mc_snum = 0;
- mx->mx_cursor.mc_top = 0;
- mx->mx_cursor.mc_flags = C_SUB | (mc->mc_flags & (C_COPYING | C_SKIPORD));
- mx->mx_dbx.md_name.iov_len = 0;
- mx->mx_dbx.md_name.iov_base = NULL;
- mx->mx_dbx.md_cmp = mc->mc_dbx->md_dcmp;
- mx->mx_dbx.md_dcmp = NULL;
- mx->mx_dbx.md_klen_min = INT_MAX;
- mx->mx_dbx.md_vlen_min = mx->mx_dbx.md_klen_max = mx->mx_dbx.md_vlen_max = 0;
- return MDBX_SUCCESS;
-}
-
-/* Final setup of a sorted-dups cursor.
- * Sets up the fields that depend on the data from the main cursor.
- * [in] mc The main cursor whose sorted-dups cursor is to be initialized.
- * [in] node The data containing the MDBX_db record for the sorted-dup database.
- */
-static int mdbx_xcursor_init1(MDBX_cursor *mc, MDBX_node *node,
- const MDBX_page *mp) {
- MDBX_xcursor *mx = mc->mc_xcursor;
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(mx == nullptr)) {
- mdbx_error("unexpected dupsort-page for non-dupsort db/cursor (dbi %u)",
- mc->mc_dbi);
- return MDBX_CORRUPTED;
- }
-
- const uint8_t flags = node_flags(node);
- switch (flags) {
- default:
- mdbx_error("invalid node flags %u", flags);
- return MDBX_CORRUPTED;
- case F_DUPDATA | F_SUBDATA:
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(node_ds(node) != sizeof(MDBX_db))) {
- mdbx_error("invalid nested-db record size %zu", node_ds(node));
- return MDBX_CORRUPTED;
- }
- memcpy(&mx->mx_db, node_data(node), sizeof(MDBX_db));
- const txnid_t pp_txnid = mp->mp_txnid;
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(mx->mx_db.md_mod_txnid > pp_txnid)) {
- mdbx_error("nested-db.md_mod_txnid (%" PRIaTXN ") > page-txnid (%" PRIaTXN
- ")",
- mx->mx_db.md_mod_txnid, pp_txnid);
- return MDBX_CORRUPTED;
- }
- mx->mx_cursor.mc_pg[0] = 0;
- mx->mx_cursor.mc_snum = 0;
- mx->mx_cursor.mc_top = 0;
- mx->mx_cursor.mc_flags = C_SUB | (mc->mc_flags & (C_COPYING | C_SKIPORD));
- break;
- case F_DUPDATA:
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(node_ds(node) <= PAGEHDRSZ)) {
- mdbx_error("invalid nested-page size %zu", node_ds(node));
- return MDBX_CORRUPTED;
- }
- MDBX_page *fp = node_data(node);
- mx->mx_db.md_depth = 1;
- mx->mx_db.md_branch_pages = 0;
- mx->mx_db.md_leaf_pages = 1;
- mx->mx_db.md_overflow_pages = 0;
- mx->mx_db.md_entries = page_numkeys(fp);
- mx->mx_db.md_root = fp->mp_pgno;
- mx->mx_db.md_mod_txnid = mp->mp_txnid;
- mx->mx_cursor.mc_snum = 1;
- mx->mx_cursor.mc_top = 0;
- mx->mx_cursor.mc_flags =
- C_INITIALIZED | C_SUB | (mc->mc_flags & (C_COPYING | C_SKIPORD));
- mx->mx_cursor.mc_pg[0] = fp;
- mx->mx_cursor.mc_ki[0] = 0;
- mx->mx_db.md_flags = flags_db2sub(mc->mc_db->md_flags);
- mx->mx_db.md_xsize =
- (mc->mc_db->md_flags & MDBX_DUPFIXED) ? fp->mp_leaf2_ksize : 0;
- break;
- }
-
- if (unlikely(mx->mx_db.md_xsize != mc->mc_db->md_xsize)) {
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(mc->mc_db->md_xsize != 0)) {
- mdbx_error("cursor mismatched nested-db md_xsize %u",
- mc->mc_db->md_xsize);
- return MDBX_CORRUPTED;
- }
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely((mc->mc_db->md_flags & MDBX_DUPFIXED) == 0)) {
- mdbx_error("mismatched nested-db md_flags %u", mc->mc_db->md_flags);
- return MDBX_CORRUPTED;
- }
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(mx->mx_db.md_xsize < mc->mc_dbx->md_vlen_min ||
- mx->mx_db.md_xsize > mc->mc_dbx->md_vlen_max)) {
- mdbx_error("mismatched nested-db.md_xsize (%u) <> min/max value-length "
- "(%zu/%zu)",
- mx->mx_db.md_xsize, mc->mc_dbx->md_vlen_min,
- mc->mc_dbx->md_vlen_max);
- return MDBX_CORRUPTED;
- }
- mc->mc_db->md_xsize = mx->mx_db.md_xsize;
- mc->mc_dbx->md_vlen_min = mc->mc_dbx->md_vlen_max = mx->mx_db.md_xsize;
- }
- mx->mx_dbx.md_klen_min = mc->mc_dbx->md_vlen_min;
- mx->mx_dbx.md_klen_max = mc->mc_dbx->md_vlen_max;
-
- mdbx_debug("Sub-db -%u root page %" PRIaPGNO, mx->mx_cursor.mc_dbi,
- mx->mx_db.md_root);
- return MDBX_SUCCESS;
-}
-
-/* Fixup a sorted-dups cursor due to underlying update.
- * Sets up some fields that depend on the data from the main cursor.
- * Almost the same as init1, but skips initialization steps if the
- * xcursor had already been used.
- * [in] mc The main cursor whose sorted-dups cursor is to be fixed up.
- * [in] src_mx The xcursor of an up-to-date cursor.
- * [in] new_dupdata True if converting from a non-F_DUPDATA item. */
-static int mdbx_xcursor_init2(MDBX_cursor *mc, MDBX_xcursor *src_mx,
- bool new_dupdata) {
- MDBX_xcursor *mx = mc->mc_xcursor;
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(mx == nullptr)) {
- mdbx_error("unexpected dupsort-page for non-dupsort db/cursor (dbi %u)",
- mc->mc_dbi);
- return MDBX_CORRUPTED;
- }
-
- if (new_dupdata) {
- mx->mx_cursor.mc_snum = 1;
- mx->mx_cursor.mc_top = 0;
- mx->mx_cursor.mc_flags |= C_INITIALIZED;
- mx->mx_cursor.mc_ki[0] = 0;
- }
-
- mx->mx_dbx.md_klen_min = src_mx->mx_dbx.md_klen_min;
- mx->mx_dbx.md_klen_max = src_mx->mx_dbx.md_klen_max;
- mx->mx_dbx.md_cmp = src_mx->mx_dbx.md_cmp;
- mx->mx_db = src_mx->mx_db;
- mx->mx_cursor.mc_pg[0] = src_mx->mx_cursor.mc_pg[0];
- if (mx->mx_cursor.mc_flags & C_INITIALIZED) {
- mdbx_debug("Sub-db -%u root page %" PRIaPGNO, mx->mx_cursor.mc_dbi,
- mx->mx_db.md_root);
- }
- return MDBX_SUCCESS;
-}
-
-static __inline int mdbx_couple_init(MDBX_cursor_couple *couple,
- const MDBX_dbi dbi, MDBX_txn *const txn,
- MDBX_db *const db, MDBX_dbx *const dbx,
- uint8_t *const dbstate) {
- couple->outer.mc_signature = MDBX_MC_LIVE;
- couple->outer.mc_next = NULL;
- couple->outer.mc_backup = NULL;
- couple->outer.mc_dbi = dbi;
- couple->outer.mc_txn = txn;
- couple->outer.mc_db = db;
- couple->outer.mc_dbx = dbx;
- couple->outer.mc_dbistate = dbstate;
- couple->outer.mc_snum = 0;
- couple->outer.mc_top = 0;
- couple->outer.mc_pg[0] = 0;
- couple->outer.mc_flags = 0;
- couple->outer.mc_ki[0] = 0;
- couple->outer.mc_xcursor = NULL;
-
- int rc = MDBX_SUCCESS;
- if (unlikely(*couple->outer.mc_dbistate & DBI_STALE)) {
- rc = mdbx_page_search(&couple->outer, NULL, MDBX_PS_ROOTONLY);
- rc = (rc != MDBX_NOTFOUND) ? rc : MDBX_SUCCESS;
- } else if (unlikely(couple->outer.mc_dbx->md_klen_max == 0)) {
- rc = mdbx_setup_dbx(couple->outer.mc_dbx, couple->outer.mc_db,
- txn->mt_env->me_psize);
- }
-
- if (couple->outer.mc_db->md_flags & MDBX_DUPSORT) {
- couple->inner.mx_cursor.mc_signature = MDBX_MC_LIVE;
- couple->outer.mc_xcursor = &couple->inner;
- rc = mdbx_xcursor_init0(&couple->outer);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- couple->inner.mx_dbx.md_klen_min = couple->outer.mc_dbx->md_vlen_min;
- couple->inner.mx_dbx.md_klen_max = couple->outer.mc_dbx->md_vlen_max;
- }
- return rc;
-}
-
-/* Initialize a cursor for a given transaction and database. */
-static int mdbx_cursor_init(MDBX_cursor *mc, MDBX_txn *txn, MDBX_dbi dbi) {
- STATIC_ASSERT(offsetof(MDBX_cursor_couple, outer) == 0);
- if (unlikely(TXN_DBI_CHANGED(txn, dbi)))
- return MDBX_BAD_DBI;
-
- return mdbx_couple_init(container_of(mc, MDBX_cursor_couple, outer), dbi, txn,
- &txn->mt_dbs[dbi], &txn->mt_dbxs[dbi],
- &txn->mt_dbistate[dbi]);
-}
-
-MDBX_cursor *mdbx_cursor_create(void *context) {
- MDBX_cursor_couple *couple = mdbx_calloc(1, sizeof(MDBX_cursor_couple));
- if (unlikely(!couple))
- return nullptr;
-
- couple->outer.mc_signature = MDBX_MC_READY4CLOSE;
- couple->outer.mc_dbi = UINT_MAX;
- couple->mc_userctx = context;
- return &couple->outer;
-}
-
-int mdbx_cursor_set_userctx(MDBX_cursor *mc, void *ctx) {
- if (unlikely(!mc))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_READY4CLOSE &&
- mc->mc_signature != MDBX_MC_LIVE))
- return MDBX_EBADSIGN;
-
- MDBX_cursor_couple *couple = container_of(mc, MDBX_cursor_couple, outer);
- couple->mc_userctx = ctx;
- return MDBX_SUCCESS;
-}
-
-void *mdbx_cursor_get_userctx(const MDBX_cursor *mc) {
- if (unlikely(!mc))
- return nullptr;
-
- if (unlikely(mc->mc_signature != MDBX_MC_READY4CLOSE &&
- mc->mc_signature != MDBX_MC_LIVE))
- return nullptr;
-
- MDBX_cursor_couple *couple = container_of(mc, MDBX_cursor_couple, outer);
- return couple->mc_userctx;
-}
-
-int mdbx_cursor_bind(MDBX_txn *txn, MDBX_cursor *mc, MDBX_dbi dbi) {
- if (unlikely(!mc))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_READY4CLOSE &&
- mc->mc_signature != MDBX_MC_LIVE))
- return MDBX_EBADSIGN;
-
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_VALID)))
- return MDBX_BAD_DBI;
-
- if (unlikely(dbi == FREE_DBI && !F_ISSET(txn->mt_flags, MDBX_TXN_RDONLY)))
- return MDBX_EACCESS;
-
- if (unlikely(mc->mc_backup)) /* Cursor from parent transaction */ {
- mdbx_cassert(mc, mc->mc_signature == MDBX_MC_LIVE);
- if (unlikely(mc->mc_dbi != dbi ||
- /* paranoia */ mc->mc_signature != MDBX_MC_LIVE ||
- mc->mc_txn != txn))
- return MDBX_EINVAL;
-
- assert(mc->mc_db == &txn->mt_dbs[dbi]);
- assert(mc->mc_dbx == &txn->mt_dbxs[dbi]);
- assert(mc->mc_dbi == dbi);
- assert(mc->mc_dbistate == &txn->mt_dbistate[dbi]);
- return likely(mc->mc_dbi == dbi &&
- /* paranoia */ mc->mc_signature == MDBX_MC_LIVE &&
- mc->mc_txn == txn)
- ? MDBX_SUCCESS
- : MDBX_EINVAL /* Disallow change DBI in nested transactions */;
- }
-
- if (mc->mc_signature == MDBX_MC_LIVE) {
- if (unlikely(!mc->mc_txn ||
- mc->mc_txn->mt_signature != MDBX_MT_SIGNATURE)) {
- mdbx_error("Wrong cursor's transaction %p 0x%x",
- __Wpedantic_format_voidptr(mc->mc_txn),
- mc->mc_txn ? mc->mc_txn->mt_signature : 0);
- return MDBX_PROBLEM;
- }
- if (mc->mc_flags & C_UNTRACK) {
- MDBX_cursor **prev = &mc->mc_txn->mt_cursors[mc->mc_dbi];
- while (*prev && *prev != mc)
- prev = &(*prev)->mc_next;
- mdbx_cassert(mc, *prev == mc);
- *prev = mc->mc_next;
- }
- mc->mc_signature = MDBX_MC_READY4CLOSE;
- mc->mc_flags = 0;
- mc->mc_dbi = UINT_MAX;
- mc->mc_next = NULL;
- mc->mc_db = NULL;
- mc->mc_dbx = NULL;
- mc->mc_dbistate = NULL;
- }
- mdbx_cassert(mc, !(mc->mc_flags & C_UNTRACK));
-
- rc = mdbx_cursor_init(mc, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- mc->mc_next = txn->mt_cursors[dbi];
- txn->mt_cursors[dbi] = mc;
- mc->mc_flags |= C_UNTRACK;
-
- return MDBX_SUCCESS;
-}
-
-int mdbx_cursor_open(MDBX_txn *txn, MDBX_dbi dbi, MDBX_cursor **ret) {
- if (unlikely(!ret))
- return MDBX_EINVAL;
- *ret = NULL;
-
- MDBX_cursor *const mc = mdbx_cursor_create(nullptr);
- if (unlikely(!mc))
- return MDBX_ENOMEM;
-
- int rc = mdbx_cursor_bind(txn, mc, dbi);
- if (unlikely(rc != MDBX_SUCCESS)) {
- mdbx_cursor_close(mc);
- return rc;
- }
-
- *ret = mc;
- return MDBX_SUCCESS;
-}
-
-int mdbx_cursor_renew(MDBX_txn *txn, MDBX_cursor *mc) {
- return likely(mc) ? mdbx_cursor_bind(txn, mc, mc->mc_dbi) : MDBX_EINVAL;
-}
-
-int mdbx_cursor_copy(const MDBX_cursor *src, MDBX_cursor *dest) {
- if (unlikely(!src))
- return MDBX_EINVAL;
- if (unlikely(src->mc_signature != MDBX_MC_LIVE))
- return (src->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- int rc = mdbx_cursor_bind(src->mc_txn, dest, src->mc_dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- assert(dest->mc_db == src->mc_db);
- assert(dest->mc_dbi == src->mc_dbi);
- assert(dest->mc_dbx == src->mc_dbx);
- assert(dest->mc_dbistate == src->mc_dbistate);
-again:
- assert(dest->mc_txn == src->mc_txn);
- dest->mc_flags ^= (dest->mc_flags ^ src->mc_flags) & ~C_UNTRACK;
- dest->mc_top = src->mc_top;
- dest->mc_snum = src->mc_snum;
- for (unsigned i = 0; i < src->mc_snum; ++i) {
- dest->mc_ki[i] = src->mc_ki[i];
- dest->mc_pg[i] = src->mc_pg[i];
- }
-
- if (src->mc_xcursor) {
- dest->mc_xcursor->mx_db = src->mc_xcursor->mx_db;
- dest->mc_xcursor->mx_dbx = src->mc_xcursor->mx_dbx;
- src = &src->mc_xcursor->mx_cursor;
- dest = &dest->mc_xcursor->mx_cursor;
- goto again;
- }
-
- return MDBX_SUCCESS;
-}
-
-void mdbx_cursor_close(MDBX_cursor *mc) {
- if (likely(mc)) {
- mdbx_ensure(NULL, mc->mc_signature == MDBX_MC_LIVE ||
- mc->mc_signature == MDBX_MC_READY4CLOSE);
- MDBX_txn *const txn = mc->mc_txn;
- if (!mc->mc_backup) {
- mc->mc_txn = NULL;
- /* Unlink from txn, if tracked. */
- if (mc->mc_flags & C_UNTRACK) {
- mdbx_ensure(txn->mt_env, check_txn(txn, 0) == MDBX_SUCCESS);
- MDBX_cursor **prev = &txn->mt_cursors[mc->mc_dbi];
- while (*prev && *prev != mc)
- prev = &(*prev)->mc_next;
- mdbx_tassert(txn, *prev == mc);
- *prev = mc->mc_next;
- }
- mc->mc_signature = 0;
- mc->mc_next = mc;
- mdbx_free(mc);
- } else {
- /* Cursor closed before nested txn ends */
- mdbx_tassert(txn, mc->mc_signature == MDBX_MC_LIVE);
- mdbx_ensure(txn->mt_env, check_txn_rw(txn, 0) == MDBX_SUCCESS);
- mc->mc_signature = MDBX_MC_WAIT4EOT;
- }
- }
-}
-
-MDBX_txn *mdbx_cursor_txn(const MDBX_cursor *mc) {
- if (unlikely(!mc || mc->mc_signature != MDBX_MC_LIVE))
- return NULL;
- MDBX_txn *txn = mc->mc_txn;
- if (unlikely(!txn || txn->mt_signature != MDBX_MT_SIGNATURE))
- return NULL;
- if (unlikely(txn->mt_flags & MDBX_TXN_FINISHED))
- return NULL;
- return txn;
-}
-
-MDBX_dbi mdbx_cursor_dbi(const MDBX_cursor *mc) {
- if (unlikely(!mc || mc->mc_signature != MDBX_MC_LIVE))
- return UINT_MAX;
- return mc->mc_dbi;
-}
-
-/* Return the count of duplicate data items for the current key */
-int mdbx_cursor_count(const MDBX_cursor *mc, size_t *countp) {
- if (unlikely(mc == NULL))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_LIVE))
- return (mc->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- int rc = check_txn(mc->mc_txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(countp == NULL || !(mc->mc_flags & C_INITIALIZED)))
- return MDBX_EINVAL;
-
- if (!mc->mc_snum) {
- *countp = 0;
- return MDBX_NOTFOUND;
- }
-
- MDBX_page *mp = mc->mc_pg[mc->mc_top];
- if ((mc->mc_flags & C_EOF) && mc->mc_ki[mc->mc_top] >= page_numkeys(mp)) {
- *countp = 0;
- return MDBX_NOTFOUND;
- }
-
- *countp = 1;
- if (mc->mc_xcursor != NULL) {
- MDBX_node *node = page_node(mp, mc->mc_ki[mc->mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- mdbx_cassert(mc, mc->mc_xcursor && (mc->mc_xcursor->mx_cursor.mc_flags &
- C_INITIALIZED));
- *countp = unlikely(mc->mc_xcursor->mx_db.md_entries > PTRDIFF_MAX)
- ? PTRDIFF_MAX
- : (size_t)mc->mc_xcursor->mx_db.md_entries;
- }
- }
- return MDBX_SUCCESS;
-}
-
-/* Replace the key for a branch node with a new key.
- * Set MDBX_TXN_ERROR on failure.
- * [in] mc Cursor pointing to the node to operate on.
- * [in] key The new key to use.
- * Returns 0 on success, non-zero on failure. */
-static int mdbx_update_key(MDBX_cursor *mc, const MDBX_val *key) {
- MDBX_page *mp;
- MDBX_node *node;
- char *base;
- size_t len;
- int delta, ksize, oksize;
- int ptr, i, nkeys, indx;
- DKBUF_DEBUG;
-
- mdbx_cassert(mc, cursor_is_tracked(mc));
- indx = mc->mc_ki[mc->mc_top];
- mp = mc->mc_pg[mc->mc_top];
- node = page_node(mp, indx);
- ptr = mp->mp_ptrs[indx];
-#if MDBX_DEBUG
- MDBX_val k2;
- k2.iov_base = node_key(node);
- k2.iov_len = node_ks(node);
- mdbx_debug("update key %u (offset %u) [%s] to [%s] on page %" PRIaPGNO, indx,
- ptr, DVAL_DEBUG(&k2), DKEY_DEBUG(key), mp->mp_pgno);
-#endif /* MDBX_DEBUG */
-
- /* Sizes must be 2-byte aligned. */
- ksize = EVEN(key->iov_len);
- oksize = EVEN(node_ks(node));
- delta = ksize - oksize;
-
- /* Shift node contents if EVEN(key length) changed. */
- if (delta) {
- if (delta > (int)page_room(mp)) {
- /* not enough space left, do a delete and split */
- mdbx_debug("Not enough room, delta = %d, splitting...", delta);
- pgno_t pgno = node_pgno(node);
- mdbx_node_del(mc, 0);
- int rc = mdbx_page_split(mc, key, NULL, pgno, MDBX_SPLIT_REPLACE);
- if (rc == MDBX_SUCCESS && mdbx_audit_enabled())
- rc = mdbx_cursor_check(mc, C_UPDATING);
- return rc;
- }
-
- nkeys = page_numkeys(mp);
- for (i = 0; i < nkeys; i++) {
- if (mp->mp_ptrs[i] <= ptr) {
- mdbx_cassert(mc, mp->mp_ptrs[i] >= delta);
- mp->mp_ptrs[i] -= (indx_t)delta;
- }
- }
-
- base = (char *)mp + mp->mp_upper + PAGEHDRSZ;
- len = ptr - mp->mp_upper + NODESIZE;
- memmove(base - delta, base, len);
- mdbx_cassert(mc, mp->mp_upper >= delta);
- mp->mp_upper -= (indx_t)delta;
-
- node = page_node(mp, indx);
- }
-
- /* But even if no shift was needed, update ksize */
- node_set_ks(node, key->iov_len);
-
- if (likely(key->iov_len /* to avoid UBSAN traps*/ != 0))
- memcpy(node_key(node), key->iov_base, key->iov_len);
- return MDBX_SUCCESS;
-}
-
-/* Move a node from csrc to cdst. */
-static int mdbx_node_move(MDBX_cursor *csrc, MDBX_cursor *cdst, bool fromleft) {
- int rc;
- DKBUF_DEBUG;
-
- MDBX_page *psrc = csrc->mc_pg[csrc->mc_top];
- MDBX_page *pdst = cdst->mc_pg[cdst->mc_top];
- mdbx_cassert(csrc, PAGETYPE(psrc) == PAGETYPE(pdst));
- mdbx_cassert(csrc, csrc->mc_dbi == cdst->mc_dbi);
- mdbx_cassert(csrc, csrc->mc_top == cdst->mc_top);
- if (unlikely(PAGETYPE(psrc) != PAGETYPE(pdst))) {
- bailout:
- mdbx_error("Wrong or mismatch pages's types (src %d, dst %d) to move node",
- PAGETYPE(psrc), PAGETYPE(pdst));
- csrc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return MDBX_PROBLEM;
- }
-
- MDBX_val key4move;
- switch (PAGETYPE(psrc)) {
- case P_BRANCH: {
- const MDBX_node *srcnode = page_node(psrc, csrc->mc_ki[csrc->mc_top]);
- mdbx_cassert(csrc, node_flags(srcnode) == 0);
- const pgno_t srcpg = node_pgno(srcnode);
- key4move.iov_len = node_ks(srcnode);
- key4move.iov_base = node_key(srcnode);
-
- if (csrc->mc_ki[csrc->mc_top] == 0) {
- const unsigned snum = csrc->mc_snum;
- mdbx_cassert(csrc, snum > 0);
- /* must find the lowest key below src */
- rc = mdbx_page_search_lowest(csrc);
- MDBX_page *lowest_page = csrc->mc_pg[csrc->mc_top];
- if (unlikely(rc))
- return rc;
- mdbx_cassert(csrc, IS_LEAF(lowest_page));
- if (unlikely(!IS_LEAF(lowest_page)))
- goto bailout;
- if (IS_LEAF2(lowest_page)) {
- key4move.iov_len = csrc->mc_db->md_xsize;
- key4move.iov_base = page_leaf2key(lowest_page, 0, key4move.iov_len);
- } else {
- const MDBX_node *lowest_node = page_node(lowest_page, 0);
- key4move.iov_len = node_ks(lowest_node);
- key4move.iov_base = node_key(lowest_node);
- }
-
- /* restore cursor after mdbx_page_search_lowest() */
- csrc->mc_snum = snum;
- csrc->mc_top = snum - 1;
- csrc->mc_ki[csrc->mc_top] = 0;
-
- /* paranoia */
- mdbx_cassert(csrc, psrc == csrc->mc_pg[csrc->mc_top]);
- mdbx_cassert(csrc, IS_BRANCH(psrc));
- if (unlikely(!IS_BRANCH(psrc)))
- goto bailout;
- }
-
- if (cdst->mc_ki[cdst->mc_top] == 0) {
- const unsigned snum = cdst->mc_snum;
- mdbx_cassert(csrc, snum > 0);
- MDBX_cursor mn;
- cursor_copy(cdst, &mn);
- /* must find the lowest key below dst */
- rc = mdbx_page_search_lowest(&mn);
- if (unlikely(rc))
- return rc;
- MDBX_page *const lowest_page = mn.mc_pg[mn.mc_top];
- mdbx_cassert(cdst, IS_LEAF(lowest_page));
- if (unlikely(!IS_LEAF(lowest_page)))
- goto bailout;
- MDBX_val key;
- if (IS_LEAF2(lowest_page)) {
- key.iov_len = mn.mc_db->md_xsize;
- key.iov_base = page_leaf2key(lowest_page, 0, key.iov_len);
- } else {
- MDBX_node *lowest_node = page_node(lowest_page, 0);
- key.iov_len = node_ks(lowest_node);
- key.iov_base = node_key(lowest_node);
- }
-
- /* restore cursor after mdbx_page_search_lowest() */
- mn.mc_snum = snum;
- mn.mc_top = snum - 1;
- mn.mc_ki[mn.mc_top] = 0;
-
- const intptr_t delta =
- EVEN(key.iov_len) - EVEN(node_ks(page_node(mn.mc_pg[mn.mc_top], 0)));
- const intptr_t needed =
- branch_size(cdst->mc_txn->mt_env, &key4move) + delta;
- const intptr_t have = page_room(pdst);
- if (unlikely(needed > have))
- return MDBX_RESULT_TRUE;
-
- if (unlikely((rc = mdbx_page_touch(csrc)) ||
- (rc = mdbx_page_touch(cdst))))
- return rc;
- psrc = csrc->mc_pg[csrc->mc_top];
- pdst = cdst->mc_pg[cdst->mc_top];
-
- WITH_CURSOR_TRACKING(mn, rc = mdbx_update_key(&mn, &key));
- if (unlikely(rc))
- return rc;
- } else {
- const size_t needed = branch_size(cdst->mc_txn->mt_env, &key4move);
- const size_t have = page_room(pdst);
- if (unlikely(needed > have))
- return MDBX_RESULT_TRUE;
-
- if (unlikely((rc = mdbx_page_touch(csrc)) ||
- (rc = mdbx_page_touch(cdst))))
- return rc;
- psrc = csrc->mc_pg[csrc->mc_top];
- pdst = cdst->mc_pg[cdst->mc_top];
- }
-
- mdbx_debug("moving %s-node %u [%s] on page %" PRIaPGNO
- " to node %u on page %" PRIaPGNO,
- "branch", csrc->mc_ki[csrc->mc_top], DKEY_DEBUG(&key4move),
- psrc->mp_pgno, cdst->mc_ki[cdst->mc_top], pdst->mp_pgno);
- /* Add the node to the destination page. */
- rc =
- mdbx_node_add_branch(cdst, cdst->mc_ki[cdst->mc_top], &key4move, srcpg);
- } break;
-
- case P_LEAF: {
- /* Mark src and dst as dirty. */
- if (unlikely((rc = mdbx_page_touch(csrc)) || (rc = mdbx_page_touch(cdst))))
- return rc;
- psrc = csrc->mc_pg[csrc->mc_top];
- pdst = cdst->mc_pg[cdst->mc_top];
- const MDBX_node *srcnode = page_node(psrc, csrc->mc_ki[csrc->mc_top]);
- MDBX_val data;
- data.iov_len = node_ds(srcnode);
- data.iov_base = node_data(srcnode);
- key4move.iov_len = node_ks(srcnode);
- key4move.iov_base = node_key(srcnode);
- mdbx_debug("moving %s-node %u [%s] on page %" PRIaPGNO
- " to node %u on page %" PRIaPGNO,
- "leaf", csrc->mc_ki[csrc->mc_top], DKEY_DEBUG(&key4move),
- psrc->mp_pgno, cdst->mc_ki[cdst->mc_top], pdst->mp_pgno);
- /* Add the node to the destination page. */
- rc = mdbx_node_add_leaf(cdst, cdst->mc_ki[cdst->mc_top], &key4move, &data,
- node_flags(srcnode));
- } break;
-
- case P_LEAF | P_LEAF2: {
- /* Mark src and dst as dirty. */
- if (unlikely((rc = mdbx_page_touch(csrc)) || (rc = mdbx_page_touch(cdst))))
- return rc;
- psrc = csrc->mc_pg[csrc->mc_top];
- pdst = cdst->mc_pg[cdst->mc_top];
- key4move.iov_len = csrc->mc_db->md_xsize;
- key4move.iov_base =
- page_leaf2key(psrc, csrc->mc_ki[csrc->mc_top], key4move.iov_len);
- mdbx_debug("moving %s-node %u [%s] on page %" PRIaPGNO
- " to node %u on page %" PRIaPGNO,
- "leaf2", csrc->mc_ki[csrc->mc_top], DKEY_DEBUG(&key4move),
- psrc->mp_pgno, cdst->mc_ki[cdst->mc_top], pdst->mp_pgno);
- /* Add the node to the destination page. */
- rc = mdbx_node_add_leaf2(cdst, cdst->mc_ki[cdst->mc_top], &key4move);
- } break;
-
- default:
- goto bailout;
- }
-
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- /* Delete the node from the source page. */
- mdbx_node_del(csrc, key4move.iov_len);
-
- mdbx_cassert(csrc, psrc == csrc->mc_pg[csrc->mc_top]);
- mdbx_cassert(cdst, pdst == cdst->mc_pg[cdst->mc_top]);
- mdbx_cassert(csrc, PAGETYPE(psrc) == PAGETYPE(pdst));
-
- {
- /* Adjust other cursors pointing to mp */
- MDBX_cursor *m2, *m3;
- const MDBX_dbi dbi = csrc->mc_dbi;
- mdbx_cassert(csrc, csrc->mc_top == cdst->mc_top);
- if (fromleft) {
- /* If we're adding on the left, bump others up */
- for (m2 = csrc->mc_txn->mt_cursors[dbi]; m2; m2 = m2->mc_next) {
- m3 = (csrc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (!(m3->mc_flags & C_INITIALIZED) || m3->mc_top < csrc->mc_top)
- continue;
- if (m3 != cdst && m3->mc_pg[csrc->mc_top] == pdst &&
- m3->mc_ki[csrc->mc_top] >= cdst->mc_ki[csrc->mc_top]) {
- m3->mc_ki[csrc->mc_top]++;
- }
- if (m3 != csrc && m3->mc_pg[csrc->mc_top] == psrc &&
- m3->mc_ki[csrc->mc_top] == csrc->mc_ki[csrc->mc_top]) {
- m3->mc_pg[csrc->mc_top] = pdst;
- m3->mc_ki[csrc->mc_top] = cdst->mc_ki[cdst->mc_top];
- mdbx_cassert(csrc, csrc->mc_top > 0);
- m3->mc_ki[csrc->mc_top - 1]++;
- }
- if (XCURSOR_INITED(m3) && IS_LEAF(psrc))
- XCURSOR_REFRESH(m3, m3->mc_pg[csrc->mc_top], m3->mc_ki[csrc->mc_top]);
- }
- } else {
- /* Adding on the right, bump others down */
- for (m2 = csrc->mc_txn->mt_cursors[dbi]; m2; m2 = m2->mc_next) {
- m3 = (csrc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (m3 == csrc)
- continue;
- if (!(m3->mc_flags & C_INITIALIZED) || m3->mc_top < csrc->mc_top)
- continue;
- if (m3->mc_pg[csrc->mc_top] == psrc) {
- if (!m3->mc_ki[csrc->mc_top]) {
- m3->mc_pg[csrc->mc_top] = pdst;
- m3->mc_ki[csrc->mc_top] = cdst->mc_ki[cdst->mc_top];
- mdbx_cassert(csrc, csrc->mc_top > 0);
- m3->mc_ki[csrc->mc_top - 1]--;
- } else {
- m3->mc_ki[csrc->mc_top]--;
- }
- if (XCURSOR_INITED(m3) && IS_LEAF(psrc))
- XCURSOR_REFRESH(m3, m3->mc_pg[csrc->mc_top],
- m3->mc_ki[csrc->mc_top]);
- }
- }
- }
- }
-
- /* Update the parent separators. */
- if (csrc->mc_ki[csrc->mc_top] == 0) {
- mdbx_cassert(csrc, csrc->mc_top > 0);
- if (csrc->mc_ki[csrc->mc_top - 1] != 0) {
- MDBX_val key;
- if (IS_LEAF2(psrc)) {
- key.iov_len = psrc->mp_leaf2_ksize;
- key.iov_base = page_leaf2key(psrc, 0, key.iov_len);
- } else {
- MDBX_node *srcnode = page_node(psrc, 0);
- key.iov_len = node_ks(srcnode);
- key.iov_base = node_key(srcnode);
- }
- mdbx_debug("update separator for source page %" PRIaPGNO " to [%s]",
- psrc->mp_pgno, DKEY_DEBUG(&key));
- MDBX_cursor mn;
- cursor_copy(csrc, &mn);
- mdbx_cassert(csrc, mn.mc_snum > 0);
- mn.mc_snum--;
- mn.mc_top--;
- /* We want mdbx_rebalance to find mn when doing fixups */
- WITH_CURSOR_TRACKING(mn, rc = mdbx_update_key(&mn, &key));
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
- if (IS_BRANCH(psrc)) {
- const MDBX_val nullkey = {0, 0};
- const indx_t ix = csrc->mc_ki[csrc->mc_top];
- csrc->mc_ki[csrc->mc_top] = 0;
- rc = mdbx_update_key(csrc, &nullkey);
- csrc->mc_ki[csrc->mc_top] = ix;
- mdbx_cassert(csrc, rc == MDBX_SUCCESS);
- }
- }
-
- if (cdst->mc_ki[cdst->mc_top] == 0) {
- mdbx_cassert(cdst, cdst->mc_top > 0);
- if (cdst->mc_ki[cdst->mc_top - 1] != 0) {
- MDBX_val key;
- if (IS_LEAF2(pdst)) {
- key.iov_len = pdst->mp_leaf2_ksize;
- key.iov_base = page_leaf2key(pdst, 0, key.iov_len);
- } else {
- MDBX_node *srcnode = page_node(pdst, 0);
- key.iov_len = node_ks(srcnode);
- key.iov_base = node_key(srcnode);
- }
- mdbx_debug("update separator for destination page %" PRIaPGNO " to [%s]",
- pdst->mp_pgno, DKEY_DEBUG(&key));
- MDBX_cursor mn;
- cursor_copy(cdst, &mn);
- mdbx_cassert(cdst, mn.mc_snum > 0);
- mn.mc_snum--;
- mn.mc_top--;
- /* We want mdbx_rebalance to find mn when doing fixups */
- WITH_CURSOR_TRACKING(mn, rc = mdbx_update_key(&mn, &key));
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
- if (IS_BRANCH(pdst)) {
- const MDBX_val nullkey = {0, 0};
- const indx_t ix = cdst->mc_ki[cdst->mc_top];
- cdst->mc_ki[cdst->mc_top] = 0;
- rc = mdbx_update_key(cdst, &nullkey);
- cdst->mc_ki[cdst->mc_top] = ix;
- mdbx_cassert(cdst, rc == MDBX_SUCCESS);
- }
- }
-
- return MDBX_SUCCESS;
-}
-
-/* Merge one page into another.
- *
- * The nodes from the page pointed to by csrc will be copied to the page
- * pointed to by cdst and then the csrc page will be freed.
- *
- * [in] csrc Cursor pointing to the source page.
- * [in] cdst Cursor pointing to the destination page.
- *
- * Returns 0 on success, non-zero on failure. */
-static int mdbx_page_merge(MDBX_cursor *csrc, MDBX_cursor *cdst) {
- MDBX_val key;
- int rc;
-
- mdbx_cassert(csrc, csrc != cdst);
- mdbx_cassert(csrc, cursor_is_tracked(csrc));
- mdbx_cassert(cdst, cursor_is_tracked(cdst));
- const MDBX_page *const psrc = csrc->mc_pg[csrc->mc_top];
- MDBX_page *pdst = cdst->mc_pg[cdst->mc_top];
- mdbx_debug("merging page %" PRIaPGNO " into %" PRIaPGNO, psrc->mp_pgno,
- pdst->mp_pgno);
-
- mdbx_cassert(csrc, PAGETYPE(psrc) == PAGETYPE(pdst));
- mdbx_cassert(csrc,
- csrc->mc_dbi == cdst->mc_dbi && csrc->mc_db == cdst->mc_db);
- mdbx_cassert(csrc, csrc->mc_snum > 1); /* can't merge root page */
- mdbx_cassert(cdst, cdst->mc_snum > 1);
- mdbx_cassert(cdst, cdst->mc_snum < cdst->mc_db->md_depth ||
- IS_LEAF(cdst->mc_pg[cdst->mc_db->md_depth - 1]));
- mdbx_cassert(csrc, csrc->mc_snum < csrc->mc_db->md_depth ||
- IS_LEAF(csrc->mc_pg[csrc->mc_db->md_depth - 1]));
- mdbx_cassert(cdst, page_room(pdst) >= page_used(cdst->mc_txn->mt_env, psrc));
- const int pagetype = PAGETYPE(psrc);
-
- /* Move all nodes from src to dst */
- const unsigned dst_nkeys = page_numkeys(pdst);
- const unsigned src_nkeys = page_numkeys(psrc);
- mdbx_cassert(cdst, dst_nkeys + src_nkeys >= (IS_LEAF(psrc) ? 1u : 2u));
- if (likely(src_nkeys)) {
- unsigned j = dst_nkeys;
- if (unlikely(pagetype & P_LEAF2)) {
- /* Mark dst as dirty. */
- if (unlikely(rc = mdbx_page_touch(cdst)))
- return rc;
-
- key.iov_len = csrc->mc_db->md_xsize;
- key.iov_base = page_data(psrc);
- unsigned i = 0;
- do {
- rc = mdbx_node_add_leaf2(cdst, j++, &key);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- key.iov_base = (char *)key.iov_base + key.iov_len;
- } while (++i != src_nkeys);
- } else {
- MDBX_node *srcnode = page_node(psrc, 0);
- key.iov_len = node_ks(srcnode);
- key.iov_base = node_key(srcnode);
- if (pagetype & P_BRANCH) {
- MDBX_cursor mn;
- cursor_copy(csrc, &mn);
- /* must find the lowest key below src */
- rc = mdbx_page_search_lowest(&mn);
- if (unlikely(rc))
- return rc;
-
- const MDBX_page *mp = mn.mc_pg[mn.mc_top];
- if (likely(!IS_LEAF2(mp))) {
- mdbx_cassert(&mn, IS_LEAF(mp));
- const MDBX_node *lowest = page_node(mp, 0);
- key.iov_len = node_ks(lowest);
- key.iov_base = node_key(lowest);
- } else {
- mdbx_cassert(&mn, mn.mc_top > csrc->mc_top);
- key.iov_len = mp->mp_leaf2_ksize;
- key.iov_base = page_leaf2key(mp, mn.mc_ki[mn.mc_top], key.iov_len);
- }
- mdbx_cassert(&mn, key.iov_len >= csrc->mc_dbx->md_klen_min);
- mdbx_cassert(&mn, key.iov_len <= csrc->mc_dbx->md_klen_max);
-
- const size_t dst_room = page_room(pdst);
- const size_t src_used = page_used(cdst->mc_txn->mt_env, psrc);
- const size_t space_needed = src_used - node_ks(srcnode) + key.iov_len;
- if (unlikely(space_needed > dst_room))
- return MDBX_RESULT_TRUE;
- }
-
- /* Mark dst as dirty. */
- if (unlikely(rc = mdbx_page_touch(cdst)))
- return rc;
-
- unsigned i = 0;
- while (true) {
- if (pagetype & P_LEAF) {
- MDBX_val data;
- data.iov_len = node_ds(srcnode);
- data.iov_base = node_data(srcnode);
- rc = mdbx_node_add_leaf(cdst, j++, &key, &data, node_flags(srcnode));
- } else {
- mdbx_cassert(csrc, node_flags(srcnode) == 0);
- rc = mdbx_node_add_branch(cdst, j++, &key, node_pgno(srcnode));
- }
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (++i == src_nkeys)
- break;
- srcnode = page_node(psrc, i);
- key.iov_len = node_ks(srcnode);
- key.iov_base = node_key(srcnode);
- }
- }
-
- pdst = cdst->mc_pg[cdst->mc_top];
- mdbx_debug("dst page %" PRIaPGNO " now has %u keys (%.1f%% filled)",
- pdst->mp_pgno, page_numkeys(pdst),
- page_fill(cdst->mc_txn->mt_env, pdst));
-
- mdbx_cassert(csrc, psrc == csrc->mc_pg[csrc->mc_top]);
- mdbx_cassert(cdst, pdst == cdst->mc_pg[cdst->mc_top]);
- }
-
- /* Unlink the src page from parent and add to free list. */
- csrc->mc_top--;
- mdbx_node_del(csrc, 0);
- if (csrc->mc_ki[csrc->mc_top] == 0) {
- const MDBX_val nullkey = {0, 0};
- rc = mdbx_update_key(csrc, &nullkey);
- if (unlikely(rc)) {
- csrc->mc_top++;
- return rc;
- }
- }
- csrc->mc_top++;
-
- mdbx_cassert(csrc, psrc == csrc->mc_pg[csrc->mc_top]);
- mdbx_cassert(cdst, pdst == cdst->mc_pg[cdst->mc_top]);
-
- {
- /* Adjust other cursors pointing to mp */
- MDBX_cursor *m2, *m3;
- const MDBX_dbi dbi = csrc->mc_dbi;
- const unsigned top = csrc->mc_top;
-
- for (m2 = csrc->mc_txn->mt_cursors[dbi]; m2; m2 = m2->mc_next) {
- m3 = (csrc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (m3 == csrc || top >= m3->mc_snum)
- continue;
- if (m3->mc_pg[top] == psrc) {
- m3->mc_pg[top] = pdst;
- mdbx_cassert(m3, dst_nkeys + m3->mc_ki[top] <= UINT16_MAX);
- m3->mc_ki[top] += (indx_t)dst_nkeys;
- m3->mc_ki[top - 1] = cdst->mc_ki[top - 1];
- } else if (m3->mc_pg[top - 1] == csrc->mc_pg[top - 1] &&
- m3->mc_ki[top - 1] > csrc->mc_ki[top - 1]) {
- m3->mc_ki[top - 1]--;
- }
- if (XCURSOR_INITED(m3) && IS_LEAF(psrc))
- XCURSOR_REFRESH(m3, m3->mc_pg[top], m3->mc_ki[top]);
- }
- }
-
- /* If not operating on GC, allow this page to be reused
- * in this txn. Otherwise just add to free list. */
- rc = mdbx_page_retire(csrc, (MDBX_page *)psrc);
- if (unlikely(rc))
- return rc;
-
- mdbx_cassert(cdst, cdst->mc_db->md_entries > 0);
- mdbx_cassert(cdst, cdst->mc_snum <= cdst->mc_db->md_depth);
- mdbx_cassert(cdst, cdst->mc_top > 0);
- mdbx_cassert(cdst, cdst->mc_snum == cdst->mc_top + 1);
- MDBX_page *const top_page = cdst->mc_pg[cdst->mc_top];
- const indx_t top_indx = cdst->mc_ki[cdst->mc_top];
- const unsigned save_snum = cdst->mc_snum;
- const uint16_t save_depth = cdst->mc_db->md_depth;
- mdbx_cursor_pop(cdst);
- rc = mdbx_rebalance(cdst);
- if (unlikely(rc))
- return rc;
-
- mdbx_cassert(cdst, cdst->mc_db->md_entries > 0);
- mdbx_cassert(cdst, cdst->mc_snum <= cdst->mc_db->md_depth);
- mdbx_cassert(cdst, cdst->mc_snum == cdst->mc_top + 1);
-
-#if MDBX_ENABLE_PGOP_STAT
- cdst->mc_txn->mt_env->me_lck->mti_pgop_stat.merge.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
- if (IS_LEAF(cdst->mc_pg[cdst->mc_top])) {
- /* LY: don't touch cursor if top-page is a LEAF */
- mdbx_cassert(cdst, IS_LEAF(cdst->mc_pg[cdst->mc_top]) ||
- PAGETYPE(cdst->mc_pg[cdst->mc_top]) == pagetype);
- return MDBX_SUCCESS;
- }
-
- mdbx_cassert(cdst, page_numkeys(top_page) == dst_nkeys + src_nkeys);
-
- if (unlikely(pagetype != PAGETYPE(top_page))) {
- /* LY: LEAF-page becomes BRANCH, unable restore cursor's stack */
- goto bailout;
- }
-
- if (top_page == cdst->mc_pg[cdst->mc_top]) {
- /* LY: don't touch cursor if prev top-page already on the top */
- mdbx_cassert(cdst, cdst->mc_ki[cdst->mc_top] == top_indx);
- mdbx_cassert(cdst, IS_LEAF(cdst->mc_pg[cdst->mc_top]) ||
- PAGETYPE(cdst->mc_pg[cdst->mc_top]) == pagetype);
- return MDBX_SUCCESS;
- }
-
- const int new_snum = save_snum - save_depth + cdst->mc_db->md_depth;
- if (unlikely(new_snum < 1 || new_snum > cdst->mc_db->md_depth)) {
- /* LY: out of range, unable restore cursor's stack */
- goto bailout;
- }
-
- if (top_page == cdst->mc_pg[new_snum - 1]) {
- mdbx_cassert(cdst, cdst->mc_ki[new_snum - 1] == top_indx);
- /* LY: restore cursor stack */
- cdst->mc_snum = (uint16_t)new_snum;
- cdst->mc_top = (uint16_t)new_snum - 1;
- mdbx_cassert(cdst, cdst->mc_snum < cdst->mc_db->md_depth ||
- IS_LEAF(cdst->mc_pg[cdst->mc_db->md_depth - 1]));
- mdbx_cassert(cdst, IS_LEAF(cdst->mc_pg[cdst->mc_top]) ||
- PAGETYPE(cdst->mc_pg[cdst->mc_top]) == pagetype);
- return MDBX_SUCCESS;
- }
-
- MDBX_page *const stub_page = (MDBX_page *)(~(uintptr_t)top_page);
- const indx_t stub_indx = top_indx;
- if (save_depth > cdst->mc_db->md_depth &&
- ((cdst->mc_pg[save_snum - 1] == top_page &&
- cdst->mc_ki[save_snum - 1] == top_indx) ||
- (cdst->mc_pg[save_snum - 1] == stub_page &&
- cdst->mc_ki[save_snum - 1] == stub_indx))) {
- /* LY: restore cursor stack */
- cdst->mc_pg[new_snum - 1] = top_page;
- cdst->mc_ki[new_snum - 1] = top_indx;
- cdst->mc_pg[new_snum] = (MDBX_page *)(~(uintptr_t)cdst->mc_pg[new_snum]);
- cdst->mc_ki[new_snum] = ~cdst->mc_ki[new_snum];
- cdst->mc_snum = (uint16_t)new_snum;
- cdst->mc_top = (uint16_t)new_snum - 1;
- mdbx_cassert(cdst, cdst->mc_snum < cdst->mc_db->md_depth ||
- IS_LEAF(cdst->mc_pg[cdst->mc_db->md_depth - 1]));
- mdbx_cassert(cdst, IS_LEAF(cdst->mc_pg[cdst->mc_top]) ||
- PAGETYPE(cdst->mc_pg[cdst->mc_top]) == pagetype);
- return MDBX_SUCCESS;
- }
-
-bailout:
- /* LY: unable restore cursor's stack */
- cdst->mc_flags &= ~C_INITIALIZED;
- return MDBX_CURSOR_FULL;
-}
-
-static void cursor_restore(const MDBX_cursor *csrc, MDBX_cursor *cdst) {
- mdbx_cassert(cdst, cdst->mc_dbi == csrc->mc_dbi);
- mdbx_cassert(cdst, cdst->mc_txn == csrc->mc_txn);
- mdbx_cassert(cdst, cdst->mc_db == csrc->mc_db);
- mdbx_cassert(cdst, cdst->mc_dbx == csrc->mc_dbx);
- mdbx_cassert(cdst, cdst->mc_dbistate == csrc->mc_dbistate);
- cdst->mc_snum = csrc->mc_snum;
- cdst->mc_top = csrc->mc_top;
- cdst->mc_flags = csrc->mc_flags;
-
- for (unsigned i = 0; i < csrc->mc_snum; i++) {
- cdst->mc_pg[i] = csrc->mc_pg[i];
- cdst->mc_ki[i] = csrc->mc_ki[i];
- }
-}
-
-/* Copy the contents of a cursor.
- * [in] csrc The cursor to copy from.
- * [out] cdst The cursor to copy to. */
-static void cursor_copy(const MDBX_cursor *csrc, MDBX_cursor *cdst) {
- mdbx_cassert(csrc, csrc->mc_txn->mt_txnid >=
- csrc->mc_txn->mt_env->me_lck->mti_oldest_reader.weak);
- cdst->mc_dbi = csrc->mc_dbi;
- cdst->mc_next = NULL;
- cdst->mc_backup = NULL;
- cdst->mc_xcursor = NULL;
- cdst->mc_txn = csrc->mc_txn;
- cdst->mc_db = csrc->mc_db;
- cdst->mc_dbx = csrc->mc_dbx;
- cdst->mc_dbistate = csrc->mc_dbistate;
- cursor_restore(csrc, cdst);
-}
-
-/* Rebalance the tree after a delete operation.
- * [in] mc Cursor pointing to the page where rebalancing should begin.
- * Returns 0 on success, non-zero on failure. */
-static int mdbx_rebalance(MDBX_cursor *mc) {
- mdbx_cassert(mc, cursor_is_tracked(mc));
- mdbx_cassert(mc, mc->mc_snum > 0);
- mdbx_cassert(mc, mc->mc_snum < mc->mc_db->md_depth ||
- IS_LEAF(mc->mc_pg[mc->mc_db->md_depth - 1]));
- const int pagetype = PAGETYPE(mc->mc_pg[mc->mc_top]);
-
- STATIC_ASSERT(P_BRANCH == 1);
- const unsigned minkeys = (pagetype & P_BRANCH) + 1;
-
- /* Pages emptier than this are candidates for merging. */
- unsigned room_threshold = likely(mc->mc_dbi != FREE_DBI)
- ? mc->mc_txn->mt_env->me_merge_threshold
- : mc->mc_txn->mt_env->me_merge_threshold_gc;
-
- const MDBX_page *const tp = mc->mc_pg[mc->mc_top];
- const unsigned numkeys = page_numkeys(tp);
- const unsigned room = page_room(tp);
- mdbx_debug("rebalancing %s page %" PRIaPGNO
- " (has %u keys, full %.1f%%, used %u, room %u bytes )",
- (pagetype & P_LEAF) ? "leaf" : "branch", tp->mp_pgno, numkeys,
- page_fill(mc->mc_txn->mt_env, tp),
- page_used(mc->mc_txn->mt_env, tp), room);
-
- if (unlikely(numkeys < minkeys)) {
- mdbx_debug("page %" PRIaPGNO " must be merged due keys < %u threshold",
- tp->mp_pgno, minkeys);
- } else if (unlikely(room > room_threshold)) {
- mdbx_debug("page %" PRIaPGNO " should be merged due room %u > %u threshold",
- tp->mp_pgno, room, room_threshold);
- } else {
- mdbx_debug("no need to rebalance page %" PRIaPGNO
- ", room %u < %u threshold",
- tp->mp_pgno, room, room_threshold);
- mdbx_cassert(mc, mc->mc_db->md_entries > 0);
- return MDBX_SUCCESS;
- }
-
- int rc;
- if (mc->mc_snum < 2) {
- MDBX_page *const mp = mc->mc_pg[0];
- const unsigned nkeys = page_numkeys(mp);
- mdbx_cassert(mc, (mc->mc_db->md_entries == 0) == (nkeys == 0));
- if (IS_SUBP(mp)) {
- mdbx_debug("%s", "Can't rebalance a subpage, ignoring");
- mdbx_cassert(mc, pagetype & P_LEAF);
- return MDBX_SUCCESS;
- }
- if (nkeys == 0) {
- mdbx_cassert(mc, IS_LEAF(mp));
- mdbx_debug("%s", "tree is completely empty");
- mdbx_cassert(mc, (*mc->mc_dbistate & DBI_DIRTY) != 0);
- mc->mc_db->md_root = P_INVALID;
- mc->mc_db->md_depth = 0;
- mdbx_cassert(mc, mc->mc_db->md_branch_pages == 0 &&
- mc->mc_db->md_overflow_pages == 0 &&
- mc->mc_db->md_leaf_pages == 1);
- /* Adjust cursors pointing to mp */
- for (MDBX_cursor *m2 = mc->mc_txn->mt_cursors[mc->mc_dbi]; m2;
- m2 = m2->mc_next) {
- MDBX_cursor *m3 =
- (mc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (m3 == mc || !(m3->mc_flags & C_INITIALIZED))
- continue;
- if (m3->mc_pg[0] == mp) {
- m3->mc_snum = 0;
- m3->mc_top = 0;
- m3->mc_flags &= ~C_INITIALIZED;
- }
- }
- mc->mc_snum = 0;
- mc->mc_top = 0;
- mc->mc_flags &= ~C_INITIALIZED;
-
- rc = mdbx_page_retire(mc, mp);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- } else if (IS_BRANCH(mp) && nkeys == 1) {
- mdbx_debug("%s", "collapsing root page!");
- mc->mc_db->md_root = node_pgno(page_node(mp, 0));
- rc = mdbx_page_get(mc, mc->mc_db->md_root, &mc->mc_pg[0],
- pp_txnid4chk(mp, mc->mc_txn));
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- mc->mc_db->md_depth--;
- mc->mc_ki[0] = mc->mc_ki[1];
- for (int i = 1; i < mc->mc_db->md_depth; i++) {
- mc->mc_pg[i] = mc->mc_pg[i + 1];
- mc->mc_ki[i] = mc->mc_ki[i + 1];
- }
-
- /* Adjust other cursors pointing to mp */
- for (MDBX_cursor *m2 = mc->mc_txn->mt_cursors[mc->mc_dbi]; m2;
- m2 = m2->mc_next) {
- MDBX_cursor *m3 =
- (mc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (m3 == mc || !(m3->mc_flags & C_INITIALIZED))
- continue;
- if (m3->mc_pg[0] == mp) {
- for (int i = 0; i < mc->mc_db->md_depth; i++) {
- m3->mc_pg[i] = m3->mc_pg[i + 1];
- m3->mc_ki[i] = m3->mc_ki[i + 1];
- }
- m3->mc_snum--;
- m3->mc_top--;
- }
- }
- mdbx_cassert(mc, IS_LEAF(mc->mc_pg[mc->mc_top]) ||
- PAGETYPE(mc->mc_pg[mc->mc_top]) == pagetype);
- mdbx_cassert(mc, mc->mc_snum < mc->mc_db->md_depth ||
- IS_LEAF(mc->mc_pg[mc->mc_db->md_depth - 1]));
-
- rc = mdbx_page_retire(mc, mp);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- } else {
- mdbx_debug("root page %" PRIaPGNO
- " doesn't need rebalancing (flags 0x%x)",
- mp->mp_pgno, mp->mp_flags);
- }
- return MDBX_SUCCESS;
- }
-
- /* The parent (branch page) must have at least 2 pointers,
- * otherwise the tree is invalid. */
- const unsigned pre_top = mc->mc_top - 1;
- mdbx_cassert(mc, IS_BRANCH(mc->mc_pg[pre_top]));
- mdbx_cassert(mc, !IS_SUBP(mc->mc_pg[0]));
- mdbx_cassert(mc, page_numkeys(mc->mc_pg[pre_top]) > 1);
-
- /* Leaf page fill factor is below the threshold.
- * Try to move keys from left or right neighbor, or
- * merge with a neighbor page. */
-
- /* Find neighbors. */
- MDBX_cursor mn;
- cursor_copy(mc, &mn);
-
- MDBX_page *left = nullptr, *right = nullptr;
- if (mn.mc_ki[pre_top] > 0) {
- rc = mdbx_page_get(
- &mn, node_pgno(page_node(mn.mc_pg[pre_top], mn.mc_ki[pre_top] - 1)),
- &left, pp_txnid4chk(mn.mc_pg[pre_top], mc->mc_txn));
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- mdbx_cassert(mc, PAGETYPE(left) == PAGETYPE(mc->mc_pg[mc->mc_top]));
- }
- if (mn.mc_ki[pre_top] + 1u < page_numkeys(mn.mc_pg[pre_top])) {
- rc = mdbx_page_get(
- &mn, node_pgno(page_node(mn.mc_pg[pre_top], mn.mc_ki[pre_top] + 1)),
- &right, pp_txnid4chk(mn.mc_pg[pre_top], mc->mc_txn));
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- mdbx_cassert(mc, PAGETYPE(right) == PAGETYPE(mc->mc_pg[mc->mc_top]));
- }
- mdbx_cassert(mc, left || right);
-
- const unsigned ki_top = mc->mc_ki[mc->mc_top];
- const unsigned ki_pre_top = mn.mc_ki[pre_top];
- const unsigned nkeys = page_numkeys(mn.mc_pg[mn.mc_top]);
-
- const unsigned left_room = left ? page_room(left) : 0;
- const unsigned right_room = right ? page_room(right) : 0;
- const unsigned left_nkeys = left ? page_numkeys(left) : 0;
- const unsigned right_nkeys = right ? page_numkeys(right) : 0;
-retry:
- if (left_room > room_threshold && left_room >= right_room) {
- /* try merge with left */
- mdbx_cassert(mc, left_nkeys >= minkeys);
- mn.mc_pg[mn.mc_top] = left;
- mn.mc_ki[mn.mc_top - 1] = (indx_t)(ki_pre_top - 1);
- mn.mc_ki[mn.mc_top] = (indx_t)(left_nkeys - 1);
- mc->mc_ki[mc->mc_top] = 0;
- const unsigned new_ki = ki_top + left_nkeys;
- mn.mc_ki[mn.mc_top] += mc->mc_ki[mn.mc_top] + 1;
- /* We want mdbx_rebalance to find mn when doing fixups */
- WITH_CURSOR_TRACKING(mn, rc = mdbx_page_merge(mc, &mn));
- if (likely(rc != MDBX_RESULT_TRUE)) {
- cursor_restore(&mn, mc);
- mc->mc_ki[mc->mc_top] = (indx_t)new_ki;
- mdbx_cassert(mc, rc || page_numkeys(mc->mc_pg[mc->mc_top]) >= minkeys);
- return rc;
- }
- }
- if (right_room > room_threshold) {
- /* try merge with right */
- mdbx_cassert(mc, right_nkeys >= minkeys);
- mn.mc_pg[mn.mc_top] = right;
- mn.mc_ki[mn.mc_top - 1] = (indx_t)(ki_pre_top + 1);
- mn.mc_ki[mn.mc_top] = 0;
- mc->mc_ki[mc->mc_top] = (indx_t)nkeys;
- WITH_CURSOR_TRACKING(mn, rc = mdbx_page_merge(&mn, mc));
- if (likely(rc != MDBX_RESULT_TRUE)) {
- mc->mc_ki[mc->mc_top] = (indx_t)ki_top;
- mdbx_cassert(mc, rc || page_numkeys(mc->mc_pg[mc->mc_top]) >= minkeys);
- return rc;
- }
- }
-
- if (left_nkeys > minkeys &&
- (right_nkeys <= left_nkeys || right_room >= left_room)) {
- /* try move from left */
- mn.mc_pg[mn.mc_top] = left;
- mn.mc_ki[mn.mc_top - 1] = (indx_t)(ki_pre_top - 1);
- mn.mc_ki[mn.mc_top] = (indx_t)(left_nkeys - 1);
- mc->mc_ki[mc->mc_top] = 0;
- WITH_CURSOR_TRACKING(mn, rc = mdbx_node_move(&mn, mc, true));
- if (likely(rc != MDBX_RESULT_TRUE)) {
- mc->mc_ki[mc->mc_top] = (indx_t)(ki_top + 1);
- mdbx_cassert(mc, rc || page_numkeys(mc->mc_pg[mc->mc_top]) >= minkeys);
- return rc;
- }
- }
- if (right_nkeys > minkeys) {
- /* try move from right */
- mn.mc_pg[mn.mc_top] = right;
- mn.mc_ki[mn.mc_top - 1] = (indx_t)(ki_pre_top + 1);
- mn.mc_ki[mn.mc_top] = 0;
- mc->mc_ki[mc->mc_top] = (indx_t)nkeys;
- WITH_CURSOR_TRACKING(mn, rc = mdbx_node_move(&mn, mc, false));
- if (likely(rc != MDBX_RESULT_TRUE)) {
- mc->mc_ki[mc->mc_top] = (indx_t)ki_top;
- mdbx_cassert(mc, rc || page_numkeys(mc->mc_pg[mc->mc_top]) >= minkeys);
- return rc;
- }
- }
-
- if (nkeys >= minkeys) {
- mc->mc_ki[mc->mc_top] = (indx_t)ki_top;
- if (!mdbx_audit_enabled())
- return MDBX_SUCCESS;
- return mdbx_cursor_check(mc, C_UPDATING);
- }
-
- if (likely(room_threshold > 0)) {
- room_threshold = 0;
- goto retry;
- }
- mdbx_error("Unable to merge/rebalance %s page %" PRIaPGNO
- " (has %u keys, full %.1f%%, used %u, room %u bytes )",
- (pagetype & P_LEAF) ? "leaf" : "branch", tp->mp_pgno, numkeys,
- page_fill(mc->mc_txn->mt_env, tp),
- page_used(mc->mc_txn->mt_env, tp), room);
- return MDBX_PROBLEM;
-}
-
-__cold static int mdbx_page_check(MDBX_cursor *const mc,
- const MDBX_page *const mp, unsigned options) {
- DKBUF;
- options |= mc->mc_flags;
- MDBX_env *const env = mc->mc_txn->mt_env;
- const unsigned nkeys = page_numkeys(mp);
- char *const end_of_page = (char *)mp + env->me_psize;
- if (unlikely(mp->mp_pgno < MIN_PAGENO || mp->mp_pgno > MAX_PAGENO))
- return bad_page(mp, "invalid pgno (%u)\n", mp->mp_pgno);
- if (IS_OVERFLOW(mp)) {
- if (unlikely(mp->mp_pages < 1 && mp->mp_pages >= MAX_PAGENO / 2))
- return bad_page(mp, "invalid overflow n-pages (%u)\n", mp->mp_pages);
- if (unlikely(mp->mp_pgno + mp->mp_pages > mc->mc_txn->mt_next_pgno))
- return bad_page(mp, "overflow page beyond (%u) next-pgno\n",
- mp->mp_pgno + mp->mp_pages);
- if (unlikely((options & (C_SUB | C_COPYING)) == C_SUB))
- return bad_page(mp,
- "unexpected overflow-page for dupsort db (flags 0x%x)\n",
- mc->mc_db->md_flags);
- return MDBX_SUCCESS;
- }
-
- int rc = MDBX_SUCCESS;
- if ((options & C_UPDATING) == 0 || !IS_MODIFIABLE(mc->mc_txn, mp)) {
- if (unlikely(nkeys < 2 && IS_BRANCH(mp)))
- rc = bad_page(mp, "branch-page nkey (%u) < 2\n", nkeys);
- }
- if (IS_LEAF2(mp) && unlikely((options & (C_SUB | C_COPYING)) == 0))
- rc = bad_page(mp, "unexpected leaf2-page (db flags 0x%x)\n",
- mc->mc_db->md_flags);
-
- MDBX_val here, prev = {0, 0};
- for (unsigned i = 0; i < nkeys; ++i) {
- if (IS_LEAF2(mp)) {
- const size_t ksize = mp->mp_leaf2_ksize;
- char *const key = page_leaf2key(mp, i, ksize);
- if (unlikely(end_of_page < key + ksize)) {
- rc = bad_page(mp, "leaf2-key beyond (%zu) page-end\n",
- key + ksize - end_of_page);
- continue;
- }
-
- if ((options & C_COPYING) == 0) {
- if (unlikely(ksize != mc->mc_dbx->md_klen_min)) {
- if (unlikely(ksize < mc->mc_dbx->md_klen_min ||
- ksize > mc->mc_dbx->md_klen_max))
- rc = bad_page(
- mp, "leaf2-key size (%zu) <> min/max key-length (%zu/%zu)\n",
- ksize, mc->mc_dbx->md_klen_min, mc->mc_dbx->md_klen_max);
- else
- mc->mc_dbx->md_klen_min = mc->mc_dbx->md_klen_max = ksize;
- }
- if ((options & C_SKIPORD) == 0) {
- here.iov_len = ksize;
- here.iov_base = key;
- if (prev.iov_base && unlikely(mc->mc_dbx->md_cmp(&prev, &here) >= 0))
- rc = bad_page(mp, "leaf2-key #%u wrong order (%s >= %s)\n", i,
- DKEY(&prev), DVAL(&here));
- prev = here;
- }
- }
- } else {
- const MDBX_node *const node = page_node(mp, i);
- const char *node_end = (char *)node + NODESIZE;
- if (unlikely(node_end > end_of_page)) {
- rc = bad_page(mp, "node[%u] (%zu) beyond page-end\n", i,
- node_end - end_of_page);
- continue;
- }
- size_t ksize = node_ks(node);
- char *key = node_key(node);
- if (unlikely(end_of_page < key + ksize)) {
- rc = bad_page(mp, "node[%u] key (%zu) beyond page-end\n", i,
- key + ksize - end_of_page);
- continue;
- }
- if ((IS_LEAF(mp) || i > 0) && (options & C_COPYING) == 0) {
- if (unlikely(ksize < mc->mc_dbx->md_klen_min ||
- ksize > mc->mc_dbx->md_klen_max))
- rc = bad_page(
- mp, "node[%u] key size (%zu) <> min/max key-length (%zu/%zu)\n",
- i, ksize, mc->mc_dbx->md_klen_min, mc->mc_dbx->md_klen_max);
- if ((options & C_SKIPORD) == 0) {
- here.iov_base = key;
- here.iov_len = ksize;
- if (prev.iov_base && unlikely(mc->mc_dbx->md_cmp(&prev, &here) >= 0))
- rc = bad_page(mp, "node[%u] key wrong order (%s >= %s)\n", i,
- DKEY(&prev), DVAL(&here));
- prev = here;
- }
- }
- if (IS_BRANCH(mp)) {
- if ((options & C_UPDATING) == 0 && i == 0 && unlikely(ksize != 0))
- rc = bad_page(mp, "branch-node[%u] wrong 0-node key-length (%zu)\n",
- i, ksize);
- if ((options & C_RETIRING) == 0) {
- const pgno_t ref = node_pgno(node);
- if (unlikely(ref < MIN_PAGENO || ref >= mc->mc_txn->mt_next_pgno))
- rc = bad_page(mp, "branch-node[%u] wrong pgno (%u)\n", i, ref);
- }
- if (unlikely(node_flags(node)))
- rc = bad_page(mp, "branch-node[%u] wrong flags (%u)\n", i,
- node_flags(node));
- continue;
- }
-
- switch (node_flags(node)) {
- default:
- rc = bad_page(mp, "invalid node[%u] flags (%u)\n", i, node_flags(node));
- break;
- case F_BIGDATA /* data on large-page */:
- case 0 /* usual */:
- case F_SUBDATA /* sub-db */:
- case F_SUBDATA | F_DUPDATA /* dupsorted sub-tree */:
- case F_DUPDATA /* short sub-page */:
- break;
- }
-
- const size_t dsize = node_ds(node);
- const char *const data = node_data(node);
- if (node_flags(node) & F_BIGDATA) {
- if (unlikely(end_of_page < data + sizeof(pgno_t))) {
- rc = bad_page(
- mp, "node-%s(%u of %u, %zu bytes) beyond (%zu) page-end\n",
- "bigdata-pgno", i, nkeys, dsize, data + dsize - end_of_page);
- continue;
- }
- if ((options & C_COPYING) == 0) {
- if (unlikely(dsize <= mc->mc_dbx->md_vlen_min ||
- dsize > mc->mc_dbx->md_vlen_max))
- rc = bad_page(
- mp,
- "big-node data size (%zu) <> min/max value-length (%zu/%zu)\n",
- dsize, mc->mc_dbx->md_vlen_min, mc->mc_dbx->md_vlen_max);
- }
- if ((options & C_RETIRING) == 0) {
- MDBX_page *lp;
- int err = mdbx_page_get(mc, node_largedata_pgno(node), &lp,
- pp_txnid4chk(mp, mc->mc_txn));
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- if (unlikely(!IS_OVERFLOW(lp))) {
- rc = bad_page(mp, "big-node refs to non-overflow page (%u)\n",
- lp->mp_pgno);
- continue;
- }
- if (unlikely(number_of_ovpages(env, dsize) > lp->mp_pages))
- rc =
- bad_page(mp, "big-node size (%zu) mismatch n-pages size (%u)\n",
- dsize, lp->mp_pages);
- }
- continue;
- }
-
- if (unlikely(end_of_page < data + dsize)) {
- rc =
- bad_page(mp, "node-%s(%u of %u, %zu bytes) beyond (%zu) page-end\n",
- "data", i, nkeys, dsize, data + dsize - end_of_page);
- continue;
- }
-
- switch (node_flags(node)) {
- default:
- /* wrong, but already handled */
- continue;
- case 0 /* usual */:
- if ((options & C_COPYING) == 0) {
- if (unlikely(dsize < mc->mc_dbx->md_vlen_min ||
- dsize > mc->mc_dbx->md_vlen_max)) {
- rc = bad_page(
- mp, "node-data size (%zu) <> min/max value-length (%zu/%zu)\n",
- dsize, mc->mc_dbx->md_vlen_min, mc->mc_dbx->md_vlen_max);
- continue;
- }
- }
- break;
- case F_SUBDATA /* sub-db */:
- if (unlikely(dsize != sizeof(MDBX_db))) {
- rc = bad_page(mp, "invalid sub-db record size (%zu)\n", dsize);
- continue;
- }
- break;
- case F_SUBDATA | F_DUPDATA /* dupsorted sub-tree */:
- if (unlikely(dsize != sizeof(MDBX_db))) {
- rc = bad_page(mp, "invalid nested-db record size (%zu)\n", dsize);
- continue;
- }
- break;
- case F_DUPDATA /* short sub-page */:
- if (unlikely(dsize <= PAGEHDRSZ)) {
- rc = bad_page(mp, "invalid nested/sub-page record size (%zu)\n",
- dsize);
- continue;
- } else {
- const MDBX_page *const sp = (MDBX_page *)data;
- const char *const end_of_subpage = data + dsize;
- const int nsubkeys = page_numkeys(sp);
- switch (sp->mp_flags & /* ignore legacy P_DIRTY flag */ ~0x10) {
- case P_LEAF | P_SUBP:
- case P_LEAF | P_LEAF2 | P_SUBP:
- break;
- default:
- rc = bad_page(mp, "invalid nested/sub-page flags (0x%02x)\n",
- sp->mp_flags);
- continue;
- }
-
- MDBX_val sub_here, sub_prev = {0, 0};
- for (int j = 0; j < nsubkeys; j++) {
- if (IS_LEAF2(sp)) {
- /* LEAF2 pages have no mp_ptrs[] or node headers */
- size_t sub_ksize = sp->mp_leaf2_ksize;
- char *sub_key = page_leaf2key(sp, j, sub_ksize);
- if (unlikely(end_of_subpage < sub_key + sub_ksize)) {
- rc = bad_page(mp, "nested-leaf2-key beyond (%zu) nested-page\n",
- sub_key + sub_ksize - end_of_subpage);
- continue;
- }
-
- if ((options & C_COPYING) == 0) {
- if (unlikely(sub_ksize != mc->mc_dbx->md_vlen_min)) {
- if (unlikely(sub_ksize < mc->mc_dbx->md_vlen_min ||
- sub_ksize > mc->mc_dbx->md_vlen_max)) {
- rc = bad_page(mp,
- "nested-leaf2-key size (%zu) <> min/max "
- "value-length (%zu/%zu)\n",
- sub_ksize, mc->mc_dbx->md_vlen_min,
- mc->mc_dbx->md_vlen_max);
- continue;
- }
- mc->mc_dbx->md_vlen_min = mc->mc_dbx->md_vlen_max = sub_ksize;
- }
- if ((options & C_SKIPORD) == 0) {
- sub_here.iov_len = sub_ksize;
- sub_here.iov_base = sub_key;
- if (sub_prev.iov_base &&
- unlikely(mc->mc_dbx->md_dcmp(&sub_prev, &sub_here) >= 0))
- rc = bad_page(
- mp, "nested-leaf2-key #%u wrong order (%s >= %s)\n", j,
- DKEY(&sub_prev), DVAL(&sub_here));
- sub_prev = sub_here;
- }
- }
- } else {
- const MDBX_node *const sub_node = page_node(sp, j);
- const char *sub_node_end = (char *)sub_node + NODESIZE;
- if (unlikely(sub_node_end > end_of_subpage)) {
- rc = bad_page(mp, "nested-node beyond (%zu) nested-page\n",
- end_of_subpage - sub_node_end);
- continue;
- }
- if (unlikely(node_flags(sub_node) != 0))
- rc = bad_page(mp, "nested-node invalid flags (%u)\n",
- node_flags(sub_node));
-
- size_t sub_ksize = node_ks(sub_node);
- char *sub_key = node_key(sub_node);
- size_t sub_dsize = node_ds(sub_node);
- /* char *sub_data = node_data(sub_node); */
-
- if ((options & C_COPYING) == 0) {
- if (unlikely(sub_ksize < mc->mc_dbx->md_vlen_min ||
- sub_ksize > mc->mc_dbx->md_vlen_max))
- rc = bad_page(mp,
- "nested-node-key size (%zu) <> min/max "
- "value-length (%zu/%zu)\n",
- sub_ksize, mc->mc_dbx->md_vlen_min,
- mc->mc_dbx->md_vlen_max);
-
- if ((options & C_SKIPORD) == 0) {
- sub_here.iov_len = sub_ksize;
- sub_here.iov_base = sub_key;
- if (sub_prev.iov_base &&
- unlikely(mc->mc_dbx->md_dcmp(&sub_prev, &sub_here) >= 0))
- rc = bad_page(
- mp, "nested-node-key #%u wrong order (%s >= %s)\n", j,
- DKEY(&sub_prev), DVAL(&sub_here));
- sub_prev = sub_here;
- }
- }
- if (unlikely(sub_dsize != 0))
- rc = bad_page(mp, "nested-node non-empty data size (%zu)\n",
- sub_dsize);
- if (unlikely(end_of_subpage < sub_key + sub_ksize))
- rc = bad_page(mp, "nested-node-key beyond (%zu) nested-page\n",
- sub_key + sub_ksize - end_of_subpage);
- }
- }
- }
- break;
- }
- }
- }
- return rc;
-}
-
-__cold static int mdbx_cursor_check(MDBX_cursor *mc, unsigned options) {
- mdbx_cassert(mc,
- mc->mc_txn->tw.dirtyroom + mc->mc_txn->tw.dirtylist->length ==
- (mc->mc_txn->mt_parent
- ? mc->mc_txn->mt_parent->tw.dirtyroom
- : mc->mc_txn->mt_env->me_options.dp_limit));
- mdbx_cassert(mc, mc->mc_top == mc->mc_snum - 1 || (options & C_UPDATING));
- if (unlikely(mc->mc_top != mc->mc_snum - 1) && (options & C_UPDATING) == 0)
- return MDBX_CURSOR_FULL;
- mdbx_cassert(mc, (options & C_UPDATING) ? mc->mc_snum <= mc->mc_db->md_depth
- : mc->mc_snum == mc->mc_db->md_depth);
- if (unlikely((options & C_UPDATING) ? mc->mc_snum > mc->mc_db->md_depth
- : mc->mc_snum != mc->mc_db->md_depth))
- return MDBX_CURSOR_FULL;
-
- for (int n = 0; n < (int)mc->mc_snum; ++n) {
- MDBX_page *mp = mc->mc_pg[n];
- const unsigned nkeys = page_numkeys(mp);
- const bool expect_branch = (n < mc->mc_db->md_depth - 1) ? true : false;
- const bool expect_nested_leaf =
- (n + 1 == mc->mc_db->md_depth - 1) ? true : false;
- const bool branch = IS_BRANCH(mp) ? true : false;
- mdbx_cassert(mc, branch == expect_branch);
- if (unlikely(branch != expect_branch))
- return MDBX_CURSOR_FULL;
- if ((options & C_UPDATING) == 0) {
- mdbx_cassert(mc,
- nkeys > mc->mc_ki[n] || (!branch && nkeys == mc->mc_ki[n] &&
- (mc->mc_flags & C_EOF) != 0));
- if (unlikely(nkeys <= mc->mc_ki[n] &&
- !(!branch && nkeys == mc->mc_ki[n] &&
- (mc->mc_flags & C_EOF) != 0)))
- return MDBX_CURSOR_FULL;
- } else {
- mdbx_cassert(mc, nkeys + 1 >= mc->mc_ki[n]);
- if (unlikely(nkeys + 1 < mc->mc_ki[n]))
- return MDBX_CURSOR_FULL;
- }
-
- int err = mdbx_page_check(mc, mp, options);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- for (unsigned i = 0; i < nkeys; ++i) {
- if (branch) {
- MDBX_node *node = page_node(mp, i);
- mdbx_cassert(mc, node_flags(node) == 0);
- if (unlikely(node_flags(node) != 0))
- return MDBX_CURSOR_FULL;
- pgno_t pgno = node_pgno(node);
- MDBX_page *np;
- int rc = mdbx_page_get(mc, pgno, &np, pp_txnid4chk(mp, mc->mc_txn));
- mdbx_cassert(mc, rc == MDBX_SUCCESS);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- const bool nested_leaf = IS_LEAF(np) ? true : false;
- mdbx_cassert(mc, nested_leaf == expect_nested_leaf);
- if (unlikely(nested_leaf != expect_nested_leaf))
- return MDBX_CURSOR_FULL;
- err = mdbx_page_check(mc, np, options);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- }
- }
- return MDBX_SUCCESS;
-}
-
-/* Complete a delete operation started by mdbx_cursor_del(). */
-static int mdbx_cursor_del0(MDBX_cursor *mc) {
- int rc;
- MDBX_page *mp;
- indx_t ki;
- unsigned nkeys;
- MDBX_dbi dbi = mc->mc_dbi;
-
- mdbx_cassert(mc, cursor_is_tracked(mc));
- mdbx_cassert(mc, IS_LEAF(mc->mc_pg[mc->mc_top]));
- ki = mc->mc_ki[mc->mc_top];
- mp = mc->mc_pg[mc->mc_top];
- mdbx_node_del(mc, mc->mc_db->md_xsize);
- mc->mc_db->md_entries--;
-
- /* Adjust other cursors pointing to mp */
- for (MDBX_cursor *m2 = mc->mc_txn->mt_cursors[dbi]; m2; m2 = m2->mc_next) {
- MDBX_cursor *m3 = (mc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (m3 == mc || !(m2->mc_flags & m3->mc_flags & C_INITIALIZED))
- continue;
- if (m3->mc_snum < mc->mc_snum)
- continue;
- if (m3->mc_pg[mc->mc_top] == mp) {
- if (m3->mc_ki[mc->mc_top] == ki) {
- m3->mc_flags |= C_DEL;
- if (mc->mc_db->md_flags & MDBX_DUPSORT) {
- /* Sub-cursor referred into dataset which is gone */
- m3->mc_xcursor->mx_cursor.mc_flags &= ~(C_INITIALIZED | C_EOF);
- }
- continue;
- } else if (m3->mc_ki[mc->mc_top] > ki) {
- m3->mc_ki[mc->mc_top]--;
- }
- if (XCURSOR_INITED(m3))
- XCURSOR_REFRESH(m3, m3->mc_pg[mc->mc_top], m3->mc_ki[mc->mc_top]);
- }
- }
-
- rc = mdbx_rebalance(mc);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
-
- if (unlikely(!mc->mc_snum)) {
- /* DB is totally empty now, just bail out.
- * Other cursors adjustments were already done
- * by mdbx_rebalance and aren't needed here. */
- mdbx_cassert(mc, mc->mc_db->md_entries == 0 && mc->mc_db->md_depth == 0 &&
- mc->mc_db->md_root == P_INVALID);
- mc->mc_flags |= C_EOF;
- return MDBX_SUCCESS;
- }
-
- ki = mc->mc_ki[mc->mc_top];
- mp = mc->mc_pg[mc->mc_top];
- mdbx_cassert(mc, IS_LEAF(mc->mc_pg[mc->mc_top]));
- nkeys = page_numkeys(mp);
- mdbx_cassert(mc, (mc->mc_db->md_entries > 0 && nkeys > 0) ||
- ((mc->mc_flags & C_SUB) && mc->mc_db->md_entries == 0 &&
- nkeys == 0));
-
- /* Adjust this and other cursors pointing to mp */
- for (MDBX_cursor *m2 = mc->mc_txn->mt_cursors[dbi]; m2; m2 = m2->mc_next) {
- MDBX_cursor *m3 = (mc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (!(m2->mc_flags & m3->mc_flags & C_INITIALIZED))
- continue;
- if (m3->mc_snum < mc->mc_snum)
- continue;
- if (m3->mc_pg[mc->mc_top] == mp) {
- /* if m3 points past last node in page, find next sibling */
- if (m3->mc_ki[mc->mc_top] >= nkeys) {
- rc = mdbx_cursor_sibling(m3, SIBLING_RIGHT);
- if (rc == MDBX_NOTFOUND) {
- m3->mc_flags |= C_EOF;
- rc = MDBX_SUCCESS;
- continue;
- }
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- if (m3->mc_ki[mc->mc_top] >= ki ||
- /* moved to right sibling */ m3->mc_pg[mc->mc_top] != mp) {
- if (m3->mc_xcursor && !(m3->mc_flags & C_EOF)) {
- MDBX_node *node =
- page_node(m3->mc_pg[m3->mc_top], m3->mc_ki[m3->mc_top]);
- /* If this node has dupdata, it may need to be reinited
- * because its data has moved.
- * If the xcursor was not inited it must be reinited.
- * Else if node points to a subDB, nothing is needed. */
- if (node_flags(node) & F_DUPDATA) {
- if (m3->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED) {
- if (!(node_flags(node) & F_SUBDATA))
- m3->mc_xcursor->mx_cursor.mc_pg[0] = node_data(node);
- } else {
- rc = mdbx_xcursor_init1(m3, node, m3->mc_pg[m3->mc_top]);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- rc = mdbx_cursor_first(&m3->mc_xcursor->mx_cursor, NULL, NULL);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- }
- m3->mc_xcursor->mx_cursor.mc_flags |= C_DEL;
- }
- m3->mc_flags |= C_DEL;
- }
- }
- }
-
- mdbx_cassert(mc, rc == MDBX_SUCCESS);
- if (mdbx_audit_enabled())
- rc = mdbx_cursor_check(mc, 0);
- return rc;
-
-bailout:
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- return rc;
-}
-
-int mdbx_del(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key,
- const MDBX_val *data) {
- int rc = check_txn_rw(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!key))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- if (unlikely(txn->mt_flags & (MDBX_TXN_RDONLY | MDBX_TXN_BLOCKED)))
- return (txn->mt_flags & MDBX_TXN_RDONLY) ? MDBX_EACCESS : MDBX_BAD_TXN;
-
- return mdbx_del0(txn, dbi, key, data, 0);
-}
-
-static int mdbx_del0(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key,
- const MDBX_val *data, unsigned flags) {
- MDBX_cursor_couple cx;
- MDBX_cursor_op op;
- MDBX_val rdata;
- int rc;
- DKBUF_DEBUG;
-
- mdbx_debug("====> delete db %u key [%s], data [%s]", dbi, DKEY_DEBUG(key),
- DVAL_DEBUG(data));
-
- rc = mdbx_cursor_init(&cx.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (data) {
- op = MDBX_GET_BOTH;
- rdata = *data;
- data = &rdata;
- } else {
- op = MDBX_SET;
- flags |= MDBX_ALLDUPS;
- }
- rc = mdbx_cursor_set(&cx.outer, (MDBX_val *)key, (MDBX_val *)data, op).err;
- if (likely(rc == MDBX_SUCCESS)) {
- /* let mdbx_page_split know about this cursor if needed:
- * delete will trigger a rebalance; if it needs to move
- * a node from one page to another, it will have to
- * update the parent's separator key(s). If the new sepkey
- * is larger than the current one, the parent page may
- * run out of space, triggering a split. We need this
- * cursor to be consistent until the end of the rebalance. */
- cx.outer.mc_next = txn->mt_cursors[dbi];
- txn->mt_cursors[dbi] = &cx.outer;
- rc = mdbx_cursor_del(&cx.outer, flags);
- txn->mt_cursors[dbi] = cx.outer.mc_next;
- }
- return rc;
-}
-
-/* Split a page and insert a new node.
- * Set MDBX_TXN_ERROR on failure.
- * [in,out] mc Cursor pointing to the page and desired insertion index.
- * The cursor will be updated to point to the actual page and index where
- * the node got inserted after the split.
- * [in] newkey The key for the newly inserted node.
- * [in] newdata The data for the newly inserted node.
- * [in] newpgno The page number, if the new node is a branch node.
- * [in] nflags The NODE_ADD_FLAGS for the new node.
- * Returns 0 on success, non-zero on failure. */
-static int mdbx_page_split(MDBX_cursor *mc, const MDBX_val *const newkey,
- MDBX_val *const newdata, pgno_t newpgno,
- unsigned nflags) {
- unsigned flags;
- int rc = MDBX_SUCCESS, foliage = 0;
- unsigned i, ptop;
- MDBX_env *const env = mc->mc_txn->mt_env;
- MDBX_val sepkey, rkey, xdata;
- MDBX_page *tmp_ki_copy = NULL;
- DKBUF;
-
- MDBX_page *const mp = mc->mc_pg[mc->mc_top];
- const unsigned newindx = mc->mc_ki[mc->mc_top];
- unsigned nkeys = page_numkeys(mp);
- if (mdbx_audit_enabled()) {
- rc = mdbx_cursor_check(mc, C_UPDATING);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
- STATIC_ASSERT(P_BRANCH == 1);
- const unsigned minkeys = (mp->mp_flags & P_BRANCH) + 1;
-
- mdbx_debug(">> splitting %s-page %" PRIaPGNO
- " and adding %zu+%zu [%s] at %i, nkeys %i",
- IS_LEAF(mp) ? "leaf" : "branch", mp->mp_pgno, newkey->iov_len,
- newdata ? newdata->iov_len : 0, DKEY_DEBUG(newkey),
- mc->mc_ki[mc->mc_top], nkeys);
- mdbx_cassert(mc, nkeys + 1 >= minkeys * 2);
-
- /* Create a new sibling page. */
- struct page_result npr = mdbx_page_new(mc, mp->mp_flags, 1);
- if (unlikely(npr.err != MDBX_SUCCESS))
- return npr.err;
- MDBX_page *const sister = npr.page;
- sister->mp_leaf2_ksize = mp->mp_leaf2_ksize;
- mdbx_debug("new sibling: page %" PRIaPGNO, sister->mp_pgno);
-
- /* Usually when splitting the root page, the cursor
- * height is 1. But when called from mdbx_update_key,
- * the cursor height may be greater because it walks
- * up the stack while finding the branch slot to update. */
- if (mc->mc_top < 1) {
- npr = mdbx_page_new(mc, P_BRANCH, 1);
- rc = npr.err;
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- MDBX_page *const pp = npr.page;
- /* shift current top to make room for new parent */
- mdbx_cassert(mc, mc->mc_snum < 2 && mc->mc_db->md_depth > 0);
-#if MDBX_DEBUG
- memset(mc->mc_pg + 3, 0, sizeof(mc->mc_pg) - sizeof(mc->mc_pg[0]) * 3);
- memset(mc->mc_ki + 3, -1, sizeof(mc->mc_ki) - sizeof(mc->mc_ki[0]) * 3);
-#endif
- mc->mc_pg[2] = mc->mc_pg[1];
- mc->mc_ki[2] = mc->mc_ki[1];
- mc->mc_pg[1] = mc->mc_pg[0];
- mc->mc_ki[1] = mc->mc_ki[0];
- mc->mc_pg[0] = pp;
- mc->mc_ki[0] = 0;
- mc->mc_db->md_root = pp->mp_pgno;
- mdbx_debug("root split! new root = %" PRIaPGNO, pp->mp_pgno);
- foliage = mc->mc_db->md_depth++;
-
- /* Add left (implicit) pointer. */
- rc = mdbx_node_add_branch(mc, 0, NULL, mp->mp_pgno);
- if (unlikely(rc != MDBX_SUCCESS)) {
- /* undo the pre-push */
- mc->mc_pg[0] = mc->mc_pg[1];
- mc->mc_ki[0] = mc->mc_ki[1];
- mc->mc_db->md_root = mp->mp_pgno;
- mc->mc_db->md_depth--;
- goto done;
- }
- mc->mc_snum++;
- mc->mc_top++;
- ptop = 0;
- if (mdbx_audit_enabled()) {
- rc = mdbx_cursor_check(mc, C_UPDATING);
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- }
- } else {
- ptop = mc->mc_top - 1;
- mdbx_debug("parent branch page is %" PRIaPGNO, mc->mc_pg[ptop]->mp_pgno);
- }
-
- MDBX_cursor mn;
- cursor_copy(mc, &mn);
- mn.mc_pg[mn.mc_top] = sister;
- mn.mc_ki[mn.mc_top] = 0;
- mn.mc_ki[ptop] = mc->mc_ki[ptop] + 1;
-
- unsigned split_indx =
- (newindx < nkeys)
- ? /* split at the middle */ (nkeys + 1) / 2
- : /* split at the end (i.e. like append-mode ) */ nkeys - minkeys + 1;
-
- mdbx_cassert(mc, !IS_BRANCH(mp) || newindx > 0);
- /* It is reasonable and possible to split the page at the begin */
- if (unlikely(newindx < minkeys)) {
- split_indx = minkeys;
- if (newindx == 0 && foliage == 0 && !(nflags & MDBX_SPLIT_REPLACE)) {
- split_indx = 0;
- /* Checking for ability of splitting by the left-side insertion
- * of a pure page with the new key */
- for (i = 0; i < mc->mc_top; ++i)
- if (mc->mc_ki[i]) {
- get_key(page_node(mc->mc_pg[i], mc->mc_ki[i]), &sepkey);
- if (mc->mc_dbx->md_cmp(newkey, &sepkey) >= 0)
- split_indx = minkeys;
- break;
- }
- if (split_indx == 0) {
- /* Save the current first key which was omitted on the parent branch
- * page and should be updated if the new first entry will be added */
- if (IS_LEAF2(mp)) {
- sepkey.iov_len = mp->mp_leaf2_ksize;
- sepkey.iov_base = page_leaf2key(mp, 0, sepkey.iov_len);
- } else
- get_key(page_node(mp, 0), &sepkey);
- mdbx_cassert(mc, mc->mc_dbx->md_cmp(newkey, &sepkey) < 0);
- /* Avoiding rare complex cases of split the parent page */
- if (page_room(mn.mc_pg[ptop]) < branch_size(env, &sepkey))
- split_indx = minkeys;
- }
- }
- }
-
- const bool pure_right = split_indx == nkeys;
- const bool pure_left = split_indx == 0;
- if (unlikely(pure_right)) {
- /* newindx == split_indx == nkeys */
- mdbx_trace("no-split, but add new pure page at the %s", "right/after");
- mdbx_cassert(mc, newindx == nkeys && split_indx == nkeys && minkeys == 1);
- sepkey = *newkey;
- } else if (unlikely(pure_left)) {
- /* newindx == split_indx == 0 */
- mdbx_trace("no-split, but add new pure page at the %s", "left/before");
- mdbx_cassert(mc, newindx == 0 && split_indx == 0 && minkeys == 1);
- mdbx_trace("old-first-key is %s", DKEY_DEBUG(&sepkey));
- } else {
- if (IS_LEAF2(sister)) {
- char *split, *ins;
- unsigned lsize, rsize, ksize;
- /* Move half of the keys to the right sibling */
- const int x = mc->mc_ki[mc->mc_top] - split_indx;
- ksize = mc->mc_db->md_xsize;
- split = page_leaf2key(mp, split_indx, ksize);
- rsize = (nkeys - split_indx) * ksize;
- lsize = (nkeys - split_indx) * sizeof(indx_t);
- mdbx_cassert(mc, mp->mp_lower >= lsize);
- mp->mp_lower -= (indx_t)lsize;
- mdbx_cassert(mc, sister->mp_lower + lsize <= UINT16_MAX);
- sister->mp_lower += (indx_t)lsize;
- mdbx_cassert(mc, mp->mp_upper + rsize - lsize <= UINT16_MAX);
- mp->mp_upper += (indx_t)(rsize - lsize);
- mdbx_cassert(mc, sister->mp_upper >= rsize - lsize);
- sister->mp_upper -= (indx_t)(rsize - lsize);
- sepkey.iov_len = ksize;
- sepkey.iov_base = (newindx != split_indx) ? split : newkey->iov_base;
- if (x < 0) {
- mdbx_cassert(mc, ksize >= sizeof(indx_t));
- ins = page_leaf2key(mp, mc->mc_ki[mc->mc_top], ksize);
- memcpy(sister->mp_ptrs, split, rsize);
- sepkey.iov_base = sister->mp_ptrs;
- memmove(ins + ksize, ins, (split_indx - mc->mc_ki[mc->mc_top]) * ksize);
- memcpy(ins, newkey->iov_base, ksize);
- mdbx_cassert(mc, UINT16_MAX - mp->mp_lower >= (int)sizeof(indx_t));
- mp->mp_lower += sizeof(indx_t);
- mdbx_cassert(mc, mp->mp_upper >= ksize - sizeof(indx_t));
- mp->mp_upper -= (indx_t)(ksize - sizeof(indx_t));
- } else {
- memcpy(sister->mp_ptrs, split, x * ksize);
- ins = page_leaf2key(sister, x, ksize);
- memcpy(ins, newkey->iov_base, ksize);
- memcpy(ins + ksize, split + x * ksize, rsize - x * ksize);
- mdbx_cassert(mc, UINT16_MAX - sister->mp_lower >= (int)sizeof(indx_t));
- sister->mp_lower += sizeof(indx_t);
- mdbx_cassert(mc, sister->mp_upper >= ksize - sizeof(indx_t));
- sister->mp_upper -= (indx_t)(ksize - sizeof(indx_t));
- mdbx_cassert(mc, x <= (int)UINT16_MAX);
- mc->mc_ki[mc->mc_top] = (indx_t)x;
- }
-
- if (mdbx_audit_enabled()) {
- rc = mdbx_cursor_check(mc, C_UPDATING);
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- rc = mdbx_cursor_check(&mn, C_UPDATING);
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- }
- } else {
- /* Maximum free space in an empty page */
- const unsigned max_space = page_space(env);
- const size_t new_size = IS_LEAF(mp) ? leaf_size(env, newkey, newdata)
- : branch_size(env, newkey);
-
- /* grab a page to hold a temporary copy */
- tmp_ki_copy = mdbx_page_malloc(mc->mc_txn, 1);
- if (unlikely(tmp_ki_copy == NULL)) {
- rc = MDBX_ENOMEM;
- goto done;
- }
-
- /* prepare to insert */
- for (unsigned j = i = 0; i < nkeys; ++i, ++j) {
- tmp_ki_copy->mp_ptrs[j] = 0;
- j += (i == newindx);
- tmp_ki_copy->mp_ptrs[j] = mp->mp_ptrs[i];
- }
- tmp_ki_copy->mp_pgno = mp->mp_pgno;
- tmp_ki_copy->mp_flags = mp->mp_flags;
- tmp_ki_copy->mp_txnid = INVALID_TXNID;
- tmp_ki_copy->mp_lower = 0;
- tmp_ki_copy->mp_upper = (indx_t)max_space;
-
- /* When items are relatively large the split point needs
- * to be checked, because being off-by-one will make the
- * difference between success or failure in mdbx_node_add.
- *
- * It's also relevant if a page happens to be laid out
- * such that one half of its nodes are all "small" and
- * the other half of its nodes are "large". If the new
- * item is also "large" and falls on the half with
- * "large" nodes, it also may not fit.
- *
- * As a final tweak, if the new item goes on the last
- * spot on the page (and thus, onto the new page), bias
- * the split so the new page is emptier than the old page.
- * This yields better packing during sequential inserts. */
-
- if (nkeys < 32 || new_size > max_space / 16) {
- /* Find split point */
- int dir;
- if (newindx <= split_indx) {
- i = 0;
- dir = 1;
- } else {
- i = nkeys;
- dir = -1;
- }
- size_t before = 0, after = new_size + page_used(env, mp);
- int best = split_indx;
- int best_offset = nkeys + 1;
-
- mdbx_trace("seek separator from %u, step %i, default %u, new-idx %u, "
- "new-size %zu",
- i, dir, split_indx, newindx, new_size);
- do {
- mdbx_cassert(mc, i <= nkeys);
- size_t size = new_size;
- if (i != newindx) {
- MDBX_node *node =
- (MDBX_node *)((char *)mp + tmp_ki_copy->mp_ptrs[i] + PAGEHDRSZ);
- size = NODESIZE + node_ks(node) + sizeof(indx_t);
- if (IS_LEAF(mp))
- size += F_ISSET(node_flags(node), F_BIGDATA) ? sizeof(pgno_t)
- : node_ds(node);
- size = EVEN(size);
- }
-
- before += size;
- after -= size;
- mdbx_trace("step %u, size %zu, before %zu, after %zu, max %u", i,
- size, before, after, max_space);
-
- if (before <= max_space && after <= max_space) {
- int offset = branchless_abs(split_indx - i);
- if (offset >= best_offset)
- break;
- best_offset = offset;
- best = i;
- }
- i += dir;
- } while (i < nkeys);
-
- split_indx = best + (dir > 0);
- split_indx = (split_indx <= nkeys - minkeys + 1) ? split_indx
- : nkeys - minkeys + 1;
- split_indx = (split_indx >= minkeys) ? split_indx : minkeys;
- mdbx_trace("chosen %u", split_indx);
- }
-
- sepkey.iov_len = newkey->iov_len;
- sepkey.iov_base = newkey->iov_base;
- if (split_indx != newindx) {
- MDBX_node *node =
- (MDBX_node *)((char *)mp + tmp_ki_copy->mp_ptrs[split_indx] +
- PAGEHDRSZ);
- sepkey.iov_len = node_ks(node);
- sepkey.iov_base = node_key(node);
- }
- }
- }
- mdbx_debug("separator is %d [%s]", split_indx, DKEY_DEBUG(&sepkey));
-
- bool did_split_parent = false;
- /* Copy separator key to the parent. */
- if (page_room(mn.mc_pg[ptop]) < branch_size(env, &sepkey)) {
- mdbx_trace("need split parent branch-page for key %s", DKEY_DEBUG(&sepkey));
- mdbx_cassert(mc, page_numkeys(mn.mc_pg[ptop]) > 2);
- mdbx_cassert(mc, !pure_left);
- const int snum = mc->mc_snum;
- const int depth = mc->mc_db->md_depth;
- mn.mc_snum--;
- mn.mc_top--;
- did_split_parent = true;
- /* We want other splits to find mn when doing fixups */
- WITH_CURSOR_TRACKING(
- mn, rc = mdbx_page_split(&mn, &sepkey, NULL, sister->mp_pgno, 0));
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- mdbx_cassert(mc, (int)mc->mc_snum - snum == mc->mc_db->md_depth - depth);
- if (mdbx_audit_enabled()) {
- rc = mdbx_cursor_check(mc, C_UPDATING);
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- }
-
- /* root split? */
- ptop += mc->mc_snum - snum;
-
- /* Right page might now have changed parent.
- * Check if left page also changed parent. */
- if (mn.mc_pg[ptop] != mc->mc_pg[ptop] &&
- mc->mc_ki[ptop] >= page_numkeys(mc->mc_pg[ptop])) {
- for (i = 0; i < ptop; i++) {
- mc->mc_pg[i] = mn.mc_pg[i];
- mc->mc_ki[i] = mn.mc_ki[i];
- }
- mc->mc_pg[ptop] = mn.mc_pg[ptop];
- if (mn.mc_ki[ptop]) {
- mc->mc_ki[ptop] = mn.mc_ki[ptop] - 1;
- } else {
- /* find right page's left sibling */
- mc->mc_ki[ptop] = mn.mc_ki[ptop];
- rc = mdbx_cursor_sibling(mc, SIBLING_LEFT);
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc == MDBX_NOTFOUND) /* improper mdbx_cursor_sibling() result */ {
- mdbx_error("unexpected %i error going left sibling", rc);
- rc = MDBX_PROBLEM;
- }
- goto done;
- }
- }
- }
- } else if (unlikely(pure_left)) {
- MDBX_page *ptop_page = mc->mc_pg[ptop];
- mdbx_debug("adding to parent page %u node[%u] left-leaf page #%u key %s",
- ptop_page->mp_pgno, mc->mc_ki[ptop], sister->mp_pgno,
- DKEY(mc->mc_ki[ptop] ? newkey : NULL));
- mc->mc_top--;
- rc = mdbx_node_add_branch(mc, mc->mc_ki[ptop],
- mc->mc_ki[ptop] ? newkey : NULL, sister->mp_pgno);
- mdbx_cassert(mc, mp == mc->mc_pg[ptop + 1] &&
- newindx == mc->mc_ki[ptop + 1] && ptop == mc->mc_top);
-
- if (likely(rc == MDBX_SUCCESS) && mc->mc_ki[ptop] == 0) {
- mdbx_debug("update prev-first key on parent %s", DKEY(&sepkey));
- MDBX_node *node = page_node(mc->mc_pg[ptop], 1);
- mdbx_cassert(mc, node_ks(node) == 0 && node_pgno(node) == mp->mp_pgno);
- mdbx_cassert(mc, mc->mc_top == ptop && mc->mc_ki[ptop] == 0);
- mc->mc_ki[ptop] = 1;
- rc = mdbx_update_key(mc, &sepkey);
- mdbx_cassert(mc, mc->mc_top == ptop && mc->mc_ki[ptop] == 1);
- mdbx_cassert(mc,
- mp == mc->mc_pg[ptop + 1] && newindx == mc->mc_ki[ptop + 1]);
- mc->mc_ki[ptop] = 0;
- }
-
- mc->mc_top++;
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
-
- MDBX_node *node = page_node(mc->mc_pg[ptop], mc->mc_ki[ptop] + 1);
- mdbx_cassert(mc, node_pgno(node) == mp->mp_pgno &&
- mc->mc_pg[ptop] == ptop_page);
- } else {
- mn.mc_top--;
- mdbx_trace("add-to-parent the right-entry[%u] for new sibling-page",
- mn.mc_ki[ptop]);
- rc = mdbx_node_add_branch(&mn, mn.mc_ki[ptop], &sepkey, sister->mp_pgno);
- mn.mc_top++;
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- }
-
- if (unlikely(pure_left | pure_right)) {
- mc->mc_pg[mc->mc_top] = sister;
- mc->mc_ki[mc->mc_top] = 0;
- switch (PAGETYPE(sister)) {
- case P_LEAF: {
- mdbx_cassert(mc, newpgno == 0 || newpgno == P_INVALID);
- rc = mdbx_node_add_leaf(mc, 0, newkey, newdata, nflags);
- } break;
- case P_LEAF | P_LEAF2: {
- mdbx_cassert(mc, (nflags & (F_BIGDATA | F_SUBDATA | F_DUPDATA)) == 0);
- mdbx_cassert(mc, newpgno == 0 || newpgno == P_INVALID);
- rc = mdbx_node_add_leaf2(mc, 0, newkey);
- } break;
- default:
- rc = bad_page(sister, "wrong page-type %u\n", PAGETYPE(sister));
- }
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
-
- if (pure_right) {
- for (i = 0; i < mc->mc_top; i++)
- mc->mc_ki[i] = mn.mc_ki[i];
- } else if (mc->mc_ki[mc->mc_top - 1] == 0) {
- for (i = 2; i <= mc->mc_top; ++i)
- if (mc->mc_ki[mc->mc_top - i]) {
- get_key(
- page_node(mc->mc_pg[mc->mc_top - i], mc->mc_ki[mc->mc_top - i]),
- &sepkey);
- if (mc->mc_dbx->md_cmp(newkey, &sepkey) < 0) {
- mc->mc_top -= i;
- mdbx_debug("update new-first on parent [%i] page %u key %s",
- mc->mc_ki[mc->mc_top], mc->mc_pg[mc->mc_top]->mp_pgno,
- DKEY(newkey));
- rc = mdbx_update_key(mc, newkey);
- mc->mc_top += i;
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- }
- break;
- }
- }
- } else if (!IS_LEAF2(mp)) {
- /* Move nodes */
- mc->mc_pg[mc->mc_top] = sister;
- i = split_indx;
- unsigned n = 0;
- pgno_t pgno = 0;
- do {
- mdbx_trace("i %u, nkeys %u => n %u, rp #%u", i, nkeys, n,
- sister->mp_pgno);
- MDBX_val *rdata = NULL;
- if (i == newindx) {
- rkey.iov_base = newkey->iov_base;
- rkey.iov_len = newkey->iov_len;
- if (IS_LEAF(mp))
- rdata = newdata;
- else
- pgno = newpgno;
- flags = nflags;
- /* Update index for the new key. */
- mc->mc_ki[mc->mc_top] = (indx_t)n;
- } else {
- MDBX_node *node =
- (MDBX_node *)((char *)mp + tmp_ki_copy->mp_ptrs[i] + PAGEHDRSZ);
- rkey.iov_base = node_key(node);
- rkey.iov_len = node_ks(node);
- if (IS_LEAF(mp)) {
- xdata.iov_base = node_data(node);
- xdata.iov_len = node_ds(node);
- rdata = &xdata;
- } else
- pgno = node_pgno(node);
- flags = node_flags(node);
- }
-
- switch (PAGETYPE(sister)) {
- case P_BRANCH: {
- mdbx_cassert(mc, 0 == (uint16_t)flags);
- /* First branch index doesn't need key data. */
- rc = mdbx_node_add_branch(mc, n, n ? &rkey : NULL, pgno);
- } break;
- case P_LEAF: {
- mdbx_cassert(mc, pgno == 0);
- mdbx_cassert(mc, rdata != NULL);
- rc = mdbx_node_add_leaf(mc, n, &rkey, rdata, flags);
- } break;
- /* case P_LEAF | P_LEAF2: {
- mdbx_cassert(mc, (nflags & (F_BIGDATA | F_SUBDATA | F_DUPDATA)) == 0);
- mdbx_cassert(mc, gno == 0);
- rc = mdbx_node_add_leaf2(mc, n, &rkey);
- } break; */
- default:
- rc = bad_page(sister, "wrong page-type %u\n", PAGETYPE(sister));
- }
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
-
- ++n;
- if (++i > nkeys) {
- i = 0;
- n = 0;
- mc->mc_pg[mc->mc_top] = tmp_ki_copy;
- mdbx_trace("switch to mp #%u", tmp_ki_copy->mp_pgno);
- }
- } while (i != split_indx);
-
- mdbx_trace("i %u, nkeys %u, n %u, pgno #%u", i, nkeys, n,
- mc->mc_pg[mc->mc_top]->mp_pgno);
-
- nkeys = page_numkeys(tmp_ki_copy);
- for (i = 0; i < nkeys; i++)
- mp->mp_ptrs[i] = tmp_ki_copy->mp_ptrs[i];
- mp->mp_lower = tmp_ki_copy->mp_lower;
- mp->mp_upper = tmp_ki_copy->mp_upper;
- memcpy(page_node(mp, nkeys - 1), page_node(tmp_ki_copy, nkeys - 1),
- env->me_psize - tmp_ki_copy->mp_upper - PAGEHDRSZ);
-
- /* reset back to original page */
- if (newindx < split_indx) {
- mc->mc_pg[mc->mc_top] = mp;
- } else {
- mc->mc_pg[mc->mc_top] = sister;
- mc->mc_ki[ptop]++;
- /* Make sure mc_ki is still valid. */
- if (mn.mc_pg[ptop] != mc->mc_pg[ptop] &&
- mc->mc_ki[ptop] >= page_numkeys(mc->mc_pg[ptop])) {
- for (i = 0; i <= ptop; i++) {
- mc->mc_pg[i] = mn.mc_pg[i];
- mc->mc_ki[i] = mn.mc_ki[i];
- }
- }
- }
- } else if (newindx >= split_indx) {
- mc->mc_pg[mc->mc_top] = sister;
- mc->mc_ki[ptop]++;
- /* Make sure mc_ki is still valid. */
- if (mn.mc_pg[ptop] != mc->mc_pg[ptop] &&
- mc->mc_ki[ptop] >= page_numkeys(mc->mc_pg[ptop])) {
- for (i = 0; i <= ptop; i++) {
- mc->mc_pg[i] = mn.mc_pg[i];
- mc->mc_ki[i] = mn.mc_ki[i];
- }
- }
- }
-
- /* Adjust other cursors pointing to mp and/or to parent page */
- nkeys = page_numkeys(mp);
- for (MDBX_cursor *m2 = mc->mc_txn->mt_cursors[mc->mc_dbi]; m2;
- m2 = m2->mc_next) {
- MDBX_cursor *m3 = (mc->mc_flags & C_SUB) ? &m2->mc_xcursor->mx_cursor : m2;
- if (m3 == mc)
- continue;
- if (!(m2->mc_flags & m3->mc_flags & C_INITIALIZED))
- continue;
- if (foliage) {
- /* sub cursors may be on different DB */
- if (m3->mc_pg[0] != mp)
- continue;
- /* root split */
- for (int k = foliage; k >= 0; k--) {
- m3->mc_ki[k + 1] = m3->mc_ki[k];
- m3->mc_pg[k + 1] = m3->mc_pg[k];
- }
- m3->mc_ki[0] = (m3->mc_ki[0] >= nkeys) ? 1 : 0;
- m3->mc_pg[0] = mc->mc_pg[0];
- m3->mc_snum++;
- m3->mc_top++;
- }
-
- if (m3->mc_top >= mc->mc_top && m3->mc_pg[mc->mc_top] == mp && !pure_left) {
- if (m3->mc_ki[mc->mc_top] >= newindx && !(nflags & MDBX_SPLIT_REPLACE))
- m3->mc_ki[mc->mc_top]++;
- if (m3->mc_ki[mc->mc_top] >= nkeys) {
- m3->mc_pg[mc->mc_top] = sister;
- mdbx_cassert(mc, m3->mc_ki[mc->mc_top] >= nkeys);
- m3->mc_ki[mc->mc_top] -= (indx_t)nkeys;
- for (i = 0; i < mc->mc_top; i++) {
- m3->mc_ki[i] = mn.mc_ki[i];
- m3->mc_pg[i] = mn.mc_pg[i];
- }
- }
- } else if (!did_split_parent && m3->mc_top >= ptop &&
- m3->mc_pg[ptop] == mc->mc_pg[ptop] &&
- m3->mc_ki[ptop] >= mc->mc_ki[ptop]) {
- m3->mc_ki[ptop]++; /* also for the `pure-left` case */
- }
- if (XCURSOR_INITED(m3) && IS_LEAF(mp))
- XCURSOR_REFRESH(m3, m3->mc_pg[mc->mc_top], m3->mc_ki[mc->mc_top]);
- }
- mdbx_trace("mp #%u left: %d, sister #%u left: %d", mp->mp_pgno, page_room(mp),
- sister->mp_pgno, page_room(sister));
-
-done:
- if (tmp_ki_copy)
- mdbx_dpage_free(env, tmp_ki_copy, 1);
-
- if (unlikely(rc != MDBX_SUCCESS))
- mc->mc_txn->mt_flags |= MDBX_TXN_ERROR;
- else {
- if (mdbx_audit_enabled())
- rc = mdbx_cursor_check(mc, C_UPDATING);
- if (unlikely(nflags & MDBX_RESERVE)) {
- MDBX_node *node = page_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
- if (!(node_flags(node) & F_BIGDATA))
- newdata->iov_base = node_data(node);
- }
-#if MDBX_ENABLE_PGOP_STAT
- env->me_lck->mti_pgop_stat.split.weak += 1;
-#endif /* MDBX_ENABLE_PGOP_STAT */
- }
-
- mdbx_debug("<< mp #%u, rc %d", mp->mp_pgno, rc);
- return rc;
-}
-
-int mdbx_put(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key, MDBX_val *data,
- unsigned flags) {
- int rc = check_txn_rw(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!key || !data))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- if (unlikely(flags & ~(MDBX_NOOVERWRITE | MDBX_NODUPDATA | MDBX_ALLDUPS |
- MDBX_ALLDUPS | MDBX_RESERVE | MDBX_APPEND |
- MDBX_APPENDDUP | MDBX_CURRENT | MDBX_MULTIPLE)))
- return MDBX_EINVAL;
-
- if (unlikely(txn->mt_flags & (MDBX_TXN_RDONLY | MDBX_TXN_BLOCKED)))
- return (txn->mt_flags & MDBX_TXN_RDONLY) ? MDBX_EACCESS : MDBX_BAD_TXN;
-
- MDBX_cursor_couple cx;
- rc = mdbx_cursor_init(&cx.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- cx.outer.mc_next = txn->mt_cursors[dbi];
- txn->mt_cursors[dbi] = &cx.outer;
-
- /* LY: support for update (explicit overwrite) */
- if (flags & MDBX_CURRENT) {
- rc = mdbx_cursor_get(&cx.outer, (MDBX_val *)key, NULL, MDBX_SET);
- if (likely(rc == MDBX_SUCCESS) &&
- (txn->mt_dbs[dbi].md_flags & MDBX_DUPSORT) &&
- (flags & MDBX_ALLDUPS) == 0) {
- /* LY: allows update (explicit overwrite) only for unique keys */
- MDBX_node *node = page_node(cx.outer.mc_pg[cx.outer.mc_top],
- cx.outer.mc_ki[cx.outer.mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- mdbx_tassert(txn, XCURSOR_INITED(&cx.outer) &&
- cx.outer.mc_xcursor->mx_db.md_entries > 1);
- rc = MDBX_EMULTIVAL;
- }
- }
- }
-
- if (likely(rc == MDBX_SUCCESS))
- rc = mdbx_cursor_put(&cx.outer, key, data, flags);
- txn->mt_cursors[dbi] = cx.outer.mc_next;
-
- return rc;
-}
-
-/**** COPYING *****************************************************************/
-
-/* State needed for a double-buffering compacting copy. */
-typedef struct mdbx_copy {
- MDBX_env *mc_env;
- MDBX_txn *mc_txn;
- mdbx_condpair_t mc_condpair;
- uint8_t *mc_wbuf[2];
- uint8_t *mc_over[2];
- size_t mc_wlen[2];
- size_t mc_olen[2];
- mdbx_filehandle_t mc_fd;
- /* Error code. Never cleared if set. Both threads can set nonzero
- * to fail the copy. Not mutex-protected, MDBX expects atomic int. */
- volatile int mc_error;
- pgno_t mc_next_pgno;
- volatile unsigned mc_head;
- volatile unsigned mc_tail;
-} mdbx_copy;
-
-/* Dedicated writer thread for compacting copy. */
-__cold static THREAD_RESULT THREAD_CALL mdbx_env_copythr(void *arg) {
- mdbx_copy *my = arg;
-
-#if defined(EPIPE) && !(defined(_WIN32) || defined(_WIN64))
- sigset_t sigset;
- sigemptyset(&sigset);
- sigaddset(&sigset, SIGPIPE);
- my->mc_error = pthread_sigmask(SIG_BLOCK, &sigset, NULL);
-#endif /* EPIPE */
-
- mdbx_condpair_lock(&my->mc_condpair);
- while (!my->mc_error) {
- while (my->mc_tail == my->mc_head && !my->mc_error) {
- int err = mdbx_condpair_wait(&my->mc_condpair, true);
- if (err != MDBX_SUCCESS) {
- my->mc_error = err;
- goto bailout;
- }
- }
- const unsigned toggle = my->mc_tail & 1;
- size_t wsize = my->mc_wlen[toggle];
- if (wsize == 0) {
- my->mc_tail += 1;
- break /* EOF */;
- }
- my->mc_wlen[toggle] = 0;
- uint8_t *ptr = my->mc_wbuf[toggle];
- again:
- if (!my->mc_error) {
- int err = mdbx_write(my->mc_fd, ptr, wsize);
- if (err != MDBX_SUCCESS) {
-#if defined(EPIPE) && !(defined(_WIN32) || defined(_WIN64))
- if (err == EPIPE) {
- /* Collect the pending SIGPIPE,
- * otherwise at least OS X gives it to the process on thread-exit. */
- int unused;
- sigwait(&sigset, &unused);
- }
-#endif /* EPIPE */
- my->mc_error = err;
- goto bailout;
- }
- }
-
- /* If there's an overflow page tail, write it too */
- wsize = my->mc_olen[toggle];
- if (wsize) {
- my->mc_olen[toggle] = 0;
- ptr = my->mc_over[toggle];
- goto again;
- }
- my->mc_tail += 1;
- mdbx_condpair_signal(&my->mc_condpair, false);
- }
-bailout:
- mdbx_condpair_unlock(&my->mc_condpair);
- return (THREAD_RESULT)0;
-}
-
-/* Give buffer and/or MDBX_EOF to writer thread, await unused buffer. */
-__cold static int mdbx_env_cthr_toggle(mdbx_copy *my) {
- mdbx_condpair_lock(&my->mc_condpair);
- mdbx_assert(my->mc_env, my->mc_head - my->mc_tail < 2 || my->mc_error);
- my->mc_head += 1;
- mdbx_condpair_signal(&my->mc_condpair, true);
- while (!my->mc_error &&
- my->mc_head - my->mc_tail == 2 /* both buffers in use */) {
- int err = mdbx_condpair_wait(&my->mc_condpair, false);
- if (err != MDBX_SUCCESS)
- my->mc_error = err;
- }
- mdbx_condpair_unlock(&my->mc_condpair);
- return my->mc_error;
-}
-
-/* Depth-first tree traversal for compacting copy.
- * [in] my control structure.
- * [in,out] pg database root.
- * [in] flags includes F_DUPDATA if it is a sorted-duplicate sub-DB. */
-__cold static int mdbx_env_cwalk(mdbx_copy *my, pgno_t *pg, int flags) {
- MDBX_cursor_couple couple;
- MDBX_page *mo, *mp, *leaf;
- char *buf, *ptr;
- int rc;
- unsigned i;
-
- /* Empty DB, nothing to do */
- if (*pg == P_INVALID)
- return MDBX_SUCCESS;
-
- memset(&couple, 0, sizeof(couple));
- couple.outer.mc_snum = 1;
- couple.outer.mc_txn = my->mc_txn;
- couple.outer.mc_flags = couple.inner.mx_cursor.mc_flags =
- C_COPYING | C_SKIPORD;
-
- rc = mdbx_page_get(&couple.outer, *pg, &couple.outer.mc_pg[0],
- my->mc_txn->mt_txnid);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- rc = mdbx_page_search_root(&couple.outer, NULL, MDBX_PS_FIRST);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- /* Make cursor pages writable */
- buf = ptr = mdbx_malloc(pgno2bytes(my->mc_env, couple.outer.mc_snum));
- if (buf == NULL)
- return MDBX_ENOMEM;
-
- for (i = 0; i < couple.outer.mc_top; i++) {
- mdbx_page_copy((MDBX_page *)ptr, couple.outer.mc_pg[i],
- my->mc_env->me_psize);
- couple.outer.mc_pg[i] = (MDBX_page *)ptr;
- ptr += my->mc_env->me_psize;
- }
-
- /* This is writable space for a leaf page. Usually not needed. */
- leaf = (MDBX_page *)ptr;
-
- while (couple.outer.mc_snum > 0) {
- mp = couple.outer.mc_pg[couple.outer.mc_top];
- unsigned n = page_numkeys(mp);
-
- if (IS_LEAF(mp)) {
- if (!IS_LEAF2(mp) && !(flags & F_DUPDATA)) {
- for (i = 0; i < n; i++) {
- MDBX_node *node = page_node(mp, i);
- if (node_flags(node) & F_BIGDATA) {
- MDBX_page *omp;
-
- /* Need writable leaf */
- if (mp != leaf) {
- couple.outer.mc_pg[couple.outer.mc_top] = leaf;
- mdbx_page_copy(leaf, mp, my->mc_env->me_psize);
- mp = leaf;
- node = page_node(mp, i);
- }
-
- const pgno_t pgno = node_largedata_pgno(node);
- poke_pgno(node_data(node), my->mc_next_pgno);
- rc = mdbx_page_get(&couple.outer, pgno, &omp,
- pp_txnid4chk(mp, my->mc_txn));
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- unsigned toggle = my->mc_head & 1;
- if (my->mc_wlen[toggle] + my->mc_env->me_psize >
- ((size_t)(MDBX_ENVCOPY_WRITEBUF))) {
- rc = mdbx_env_cthr_toggle(my);
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- toggle = my->mc_head & 1;
- }
- mo = (MDBX_page *)(my->mc_wbuf[toggle] + my->mc_wlen[toggle]);
- memcpy(mo, omp, my->mc_env->me_psize);
- mo->mp_pgno = my->mc_next_pgno;
- my->mc_next_pgno += omp->mp_pages;
- my->mc_wlen[toggle] += my->mc_env->me_psize;
- if (omp->mp_pages > 1) {
- my->mc_olen[toggle] = pgno2bytes(my->mc_env, omp->mp_pages - 1);
- my->mc_over[toggle] = (uint8_t *)omp + my->mc_env->me_psize;
- rc = mdbx_env_cthr_toggle(my);
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- toggle = my->mc_head & 1;
- }
- } else if (node_flags(node) & F_SUBDATA) {
- if (!MDBX_DISABLE_PAGECHECKS &&
- unlikely(node_ds(node) != sizeof(MDBX_db))) {
- rc = MDBX_CORRUPTED;
- goto done;
- }
-
- /* Need writable leaf */
- if (mp != leaf) {
- couple.outer.mc_pg[couple.outer.mc_top] = leaf;
- mdbx_page_copy(leaf, mp, my->mc_env->me_psize);
- mp = leaf;
- node = page_node(mp, i);
- }
-
- MDBX_db db;
- memcpy(&db, node_data(node), sizeof(MDBX_db));
- rc = mdbx_env_cwalk(my, &db.md_root, node_flags(node) & F_DUPDATA);
- if (rc)
- goto done;
- memcpy(node_data(node), &db, sizeof(MDBX_db));
- }
- }
- }
- } else {
- couple.outer.mc_ki[couple.outer.mc_top]++;
- if (couple.outer.mc_ki[couple.outer.mc_top] < n) {
- again:
- rc = mdbx_page_get(
- &couple.outer,
- node_pgno(page_node(mp, couple.outer.mc_ki[couple.outer.mc_top])),
- &mp, pp_txnid4chk(mp, my->mc_txn));
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- couple.outer.mc_top++;
- couple.outer.mc_snum++;
- couple.outer.mc_ki[couple.outer.mc_top] = 0;
- if (IS_BRANCH(mp)) {
- /* Whenever we advance to a sibling branch page,
- * we must proceed all the way down to its first leaf. */
- mdbx_page_copy(couple.outer.mc_pg[couple.outer.mc_top], mp,
- my->mc_env->me_psize);
- goto again;
- } else
- couple.outer.mc_pg[couple.outer.mc_top] = mp;
- continue;
- }
- }
- unsigned toggle = my->mc_head & 1;
- if (my->mc_wlen[toggle] + my->mc_wlen[toggle] >
- ((size_t)(MDBX_ENVCOPY_WRITEBUF))) {
- rc = mdbx_env_cthr_toggle(my);
- if (unlikely(rc != MDBX_SUCCESS))
- goto done;
- toggle = my->mc_head & 1;
- }
- mo = (MDBX_page *)(my->mc_wbuf[toggle] + my->mc_wlen[toggle]);
- mdbx_page_copy(mo, mp, my->mc_env->me_psize);
- mo->mp_pgno = my->mc_next_pgno++;
- my->mc_wlen[toggle] += my->mc_env->me_psize;
- if (couple.outer.mc_top) {
- /* Update parent if there is one */
- node_set_pgno(page_node(couple.outer.mc_pg[couple.outer.mc_top - 1],
- couple.outer.mc_ki[couple.outer.mc_top - 1]),
- mo->mp_pgno);
- mdbx_cursor_pop(&couple.outer);
- } else {
- /* Otherwise we're done */
- *pg = mo->mp_pgno;
- break;
- }
- }
-done:
- mdbx_free(buf);
- return rc;
-}
-
-__cold static void compact_fixup_meta(MDBX_env *env, MDBX_meta *meta) {
- /* Calculate filesize taking in account shrink/growing thresholds */
- if (meta->mm_geo.next != meta->mm_geo.now) {
- meta->mm_geo.now = meta->mm_geo.next;
- const pgno_t aligner = pv2pages(
- meta->mm_geo.grow_pv ? meta->mm_geo.grow_pv : meta->mm_geo.shrink_pv);
- if (aligner) {
- const pgno_t aligned = pgno_align2os_pgno(
- env, meta->mm_geo.next + aligner - meta->mm_geo.next % aligner);
- meta->mm_geo.now = aligned;
- }
- }
-
- if (meta->mm_geo.now < meta->mm_geo.lower)
- meta->mm_geo.now = meta->mm_geo.lower;
- if (meta->mm_geo.now > meta->mm_geo.upper)
- meta->mm_geo.now = meta->mm_geo.upper;
-
- /* Update signature */
- assert(meta->mm_geo.now >= meta->mm_geo.next);
- unaligned_poke_u64(4, meta->mm_datasync_sign, meta_sign(meta));
-}
-
-/* Make resizeable */
-__cold static void make_sizeable(MDBX_meta *meta) {
- meta->mm_geo.lower = MIN_PAGENO;
- if (meta->mm_geo.grow_pv == 0) {
- const pgno_t step = 1 + (meta->mm_geo.upper - meta->mm_geo.lower) / 42;
- meta->mm_geo.grow_pv = pages2pv(step);
- }
- if (meta->mm_geo.shrink_pv == 0) {
- const pgno_t step = pv2pages(meta->mm_geo.grow_pv) << 1;
- meta->mm_geo.shrink_pv = pages2pv(step);
- }
-}
-
-/* Copy environment with compaction. */
-__cold static int mdbx_env_compact(MDBX_env *env, MDBX_txn *read_txn,
- mdbx_filehandle_t fd, uint8_t *buffer,
- const bool dest_is_pipe, const int flags) {
- const size_t meta_bytes = pgno2bytes(env, NUM_METAS);
- uint8_t *const data_buffer =
- buffer + ceil_powerof2(meta_bytes, env->me_os_psize);
- MDBX_meta *const meta = mdbx_init_metas(env, buffer);
- meta_set_txnid(env, meta, read_txn->mt_txnid);
-
- if (flags & MDBX_CP_FORCE_DYNAMIC_SIZE)
- make_sizeable(meta);
-
- /* copy canary sequences if present */
- if (read_txn->mt_canary.v) {
- meta->mm_canary = read_txn->mt_canary;
- meta->mm_canary.v = constmeta_txnid(env, meta);
- }
-
- /* Set metapage 1 with current main DB */
- pgno_t new_root, root = read_txn->mt_dbs[MAIN_DBI].md_root;
- if ((new_root = root) == P_INVALID) {
- /* When the DB is empty, handle it specially to
- * fix any breakage like page leaks from ITS#8174. */
- meta->mm_dbs[MAIN_DBI].md_flags = read_txn->mt_dbs[MAIN_DBI].md_flags;
- compact_fixup_meta(env, meta);
- if (dest_is_pipe) {
- int rc = mdbx_write(fd, buffer, meta_bytes);
- if (rc != MDBX_SUCCESS)
- return rc;
- }
- } else {
- /* Count free pages + GC pages. Subtract from last_pg
- * to find the new last_pg, which also becomes the new root. */
- pgno_t freecount = 0;
- MDBX_cursor_couple couple;
- MDBX_val key, data;
-
- int rc = mdbx_cursor_init(&couple.outer, read_txn, FREE_DBI);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- while ((rc = mdbx_cursor_get(&couple.outer, &key, &data, MDBX_NEXT)) == 0)
- freecount += *(pgno_t *)data.iov_base;
- if (unlikely(rc != MDBX_NOTFOUND))
- return rc;
-
- freecount += read_txn->mt_dbs[FREE_DBI].md_branch_pages +
- read_txn->mt_dbs[FREE_DBI].md_leaf_pages +
- read_txn->mt_dbs[FREE_DBI].md_overflow_pages;
-
- new_root = read_txn->mt_next_pgno - 1 - freecount;
- meta->mm_geo.next = new_root + 1;
- meta->mm_dbs[MAIN_DBI] = read_txn->mt_dbs[MAIN_DBI];
- meta->mm_dbs[MAIN_DBI].md_root = new_root;
-
- mdbx_copy ctx;
- memset(&ctx, 0, sizeof(ctx));
- rc = mdbx_condpair_init(&ctx.mc_condpair);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- memset(data_buffer, 0, ((size_t)(MDBX_ENVCOPY_WRITEBUF)) * 2);
- ctx.mc_wbuf[0] = data_buffer;
- ctx.mc_wbuf[1] = data_buffer + ((size_t)(MDBX_ENVCOPY_WRITEBUF));
- ctx.mc_next_pgno = NUM_METAS;
- ctx.mc_env = env;
- ctx.mc_fd = fd;
- ctx.mc_txn = read_txn;
-
- mdbx_thread_t thread;
- int thread_err = mdbx_thread_create(&thread, mdbx_env_copythr, &ctx);
- if (likely(thread_err == MDBX_SUCCESS)) {
- if (dest_is_pipe) {
- compact_fixup_meta(env, meta);
- rc = mdbx_write(fd, buffer, meta_bytes);
- }
- if (rc == MDBX_SUCCESS)
- rc = mdbx_env_cwalk(&ctx, &root, 0);
- mdbx_env_cthr_toggle(&ctx);
- mdbx_env_cthr_toggle(&ctx);
- thread_err = mdbx_thread_join(thread);
- mdbx_assert(env, (ctx.mc_tail == ctx.mc_head &&
- ctx.mc_wlen[ctx.mc_head & 1] == 0) ||
- ctx.mc_error);
- mdbx_condpair_destroy(&ctx.mc_condpair);
- }
- if (unlikely(thread_err != MDBX_SUCCESS))
- return thread_err;
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- if (unlikely(ctx.mc_error != MDBX_SUCCESS))
- return ctx.mc_error;
-
- if (dest_is_pipe) {
- if (unlikely(root != new_root)) {
- mdbx_error("post-compactification root %" PRIaPGNO
- " NE expected %" PRIaPGNO
- " (source DB corrupted or has a page leak(s))",
- root, new_root);
- return MDBX_CORRUPTED; /* page leak or corrupt DB */
- }
- } else {
- if (unlikely(root > new_root)) {
- mdbx_error("post-compactification root %" PRIaPGNO
- " GT expected %" PRIaPGNO " (source DB corrupted)",
- root, new_root);
- return MDBX_CORRUPTED; /* page leak or corrupt DB */
- }
- if (unlikely(root < new_root)) {
- mdbx_warning("post-compactification root %" PRIaPGNO
- " LT expected %" PRIaPGNO " (page leak(s) in source DB)",
- root, new_root);
- /* fixup meta */
- meta->mm_dbs[MAIN_DBI].md_root = root;
- meta->mm_geo.next = root + 1;
- }
- compact_fixup_meta(env, meta);
- }
- }
-
- /* Extend file if required */
- if (meta->mm_geo.now != meta->mm_geo.next) {
- const size_t whole_size = pgno2bytes(env, meta->mm_geo.now);
- if (!dest_is_pipe)
- return mdbx_ftruncate(fd, whole_size);
-
- const size_t used_size = pgno2bytes(env, meta->mm_geo.next);
- memset(data_buffer, 0, ((size_t)(MDBX_ENVCOPY_WRITEBUF)));
- for (size_t offset = used_size; offset < whole_size;) {
- const size_t chunk =
- (((size_t)(MDBX_ENVCOPY_WRITEBUF)) < whole_size - offset)
- ? ((size_t)(MDBX_ENVCOPY_WRITEBUF))
- : whole_size - offset;
- /* copy to avoid EFAULT in case swapped-out */
- int rc = mdbx_write(fd, data_buffer, chunk);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- offset += chunk;
- }
- }
- return MDBX_SUCCESS;
-}
-
-/* Copy environment as-is. */
-__cold static int mdbx_env_copy_asis(MDBX_env *env, MDBX_txn *read_txn,
- mdbx_filehandle_t fd, uint8_t *buffer,
- const bool dest_is_pipe, const int flags) {
- /* We must start the actual read txn after blocking writers */
- int rc = mdbx_txn_end(read_txn, MDBX_END_RESET_TMP);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- /* Temporarily block writers until we snapshot the meta pages */
- rc = mdbx_txn_lock(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- rc = mdbx_txn_renew0(read_txn, MDBX_TXN_RDONLY);
- if (unlikely(rc != MDBX_SUCCESS)) {
- mdbx_txn_unlock(env);
- return rc;
- }
-
- mdbx_jitter4testing(false);
- const size_t meta_bytes = pgno2bytes(env, NUM_METAS);
- /* Make a snapshot of meta-pages,
- * but writing ones after the data was flushed */
- memcpy(buffer, env->me_map, meta_bytes);
- MDBX_meta *const headcopy = /* LY: get pointer to the snapshot copy */
- (MDBX_meta *)(buffer + ((uint8_t *)meta_prefer_last(env) - env->me_map));
- mdbx_txn_unlock(env);
-
- if (flags & MDBX_CP_FORCE_DYNAMIC_SIZE)
- make_sizeable(headcopy);
- /* Update signature to steady */
- unaligned_poke_u64(4, headcopy->mm_datasync_sign, meta_sign(headcopy));
-
- /* Copy the data */
- const size_t whole_size = pgno_align2os_bytes(env, read_txn->mt_end_pgno);
- const size_t used_size = pgno2bytes(env, read_txn->mt_next_pgno);
- mdbx_jitter4testing(false);
-
- if (dest_is_pipe)
- rc = mdbx_write(fd, buffer, meta_bytes);
-
- uint8_t *const data_buffer =
- buffer + ceil_powerof2(meta_bytes, env->me_os_psize);
-#if MDBX_USE_COPYFILERANGE
- static bool copyfilerange_unavailable;
- bool not_the_same_filesystem = false;
-#endif /* MDBX_USE_COPYFILERANGE */
- for (size_t offset = meta_bytes; rc == MDBX_SUCCESS && offset < used_size;) {
-#if MDBX_USE_SENDFILE
- static bool sendfile_unavailable;
- if (dest_is_pipe && likely(!sendfile_unavailable)) {
- off_t in_offset = offset;
- const ssize_t written =
- sendfile(fd, env->me_lazy_fd, &in_offset, used_size - offset);
- if (likely(written > 0)) {
- offset = in_offset;
- continue;
- }
- rc = MDBX_ENODATA;
- if (written == 0 || ignore_enosys(rc = errno) != MDBX_RESULT_TRUE)
- break;
- sendfile_unavailable = true;
- }
-#endif /* MDBX_USE_SENDFILE */
-
-#if MDBX_USE_COPYFILERANGE
- if (!dest_is_pipe && !not_the_same_filesystem &&
- likely(!copyfilerange_unavailable)) {
- off_t in_offset = offset, out_offset = offset;
- ssize_t bytes_copied = copy_file_range(
- env->me_lazy_fd, &in_offset, fd, &out_offset, used_size - offset, 0);
- if (likely(bytes_copied > 0)) {
- offset = in_offset;
- continue;
- }
- rc = MDBX_ENODATA;
- if (bytes_copied == 0)
- break;
- rc = errno;
- if (rc == EXDEV)
- not_the_same_filesystem = true;
- else if (ignore_enosys(rc) == MDBX_RESULT_TRUE)
- copyfilerange_unavailable = true;
- else
- break;
- }
-#endif /* MDBX_USE_COPYFILERANGE */
-
- /* fallback to portable */
- const size_t chunk =
- (((size_t)(MDBX_ENVCOPY_WRITEBUF)) < used_size - offset)
- ? ((size_t)(MDBX_ENVCOPY_WRITEBUF))
- : used_size - offset;
- /* copy to avoid EFAULT in case swapped-out */
- memcpy(data_buffer, env->me_map + offset, chunk);
- rc = mdbx_write(fd, data_buffer, chunk);
- offset += chunk;
- }
-
- /* Extend file if required */
- if (likely(rc == MDBX_SUCCESS) && whole_size != used_size) {
- if (!dest_is_pipe)
- rc = mdbx_ftruncate(fd, whole_size);
- else {
- memset(data_buffer, 0, ((size_t)(MDBX_ENVCOPY_WRITEBUF)));
- for (size_t offset = used_size;
- rc == MDBX_SUCCESS && offset < whole_size;) {
- const size_t chunk =
- (((size_t)(MDBX_ENVCOPY_WRITEBUF)) < whole_size - offset)
- ? ((size_t)(MDBX_ENVCOPY_WRITEBUF))
- : whole_size - offset;
- /* copy to avoid EFAULT in case swapped-out */
- rc = mdbx_write(fd, data_buffer, chunk);
- offset += chunk;
- }
- }
- }
-
- return rc;
-}
-
-__cold int mdbx_env_copy2fd(MDBX_env *env, mdbx_filehandle_t fd,
- unsigned flags) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- const int dest_is_pipe = mdbx_is_pipe(fd);
- if (MDBX_IS_ERROR(dest_is_pipe))
- return dest_is_pipe;
-
- if (!dest_is_pipe) {
- rc = mdbx_fseek(fd, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
-
- const size_t buffer_size =
- pgno_align2os_bytes(env, NUM_METAS) +
- ceil_powerof2(((flags & MDBX_CP_COMPACT)
- ? ((size_t)(MDBX_ENVCOPY_WRITEBUF)) * 2
- : ((size_t)(MDBX_ENVCOPY_WRITEBUF))),
- env->me_os_psize);
-
- uint8_t *buffer = NULL;
- rc = mdbx_memalign_alloc(env->me_os_psize, buffer_size, (void **)&buffer);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- MDBX_txn *read_txn = NULL;
- /* Do the lock/unlock of the reader mutex before starting the
- * write txn. Otherwise other read txns could block writers. */
- rc = mdbx_txn_begin(env, NULL, MDBX_TXN_RDONLY, &read_txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- mdbx_memalign_free(buffer);
- return rc;
- }
-
- if (!dest_is_pipe) {
- /* Firstly write a stub to meta-pages.
- * Now we sure to incomplete copy will not be used. */
- memset(buffer, -1, pgno2bytes(env, NUM_METAS));
- rc = mdbx_write(fd, buffer, pgno2bytes(env, NUM_METAS));
- }
-
- if (likely(rc == MDBX_SUCCESS)) {
- memset(buffer, 0, pgno2bytes(env, NUM_METAS));
- rc = ((flags & MDBX_CP_COMPACT) ? mdbx_env_compact : mdbx_env_copy_asis)(
- env, read_txn, fd, buffer, dest_is_pipe, flags);
- }
- mdbx_txn_abort(read_txn);
-
- if (!dest_is_pipe) {
- if (likely(rc == MDBX_SUCCESS))
- rc = mdbx_fsync(fd, MDBX_SYNC_DATA | MDBX_SYNC_SIZE);
-
- /* Write actual meta */
- if (likely(rc == MDBX_SUCCESS))
- rc = mdbx_pwrite(fd, buffer, pgno2bytes(env, NUM_METAS), 0);
-
- if (likely(rc == MDBX_SUCCESS))
- rc = mdbx_fsync(fd, MDBX_SYNC_DATA | MDBX_SYNC_IODQ);
- }
-
- mdbx_memalign_free(buffer);
- return rc;
-}
-
-__cold int mdbx_env_copy(MDBX_env *env, const char *dest_path,
- MDBX_copy_flags_t flags) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!dest_path))
- return MDBX_EINVAL;
-
- /* The destination path must exist, but the destination file must not.
- * We don't want the OS to cache the writes, since the source data is
- * already in the OS cache. */
- mdbx_filehandle_t newfd;
- rc = mdbx_openfile(MDBX_OPEN_COPY, env, dest_path, &newfd,
-#if defined(_WIN32) || defined(_WIN64)
- (mdbx_mode_t)-1
-#else
- S_IRUSR | S_IWUSR | S_IRGRP | S_IWGRP
-#endif
- );
-
- if (rc == MDBX_SUCCESS) {
-#if defined(_WIN32) || defined(_WIN64)
- OVERLAPPED ov;
- memset(&ov, 0, sizeof(ov));
- if (!LockFileEx(newfd, LOCKFILE_EXCLUSIVE_LOCK | LOCKFILE_FAIL_IMMEDIATELY,
- 0, 0, INT32_MAX, &ov))
- rc = GetLastError();
-#else
- struct flock lock_op;
- memset(&lock_op, 0, sizeof(lock_op));
- lock_op.l_type = F_WRLCK;
- lock_op.l_whence = SEEK_SET;
- lock_op.l_start = 0;
- lock_op.l_len =
- (sizeof(lock_op.l_len) > 4 ? INT64_MAX : INT32_MAX) & ~(size_t)0xffff;
- if (fcntl(newfd, F_SETLK, &lock_op)
-#if (defined(__linux__) || defined(__gnu_linux__)) && defined(LOCK_EX) && \
- (!defined(__ANDROID_API__) || __ANDROID_API__ >= 24)
- || flock(newfd, LOCK_EX | LOCK_NB)
-#endif /* Linux */
- )
- rc = errno;
-#endif /* Windows / POSIX */
- }
-
- if (rc == MDBX_SUCCESS)
- rc = mdbx_env_copy2fd(env, newfd, flags);
-
- if (newfd != INVALID_HANDLE_VALUE) {
- int err = mdbx_closefile(newfd);
- if (rc == MDBX_SUCCESS && err != rc)
- rc = err;
- if (rc != MDBX_SUCCESS)
- (void)mdbx_removefile(dest_path);
- }
-
- return rc;
-}
-
-/******************************************************************************/
-
-__cold int mdbx_env_set_flags(MDBX_env *env, MDBX_env_flags_t flags,
- bool onoff) {
- int rc = check_env(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(flags &
- ((env->me_flags & MDBX_ENV_ACTIVE) ? ~ENV_CHANGEABLE_FLAGS
- : ~ENV_USABLE_FLAGS)))
- return MDBX_EPERM;
-
- if (unlikely(env->me_flags & MDBX_RDONLY))
- return MDBX_EACCESS;
-
- if ((env->me_flags & MDBX_ENV_ACTIVE) &&
- unlikely(env->me_txn0->mt_owner == mdbx_thread_self()))
- return MDBX_BUSY;
-
- const bool lock_needed = (env->me_flags & MDBX_ENV_ACTIVE) &&
- env->me_txn0->mt_owner != mdbx_thread_self();
- bool should_unlock = false;
- if (lock_needed) {
- rc = mdbx_txn_lock(env, false);
- if (unlikely(rc))
- return rc;
- should_unlock = true;
- }
-
- if (onoff)
- env->me_flags = merge_sync_flags(env->me_flags, flags);
- else
- env->me_flags &= ~flags;
-
- if (should_unlock)
- mdbx_txn_unlock(env);
- return MDBX_SUCCESS;
-}
-
-__cold int mdbx_env_get_flags(const MDBX_env *env, unsigned *arg) {
- int rc = check_env(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!arg))
- return MDBX_EINVAL;
-
- *arg = env->me_flags & ENV_USABLE_FLAGS;
- return MDBX_SUCCESS;
-}
-
-__cold int mdbx_env_set_userctx(MDBX_env *env, void *ctx) {
- int rc = check_env(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- env->me_userctx = ctx;
- return MDBX_SUCCESS;
-}
-
-__cold void *mdbx_env_get_userctx(const MDBX_env *env) {
- return env ? env->me_userctx : NULL;
-}
-
-__cold int mdbx_env_set_assert(MDBX_env *env, MDBX_assert_func *func) {
- int rc = check_env(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
-#if MDBX_DEBUG
- env->me_assert_func = func;
- return MDBX_SUCCESS;
-#else
- (void)func;
- return MDBX_ENOSYS;
-#endif
-}
-
-__cold int mdbx_env_get_path(const MDBX_env *env, const char **arg) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!arg))
- return MDBX_EINVAL;
-
- *arg = env->me_pathname;
- return MDBX_SUCCESS;
-}
-
-__cold int mdbx_env_get_fd(const MDBX_env *env, mdbx_filehandle_t *arg) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!arg))
- return MDBX_EINVAL;
-
- *arg = env->me_lazy_fd;
- return MDBX_SUCCESS;
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-__cold int mdbx_env_stat(const MDBX_env *env, MDBX_stat *stat, size_t bytes) {
- return __inline_mdbx_env_stat(env, stat, bytes);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-static void stat_get(const MDBX_db *db, MDBX_stat *st, size_t bytes) {
- st->ms_depth = db->md_depth;
- st->ms_branch_pages = db->md_branch_pages;
- st->ms_leaf_pages = db->md_leaf_pages;
- st->ms_overflow_pages = db->md_overflow_pages;
- st->ms_entries = db->md_entries;
- if (likely(bytes >=
- offsetof(MDBX_stat, ms_mod_txnid) + sizeof(st->ms_mod_txnid)))
- st->ms_mod_txnid = db->md_mod_txnid;
-}
-
-static void stat_add(const MDBX_db *db, MDBX_stat *const st,
- const size_t bytes) {
- st->ms_depth += db->md_depth;
- st->ms_branch_pages += db->md_branch_pages;
- st->ms_leaf_pages += db->md_leaf_pages;
- st->ms_overflow_pages += db->md_overflow_pages;
- st->ms_entries += db->md_entries;
- if (likely(bytes >=
- offsetof(MDBX_stat, ms_mod_txnid) + sizeof(st->ms_mod_txnid)))
- st->ms_mod_txnid = (st->ms_mod_txnid > db->md_mod_txnid) ? st->ms_mod_txnid
- : db->md_mod_txnid;
-}
-
-__cold static int stat_acc(const MDBX_txn *txn, MDBX_stat *st, size_t bytes) {
- int err = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- st->ms_psize = txn->mt_env->me_psize;
-#if 1
- /* assuming GC is internal and not subject for accounting */
- stat_get(&txn->mt_dbs[MAIN_DBI], st, bytes);
-#else
- stat_get(&txn->mt_dbs[FREE_DBI], st, bytes);
- stat_add(&txn->mt_dbs[MAIN_DBI], st, bytes);
-#endif
-
- /* account opened named subDBs */
- for (MDBX_dbi dbi = CORE_DBS; dbi < txn->mt_numdbs; dbi++)
- if ((txn->mt_dbistate[dbi] & (DBI_VALID | DBI_STALE)) == DBI_VALID)
- stat_add(txn->mt_dbs + dbi, st, bytes);
-
- if (!(txn->mt_dbs[MAIN_DBI].md_flags & (MDBX_DUPSORT | MDBX_INTEGERKEY)) &&
- txn->mt_dbs[MAIN_DBI].md_entries /* TODO: use `md_subs` field */) {
- MDBX_cursor_couple cx;
- err = mdbx_cursor_init(&cx.outer, (MDBX_txn *)txn, MAIN_DBI);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- /* scan and account not opened named subDBs */
- err = mdbx_page_search(&cx.outer, NULL, MDBX_PS_FIRST);
- while (err == MDBX_SUCCESS) {
- const MDBX_page *mp = cx.outer.mc_pg[cx.outer.mc_top];
- for (unsigned i = 0; i < page_numkeys(mp); i++) {
- const MDBX_node *node = page_node(mp, i);
- if (node_flags(node) != F_SUBDATA)
- continue;
- if (unlikely(node_ds(node) != sizeof(MDBX_db)))
- return MDBX_CORRUPTED;
-
- /* skip opened and already accounted */
- for (MDBX_dbi dbi = CORE_DBS; dbi < txn->mt_numdbs; dbi++)
- if ((txn->mt_dbistate[dbi] & (DBI_VALID | DBI_STALE)) == DBI_VALID &&
- node_ks(node) == txn->mt_dbxs[dbi].md_name.iov_len &&
- memcmp(node_key(node), txn->mt_dbxs[dbi].md_name.iov_base,
- node_ks(node)) == 0) {
- node = NULL;
- break;
- }
-
- if (node) {
- MDBX_db db;
- memcpy(&db, node_data(node), sizeof(db));
- stat_add(&db, st, bytes);
- }
- }
- err = mdbx_cursor_sibling(&cx.outer, SIBLING_RIGHT);
- }
- if (unlikely(err != MDBX_NOTFOUND))
- return err;
- }
-
- return MDBX_SUCCESS;
-}
-
-__cold int mdbx_env_stat_ex(const MDBX_env *env, const MDBX_txn *txn,
- MDBX_stat *dest, size_t bytes) {
- if (unlikely(!dest))
- return MDBX_EINVAL;
- const size_t size_before_modtxnid = offsetof(MDBX_stat, ms_mod_txnid);
- if (unlikely(bytes != sizeof(MDBX_stat)) && bytes != size_before_modtxnid)
- return MDBX_EINVAL;
-
- if (likely(txn)) {
- if (env && unlikely(txn->mt_env != env))
- return MDBX_EINVAL;
- return stat_acc(txn, dest, bytes);
- }
-
- int err = check_env(env, true);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- if (env->me_txn0 && env->me_txn0->mt_owner == mdbx_thread_self())
- /* inside write-txn */
- return stat_acc(env->me_txn, dest, bytes);
-
- MDBX_txn *tmp_txn;
- err = mdbx_txn_begin((MDBX_env *)env, NULL, MDBX_TXN_RDONLY, &tmp_txn);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- const int rc = stat_acc(tmp_txn, dest, bytes);
- err = mdbx_txn_abort(tmp_txn);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- return rc;
-}
-
-__cold int mdbx_dbi_dupsort_depthmask(MDBX_txn *txn, MDBX_dbi dbi,
- uint32_t *mask) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!mask))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_VALID)))
- return MDBX_BAD_DBI;
-
- MDBX_cursor_couple cx;
- rc = mdbx_cursor_init(&cx.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- if ((cx.outer.mc_db->md_flags & MDBX_DUPSORT) == 0)
- return MDBX_RESULT_TRUE;
-
- MDBX_val key, data;
- rc = mdbx_cursor_first(&cx.outer, &key, &data);
- *mask = 0;
- while (rc == MDBX_SUCCESS) {
- const MDBX_node *node = page_node(cx.outer.mc_pg[cx.outer.mc_top],
- cx.outer.mc_ki[cx.outer.mc_top]);
- const MDBX_db *db = node_data(node);
- const unsigned flags = node_flags(node);
- switch (flags) {
- case F_BIGDATA:
- case 0:
- /* single-value entry, deep = 0 */
- *mask |= 1 << 0;
- break;
- case F_DUPDATA:
- /* single sub-page, deep = 1 */
- *mask |= 1 << 1;
- break;
- case F_DUPDATA | F_SUBDATA:
- /* sub-tree */
- *mask |= 1 << UNALIGNED_PEEK_16(db, MDBX_db, md_depth);
- break;
- default:
- mdbx_error("wrong node-flags %u", flags);
- return MDBX_CORRUPTED;
- }
- rc = mdbx_cursor_next(&cx.outer, &key, &data, MDBX_NEXT_NODUP);
- }
-
- return (rc == MDBX_NOTFOUND) ? MDBX_SUCCESS : rc;
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-__cold int mdbx_env_info(const MDBX_env *env, MDBX_envinfo *info,
- size_t bytes) {
- return __inline_mdbx_env_info(env, info, bytes);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-__cold static int fetch_envinfo_ex(const MDBX_env *env, const MDBX_txn *txn,
- MDBX_envinfo *arg, const size_t bytes) {
-
- const size_t size_before_bootid = offsetof(MDBX_envinfo, mi_bootid);
- const size_t size_before_pgop_stat = offsetof(MDBX_envinfo, mi_pgop_stat);
-
- /* is the environment open?
- * (todo4recovery://erased_by_github/libmdbx/issues/171) */
- if (unlikely(!env->me_map)) {
- /* environment not yet opened */
-#if 1
- /* default behavior: returns the available info but zeroed the rest */
- memset(arg, 0, bytes);
- arg->mi_geo.lower = env->me_dbgeo.lower;
- arg->mi_geo.upper = env->me_dbgeo.upper;
- arg->mi_geo.shrink = env->me_dbgeo.shrink;
- arg->mi_geo.grow = env->me_dbgeo.grow;
- arg->mi_geo.current = env->me_dbgeo.now;
- arg->mi_maxreaders = env->me_maxreaders;
- arg->mi_dxb_pagesize = env->me_psize;
- arg->mi_sys_pagesize = env->me_os_psize;
- if (likely(bytes > size_before_bootid)) {
- arg->mi_bootid.current.x = bootid.x;
- arg->mi_bootid.current.y = bootid.y;
- }
- return MDBX_SUCCESS;
-#else
- /* some users may prefer this behavior: return appropriate error */
- return MDBX_EPERM;
-#endif
- }
-
- const MDBX_meta *const meta0 = METAPAGE(env, 0);
- const MDBX_meta *const meta1 = METAPAGE(env, 1);
- const MDBX_meta *const meta2 = METAPAGE(env, 2);
- if (unlikely(env->me_flags & MDBX_FATAL_ERROR))
- return MDBX_PANIC;
-
- volatile const MDBX_meta *const recent_meta = meta_prefer_last(env);
- arg->mi_recent_txnid = meta_txnid(env, recent_meta);
- arg->mi_meta0_txnid = meta_txnid(env, meta0);
- arg->mi_meta0_sign = unaligned_peek_u64(4, meta0->mm_datasync_sign);
- arg->mi_meta1_txnid = meta_txnid(env, meta1);
- arg->mi_meta1_sign = unaligned_peek_u64(4, meta1->mm_datasync_sign);
- arg->mi_meta2_txnid = meta_txnid(env, meta2);
- arg->mi_meta2_sign = unaligned_peek_u64(4, meta2->mm_datasync_sign);
- if (likely(bytes > size_before_bootid)) {
- memcpy(&arg->mi_bootid.meta0, &meta0->mm_bootid, 16);
- memcpy(&arg->mi_bootid.meta1, &meta1->mm_bootid, 16);
- memcpy(&arg->mi_bootid.meta2, &meta2->mm_bootid, 16);
- }
-
- volatile const MDBX_meta *txn_meta = recent_meta;
- arg->mi_last_pgno = txn_meta->mm_geo.next - 1;
- arg->mi_geo.current = pgno2bytes(env, txn_meta->mm_geo.now);
- if (txn) {
- arg->mi_last_pgno = txn->mt_next_pgno - 1;
- arg->mi_geo.current = pgno2bytes(env, txn->mt_end_pgno);
-
- const txnid_t wanna_meta_txnid = (txn->mt_flags & MDBX_TXN_RDONLY)
- ? txn->mt_txnid
- : txn->mt_txnid - xMDBX_TXNID_STEP;
- txn_meta = (arg->mi_meta0_txnid == wanna_meta_txnid) ? meta0 : txn_meta;
- txn_meta = (arg->mi_meta1_txnid == wanna_meta_txnid) ? meta1 : txn_meta;
- txn_meta = (arg->mi_meta2_txnid == wanna_meta_txnid) ? meta2 : txn_meta;
- }
- arg->mi_geo.lower = pgno2bytes(env, txn_meta->mm_geo.lower);
- arg->mi_geo.upper = pgno2bytes(env, txn_meta->mm_geo.upper);
- arg->mi_geo.shrink = pgno2bytes(env, pv2pages(txn_meta->mm_geo.shrink_pv));
- arg->mi_geo.grow = pgno2bytes(env, pv2pages(txn_meta->mm_geo.grow_pv));
- const pgno_t unsynced_pages =
- atomic_load32(&env->me_lck->mti_unsynced_pages, mo_Relaxed) +
- (atomic_load32(&env->me_lck->mti_meta_sync_txnid, mo_Relaxed) !=
- (uint32_t)arg->mi_recent_txnid);
-
- arg->mi_mapsize = env->me_dxb_mmap.limit;
-
- const MDBX_lockinfo *const lck = env->me_lck;
- arg->mi_maxreaders = env->me_maxreaders;
- arg->mi_numreaders = env->me_lck_mmap.lck
- ? atomic_load32(&lck->mti_numreaders, mo_Relaxed)
- : INT32_MAX;
- arg->mi_dxb_pagesize = env->me_psize;
- arg->mi_sys_pagesize = env->me_os_psize;
-
- if (likely(bytes > size_before_bootid)) {
- arg->mi_unsync_volume = pgno2bytes(env, unsynced_pages);
- const uint64_t monotime_now = mdbx_osal_monotime();
- uint64_t ts = atomic_load64(&lck->mti_sync_timestamp, mo_Relaxed);
- arg->mi_since_sync_seconds16dot16 =
- ts ? mdbx_osal_monotime_to_16dot16(monotime_now - ts) : 0;
- ts = atomic_load64(&lck->mti_reader_check_timestamp, mo_Relaxed);
- arg->mi_since_reader_check_seconds16dot16 =
- ts ? mdbx_osal_monotime_to_16dot16(monotime_now - ts) : 0;
- arg->mi_autosync_threshold = pgno2bytes(
- env, atomic_load32(&lck->mti_autosync_threshold, mo_Relaxed));
- arg->mi_autosync_period_seconds16dot16 = mdbx_osal_monotime_to_16dot16(
- atomic_load64(&lck->mti_autosync_period, mo_Relaxed));
- arg->mi_bootid.current.x = bootid.x;
- arg->mi_bootid.current.y = bootid.y;
- arg->mi_mode = env->me_lck_mmap.lck ? lck->mti_envmode.weak : env->me_flags;
- }
-
- if (likely(bytes > size_before_pgop_stat)) {
-#if MDBX_ENABLE_PGOP_STAT
- arg->mi_pgop_stat.newly =
- atomic_load64(&lck->mti_pgop_stat.newly, mo_Relaxed);
- arg->mi_pgop_stat.cow = atomic_load64(&lck->mti_pgop_stat.cow, mo_Relaxed);
- arg->mi_pgop_stat.clone =
- atomic_load64(&lck->mti_pgop_stat.clone, mo_Relaxed);
- arg->mi_pgop_stat.split =
- atomic_load64(&lck->mti_pgop_stat.split, mo_Relaxed);
- arg->mi_pgop_stat.merge =
- atomic_load64(&lck->mti_pgop_stat.merge, mo_Relaxed);
- arg->mi_pgop_stat.spill =
- atomic_load64(&lck->mti_pgop_stat.spill, mo_Relaxed);
- arg->mi_pgop_stat.unspill =
- atomic_load64(&lck->mti_pgop_stat.unspill, mo_Relaxed);
- arg->mi_pgop_stat.wops =
- atomic_load64(&lck->mti_pgop_stat.wops, mo_Relaxed);
-#else
- memset(&arg->mi_pgop_stat, 0, sizeof(arg->mi_pgop_stat));
-#endif /* MDBX_ENABLE_PGOP_STAT*/
- }
-
- arg->mi_self_latter_reader_txnid = arg->mi_latter_reader_txnid =
- arg->mi_recent_txnid;
- for (unsigned i = 0; i < arg->mi_numreaders; ++i) {
- const uint32_t pid =
- atomic_load32(&lck->mti_readers[i].mr_pid, mo_AcquireRelease);
- if (pid) {
- const txnid_t txnid = safe64_read(&lck->mti_readers[i].mr_txnid);
- if (arg->mi_latter_reader_txnid > txnid)
- arg->mi_latter_reader_txnid = txnid;
- if (pid == env->me_pid && arg->mi_self_latter_reader_txnid > txnid)
- arg->mi_self_latter_reader_txnid = txnid;
- }
- }
-
- mdbx_compiler_barrier();
- return MDBX_SUCCESS;
-}
-
-__cold int mdbx_env_info_ex(const MDBX_env *env, const MDBX_txn *txn,
- MDBX_envinfo *arg, size_t bytes) {
- if (unlikely((env == NULL && txn == NULL) || arg == NULL))
- return MDBX_EINVAL;
-
- if (txn) {
- int err = check_txn(txn, MDBX_TXN_BLOCKED - MDBX_TXN_ERROR);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- }
- if (env) {
- int err = check_env(env, false);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- if (txn && unlikely(txn->mt_env != env))
- return MDBX_EINVAL;
- } else {
- env = txn->mt_env;
- }
-
- const size_t size_before_bootid = offsetof(MDBX_envinfo, mi_bootid);
- const size_t size_before_pgop_stat = offsetof(MDBX_envinfo, mi_pgop_stat);
- if (unlikely(bytes != sizeof(MDBX_envinfo)) && bytes != size_before_bootid &&
- bytes != size_before_pgop_stat)
- return MDBX_EINVAL;
-
- MDBX_envinfo snap;
- int rc = fetch_envinfo_ex(env, txn, &snap, sizeof(snap));
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- while (1) {
- rc = fetch_envinfo_ex(env, txn, arg, bytes);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- snap.mi_since_sync_seconds16dot16 = arg->mi_since_sync_seconds16dot16;
- snap.mi_since_reader_check_seconds16dot16 =
- arg->mi_since_reader_check_seconds16dot16;
- if (likely(memcmp(&snap, arg, bytes) == 0))
- return MDBX_SUCCESS;
- memcpy(&snap, arg, bytes);
- }
-}
-
-static __inline MDBX_cmp_func *get_default_keycmp(unsigned flags) {
- return (flags & MDBX_REVERSEKEY) ? cmp_reverse
- : (flags & MDBX_INTEGERKEY) ? cmp_int_align2
- : cmp_lexical;
-}
-
-static __inline MDBX_cmp_func *get_default_datacmp(unsigned flags) {
- return !(flags & MDBX_DUPSORT)
- ? cmp_lenfast
- : ((flags & MDBX_INTEGERDUP)
- ? cmp_int_unaligned
- : ((flags & MDBX_REVERSEDUP) ? cmp_reverse : cmp_lexical));
-}
-
-static int mdbx_dbi_bind(MDBX_txn *txn, const MDBX_dbi dbi, unsigned user_flags,
- MDBX_cmp_func *keycmp, MDBX_cmp_func *datacmp) {
- /* LY: so, accepting only three cases for the table's flags:
- * 1) user_flags and both comparators are zero
- * = assume that a by-default mode/flags is requested for reading;
- * 2) user_flags exactly the same
- * = assume that the target mode/flags are requested properly;
- * 3) user_flags differs, but table is empty and MDBX_CREATE is provided
- * = assume that a properly create request with custom flags;
- */
- if ((user_flags ^ txn->mt_dbs[dbi].md_flags) & DB_PERSISTENT_FLAGS) {
- /* flags are differs, check other conditions */
- if ((!user_flags && (!keycmp || keycmp == txn->mt_dbxs[dbi].md_cmp) &&
- (!datacmp || datacmp == txn->mt_dbxs[dbi].md_dcmp)) ||
- user_flags == MDBX_ACCEDE) {
- /* no comparators were provided and flags are zero,
- * seems that is case #1 above */
- user_flags = txn->mt_dbs[dbi].md_flags;
- } else if ((user_flags & MDBX_CREATE) && txn->mt_dbs[dbi].md_entries == 0) {
- if (txn->mt_flags & MDBX_TXN_RDONLY)
- return /* FIXME: return extended info */ MDBX_EACCESS;
- /* make sure flags changes get committed */
- txn->mt_dbs[dbi].md_flags = user_flags & DB_PERSISTENT_FLAGS;
- txn->mt_flags |= MDBX_TXN_DIRTY;
- } else {
- return /* FIXME: return extended info */ MDBX_INCOMPATIBLE;
- }
- }
-
- if (!keycmp)
- keycmp = txn->mt_dbxs[dbi].md_cmp ? txn->mt_dbxs[dbi].md_cmp
- : get_default_keycmp(user_flags);
- if (txn->mt_dbxs[dbi].md_cmp != keycmp) {
- if (txn->mt_dbxs[dbi].md_cmp)
- return MDBX_EINVAL;
- txn->mt_dbxs[dbi].md_cmp = keycmp;
- }
-
- if (!datacmp)
- datacmp = txn->mt_dbxs[dbi].md_dcmp ? txn->mt_dbxs[dbi].md_dcmp
- : get_default_datacmp(user_flags);
- if (txn->mt_dbxs[dbi].md_dcmp != datacmp) {
- if (txn->mt_dbxs[dbi].md_dcmp)
- return MDBX_EINVAL;
- txn->mt_dbxs[dbi].md_dcmp = datacmp;
- }
-
- return MDBX_SUCCESS;
-}
-
-static int dbi_open(MDBX_txn *txn, const char *table_name, unsigned user_flags,
- MDBX_dbi *dbi, MDBX_cmp_func *keycmp,
- MDBX_cmp_func *datacmp) {
- int rc = MDBX_EINVAL;
- if (unlikely(!dbi))
- return rc;
-
- if (unlikely((user_flags & ~DB_USABLE_FLAGS) != 0)) {
- early_bailout:
- *dbi = 0;
- return rc;
- }
-
- rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- goto early_bailout;
-
- switch (user_flags & (MDBX_INTEGERDUP | MDBX_DUPFIXED | MDBX_DUPSORT |
- MDBX_REVERSEDUP | MDBX_ACCEDE)) {
- case MDBX_ACCEDE:
- if ((user_flags & MDBX_CREATE) == 0)
- break;
- __fallthrough /* fall through */;
- default:
- rc = MDBX_EINVAL;
- goto early_bailout;
-
- case MDBX_DUPSORT:
- case MDBX_DUPSORT | MDBX_REVERSEDUP:
- case MDBX_DUPSORT | MDBX_DUPFIXED:
- case MDBX_DUPSORT | MDBX_DUPFIXED | MDBX_REVERSEDUP:
- case MDBX_DUPSORT | MDBX_DUPFIXED | MDBX_INTEGERDUP:
- case MDBX_DUPSORT | MDBX_DUPFIXED | MDBX_INTEGERDUP | MDBX_REVERSEDUP:
- case 0:
- break;
- }
-
- /* main table? */
- if (!table_name) {
- rc = mdbx_dbi_bind(txn, MAIN_DBI, user_flags, keycmp, datacmp);
- if (unlikely(rc != MDBX_SUCCESS))
- goto early_bailout;
- *dbi = MAIN_DBI;
- return rc;
- }
-
- MDBX_env *env = txn->mt_env;
- size_t len = strlen(table_name);
- if (len > env->me_leaf_nodemax - NODESIZE - sizeof(MDBX_db))
- return MDBX_EINVAL;
-
- if (txn->mt_dbxs[MAIN_DBI].md_cmp == NULL) {
- txn->mt_dbxs[MAIN_DBI].md_cmp =
- get_default_keycmp(txn->mt_dbs[MAIN_DBI].md_flags);
- txn->mt_dbxs[MAIN_DBI].md_dcmp =
- get_default_datacmp(txn->mt_dbs[MAIN_DBI].md_flags);
- }
-
- /* Is the DB already open? */
- MDBX_dbi scan, slot;
- for (slot = scan = txn->mt_numdbs; --scan >= CORE_DBS;) {
- if (!txn->mt_dbxs[scan].md_name.iov_len) {
- /* Remember this free slot */
- slot = scan;
- continue;
- }
- if (len == txn->mt_dbxs[scan].md_name.iov_len &&
- !strncmp(table_name, txn->mt_dbxs[scan].md_name.iov_base, len)) {
- rc = mdbx_dbi_bind(txn, scan, user_flags, keycmp, datacmp);
- if (unlikely(rc != MDBX_SUCCESS))
- goto early_bailout;
- *dbi = scan;
- return rc;
- }
- }
-
- /* Fail, if no free slot and max hit */
- if (unlikely(slot >= env->me_maxdbs)) {
- rc = MDBX_DBS_FULL;
- goto early_bailout;
- }
-
- /* Cannot mix named table with some main-table flags */
- if (unlikely(txn->mt_dbs[MAIN_DBI].md_flags &
- (MDBX_DUPSORT | MDBX_INTEGERKEY))) {
- rc = (user_flags & MDBX_CREATE) ? MDBX_INCOMPATIBLE : MDBX_NOTFOUND;
- goto early_bailout;
- }
-
- /* Find the DB info */
- MDBX_val key, data;
- key.iov_len = len;
- key.iov_base = (void *)table_name;
- MDBX_cursor_couple couple;
- rc = mdbx_cursor_init(&couple.outer, txn, MAIN_DBI);
- if (unlikely(rc != MDBX_SUCCESS))
- goto early_bailout;
- rc = mdbx_cursor_set(&couple.outer, &key, &data, MDBX_SET).err;
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc != MDBX_NOTFOUND || !(user_flags & MDBX_CREATE))
- goto early_bailout;
- } else {
- /* make sure this is actually a table */
- MDBX_node *node = page_node(couple.outer.mc_pg[couple.outer.mc_top],
- couple.outer.mc_ki[couple.outer.mc_top]);
- if (unlikely((node_flags(node) & (F_DUPDATA | F_SUBDATA)) != F_SUBDATA)) {
- rc = MDBX_INCOMPATIBLE;
- goto early_bailout;
- }
- if (!MDBX_DISABLE_PAGECHECKS && unlikely(data.iov_len != sizeof(MDBX_db))) {
- rc = MDBX_CORRUPTED;
- goto early_bailout;
- }
- }
-
- if (rc != MDBX_SUCCESS && unlikely(txn->mt_flags & MDBX_TXN_RDONLY)) {
- rc = MDBX_EACCESS;
- goto early_bailout;
- }
-
- /* Done here so we cannot fail after creating a new DB */
- char *namedup = mdbx_strdup(table_name);
- if (unlikely(!namedup)) {
- rc = MDBX_ENOMEM;
- goto early_bailout;
- }
-
- int err = mdbx_fastmutex_acquire(&env->me_dbi_lock);
- if (unlikely(err != MDBX_SUCCESS)) {
- rc = err;
- mdbx_free(namedup);
- goto early_bailout;
- }
-
- /* Import handles from env */
- dbi_import_locked(txn);
-
- /* Rescan after mutex acquisition & import handles */
- for (slot = scan = txn->mt_numdbs; --scan >= CORE_DBS;) {
- if (!txn->mt_dbxs[scan].md_name.iov_len) {
- /* Remember this free slot */
- slot = scan;
- continue;
- }
- if (len == txn->mt_dbxs[scan].md_name.iov_len &&
- !strncmp(table_name, txn->mt_dbxs[scan].md_name.iov_base, len)) {
- rc = mdbx_dbi_bind(txn, scan, user_flags, keycmp, datacmp);
- if (unlikely(rc != MDBX_SUCCESS))
- goto later_bailout;
- *dbi = scan;
- goto later_exit;
- }
- }
-
- if (unlikely(slot >= env->me_maxdbs)) {
- rc = MDBX_DBS_FULL;
- goto later_bailout;
- }
-
- unsigned dbiflags = DBI_FRESH | DBI_VALID | DBI_USRVALID;
- MDBX_db db_dummy;
- if (unlikely(rc)) {
- /* MDBX_NOTFOUND and MDBX_CREATE: Create new DB */
- mdbx_tassert(txn, rc == MDBX_NOTFOUND);
- memset(&db_dummy, 0, sizeof(db_dummy));
- db_dummy.md_root = P_INVALID;
- db_dummy.md_mod_txnid = txn->mt_txnid;
- db_dummy.md_flags = user_flags & DB_PERSISTENT_FLAGS;
- data.iov_len = sizeof(db_dummy);
- data.iov_base = &db_dummy;
- WITH_CURSOR_TRACKING(couple.outer,
- rc = mdbx_cursor_put(&couple.outer, &key, &data,
- F_SUBDATA | MDBX_NOOVERWRITE));
-
- if (unlikely(rc != MDBX_SUCCESS))
- goto later_bailout;
-
- dbiflags |= DBI_DIRTY | DBI_CREAT;
- txn->mt_flags |= MDBX_TXN_DIRTY;
- mdbx_tassert(txn, (txn->mt_dbistate[MAIN_DBI] & DBI_DIRTY) != 0);
- }
-
- /* Got info, register DBI in this txn */
- memset(txn->mt_dbxs + slot, 0, sizeof(MDBX_dbx));
- memcpy(&txn->mt_dbs[slot], data.iov_base, sizeof(MDBX_db));
- env->me_dbflags[slot] = 0;
- rc = mdbx_dbi_bind(txn, slot, user_flags, keycmp, datacmp);
- if (unlikely(rc != MDBX_SUCCESS)) {
- mdbx_tassert(txn, (dbiflags & DBI_CREAT) == 0);
- later_bailout:
- *dbi = 0;
- later_exit:
- mdbx_free(namedup);
- } else {
- txn->mt_dbistate[slot] = (uint8_t)dbiflags;
- txn->mt_dbxs[slot].md_name.iov_base = namedup;
- txn->mt_dbxs[slot].md_name.iov_len = len;
- txn->mt_dbiseqs[slot] = ++env->me_dbiseqs[slot];
- if (!(dbiflags & DBI_CREAT))
- env->me_dbflags[slot] = txn->mt_dbs[slot].md_flags | DB_VALID;
- if (txn->mt_numdbs == slot) {
- mdbx_compiler_barrier();
- txn->mt_numdbs = env->me_numdbs = slot + 1;
- txn->mt_cursors[slot] = NULL;
- }
- mdbx_assert(env, env->me_numdbs > slot);
- *dbi = slot;
- }
-
- mdbx_ensure(env, mdbx_fastmutex_release(&env->me_dbi_lock) == MDBX_SUCCESS);
- return rc;
-}
-
-int mdbx_dbi_open(MDBX_txn *txn, const char *table_name,
- MDBX_db_flags_t table_flags, MDBX_dbi *dbi) {
- return dbi_open(txn, table_name, table_flags, dbi, nullptr, nullptr);
-}
-
-int mdbx_dbi_open_ex(MDBX_txn *txn, const char *table_name,
- MDBX_db_flags_t table_flags, MDBX_dbi *dbi,
- MDBX_cmp_func *keycmp, MDBX_cmp_func *datacmp) {
- return dbi_open(txn, table_name, table_flags, dbi, keycmp, datacmp);
-}
-
-__cold int mdbx_dbi_stat(MDBX_txn *txn, MDBX_dbi dbi, MDBX_stat *dest,
- size_t bytes) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!dest))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_VALID)))
- return MDBX_BAD_DBI;
-
- const size_t size_before_modtxnid = offsetof(MDBX_stat, ms_mod_txnid);
- if (unlikely(bytes != sizeof(MDBX_stat)) && bytes != size_before_modtxnid)
- return MDBX_EINVAL;
-
- if (unlikely(txn->mt_flags & MDBX_TXN_BLOCKED))
- return MDBX_BAD_TXN;
-
- if (unlikely(txn->mt_dbistate[dbi] & DBI_STALE)) {
- rc = mdbx_fetch_sdb(txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
-
- dest->ms_psize = txn->mt_env->me_psize;
- stat_get(&txn->mt_dbs[dbi], dest, bytes);
- return MDBX_SUCCESS;
-}
-
-static int mdbx_dbi_close_locked(MDBX_env *env, MDBX_dbi dbi) {
- mdbx_assert(env, dbi >= CORE_DBS);
- if (unlikely(dbi >= env->me_numdbs))
- return MDBX_BAD_DBI;
-
- char *ptr = env->me_dbxs[dbi].md_name.iov_base;
- /* If there was no name, this was already closed */
- if (unlikely(!ptr))
- return MDBX_BAD_DBI;
-
- env->me_dbflags[dbi] = 0;
- env->me_dbiseqs[dbi]++;
- env->me_dbxs[dbi].md_name.iov_len = 0;
- mdbx_memory_fence(mo_AcquireRelease, true);
- env->me_dbxs[dbi].md_name.iov_base = NULL;
- mdbx_free(ptr);
-
- if (env->me_numdbs == dbi + 1) {
- unsigned i = env->me_numdbs;
- do
- --i;
- while (i > CORE_DBS && !env->me_dbxs[i - 1].md_name.iov_base);
- env->me_numdbs = i;
- }
-
- return MDBX_SUCCESS;
-}
-
-int mdbx_dbi_close(MDBX_env *env, MDBX_dbi dbi) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(dbi < CORE_DBS || dbi >= env->me_maxdbs))
- return MDBX_BAD_DBI;
-
- rc = mdbx_fastmutex_acquire(&env->me_dbi_lock);
- if (likely(rc == MDBX_SUCCESS)) {
- rc = (dbi < env->me_maxdbs && (env->me_dbflags[dbi] & DB_VALID))
- ? mdbx_dbi_close_locked(env, dbi)
- : MDBX_BAD_DBI;
- mdbx_ensure(env, mdbx_fastmutex_release(&env->me_dbi_lock) == MDBX_SUCCESS);
- }
- return rc;
-}
-
-int mdbx_dbi_flags_ex(MDBX_txn *txn, MDBX_dbi dbi, unsigned *flags,
- unsigned *state) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED - MDBX_TXN_ERROR);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!flags || !state))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_VALID)))
- return MDBX_BAD_DBI;
-
- *flags = txn->mt_dbs[dbi].md_flags & DB_PERSISTENT_FLAGS;
- *state =
- txn->mt_dbistate[dbi] & (DBI_FRESH | DBI_CREAT | DBI_DIRTY | DBI_STALE);
-
- return MDBX_SUCCESS;
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-int mdbx_dbi_flags(MDBX_txn *txn, MDBX_dbi dbi, unsigned *flags) {
- return __inline_mdbx_dbi_flags(txn, dbi, flags);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-static int mdbx_drop_tree(MDBX_cursor *mc, const bool may_have_subDBs) {
- int rc = mdbx_page_search(mc, NULL, MDBX_PS_FIRST);
- if (likely(rc == MDBX_SUCCESS)) {
- MDBX_txn *txn = mc->mc_txn;
-
- /* DUPSORT sub-DBs have no ovpages/DBs. Omit scanning leaves.
- * This also avoids any P_LEAF2 pages, which have no nodes.
- * Also if the DB doesn't have sub-DBs and has no overflow
- * pages, omit scanning leaves. */
- if (!(may_have_subDBs | mc->mc_db->md_overflow_pages))
- mdbx_cursor_pop(mc);
-
- rc = mdbx_pnl_need(&txn->tw.retired_pages,
- mc->mc_db->md_branch_pages + mc->mc_db->md_leaf_pages +
- mc->mc_db->md_overflow_pages);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
-
- MDBX_cursor mx;
- cursor_copy(mc, &mx);
- while (mc->mc_snum > 0) {
- MDBX_page *const mp = mc->mc_pg[mc->mc_top];
- const unsigned nkeys = page_numkeys(mp);
- if (IS_LEAF(mp)) {
- mdbx_cassert(mc, mc->mc_snum == mc->mc_db->md_depth);
- for (unsigned i = 0; i < nkeys; i++) {
- MDBX_node *node = page_node(mp, i);
- if (node_flags(node) & F_BIGDATA) {
- rc = mdbx_page_retire_ex(mc, node_largedata_pgno(node), NULL, 0);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- if (!(may_have_subDBs | mc->mc_db->md_overflow_pages))
- goto pop;
- } else if (node_flags(node) & F_SUBDATA) {
- if (unlikely((node_flags(node) & F_DUPDATA) == 0)) {
- rc = /* disallowing implicit subDB deletion */ MDBX_INCOMPATIBLE;
- goto bailout;
- }
- rc = mdbx_xcursor_init1(mc, node, mp);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- rc = mdbx_drop_tree(&mc->mc_xcursor->mx_cursor, false);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- }
- } else {
- mdbx_cassert(mc, mc->mc_snum < mc->mc_db->md_depth);
- if (mdbx_audit_enabled())
- mc->mc_flags |= C_RETIRING;
- const int pagetype =
- (IS_FROZEN(txn, mp) ? P_FROZEN : 0) +
- ((mc->mc_snum + 1 == mc->mc_db->md_depth) ? P_LEAF : P_BRANCH);
- for (unsigned i = 0; i < nkeys; i++) {
- MDBX_node *node = page_node(mp, i);
- mdbx_tassert(txn, (node_flags(node) &
- (F_BIGDATA | F_SUBDATA | F_DUPDATA)) == 0);
- const pgno_t pgno = node_pgno(node);
- rc = mdbx_page_retire_ex(mc, pgno, NULL, pagetype);
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- }
- if (mdbx_audit_enabled())
- mc->mc_flags -= C_RETIRING;
- }
- if (!mc->mc_top)
- break;
- mdbx_cassert(mc, nkeys > 0);
- mc->mc_ki[mc->mc_top] = (indx_t)nkeys;
- rc = mdbx_cursor_sibling(mc, SIBLING_RIGHT);
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (unlikely(rc != MDBX_NOTFOUND))
- goto bailout;
- /* no more siblings, go back to beginning
- * of previous level. */
- pop:
- mdbx_cursor_pop(mc);
- mc->mc_ki[0] = 0;
- for (unsigned i = 1; i < mc->mc_snum; i++) {
- mc->mc_ki[i] = 0;
- mc->mc_pg[i] = mx.mc_pg[i];
- }
- }
- }
- rc = mdbx_page_retire(mc, mc->mc_pg[0]);
- bailout:
- if (unlikely(rc != MDBX_SUCCESS))
- txn->mt_flags |= MDBX_TXN_ERROR;
- } else if (rc == MDBX_NOTFOUND) {
- rc = MDBX_SUCCESS;
- }
- mc->mc_flags &= ~C_INITIALIZED;
- return rc;
-}
-
-int mdbx_drop(MDBX_txn *txn, MDBX_dbi dbi, bool del) {
- int rc = check_txn_rw(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- MDBX_cursor *mc;
- rc = mdbx_cursor_open(txn, dbi, &mc);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- rc = mdbx_drop_tree(mc, dbi == MAIN_DBI ||
- (mc->mc_db->md_flags & MDBX_DUPSORT) != 0);
- /* Invalidate the dropped DB's cursors */
- for (MDBX_cursor *m2 = txn->mt_cursors[dbi]; m2; m2 = m2->mc_next)
- m2->mc_flags &= ~(C_INITIALIZED | C_EOF);
- if (unlikely(rc))
- goto bailout;
-
- /* Can't delete the main DB */
- if (del && dbi >= CORE_DBS) {
- rc = mdbx_del0(txn, MAIN_DBI, &mc->mc_dbx->md_name, NULL, F_SUBDATA);
- if (likely(rc == MDBX_SUCCESS)) {
- mdbx_tassert(txn, txn->mt_dbistate[MAIN_DBI] & DBI_DIRTY);
- mdbx_tassert(txn, txn->mt_flags & MDBX_TXN_DIRTY);
- txn->mt_dbistate[dbi] = DBI_STALE;
- MDBX_env *env = txn->mt_env;
- rc = mdbx_fastmutex_acquire(&env->me_dbi_lock);
- if (unlikely(rc != MDBX_SUCCESS)) {
- txn->mt_flags |= MDBX_TXN_ERROR;
- goto bailout;
- }
- mdbx_dbi_close_locked(env, dbi);
- mdbx_ensure(env,
- mdbx_fastmutex_release(&env->me_dbi_lock) == MDBX_SUCCESS);
- } else {
- txn->mt_flags |= MDBX_TXN_ERROR;
- }
- } else {
- /* reset the DB record, mark it dirty */
- txn->mt_dbistate[dbi] |= DBI_DIRTY;
- txn->mt_dbs[dbi].md_depth = 0;
- txn->mt_dbs[dbi].md_branch_pages = 0;
- txn->mt_dbs[dbi].md_leaf_pages = 0;
- txn->mt_dbs[dbi].md_overflow_pages = 0;
- txn->mt_dbs[dbi].md_entries = 0;
- txn->mt_dbs[dbi].md_root = P_INVALID;
- txn->mt_dbs[dbi].md_seq = 0;
- /* txn->mt_dbs[dbi].md_mod_txnid = txn->mt_txnid; */
- txn->mt_flags |= MDBX_TXN_DIRTY;
- }
-
-bailout:
- mdbx_cursor_close(mc);
- return rc;
-}
-
-int mdbx_set_compare(MDBX_txn *txn, MDBX_dbi dbi, MDBX_cmp_func *cmp) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED - MDBX_TXN_ERROR);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- txn->mt_dbxs[dbi].md_cmp = cmp;
- return MDBX_SUCCESS;
-}
-
-int mdbx_set_dupsort(MDBX_txn *txn, MDBX_dbi dbi, MDBX_cmp_func *cmp) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED - MDBX_TXN_ERROR);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- txn->mt_dbxs[dbi].md_dcmp = cmp;
- return MDBX_SUCCESS;
-}
-
-__cold int mdbx_reader_list(const MDBX_env *env, MDBX_reader_list_func *func,
- void *ctx) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!func))
- return MDBX_EINVAL;
-
- rc = MDBX_RESULT_TRUE;
- int serial = 0;
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (likely(lck)) {
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- for (unsigned i = 0; i < snap_nreaders; i++) {
- const MDBX_reader *r = lck->mti_readers + i;
- retry_reader:;
- const uint32_t pid = atomic_load32(&r->mr_pid, mo_AcquireRelease);
- if (!pid)
- continue;
- txnid_t txnid = safe64_read(&r->mr_txnid);
- const uint64_t tid = atomic_load64(&r->mr_tid, mo_Relaxed);
- const pgno_t pages_used =
- atomic_load32(&r->mr_snapshot_pages_used, mo_Relaxed);
- const uint64_t reader_pages_retired =
- atomic_load64(&r->mr_snapshot_pages_retired, mo_Relaxed);
- if (unlikely(
- txnid != safe64_read(&r->mr_txnid) ||
- pid != atomic_load32(&r->mr_pid, mo_AcquireRelease) ||
- tid != atomic_load64(&r->mr_tid, mo_Relaxed) ||
- pages_used !=
- atomic_load32(&r->mr_snapshot_pages_used, mo_Relaxed) ||
- reader_pages_retired !=
- atomic_load64(&r->mr_snapshot_pages_retired, mo_Relaxed)))
- goto retry_reader;
-
- mdbx_assert(env, txnid > 0);
- if (txnid >= SAFE64_INVALID_THRESHOLD)
- txnid = 0;
-
- size_t bytes_used = 0;
- size_t bytes_retained = 0;
- uint64_t lag = 0;
- if (txnid) {
- retry_header:;
- volatile const MDBX_meta *const recent_meta = meta_prefer_last(env);
- const uint64_t head_pages_retired =
- unaligned_peek_u64_volatile(4, recent_meta->mm_pages_retired);
- const txnid_t head_txnid = meta_txnid(env, recent_meta);
- mdbx_compiler_barrier();
- if (unlikely(recent_meta != meta_prefer_last(env) ||
- head_pages_retired !=
- unaligned_peek_u64_volatile(
- 4, recent_meta->mm_pages_retired)) ||
- head_txnid != meta_txnid(env, recent_meta))
- goto retry_header;
-
- lag = (head_txnid - txnid) / xMDBX_TXNID_STEP;
- bytes_used = pgno2bytes(env, pages_used);
- bytes_retained = (head_pages_retired > reader_pages_retired)
- ? pgno2bytes(env, (pgno_t)(head_pages_retired -
- reader_pages_retired))
- : 0;
- }
- rc = func(ctx, ++serial, i, pid, (mdbx_tid_t)tid, txnid, lag, bytes_used,
- bytes_retained);
- if (unlikely(rc != MDBX_SUCCESS))
- break;
- }
- }
-
- return rc;
-}
-
-/* Insert pid into list if not already present.
- * return -1 if already present. */
-__cold static bool mdbx_pid_insert(uint32_t *ids, uint32_t pid) {
- /* binary search of pid in list */
- unsigned base = 0;
- unsigned cursor = 1;
- int val = 0;
- unsigned n = ids[0];
-
- while (n > 0) {
- unsigned pivot = n >> 1;
- cursor = base + pivot + 1;
- val = pid - ids[cursor];
-
- if (val < 0) {
- n = pivot;
- } else if (val > 0) {
- base = cursor;
- n -= pivot + 1;
- } else {
- /* found, so it's a duplicate */
- return false;
- }
- }
-
- if (val > 0)
- ++cursor;
-
- ids[0]++;
- for (n = ids[0]; n > cursor; n--)
- ids[n] = ids[n - 1];
- ids[n] = pid;
- return true;
-}
-
-__cold int mdbx_reader_check(MDBX_env *env, int *dead) {
- if (dead)
- *dead = 0;
- return mdbx_cleanup_dead_readers(env, false, dead);
-}
-
-/* Return:
- * MDBX_RESULT_TRUE - done and mutex recovered
- * MDBX_SUCCESS - done
- * Otherwise errcode. */
-__cold MDBX_INTERNAL_FUNC int
-mdbx_cleanup_dead_readers(MDBX_env *env, int rdt_locked, int *dead) {
- int rc = check_env(env, true);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- mdbx_assert(env, rdt_locked >= 0);
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (unlikely(lck == NULL)) {
- /* exclusive mode */
- if (dead)
- *dead = 0;
- return MDBX_SUCCESS;
- }
-
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- uint32_t pidsbuf_onstask[142];
- uint32_t *const pids =
- (snap_nreaders < ARRAY_LENGTH(pidsbuf_onstask))
- ? pidsbuf_onstask
- : mdbx_malloc((snap_nreaders + 1) * sizeof(uint32_t));
- if (unlikely(!pids))
- return MDBX_ENOMEM;
-
- pids[0] = 0;
- int count = 0;
- for (unsigned i = 0; i < snap_nreaders; i++) {
- const uint32_t pid =
- atomic_load32(&lck->mti_readers[i].mr_pid, mo_AcquireRelease);
- if (pid == 0)
- continue /* skip empty */;
- if (pid == env->me_pid)
- continue /* skip self */;
- if (!mdbx_pid_insert(pids, pid))
- continue /* such pid already processed */;
-
- int err = mdbx_rpid_check(env, pid);
- if (err == MDBX_RESULT_TRUE)
- continue /* reader is live */;
-
- if (err != MDBX_SUCCESS) {
- rc = err;
- break /* mdbx_rpid_check() failed */;
- }
-
- /* stale reader found */
- if (!rdt_locked) {
- err = mdbx_rdt_lock(env);
- if (MDBX_IS_ERROR(err)) {
- rc = err;
- break;
- }
-
- rdt_locked = -1;
- if (err == MDBX_RESULT_TRUE) {
- /* mutex recovered, the mdbx_ipclock_failed() checked all readers */
- rc = MDBX_RESULT_TRUE;
- break;
- }
-
- /* a other process may have clean and reused slot, recheck */
- if (lck->mti_readers[i].mr_pid.weak != pid)
- continue;
-
- err = mdbx_rpid_check(env, pid);
- if (MDBX_IS_ERROR(err)) {
- rc = err;
- break;
- }
-
- if (err != MDBX_SUCCESS)
- continue /* the race with other process, slot reused */;
- }
-
- /* clean it */
- for (unsigned j = i; j < snap_nreaders; j++) {
- if (lck->mti_readers[j].mr_pid.weak == pid) {
- mdbx_debug("clear stale reader pid %" PRIuPTR " txn %" PRIaTXN,
- (size_t)pid, lck->mti_readers[j].mr_txnid.weak);
- atomic_store32(&lck->mti_readers[j].mr_pid, 0, mo_Relaxed);
- atomic_store32(&lck->mti_readers_refresh_flag, true, mo_AcquireRelease);
- count++;
- }
- }
- }
-
- if (likely(!MDBX_IS_ERROR(rc)))
- atomic_store64(&lck->mti_reader_check_timestamp, mdbx_osal_monotime(),
- mo_Relaxed);
-
- if (rdt_locked < 0)
- mdbx_rdt_unlock(env);
-
- if (pids != pidsbuf_onstask)
- mdbx_free(pids);
-
- if (dead)
- *dead = count;
- return rc;
-}
-
-__cold int mdbx_setup_debug(int loglevel, int flags, MDBX_debug_func *logger) {
- const int rc = mdbx_runtime_flags | (mdbx_loglevel << 16);
-
- if (loglevel != MDBX_LOG_DONTCHANGE)
- mdbx_loglevel = (uint8_t)loglevel;
-
- if (flags != MDBX_DBG_DONTCHANGE) {
- flags &=
-#if MDBX_DEBUG
- MDBX_DBG_ASSERT | MDBX_DBG_AUDIT | MDBX_DBG_JITTER |
-#endif
- MDBX_DBG_DUMP | MDBX_DBG_LEGACY_MULTIOPEN | MDBX_DBG_LEGACY_OVERLAP |
- MDBX_DBG_DONT_UPGRADE;
- mdbx_runtime_flags = (uint8_t)flags;
- }
-
- if (logger != MDBX_LOGGER_DONTCHANGE)
- mdbx_debug_logger = logger;
- return rc;
-}
-
-__cold static txnid_t mdbx_kick_longlived_readers(MDBX_env *env,
- const txnid_t laggard) {
- mdbx_debug("DB size maxed out by reading #%" PRIaTXN, laggard);
-
- int retry;
- for (retry = 0; retry < INT_MAX; ++retry) {
- txnid_t oldest = mdbx_recent_steady_txnid(env);
- mdbx_assert(env, oldest < env->me_txn0->mt_txnid);
- mdbx_assert(env, oldest >= laggard);
- mdbx_assert(env, oldest >= env->me_lck->mti_oldest_reader.weak);
- MDBX_lockinfo *const lck = env->me_lck_mmap.lck;
- if (oldest == laggard || unlikely(!lck /* without-LCK mode */))
- return oldest;
-
- if (MDBX_IS_ERROR(mdbx_cleanup_dead_readers(env, false, NULL)))
- break;
-
- MDBX_reader *asleep = nullptr;
- uint64_t oldest_retired = UINT64_MAX;
- const unsigned snap_nreaders =
- atomic_load32(&lck->mti_numreaders, mo_AcquireRelease);
- for (unsigned i = 0; i < snap_nreaders; ++i) {
- retry:
- if (atomic_load32(&lck->mti_readers[i].mr_pid, mo_AcquireRelease)) {
- /* mdbx_jitter4testing(true); */
- const uint64_t snap_retired = atomic_load64(
- &lck->mti_readers[i].mr_snapshot_pages_retired, mo_Relaxed);
- const txnid_t snap_txnid = safe64_read(&lck->mti_readers[i].mr_txnid);
- if (unlikely(snap_retired !=
- atomic_load64(
- &lck->mti_readers[i].mr_snapshot_pages_retired,
- mo_AcquireRelease) ||
- snap_txnid != safe64_read(&lck->mti_readers[i].mr_txnid)))
- goto retry;
- if (oldest > snap_txnid &&
- laggard <= /* ignore pending updates */ snap_txnid) {
- oldest = snap_txnid;
- oldest_retired = snap_retired;
- asleep = &lck->mti_readers[i];
- }
- }
- }
-
- if (laggard < oldest || !asleep) {
- if (retry && env->me_hsr_callback) {
- /* LY: notify end of hsr-loop */
- const txnid_t gap = oldest - laggard;
- env->me_hsr_callback(env, env->me_txn, 0, 0, laggard,
- (gap < UINT_MAX) ? (unsigned)gap : UINT_MAX, 0,
- -retry);
- }
- mdbx_notice("hsr-kick: update oldest %" PRIaTXN " -> %" PRIaTXN,
- lck->mti_oldest_reader.weak, oldest);
- mdbx_assert(env, lck->mti_oldest_reader.weak <= oldest);
- return atomic_store64(&lck->mti_oldest_reader, oldest, mo_Relaxed);
- }
-
- if (!env->me_hsr_callback)
- break;
-
- uint32_t pid = atomic_load32(&asleep->mr_pid, mo_AcquireRelease);
- uint64_t tid = asleep->mr_tid.weak;
- if (safe64_read(&asleep->mr_txnid) != laggard || pid <= 0)
- continue;
-
- const MDBX_meta *head_meta = constmeta_prefer_last(env);
- const txnid_t gap =
- (constmeta_txnid(env, head_meta) - laggard) / xMDBX_TXNID_STEP;
- const uint64_t head_retired =
- unaligned_peek_u64(4, head_meta->mm_pages_retired);
- const size_t space =
- (head_retired > oldest_retired)
- ? pgno2bytes(env, (pgno_t)(head_retired - oldest_retired))
- : 0;
- int rc = env->me_hsr_callback(
- env, env->me_txn, pid, (mdbx_tid_t)tid, laggard,
- (gap < UINT_MAX) ? (unsigned)gap : UINT_MAX, space, retry);
- if (rc < 0)
- break;
-
- if (rc > 0) {
- if (rc == 1) {
- safe64_reset_compare(&asleep->mr_txnid, laggard);
- } else {
- safe64_reset(&asleep->mr_txnid, true);
- atomic_store64(&asleep->mr_tid, 0, mo_Relaxed);
- atomic_store32(&asleep->mr_pid, 0, mo_Relaxed);
- }
- atomic_store32(&lck->mti_readers_refresh_flag, true, mo_Relaxed);
- }
- }
-
- if (retry && env->me_hsr_callback) {
- /* LY: notify end of hsr-loop */
- env->me_hsr_callback(env, env->me_txn, 0, 0, laggard, 0, 0, -retry);
- }
- return mdbx_find_oldest(env->me_txn);
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-__cold int mdbx_env_set_syncbytes(MDBX_env *env, size_t threshold) {
- return __inline_mdbx_env_set_syncbytes(env, threshold);
-}
-
-__cold int mdbx_env_set_syncperiod(MDBX_env *env, unsigned seconds_16dot16) {
- return __inline_mdbx_env_set_syncperiod(env, seconds_16dot16);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-__cold int mdbx_env_set_hsr(MDBX_env *env, MDBX_hsr_func *hsr) {
- int rc = check_env(env, false);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- env->me_hsr_callback = hsr;
- return MDBX_SUCCESS;
-}
-
-__cold MDBX_hsr_func *mdbx_env_get_hsr(const MDBX_env *env) {
- return likely(env && env->me_signature.weak == MDBX_ME_SIGNATURE)
- ? env->me_hsr_callback
- : NULL;
-}
-
-#ifdef __SANITIZE_THREAD__
-/* LY: avoid tsan-trap by me_txn, mm_last_pg and mt_next_pgno */
-__attribute__((__no_sanitize_thread__, __noinline__))
-#endif
-int mdbx_txn_straggler(const MDBX_txn *txn, int *percent)
-{
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return (rc > 0) ? -rc : rc;
-
- MDBX_env *env = txn->mt_env;
- if (unlikely((txn->mt_flags & MDBX_TXN_RDONLY) == 0)) {
- if (percent)
- *percent =
- (int)((txn->mt_next_pgno * UINT64_C(100) + txn->mt_end_pgno / 2) /
- txn->mt_end_pgno);
- return 0;
- }
-
- txnid_t recent = 0;
- volatile const MDBX_meta *meta = nullptr;
- do {
- meta = meta_prefer_last(env);
- recent = meta_txnid(env, meta);
- if (percent) {
- const pgno_t maxpg = meta->mm_geo.now;
- *percent = (int)((meta->mm_geo.next * UINT64_C(100) + maxpg / 2) / maxpg);
- }
- } while (unlikely(recent != meta_txnid(env, meta)));
-
- txnid_t lag = (recent - txn->mt_txnid) / xMDBX_TXNID_STEP;
- return (lag > INT_MAX) ? INT_MAX : (int)lag;
-}
-
-typedef struct mdbx_walk_ctx {
- void *mw_user;
- MDBX_pgvisitor_func *mw_visitor;
- MDBX_txn *mw_txn;
- MDBX_cursor *mw_cursor;
- bool mw_dont_check_keys_ordering;
-} mdbx_walk_ctx_t;
-
-__cold static int mdbx_walk_sdb(mdbx_walk_ctx_t *ctx, MDBX_db *const db,
- const char *name, int deep);
-
-static MDBX_page_type_t walk_page_type(const MDBX_page *mp) {
- if (mp)
- switch (mp->mp_flags) {
- case P_BRANCH:
- return MDBX_page_branch;
- case P_LEAF:
- return MDBX_page_leaf;
- case P_LEAF | P_LEAF2:
- return MDBX_page_dupfixed_leaf;
- case P_OVERFLOW:
- return MDBX_page_large;
- case P_META:
- return MDBX_page_meta;
- }
- return MDBX_page_broken;
-}
-
-/* Depth-first tree traversal. */
-__cold static int mdbx_walk_tree(mdbx_walk_ctx_t *ctx, const pgno_t pgno,
- const char *name, int deep,
- txnid_t parent_txnid) {
- assert(pgno != P_INVALID);
- MDBX_page *mp = nullptr;
- int rc, err = mdbx_page_get(ctx->mw_cursor, pgno, &mp, parent_txnid);
- if (err == MDBX_SUCCESS)
- err = mdbx_page_check(ctx->mw_cursor, mp, 0);
-
- MDBX_page_type_t type = walk_page_type(mp);
- const int nentries = (mp && !IS_OVERFLOW(mp)) ? page_numkeys(mp) : 1;
- unsigned npages = (mp && IS_OVERFLOW(mp)) ? mp->mp_pages : 1;
- size_t pagesize = pgno2bytes(ctx->mw_txn->mt_env, npages);
- size_t header_size = (mp && !IS_LEAF2(mp) && !IS_OVERFLOW(mp))
- ? PAGEHDRSZ + mp->mp_lower
- : PAGEHDRSZ;
- size_t payload_size = 0;
- size_t unused_size =
- (mp && !IS_OVERFLOW(mp) ? page_room(mp) : pagesize - header_size) -
- payload_size;
- size_t align_bytes = 0;
-
- if (err == MDBX_SUCCESS) {
- /* LY: Don't use mask here, e.g bitwise
- * (P_BRANCH|P_LEAF|P_LEAF2|P_META|P_OVERFLOW|P_SUBP).
- * Pages should not me marked dirty/loose or otherwise. */
- switch (mp->mp_flags) {
- default:
- err = MDBX_CORRUPTED;
- break;
- case P_BRANCH:
- if (unlikely(nentries < 2))
- err = MDBX_CORRUPTED;
- case P_LEAF:
- case P_LEAF | P_LEAF2:
- break;
- }
- }
-
- for (int i = 0; err == MDBX_SUCCESS && i < nentries;
- align_bytes += ((payload_size + align_bytes) & 1), i++) {
- if (type == MDBX_page_dupfixed_leaf) {
- /* LEAF2 pages have no mp_ptrs[] or node headers */
- payload_size += mp->mp_leaf2_ksize;
- continue;
- }
-
- MDBX_node *node = page_node(mp, i);
- payload_size += NODESIZE + node_ks(node);
-
- if (type == MDBX_page_branch) {
- assert(i > 0 || node_ks(node) == 0);
- continue;
- }
-
- assert(type == MDBX_page_leaf);
- switch (node_flags(node)) {
- case 0 /* usual node */:
- payload_size += node_ds(node);
- break;
-
- case F_BIGDATA /* long data on the large/overflow page */: {
- payload_size += sizeof(pgno_t);
- const pgno_t large_pgno = node_largedata_pgno(node);
- const size_t over_payload = node_ds(node);
- const size_t over_header = PAGEHDRSZ;
- npages = 1;
-
- MDBX_page *op;
- err = mdbx_page_get(ctx->mw_cursor, large_pgno, &op,
- pp_txnid4chk(mp, ctx->mw_txn));
- if (err == MDBX_SUCCESS)
- err = mdbx_page_check(ctx->mw_cursor, op, 0);
- if (err == MDBX_SUCCESS) {
- /* LY: Don't use mask here, e.g bitwise
- * (P_BRANCH|P_LEAF|P_LEAF2|P_META|P_OVERFLOW|P_SUBP).
- * Pages should not me marked dirty/loose or otherwise. */
- if (unlikely(P_OVERFLOW != op->mp_flags))
- err = bad_page(mp, "wrong page type %d for large data", op->mp_flags);
- else
- npages = op->mp_pages;
- }
-
- pagesize = pgno2bytes(ctx->mw_txn->mt_env, npages);
- const size_t over_unused = pagesize - over_payload - over_header;
- rc = ctx->mw_visitor(large_pgno, npages, ctx->mw_user, deep, name,
- pagesize, MDBX_page_large, err, 1, over_payload,
- over_header, over_unused);
- if (unlikely(rc != MDBX_SUCCESS))
- return (rc == MDBX_RESULT_TRUE) ? MDBX_SUCCESS : rc;
- } break;
-
- case F_SUBDATA /* sub-db */: {
- const size_t namelen = node_ks(node);
- payload_size += node_ds(node);
- if (unlikely(namelen == 0 || node_ds(node) != sizeof(MDBX_db)))
- err = MDBX_CORRUPTED;
- } break;
-
- case F_SUBDATA | F_DUPDATA /* dupsorted sub-tree */:
- payload_size += sizeof(MDBX_db);
- if (unlikely(node_ds(node) != sizeof(MDBX_db)))
- err = MDBX_CORRUPTED;
- break;
-
- case F_DUPDATA /* short sub-page */: {
- if (unlikely(node_ds(node) <= PAGEHDRSZ)) {
- err = MDBX_CORRUPTED;
- break;
- }
-
- MDBX_page *sp = node_data(node);
- const int nsubkeys = page_numkeys(sp);
- size_t subheader_size =
- IS_LEAF2(sp) ? PAGEHDRSZ : PAGEHDRSZ + sp->mp_lower;
- size_t subunused_size = page_room(sp);
- size_t subpayload_size = 0;
- size_t subalign_bytes = 0;
- MDBX_page_type_t subtype;
-
- switch (sp->mp_flags & /* ignore legacy P_DIRTY flag */ ~0x10) {
- case P_LEAF | P_SUBP:
- subtype = MDBX_subpage_leaf;
- break;
- case P_LEAF | P_LEAF2 | P_SUBP:
- subtype = MDBX_subpage_dupfixed_leaf;
- break;
- default:
- subtype = MDBX_subpage_broken;
- err = MDBX_CORRUPTED;
- }
-
- for (int j = 0; err == MDBX_SUCCESS && j < nsubkeys;
- subalign_bytes += ((subpayload_size + subalign_bytes) & 1), j++) {
-
- if (subtype == MDBX_subpage_dupfixed_leaf) {
- /* LEAF2 pages have no mp_ptrs[] or node headers */
- subpayload_size += sp->mp_leaf2_ksize;
- } else {
- assert(subtype == MDBX_subpage_leaf);
- MDBX_node *subnode = page_node(sp, j);
- subpayload_size += NODESIZE + node_ks(subnode) + node_ds(subnode);
- if (unlikely(node_flags(subnode) != 0))
- err = MDBX_CORRUPTED;
- }
- }
-
- rc = ctx->mw_visitor(pgno, 0, ctx->mw_user, deep + 1, name, node_ds(node),
- subtype, err, nsubkeys, subpayload_size,
- subheader_size, subunused_size + subalign_bytes);
- if (unlikely(rc != MDBX_SUCCESS))
- return (rc == MDBX_RESULT_TRUE) ? MDBX_SUCCESS : rc;
- header_size += subheader_size;
- unused_size += subunused_size;
- payload_size += subpayload_size;
- align_bytes += subalign_bytes;
- } break;
-
- default:
- err = MDBX_CORRUPTED;
- }
- }
-
- rc = ctx->mw_visitor(pgno, 1, ctx->mw_user, deep, name,
- ctx->mw_txn->mt_env->me_psize, type, err, nentries,
- payload_size, header_size, unused_size + align_bytes);
- if (unlikely(rc != MDBX_SUCCESS))
- return (rc == MDBX_RESULT_TRUE) ? MDBX_SUCCESS : rc;
-
- for (int i = 0; err == MDBX_SUCCESS && i < nentries; i++) {
- if (type == MDBX_page_dupfixed_leaf)
- continue;
-
- MDBX_node *node = page_node(mp, i);
- if (type == MDBX_page_branch) {
- err = mdbx_walk_tree(ctx, node_pgno(node), name, deep + 1,
- pp_txnid4chk(mp, ctx->mw_txn));
- if (unlikely(err != MDBX_SUCCESS)) {
- if (err == MDBX_RESULT_TRUE)
- break;
- return err;
- }
- continue;
- }
-
- assert(type == MDBX_page_leaf);
- MDBX_db db;
- switch (node_flags(node)) {
- default:
- continue;
-
- case F_SUBDATA /* sub-db */: {
- const size_t namelen = node_ks(node);
- if (unlikely(namelen == 0 || node_ds(node) != sizeof(MDBX_db))) {
- err = MDBX_CORRUPTED;
- break;
- }
-
- char namebuf_onstask[64];
- char *const sub_name = (namelen < sizeof(namebuf_onstask))
- ? namebuf_onstask
- : mdbx_malloc(namelen + 1);
- if (sub_name) {
- memcpy(sub_name, node_key(node), namelen);
- sub_name[namelen] = 0;
- memcpy(&db, node_data(node), sizeof(db));
- err = mdbx_walk_sdb(ctx, &db, sub_name, deep + 1);
- if (sub_name != namebuf_onstask)
- mdbx_free(sub_name);
- } else {
- err = MDBX_ENOMEM;
- }
- } break;
-
- case F_SUBDATA | F_DUPDATA /* dupsorted sub-tree */:
- if (unlikely(node_ds(node) != sizeof(MDBX_db) ||
- ctx->mw_cursor->mc_xcursor == NULL))
- err = MDBX_CORRUPTED;
- else {
- memcpy(&db, node_data(node), sizeof(db));
- assert(ctx->mw_cursor->mc_xcursor ==
- &container_of(ctx->mw_cursor, MDBX_cursor_couple, outer)->inner);
- ctx->mw_cursor = &ctx->mw_cursor->mc_xcursor->mx_cursor;
- err = mdbx_walk_tree(ctx, db.md_root, name, deep + 1,
- pp_txnid4chk(mp, ctx->mw_txn));
- MDBX_xcursor *inner_xcursor =
- container_of(ctx->mw_cursor, MDBX_xcursor, mx_cursor);
- MDBX_cursor_couple *couple =
- container_of(inner_xcursor, MDBX_cursor_couple, inner);
- ctx->mw_cursor = &couple->outer;
- }
- break;
- }
- }
-
- return MDBX_SUCCESS;
-}
-
-__cold static int mdbx_walk_sdb(mdbx_walk_ctx_t *ctx, MDBX_db *const db,
- const char *name, int deep) {
- if (unlikely(db->md_root == P_INVALID))
- return MDBX_SUCCESS; /* empty db */
-
- MDBX_cursor_couple couple;
- MDBX_dbx dbx = {.md_klen_min = INT_MAX};
- uint8_t dbistate = DBI_VALID | DBI_AUDITED;
- int rc = mdbx_couple_init(&couple, ~0u, ctx->mw_txn, db, &dbx, &dbistate);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (ctx->mw_dont_check_keys_ordering) {
- couple.outer.mc_flags |= C_SKIPORD;
- couple.inner.mx_cursor.mc_flags |= C_SKIPORD;
- }
- couple.outer.mc_next = ctx->mw_cursor;
- ctx->mw_cursor = &couple.outer;
- rc = mdbx_walk_tree(ctx, db->md_root, name, deep, ctx->mw_txn->mt_txnid);
- ctx->mw_cursor = couple.outer.mc_next;
- return rc;
-}
-
-__cold int mdbx_env_pgwalk(MDBX_txn *txn, MDBX_pgvisitor_func *visitor,
- void *user, bool dont_check_keys_ordering) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- mdbx_walk_ctx_t ctx;
- memset(&ctx, 0, sizeof(ctx));
- ctx.mw_txn = txn;
- ctx.mw_user = user;
- ctx.mw_visitor = visitor;
- ctx.mw_dont_check_keys_ordering = dont_check_keys_ordering;
-
- rc = visitor(0, NUM_METAS, user, 0, MDBX_PGWALK_META,
- pgno2bytes(txn->mt_env, NUM_METAS), MDBX_page_meta, MDBX_SUCCESS,
- NUM_METAS, sizeof(MDBX_meta) * NUM_METAS, PAGEHDRSZ * NUM_METAS,
- (txn->mt_env->me_psize - sizeof(MDBX_meta) - PAGEHDRSZ) *
- NUM_METAS);
- if (!MDBX_IS_ERROR(rc))
- rc = mdbx_walk_sdb(&ctx, &txn->mt_dbs[FREE_DBI], MDBX_PGWALK_GC, 0);
- if (!MDBX_IS_ERROR(rc))
- rc = mdbx_walk_sdb(&ctx, &txn->mt_dbs[MAIN_DBI], MDBX_PGWALK_MAIN, 0);
- return rc;
-}
-
-int mdbx_canary_put(MDBX_txn *txn, const MDBX_canary *canary) {
- int rc = check_txn_rw(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (likely(canary)) {
- if (txn->mt_canary.x == canary->x && txn->mt_canary.y == canary->y &&
- txn->mt_canary.z == canary->z)
- return MDBX_SUCCESS;
- txn->mt_canary.x = canary->x;
- txn->mt_canary.y = canary->y;
- txn->mt_canary.z = canary->z;
- }
- txn->mt_canary.v = txn->mt_txnid;
- txn->mt_flags |= MDBX_TXN_DIRTY;
-
- return MDBX_SUCCESS;
-}
-
-int mdbx_canary_get(const MDBX_txn *txn, MDBX_canary *canary) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(canary == NULL))
- return MDBX_EINVAL;
-
- *canary = txn->mt_canary;
- return MDBX_SUCCESS;
-}
-
-int mdbx_cursor_on_first(const MDBX_cursor *mc) {
- if (unlikely(mc == NULL))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_LIVE))
- return (mc->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- if (!(mc->mc_flags & C_INITIALIZED))
- return mc->mc_db->md_entries ? MDBX_RESULT_FALSE : MDBX_RESULT_TRUE;
-
- for (unsigned i = 0; i < mc->mc_snum; ++i) {
- if (mc->mc_ki[i])
- return MDBX_RESULT_FALSE;
- }
-
- return MDBX_RESULT_TRUE;
-}
-
-int mdbx_cursor_on_last(const MDBX_cursor *mc) {
- if (unlikely(mc == NULL))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_LIVE))
- return (mc->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- if (!(mc->mc_flags & C_INITIALIZED))
- return mc->mc_db->md_entries ? MDBX_RESULT_FALSE : MDBX_RESULT_TRUE;
-
- for (unsigned i = 0; i < mc->mc_snum; ++i) {
- unsigned nkeys = page_numkeys(mc->mc_pg[i]);
- if (mc->mc_ki[i] < nkeys - 1)
- return MDBX_RESULT_FALSE;
- }
-
- return MDBX_RESULT_TRUE;
-}
-
-int mdbx_cursor_eof(const MDBX_cursor *mc) {
- if (unlikely(mc == NULL))
- return MDBX_EINVAL;
-
- if (unlikely(mc->mc_signature != MDBX_MC_LIVE))
- return (mc->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- return ((mc->mc_flags & (C_INITIALIZED | C_EOF)) == C_INITIALIZED &&
- mc->mc_snum &&
- mc->mc_ki[mc->mc_top] < page_numkeys(mc->mc_pg[mc->mc_top]))
- ? MDBX_RESULT_FALSE
- : MDBX_RESULT_TRUE;
-}
-
-//------------------------------------------------------------------------------
-
-struct diff_result {
- ptrdiff_t diff;
- unsigned level;
- int root_nkeys;
-};
-
-/* calculates: r = x - y */
-__hot static int cursor_diff(const MDBX_cursor *const __restrict x,
- const MDBX_cursor *const __restrict y,
- struct diff_result *const __restrict r) {
- r->diff = 0;
- r->level = 0;
- r->root_nkeys = 0;
-
- if (unlikely(x->mc_signature != MDBX_MC_LIVE))
- return (x->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- if (unlikely(y->mc_signature != MDBX_MC_LIVE))
- return (y->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- int rc = check_txn(x->mc_txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(x->mc_txn != y->mc_txn))
- return MDBX_BAD_TXN;
-
- if (unlikely(y->mc_dbi != x->mc_dbi))
- return MDBX_EINVAL;
-
- if (unlikely(!(y->mc_flags & x->mc_flags & C_INITIALIZED)))
- return MDBX_ENODATA;
-
- while (likely(r->level < y->mc_snum && r->level < x->mc_snum)) {
- if (unlikely(y->mc_pg[r->level] != x->mc_pg[r->level])) {
- mdbx_error("Mismatch cursors's pages at %u level", r->level);
- return MDBX_PROBLEM;
- }
-
- int nkeys = page_numkeys(y->mc_pg[r->level]);
- assert(nkeys > 0);
- if (r->level == 0)
- r->root_nkeys = nkeys;
-
- const int limit_ki = nkeys - 1;
- const int x_ki = x->mc_ki[r->level];
- const int y_ki = y->mc_ki[r->level];
- r->diff = ((x_ki < limit_ki) ? x_ki : limit_ki) -
- ((y_ki < limit_ki) ? y_ki : limit_ki);
- if (r->diff == 0) {
- r->level += 1;
- continue;
- }
-
- while (unlikely(r->diff == 1) &&
- likely(r->level + 1 < y->mc_snum && r->level + 1 < x->mc_snum)) {
- r->level += 1;
- /* DB'PAGEs: 0------------------>MAX
- *
- * CURSORs: y < x
- * STACK[i ]: |
- * STACK[+1]: ...y++N|0++x...
- */
- nkeys = page_numkeys(y->mc_pg[r->level]);
- r->diff = (nkeys - y->mc_ki[r->level]) + x->mc_ki[r->level];
- assert(r->diff > 0);
- }
-
- while (unlikely(r->diff == -1) &&
- likely(r->level + 1 < y->mc_snum && r->level + 1 < x->mc_snum)) {
- r->level += 1;
- /* DB'PAGEs: 0------------------>MAX
- *
- * CURSORs: x < y
- * STACK[i ]: |
- * STACK[+1]: ...x--N|0--y...
- */
- nkeys = page_numkeys(x->mc_pg[r->level]);
- r->diff = -(nkeys - x->mc_ki[r->level]) - y->mc_ki[r->level];
- assert(r->diff < 0);
- }
-
- return MDBX_SUCCESS;
- }
-
- r->diff = CMP2INT(x->mc_flags & C_EOF, y->mc_flags & C_EOF);
- return MDBX_SUCCESS;
-}
-
-__hot static ptrdiff_t estimate(const MDBX_db *db,
- struct diff_result *const __restrict dr) {
- /* root: branch-page => scale = leaf-factor * branch-factor^(N-1)
- * level-1: branch-page(s) => scale = leaf-factor * branch-factor^2
- * level-2: branch-page(s) => scale = leaf-factor * branch-factor
- * level-N: branch-page(s) => scale = leaf-factor
- * leaf-level: leaf-page(s) => scale = 1
- */
- ptrdiff_t btree_power = (ptrdiff_t)db->md_depth - 2 - (ptrdiff_t)dr->level;
- if (btree_power < 0)
- return dr->diff;
-
- ptrdiff_t estimated =
- (ptrdiff_t)db->md_entries * dr->diff / (ptrdiff_t)db->md_leaf_pages;
- if (btree_power == 0)
- return estimated;
-
- if (db->md_depth < 4) {
- assert(dr->level == 0 && btree_power == 1);
- return (ptrdiff_t)db->md_entries * dr->diff / (ptrdiff_t)dr->root_nkeys;
- }
-
- /* average_branchpage_fillfactor = total(branch_entries) / branch_pages
- total(branch_entries) = leaf_pages + branch_pages - 1 (root page) */
- const size_t log2_fixedpoint = sizeof(size_t) - 1;
- const size_t half = UINT64_C(1) << (log2_fixedpoint - 1);
- const size_t factor =
- ((db->md_leaf_pages + db->md_branch_pages - 1) << log2_fixedpoint) /
- db->md_branch_pages;
- while (1) {
- switch ((size_t)btree_power) {
- default: {
- const size_t square = (factor * factor + half) >> log2_fixedpoint;
- const size_t quad = (square * square + half) >> log2_fixedpoint;
- do {
- estimated = estimated * quad + half;
- estimated >>= log2_fixedpoint;
- btree_power -= 4;
- } while (btree_power >= 4);
- continue;
- }
- case 3:
- estimated = estimated * factor + half;
- estimated >>= log2_fixedpoint;
- __fallthrough /* fall through */;
- case 2:
- estimated = estimated * factor + half;
- estimated >>= log2_fixedpoint;
- __fallthrough /* fall through */;
- case 1:
- estimated = estimated * factor + half;
- estimated >>= log2_fixedpoint;
- __fallthrough /* fall through */;
- case 0:
- if (unlikely(estimated > (ptrdiff_t)db->md_entries))
- return (ptrdiff_t)db->md_entries;
- if (unlikely(estimated < -(ptrdiff_t)db->md_entries))
- return -(ptrdiff_t)db->md_entries;
- return estimated;
- }
- }
-}
-
-int mdbx_estimate_distance(const MDBX_cursor *first, const MDBX_cursor *last,
- ptrdiff_t *distance_items) {
- if (unlikely(first == NULL || last == NULL || distance_items == NULL))
- return MDBX_EINVAL;
-
- *distance_items = 0;
- struct diff_result dr;
- int rc = cursor_diff(last, first, &dr);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(dr.diff == 0) &&
- F_ISSET(first->mc_db->md_flags & last->mc_db->md_flags,
- MDBX_DUPSORT | C_INITIALIZED)) {
- first = &first->mc_xcursor->mx_cursor;
- last = &last->mc_xcursor->mx_cursor;
- rc = cursor_diff(first, last, &dr);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
-
- if (likely(dr.diff != 0))
- *distance_items = estimate(first->mc_db, &dr);
-
- return MDBX_SUCCESS;
-}
-
-int mdbx_estimate_move(const MDBX_cursor *cursor, MDBX_val *key, MDBX_val *data,
- MDBX_cursor_op move_op, ptrdiff_t *distance_items) {
- if (unlikely(cursor == NULL || distance_items == NULL ||
- move_op == MDBX_GET_CURRENT || move_op == MDBX_GET_MULTIPLE))
- return MDBX_EINVAL;
-
- if (unlikely(cursor->mc_signature != MDBX_MC_LIVE))
- return (cursor->mc_signature == MDBX_MC_READY4CLOSE) ? MDBX_EINVAL
- : MDBX_EBADSIGN;
-
- int rc = check_txn(cursor->mc_txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (!(cursor->mc_flags & C_INITIALIZED))
- return MDBX_ENODATA;
-
- MDBX_cursor_couple next;
- cursor_copy(cursor, &next.outer);
- if (cursor->mc_db->md_flags & MDBX_DUPSORT) {
- next.outer.mc_xcursor = &next.inner;
- rc = mdbx_xcursor_init0(&next.outer);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- MDBX_xcursor *mx = &container_of(cursor, MDBX_cursor_couple, outer)->inner;
- cursor_copy(&mx->mx_cursor, &next.inner.mx_cursor);
- }
-
- MDBX_val stub = {0, 0};
- if (data == NULL) {
- const unsigned mask =
- 1 << MDBX_GET_BOTH | 1 << MDBX_GET_BOTH_RANGE | 1 << MDBX_SET_KEY;
- if (unlikely(mask & (1 << move_op)))
- return MDBX_EINVAL;
- data = &stub;
- }
-
- if (key == NULL) {
- const unsigned mask = 1 << MDBX_GET_BOTH | 1 << MDBX_GET_BOTH_RANGE |
- 1 << MDBX_SET_KEY | 1 << MDBX_SET |
- 1 << MDBX_SET_RANGE;
- if (unlikely(mask & (1 << move_op)))
- return MDBX_EINVAL;
- key = &stub;
- }
-
- next.outer.mc_signature = MDBX_MC_LIVE;
- rc = mdbx_cursor_get(&next.outer, key, data, move_op);
- if (unlikely(rc != MDBX_SUCCESS &&
- (rc != MDBX_NOTFOUND || !(next.outer.mc_flags & C_INITIALIZED))))
- return rc;
-
- return mdbx_estimate_distance(cursor, &next.outer, distance_items);
-}
-
-int mdbx_estimate_range(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *begin_key,
- MDBX_val *begin_data, MDBX_val *end_key,
- MDBX_val *end_data, ptrdiff_t *size_items) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!size_items))
- return MDBX_EINVAL;
-
- if (unlikely(begin_data && (begin_key == NULL || begin_key == MDBX_EPSILON)))
- return MDBX_EINVAL;
-
- if (unlikely(end_data && (end_key == NULL || end_key == MDBX_EPSILON)))
- return MDBX_EINVAL;
-
- if (unlikely(begin_key == MDBX_EPSILON && end_key == MDBX_EPSILON))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- MDBX_cursor_couple begin;
- /* LY: first, initialize cursor to refresh a DB in case it have DB_STALE */
- rc = mdbx_cursor_init(&begin.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(begin.outer.mc_db->md_entries == 0)) {
- *size_items = 0;
- return MDBX_SUCCESS;
- }
-
- if (!begin_key) {
- if (unlikely(!end_key)) {
- /* LY: FIRST..LAST case */
- *size_items = (ptrdiff_t)begin.outer.mc_db->md_entries;
- return MDBX_SUCCESS;
- }
- MDBX_val stub = {0, 0};
- rc = mdbx_cursor_first(&begin.outer, &stub, &stub);
- if (unlikely(end_key == MDBX_EPSILON)) {
- /* LY: FIRST..+epsilon case */
- return (rc == MDBX_SUCCESS)
- ? mdbx_cursor_count(&begin.outer, (size_t *)size_items)
- : rc;
- }
- } else {
- if (unlikely(begin_key == MDBX_EPSILON)) {
- if (end_key == NULL) {
- /* LY: -epsilon..LAST case */
- MDBX_val stub = {0, 0};
- rc = mdbx_cursor_last(&begin.outer, &stub, &stub);
- return (rc == MDBX_SUCCESS)
- ? mdbx_cursor_count(&begin.outer, (size_t *)size_items)
- : rc;
- }
- /* LY: -epsilon..value case */
- assert(end_key != MDBX_EPSILON);
- begin_key = end_key;
- } else if (unlikely(end_key == MDBX_EPSILON)) {
- /* LY: value..+epsilon case */
- assert(begin_key != MDBX_EPSILON);
- end_key = begin_key;
- }
- if (end_key && !begin_data && !end_data &&
- (begin_key == end_key ||
- begin.outer.mc_dbx->md_cmp(begin_key, end_key) == 0)) {
- /* LY: single key case */
- rc = mdbx_cursor_set(&begin.outer, begin_key, NULL, MDBX_SET).err;
- if (unlikely(rc != MDBX_SUCCESS)) {
- *size_items = 0;
- return (rc == MDBX_NOTFOUND) ? MDBX_SUCCESS : rc;
- }
- *size_items = 1;
- if (begin.outer.mc_xcursor != NULL) {
- MDBX_node *node = page_node(begin.outer.mc_pg[begin.outer.mc_top],
- begin.outer.mc_ki[begin.outer.mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- /* LY: return the number of duplicates for given key */
- mdbx_tassert(txn,
- begin.outer.mc_xcursor == &begin.inner &&
- (begin.inner.mx_cursor.mc_flags & C_INITIALIZED));
- *size_items =
- (sizeof(*size_items) >= sizeof(begin.inner.mx_db.md_entries) ||
- begin.inner.mx_db.md_entries <= PTRDIFF_MAX)
- ? (size_t)begin.inner.mx_db.md_entries
- : PTRDIFF_MAX;
- }
- }
- return MDBX_SUCCESS;
- } else {
- rc = mdbx_cursor_set(&begin.outer, begin_key, begin_data,
- begin_data ? MDBX_GET_BOTH_RANGE : MDBX_SET_RANGE)
- .err;
- }
- }
-
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc != MDBX_NOTFOUND || !(begin.outer.mc_flags & C_INITIALIZED))
- return rc;
- }
-
- MDBX_cursor_couple end;
- rc = mdbx_cursor_init(&end.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- if (!end_key) {
- MDBX_val stub = {0, 0};
- rc = mdbx_cursor_last(&end.outer, &stub, &stub);
- } else {
- rc = mdbx_cursor_set(&end.outer, end_key, end_data,
- end_data ? MDBX_GET_BOTH_RANGE : MDBX_SET_RANGE)
- .err;
- }
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc != MDBX_NOTFOUND || !(end.outer.mc_flags & C_INITIALIZED))
- return rc;
- }
-
- rc = mdbx_estimate_distance(&begin.outer, &end.outer, size_items);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- assert(*size_items >= -(ptrdiff_t)begin.outer.mc_db->md_entries &&
- *size_items <= (ptrdiff_t)begin.outer.mc_db->md_entries);
-
-#if 0 /* LY: Was decided to returns as-is (i.e. negative) the estimation \
- * results for an inverted ranges. */
-
- /* Commit 8ddfd1f34ad7cf7a3c4aa75d2e248ca7e639ed63
- Change-Id: If59eccf7311123ab6384c4b93f9b1fed5a0a10d1 */
-
- if (*size_items < 0) {
- /* LY: inverted range case */
- *size_items += (ptrdiff_t)begin.outer.mc_db->md_entries;
- } else if (*size_items == 0 && begin_key && end_key) {
- int cmp = begin.outer.mc_dbx->md_cmp(&origin_begin_key, &origin_end_key);
- if (cmp == 0 && (begin.inner.mx_cursor.mc_flags & C_INITIALIZED) &&
- begin_data && end_data)
- cmp = begin.outer.mc_dbx->md_dcmp(&origin_begin_data, &origin_end_data);
- if (cmp > 0) {
- /* LY: inverted range case with empty scope */
- *size_items = (ptrdiff_t)begin.outer.mc_db->md_entries;
- }
- }
- assert(*size_items >= 0 &&
- *size_items <= (ptrdiff_t)begin.outer.mc_db->md_entries);
-#endif
-
- return MDBX_SUCCESS;
-}
-
-//------------------------------------------------------------------------------
-
-/* ПозволÑет обновить или удалить ÑущеÑтвующую запиÑÑŒ Ñ Ð¿Ð¾Ð»ÑƒÑ‡ÐµÐ½Ð¸ÐµÐ¼
- * в old_data предыдущего Ð·Ð½Ð°Ñ‡ÐµÐ½Ð¸Ñ Ð´Ð°Ð½Ð½Ñ‹Ñ…. При Ñтом еÑли new_data равен
- * нулю, то выполнÑетÑÑ ÑƒÐ´Ð°Ð»ÐµÐ½Ð¸Ðµ, иначе обновление/вÑтавка.
- *
- * Текущее значение может находитьÑÑ Ð² уже измененной (грÑзной) Ñтранице.
- * Ð’ Ñтом Ñлучае Ñтраница будет перезапиÑана при обновлении, а Ñамо Ñтарое
- * значение утрачено. ПоÑтому иÑходно в old_data должен быть передан
- * дополнительный буфер Ð´Ð»Ñ ÐºÐ¾Ð¿Ð¸Ñ€Ð¾Ð²Ð°Ð½Ð¸Ñ Ñтарого значениÑ.
- * ЕÑли переданный буфер Ñлишком мал, то Ñ„ÑƒÐ½ÐºÑ†Ð¸Ñ Ð²ÐµÑ€Ð½ÐµÑ‚ -1, уÑтановив
- * old_data->iov_len в ÑоответÑтвующее значение.
- *
- * Ð”Ð»Ñ Ð½Ðµ-уникальных ключей также возможен второй Ñценарий иÑпользованиÑ,
- * когда поÑредÑтвом old_data из запиÑей Ñ Ð¾Ð´Ð¸Ð½Ð°ÐºÐ¾Ð²Ñ‹Ð¼ ключом длÑ
- * удалениÑ/Ð¾Ð±Ð½Ð¾Ð²Ð»ÐµÐ½Ð¸Ñ Ð²Ñ‹Ð±Ð¸Ñ€Ð°ÐµÑ‚ÑÑ ÐºÐ¾Ð½ÐºÑ€ÐµÑ‚Ð½Ð°Ñ. Ð”Ð»Ñ Ð²Ñ‹Ð±Ð¾Ñ€Ð° Ñтого ÑценариÑ
- * во flags Ñледует одновременно указать MDBX_CURRENT и MDBX_NOOVERWRITE.
- * Именно Ñта ÐºÐ¾Ð¼Ð±Ð¸Ð½Ð°Ñ†Ð¸Ñ Ð²Ñ‹Ð±Ñ€Ð°Ð½Ð°, так как она лишена ÑмыÑла, и Ñтим позволÑет
- * идентифицировать Ð·Ð°Ð¿Ñ€Ð¾Ñ Ñ‚Ð°ÐºÐ¾Ð³Ð¾ ÑценариÑ.
- *
- * Ð¤ÑƒÐ½ÐºÑ†Ð¸Ñ Ð¼Ð¾Ð¶ÐµÑ‚ быть замещена ÑоответÑтвующими операциÑми Ñ ÐºÑƒÑ€Ñорами
- * поÑле двух доработок (TODO):
- * - внешнÑÑ Ð°Ð»Ð»Ð¾ÐºÐ°Ñ†Ð¸Ñ ÐºÑƒÑ€Ñоров, в том чиÑле на Ñтеке (без malloc).
- * - Ð¿Ð¾Ð»ÑƒÑ‡ÐµÐ½Ð¸Ñ dirty-ÑтатуÑа Ñтраницы по адреÑу (знать о MUTABLE/WRITEABLE).
- */
-
-int mdbx_replace_ex(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key,
- MDBX_val *new_data, MDBX_val *old_data,
- MDBX_put_flags_t flags, MDBX_preserve_func preserver,
- void *preserver_context) {
- int rc = check_txn_rw(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!key || !old_data || old_data == new_data))
- return MDBX_EINVAL;
-
- if (unlikely(old_data->iov_base == NULL && old_data->iov_len))
- return MDBX_EINVAL;
-
- if (unlikely(new_data == NULL &&
- (flags & (MDBX_CURRENT | MDBX_RESERVE)) != MDBX_CURRENT))
- return MDBX_EINVAL;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- if (unlikely(flags &
- ~(MDBX_NOOVERWRITE | MDBX_NODUPDATA | MDBX_ALLDUPS |
- MDBX_RESERVE | MDBX_APPEND | MDBX_APPENDDUP | MDBX_CURRENT)))
- return MDBX_EINVAL;
-
- MDBX_cursor_couple cx;
- rc = mdbx_cursor_init(&cx.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- cx.outer.mc_next = txn->mt_cursors[dbi];
- txn->mt_cursors[dbi] = &cx.outer;
-
- MDBX_val present_key = *key;
- if (F_ISSET(flags, MDBX_CURRENT | MDBX_NOOVERWRITE)) {
- /* в old_data значение Ð´Ð»Ñ Ð²Ñ‹Ð±Ð¾Ñ€Ð° конкретного дубликата */
- if (unlikely(!(txn->mt_dbs[dbi].md_flags & MDBX_DUPSORT))) {
- rc = MDBX_EINVAL;
- goto bailout;
- }
-
- /* убираем лишний бит, он был признаком запрошенного режима */
- flags -= MDBX_NOOVERWRITE;
-
- rc = mdbx_cursor_get(&cx.outer, &present_key, old_data, MDBX_GET_BOTH);
- if (rc != MDBX_SUCCESS)
- goto bailout;
- } else {
- /* в old_data буфер Ð´Ð»Ñ ÑÐ¾Ñ…Ñ€Ð°Ð½ÐµÐ½Ð¸Ñ Ð¿Ñ€ÐµÐ´Ñ‹Ð´ÑƒÑ‰ÐµÐ³Ð¾ Ð·Ð½Ð°Ñ‡ÐµÐ½Ð¸Ñ */
- if (unlikely(new_data && old_data->iov_base == new_data->iov_base))
- return MDBX_EINVAL;
- MDBX_val present_data;
- rc = mdbx_cursor_get(&cx.outer, &present_key, &present_data, MDBX_SET_KEY);
- if (unlikely(rc != MDBX_SUCCESS)) {
- old_data->iov_base = NULL;
- old_data->iov_len = 0;
- if (rc != MDBX_NOTFOUND || (flags & MDBX_CURRENT))
- goto bailout;
- } else if (flags & MDBX_NOOVERWRITE) {
- rc = MDBX_KEYEXIST;
- *old_data = present_data;
- goto bailout;
- } else {
- MDBX_page *page = cx.outer.mc_pg[cx.outer.mc_top];
- if (txn->mt_dbs[dbi].md_flags & MDBX_DUPSORT) {
- if (flags & MDBX_CURRENT) {
- /* disallow update/delete for multi-values */
- MDBX_node *node = page_node(page, cx.outer.mc_ki[cx.outer.mc_top]);
- if (F_ISSET(node_flags(node), F_DUPDATA)) {
- mdbx_tassert(txn, XCURSOR_INITED(&cx.outer) &&
- cx.outer.mc_xcursor->mx_db.md_entries > 1);
- if (cx.outer.mc_xcursor->mx_db.md_entries > 1) {
- rc = MDBX_EMULTIVAL;
- goto bailout;
- }
- }
- /* Ð’ оригинальной LMDB флажок MDBX_CURRENT здеÑÑŒ приведет
- * к замене данных без учета MDBX_DUPSORT Ñортировки,
- * но здеÑÑŒ Ñто в любом Ñлучае допуÑтимо, так как мы
- * проверили что Ð´Ð»Ñ ÐºÐ»ÑŽÑ‡Ð° еÑÑ‚ÑŒ только одно значение. */
- }
- }
-
- if (IS_MODIFIABLE(txn, page)) {
- if (new_data && cmp_lenfast(&present_data, new_data) == 0) {
- /* еÑли данные Ñовпадают, то ничего делать не надо */
- *old_data = *new_data;
- goto bailout;
- }
- rc = preserver ? preserver(preserver_context, old_data,
- present_data.iov_base, present_data.iov_len)
- : MDBX_SUCCESS;
- if (unlikely(rc != MDBX_SUCCESS))
- goto bailout;
- } else {
- *old_data = present_data;
- }
- flags |= MDBX_CURRENT;
- }
- }
-
- if (likely(new_data))
- rc = mdbx_cursor_put(&cx.outer, key, new_data, flags);
- else
- rc = mdbx_cursor_del(&cx.outer, flags & MDBX_ALLDUPS);
-
-bailout:
- txn->mt_cursors[dbi] = cx.outer.mc_next;
- return rc;
-}
-
-static int default_value_preserver(void *context, MDBX_val *target,
- const void *src, size_t bytes) {
- (void)context;
- if (unlikely(target->iov_len < bytes)) {
- target->iov_base = nullptr;
- target->iov_len = bytes;
- return MDBX_RESULT_TRUE;
- }
- memcpy(target->iov_base, src, target->iov_len = bytes);
- return MDBX_SUCCESS;
-}
-
-int mdbx_replace(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key,
- MDBX_val *new_data, MDBX_val *old_data,
- MDBX_put_flags_t flags) {
- return mdbx_replace_ex(txn, dbi, key, new_data, old_data, flags,
- default_value_preserver, nullptr);
-}
-
-/* Ð¤ÑƒÐ½ÐºÑ†Ð¸Ñ Ñообщает находитÑÑ Ð»Ð¸ указанный Ð°Ð´Ñ€ÐµÑ Ð² "грÑзной" Ñтранице у
- * заданной пишущей транзакции. Ð’ конечном Ñчете Ñто позволÑет избавитьÑÑ Ð¾Ñ‚
- * лишнего ÐºÐ¾Ð¿Ð¸Ñ€Ð¾Ð²Ð°Ð½Ð¸Ñ Ð´Ð°Ð½Ð½Ñ‹Ñ… из ÐЕ-грÑзных Ñтраниц.
- *
- * "ГрÑзные" Ñтраницы - Ñто те, которые уже были изменены в ходе пишущей
- * транзакции. СоответÑтвенно, какие-либо дальнейшие Ð¸Ð·Ð¼ÐµÐ½ÐµÐ½Ð¸Ñ Ð¼Ð¾Ð³ÑƒÑ‚ привеÑти
- * к перезапиÑи таких Ñтраниц. ПоÑтому вÑе функции, выполнÑющие изменениÑ, в
- * качеÑтве аргументов ÐЕ должны получать указатели на данные в таких
- * Ñтраницах. Ð’ Ñвою очередь "ÐЕ грÑзные" Ñтраницы перед модификацией будут
- * Ñкопированы.
- *
- * Другими Ñловами, данные из "грÑзных" Ñтраниц должны быть либо Ñкопированы
- * перед передачей в качеÑтве аргументов Ð´Ð»Ñ Ð´Ð°Ð»ÑŒÐ½ÐµÐ¹ÑˆÐ¸Ñ… модификаций, либо
- * отвергнуты на Ñтадии проверки корректноÑти аргументов.
- *
- * Таким образом, Ñ„ÑƒÐ½ÐºÑ†Ð¸Ñ Ð¿Ð¾Ð·Ð²Ð¾Ð»Ñет как избавитÑÑ Ð¾Ñ‚ лишнего копированиÑ,
- * так и выполнить более полную проверку аргументов.
- *
- * Ð’ÐЖÐО: Передаваемый указатель должен указывать на начало данных. Только
- * так гарантируетÑÑ Ñ‡Ñ‚Ð¾ актуальный заголовок Ñтраницы будет физичеÑки
- * раÑположен в той-же Ñтранице памÑти, в том чиÑле Ð´Ð»Ñ Ð¼Ð½Ð¾Ð³Ð¾Ñтраничных
- * P_OVERFLOW Ñтраниц Ñ Ð´Ð»Ð¸Ð½Ð½Ñ‹Ð¼Ð¸ данными. */
-int mdbx_is_dirty(const MDBX_txn *txn, const void *ptr) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- const MDBX_env *env = txn->mt_env;
- const ptrdiff_t offset = (uint8_t *)ptr - env->me_map;
- if (offset >= 0) {
- const pgno_t pgno = bytes2pgno(env, offset);
- if (likely(pgno < txn->mt_next_pgno)) {
- const MDBX_page *page = pgno2page(env, pgno);
- if (unlikely(page->mp_pgno != pgno ||
- (page->mp_flags & P_ILL_BITS) != 0)) {
- /* The ptr pointed into middle of a large page,
- * not to the beginning of a data. */
- return MDBX_EINVAL;
- }
- return ((txn->mt_flags & MDBX_TXN_RDONLY) || !IS_MODIFIABLE(txn, page))
- ? MDBX_RESULT_FALSE
- : MDBX_RESULT_TRUE;
- }
- if ((size_t)offset < env->me_dxb_mmap.limit) {
- /* Указатель адреÑует что-то в пределах mmap, но за границей
- * раÑпределенных Ñтраниц. Такое может ÑлучитÑÑ ÐµÑли mdbx_is_dirty()
- * вызываетÑÑ Ð¿Ð¾Ñле операции, в ходе которой грÑÐ·Ð½Ð°Ñ Ñтраница была
- * возвращена в нераÑпределенное проÑтранÑтво. */
- return (txn->mt_flags & MDBX_TXN_RDONLY) ? MDBX_EINVAL : MDBX_RESULT_TRUE;
- }
- }
-
- /* Страница вне иÑпользуемого mmap-диапазона, Ñ‚.е. либо в функцию был
- * передан некорректный адреÑ, либо Ð°Ð´Ñ€ÐµÑ Ð² теневой Ñтранице, ÐºÐ¾Ñ‚Ð¾Ñ€Ð°Ñ Ð±Ñ‹Ð»Ð°
- * выделена поÑредÑтвом malloc().
- *
- * Ð”Ð»Ñ Ñ€ÐµÐ¶Ð¸Ð¼Ð° MDBX_WRITE_MAP режима Ñтраница однозначно "не грÑзнаÑ",
- * а Ð´Ð»Ñ Ñ€ÐµÐ¶Ð¸Ð¼Ð¾Ð² без MDBX_WRITE_MAP однозначно "не чиÑтаÑ". */
- return (txn->mt_flags & (MDBX_WRITEMAP | MDBX_TXN_RDONLY)) ? MDBX_EINVAL
- : MDBX_RESULT_TRUE;
-}
-
-int mdbx_dbi_sequence(MDBX_txn *txn, MDBX_dbi dbi, uint64_t *result,
- uint64_t increment) {
- int rc = check_txn(txn, MDBX_TXN_BLOCKED);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (unlikely(!check_dbi(txn, dbi, DBI_USRVALID)))
- return MDBX_BAD_DBI;
-
- if (unlikely(txn->mt_dbistate[dbi] & DBI_STALE)) {
- rc = mdbx_fetch_sdb(txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- }
-
- MDBX_db *dbs = &txn->mt_dbs[dbi];
- if (likely(result))
- *result = dbs->md_seq;
-
- if (likely(increment > 0)) {
- if (unlikely(txn->mt_flags & MDBX_TXN_RDONLY))
- return MDBX_EACCESS;
-
- uint64_t new = dbs->md_seq + increment;
- if (unlikely(new < increment))
- return MDBX_RESULT_TRUE;
-
- mdbx_tassert(txn, new > dbs->md_seq);
- dbs->md_seq = new;
- txn->mt_flags |= MDBX_TXN_DIRTY;
- txn->mt_dbistate[dbi] |= DBI_DIRTY;
- }
-
- return MDBX_SUCCESS;
-}
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-__cold MDBX_NOTHROW_CONST_FUNCTION intptr_t mdbx_limits_pgsize_min(void) {
- return __inline_mdbx_limits_pgsize_min();
-}
-
-__cold MDBX_NOTHROW_CONST_FUNCTION intptr_t mdbx_limits_pgsize_max(void) {
- return __inline_mdbx_limits_pgsize_max();
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-__cold intptr_t mdbx_limits_dbsize_min(intptr_t pagesize) {
- if (pagesize < 1)
- pagesize = (intptr_t)mdbx_default_pagesize();
- else if (unlikely(pagesize < (intptr_t)MIN_PAGESIZE ||
- pagesize > (intptr_t)MAX_PAGESIZE ||
- !is_powerof2((size_t)pagesize)))
- return -1;
-
- return MIN_PAGENO * pagesize;
-}
-
-__cold intptr_t mdbx_limits_dbsize_max(intptr_t pagesize) {
- if (pagesize < 1)
- pagesize = (intptr_t)mdbx_default_pagesize();
- else if (unlikely(pagesize < (intptr_t)MIN_PAGESIZE ||
- pagesize > (intptr_t)MAX_PAGESIZE ||
- !is_powerof2((size_t)pagesize)))
- return -1;
-
- STATIC_ASSERT(MAX_MAPSIZE < INTPTR_MAX);
- const uint64_t limit = (1 + (uint64_t)MAX_PAGENO) * pagesize;
- return (limit < MAX_MAPSIZE) ? (intptr_t)limit : (intptr_t)MAX_MAPSIZE;
-}
-
-__cold intptr_t mdbx_limits_txnsize_max(intptr_t pagesize) {
- if (pagesize < 1)
- pagesize = (intptr_t)mdbx_default_pagesize();
- else if (unlikely(pagesize < (intptr_t)MIN_PAGESIZE ||
- pagesize > (intptr_t)MAX_PAGESIZE ||
- !is_powerof2((size_t)pagesize)))
- return -1;
-
- STATIC_ASSERT(MAX_MAPSIZE < INTPTR_MAX);
- const uint64_t pgl_limit =
- pagesize * (uint64_t)(MDBX_PGL_LIMIT / 1.6180339887498948482);
- const uint64_t map_limit = (uint64_t)(MAX_MAPSIZE / 1.6180339887498948482);
- return (pgl_limit < map_limit) ? (intptr_t)pgl_limit : (intptr_t)map_limit;
-}
-
-/*** Key-making functions to avoid custom comparators *************************/
-
-static __always_inline double key2double(const int64_t key) {
- union {
- uint64_t u;
- double f;
- } casting;
-
- casting.u = (key < 0) ? key + UINT64_C(0x8000000000000000)
- : UINT64_C(0xffffFFFFffffFFFF) - key;
- return casting.f;
-}
-
-static __always_inline uint64_t double2key(const double *const ptr) {
- STATIC_ASSERT(sizeof(double) == sizeof(int64_t));
- const int64_t i = *(const int64_t *)ptr;
- const uint64_t u = (i < 0) ? UINT64_C(0xffffFFFFffffFFFF) - i
- : i + UINT64_C(0x8000000000000000);
- if (mdbx_assert_enabled()) {
- const double f = key2double(u);
- assert(memcmp(&f, ptr, 8) == 0);
- }
- return u;
-}
-
-static __always_inline float key2float(const int32_t key) {
- union {
- uint32_t u;
- float f;
- } casting;
-
- casting.u =
- (key < 0) ? key + UINT32_C(0x80000000) : UINT32_C(0xffffFFFF) - key;
- return casting.f;
-}
-
-static __always_inline uint32_t float2key(const float *const ptr) {
- STATIC_ASSERT(sizeof(float) == sizeof(int32_t));
- const int32_t i = *(const int32_t *)ptr;
- const uint32_t u =
- (i < 0) ? UINT32_C(0xffffFFFF) - i : i + UINT32_C(0x80000000);
- if (mdbx_assert_enabled()) {
- const float f = key2float(u);
- assert(memcmp(&f, ptr, 4) == 0);
- }
- return u;
-}
-
-uint64_t mdbx_key_from_double(const double ieee754_64bit) {
- return double2key(&ieee754_64bit);
-}
-
-uint64_t mdbx_key_from_ptrdouble(const double *const ieee754_64bit) {
- return double2key(ieee754_64bit);
-}
-
-uint32_t mdbx_key_from_float(const float ieee754_32bit) {
- return float2key(&ieee754_32bit);
-}
-
-uint32_t mdbx_key_from_ptrfloat(const float *const ieee754_32bit) {
- return float2key(ieee754_32bit);
-}
-
-#ifndef LIBMDBX_NO_EXPORTS_LEGACY_API
-MDBX_NOTHROW_CONST_FUNCTION uint64_t mdbx_key_from_int64(const int64_t i64) {
- return __inline_mdbx_key_from_int64(i64);
-}
-
-MDBX_NOTHROW_CONST_FUNCTION uint32_t mdbx_key_from_int32(const int32_t i32) {
- return __inline_mdbx_key_from_int32(i32);
-}
-#endif /* LIBMDBX_NO_EXPORTS_LEGACY_API */
-
-#define IEEE754_DOUBLE_MANTISSA_SIZE 52
-#define IEEE754_DOUBLE_EXPONENTA_BIAS 0x3FF
-#define IEEE754_DOUBLE_EXPONENTA_MAX 0x7FF
-#define IEEE754_DOUBLE_IMPLICIT_LEAD UINT64_C(0x0010000000000000)
-#define IEEE754_DOUBLE_MANTISSA_MASK UINT64_C(0x000FFFFFFFFFFFFF)
-#define IEEE754_DOUBLE_MANTISSA_AMAX UINT64_C(0x001FFFFFFFFFFFFF)
-
-static __inline int clz64(uint64_t value) {
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_clzl)
- if (sizeof(value) == sizeof(int))
- return __builtin_clz(value);
- if (sizeof(value) == sizeof(long))
- return __builtin_clzl(value);
-#if (defined(__SIZEOF_LONG_LONG__) && __SIZEOF_LONG_LONG__ == 8) || \
- __has_builtin(__builtin_clzll)
- return __builtin_clzll(value);
-#endif /* have(long long) && long long == uint64_t */
-#endif /* GNU C */
-
-#if defined(_MSC_VER)
- unsigned long index;
-#if defined(_M_AMD64) || defined(_M_ARM64) || defined(_M_X64)
- _BitScanReverse64(&index, value);
- return 63 - index;
-#else
- if (value > UINT32_MAX) {
- _BitScanReverse(&index, (uint32_t)(value >> 32));
- return 31 - index;
- }
- _BitScanReverse(&index, (uint32_t)value);
- return 63 - index;
-#endif
-#endif /* MSVC */
-
- value |= value >> 1;
- value |= value >> 2;
- value |= value >> 4;
- value |= value >> 8;
- value |= value >> 16;
- value |= value >> 32;
- static const uint8_t debruijn_clz64[64] = {
- 63, 16, 62, 7, 15, 36, 61, 3, 6, 14, 22, 26, 35, 47, 60, 2,
- 9, 5, 28, 11, 13, 21, 42, 19, 25, 31, 34, 40, 46, 52, 59, 1,
- 17, 8, 37, 4, 23, 27, 48, 10, 29, 12, 43, 20, 32, 41, 53, 18,
- 38, 24, 49, 30, 44, 33, 54, 39, 50, 45, 55, 51, 56, 57, 58, 0};
- return debruijn_clz64[value * UINT64_C(0x03F79D71B4CB0A89) >> 58];
-}
-
-static __inline uint64_t round_mantissa(const uint64_t u64, int shift) {
- assert(shift < 0 && u64 > 0);
- shift = -shift;
- const unsigned half = 1 << (shift - 1);
- const unsigned lsb = 1 & (unsigned)(u64 >> shift);
- const unsigned tie2even = 1 ^ lsb;
- return (u64 + half - tie2even) >> shift;
-}
-
-uint64_t mdbx_key_from_jsonInteger(const int64_t json_integer) {
- const uint64_t bias = UINT64_C(0x8000000000000000);
- if (json_integer > 0) {
- const uint64_t u64 = json_integer;
- int shift = clz64(u64) - (64 - IEEE754_DOUBLE_MANTISSA_SIZE - 1);
- uint64_t mantissa = u64 << shift;
- if (unlikely(shift < 0)) {
- mantissa = round_mantissa(u64, shift);
- if (mantissa > IEEE754_DOUBLE_MANTISSA_AMAX)
- mantissa = round_mantissa(u64, --shift);
- }
-
- assert(mantissa >= IEEE754_DOUBLE_IMPLICIT_LEAD &&
- mantissa <= IEEE754_DOUBLE_MANTISSA_AMAX);
- const uint64_t exponent =
- IEEE754_DOUBLE_EXPONENTA_BIAS + IEEE754_DOUBLE_MANTISSA_SIZE - shift;
- assert(exponent > 0 && exponent <= IEEE754_DOUBLE_EXPONENTA_MAX);
- const uint64_t key = bias + (exponent << IEEE754_DOUBLE_MANTISSA_SIZE) +
- (mantissa - IEEE754_DOUBLE_IMPLICIT_LEAD);
-#if !defined(_MSC_VER) || \
- defined( \
- _DEBUG) /* Workaround for MSVC error LNK2019: unresolved external \
- symbol __except1 referenced in function __ftol3_except */
- assert(key == mdbx_key_from_double((double)json_integer));
-#endif /* Workaround for MSVC */
- return key;
- }
-
- if (json_integer < 0) {
- const uint64_t u64 = -json_integer;
- int shift = clz64(u64) - (64 - IEEE754_DOUBLE_MANTISSA_SIZE - 1);
- uint64_t mantissa = u64 << shift;
- if (unlikely(shift < 0)) {
- mantissa = round_mantissa(u64, shift);
- if (mantissa > IEEE754_DOUBLE_MANTISSA_AMAX)
- mantissa = round_mantissa(u64, --shift);
- }
-
- assert(mantissa >= IEEE754_DOUBLE_IMPLICIT_LEAD &&
- mantissa <= IEEE754_DOUBLE_MANTISSA_AMAX);
- const uint64_t exponent =
- IEEE754_DOUBLE_EXPONENTA_BIAS + IEEE754_DOUBLE_MANTISSA_SIZE - shift;
- assert(exponent > 0 && exponent <= IEEE754_DOUBLE_EXPONENTA_MAX);
- const uint64_t key = bias - 1 - (exponent << IEEE754_DOUBLE_MANTISSA_SIZE) -
- (mantissa - IEEE754_DOUBLE_IMPLICIT_LEAD);
-#if !defined(_MSC_VER) || \
- defined( \
- _DEBUG) /* Workaround for MSVC error LNK2019: unresolved external \
- symbol __except1 referenced in function __ftol3_except */
- assert(key == mdbx_key_from_double((double)json_integer));
-#endif /* Workaround for MSVC */
- return key;
- }
-
- return bias;
-}
-
-int64_t mdbx_jsonInteger_from_key(const MDBX_val v) {
- assert(v.iov_len == 8);
- const uint64_t key = unaligned_peek_u64(2, v.iov_base);
- const uint64_t bias = UINT64_C(0x8000000000000000);
- const uint64_t covalent = (key > bias) ? key - bias : bias - key - 1;
- const int shift = IEEE754_DOUBLE_EXPONENTA_BIAS + 63 -
- (IEEE754_DOUBLE_EXPONENTA_MAX &
- (int)(covalent >> IEEE754_DOUBLE_MANTISSA_SIZE));
- if (unlikely(shift < 1))
- return (key < bias) ? INT64_MIN : INT64_MAX;
- if (unlikely(shift > 63))
- return 0;
-
- const uint64_t unscaled = ((covalent & IEEE754_DOUBLE_MANTISSA_MASK)
- << (63 - IEEE754_DOUBLE_MANTISSA_SIZE)) +
- bias;
- const int64_t absolute = unscaled >> shift;
- const int64_t value = (key < bias) ? -absolute : absolute;
- assert(key == mdbx_key_from_jsonInteger(value) ||
- (mdbx_key_from_jsonInteger(value - 1) < key &&
- key < mdbx_key_from_jsonInteger(value + 1)));
- return value;
-}
-
-double mdbx_double_from_key(const MDBX_val v) {
- assert(v.iov_len == 8);
- return key2double(unaligned_peek_u64(2, v.iov_base));
-}
-
-float mdbx_float_from_key(const MDBX_val v) {
- assert(v.iov_len == 4);
- return key2float(unaligned_peek_u32(2, v.iov_base));
-}
-
-int32_t mdbx_int32_from_key(const MDBX_val v) {
- assert(v.iov_len == 4);
- return (int32_t)(unaligned_peek_u32(2, v.iov_base) - UINT32_C(0x80000000));
-}
-
-int64_t mdbx_int64_from_key(const MDBX_val v) {
- assert(v.iov_len == 8);
- return (int64_t)(unaligned_peek_u64(2, v.iov_base) -
- UINT64_C(0x8000000000000000));
-}
-
-__cold MDBX_cmp_func *mdbx_get_keycmp(unsigned flags) {
- return get_default_keycmp(flags);
-}
-
-__cold MDBX_cmp_func *mdbx_get_datacmp(unsigned flags) {
- return get_default_datacmp(flags);
-}
-
-__cold int mdbx_env_set_option(MDBX_env *env, const MDBX_option_t option,
- uint64_t value) {
- int err = check_env(env, false);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- const bool lock_needed = ((env->me_flags & MDBX_ENV_ACTIVE) && env->me_txn0 &&
- env->me_txn0->mt_owner != mdbx_thread_self());
- bool should_unlock = false;
- switch (option) {
- case MDBX_opt_sync_bytes:
- if (value == UINT64_MAX)
- value = SIZE_MAX - 65536;
- if (unlikely(env->me_flags & MDBX_RDONLY))
- return MDBX_EACCESS;
- if (unlikely(!(env->me_flags & MDBX_ENV_ACTIVE)))
- return MDBX_EPERM;
- if (unlikely(value > SIZE_MAX - 65536))
- return MDBX_TOO_LARGE;
- if (atomic_store32(&env->me_lck->mti_autosync_threshold,
- bytes2pgno(env, (size_t)value + env->me_psize - 1),
- mo_Relaxed) != 0 &&
- (env->me_flags & MDBX_ENV_ACTIVE)) {
- err = mdbx_env_sync_poll(env);
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
- err = MDBX_SUCCESS;
- }
- break;
-
- case MDBX_opt_sync_period:
- if (value == UINT64_MAX)
- value = UINT32_MAX;
- if (unlikely(env->me_flags & MDBX_RDONLY))
- return MDBX_EACCESS;
- if (unlikely(!(env->me_flags & MDBX_ENV_ACTIVE)))
- return MDBX_EPERM;
- if (unlikely(value > UINT32_MAX))
- return MDBX_TOO_LARGE;
- if (atomic_store64(&env->me_lck->mti_autosync_period,
- mdbx_osal_16dot16_to_monotime((uint32_t)value),
- mo_Relaxed) != 0 &&
- (env->me_flags & MDBX_ENV_ACTIVE)) {
- err = mdbx_env_sync_poll(env);
- if (unlikely(MDBX_IS_ERROR(err)))
- return err;
- err = MDBX_SUCCESS;
- }
- break;
-
- case MDBX_opt_max_db:
- if (value == UINT64_MAX)
- value = MDBX_MAX_DBI;
- if (unlikely(value > MDBX_MAX_DBI))
- return MDBX_EINVAL;
- if (unlikely(env->me_map))
- return MDBX_EPERM;
- env->me_maxdbs = (unsigned)value + CORE_DBS;
- break;
-
- case MDBX_opt_max_readers:
- if (value == UINT64_MAX)
- value = MDBX_READERS_LIMIT;
- if (unlikely(value < 1 || value > MDBX_READERS_LIMIT))
- return MDBX_EINVAL;
- if (unlikely(env->me_map))
- return MDBX_EPERM;
- env->me_maxreaders = (unsigned)value;
- break;
-
- case MDBX_opt_dp_reserve_limit:
- if (value == UINT64_MAX)
- value = INT_MAX;
- if (unlikely(value > INT_MAX))
- return MDBX_EINVAL;
- if (env->me_options.dp_reserve_limit != (unsigned)value) {
- if (lock_needed) {
- err = mdbx_txn_lock(env, false);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- should_unlock = true;
- }
- env->me_options.dp_reserve_limit = (unsigned)value;
- while (env->me_dp_reserve_len > env->me_options.dp_reserve_limit) {
- mdbx_assert(env, env->me_dp_reserve != NULL);
- MDBX_page *dp = env->me_dp_reserve;
- MDBX_ASAN_UNPOISON_MEMORY_REGION(dp, env->me_psize);
- VALGRIND_MAKE_MEM_DEFINED(&dp->mp_next, sizeof(dp->mp_next));
- env->me_dp_reserve = dp->mp_next;
- VALGRIND_MEMPOOL_FREE(env, dp);
- mdbx_free(dp);
- env->me_dp_reserve_len -= 1;
- }
- }
- break;
-
- case MDBX_opt_rp_augment_limit:
- if (value == UINT64_MAX)
- value = MDBX_PGL_LIMIT;
- if (unlikely(value > MDBX_PGL_LIMIT))
- return MDBX_EINVAL;
- env->me_options.rp_augment_limit = (unsigned)value;
- break;
-
- case MDBX_opt_txn_dp_limit:
- case MDBX_opt_txn_dp_initial:
- if (value == UINT64_MAX)
- value = MDBX_PGL_LIMIT;
- if (unlikely(value > MDBX_PGL_LIMIT || value < CURSOR_STACK * 4))
- return MDBX_EINVAL;
- if (unlikely(env->me_flags & MDBX_RDONLY))
- return MDBX_EACCESS;
- if (lock_needed) {
- err = mdbx_txn_lock(env, false);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- should_unlock = true;
- }
- if (env->me_txn)
- err = MDBX_EPERM /* unable change during transaction */;
- else {
- const pgno_t value32 = (pgno_t)value;
- if (option == MDBX_opt_txn_dp_initial &&
- env->me_options.dp_initial != value32) {
- env->me_options.dp_initial = value32;
- if (env->me_options.dp_limit < value32) {
- env->me_options.dp_limit = value32;
- env->me_options.flags.non_auto.dp_limit = 1;
- }
- }
- if (option == MDBX_opt_txn_dp_limit &&
- env->me_options.dp_limit != value32) {
- env->me_options.dp_limit = value32;
- env->me_options.flags.non_auto.dp_limit = 1;
- if (env->me_options.dp_initial > value32)
- env->me_options.dp_initial = value32;
- }
- }
- break;
-
- case MDBX_opt_spill_max_denominator:
- if (value == UINT64_MAX)
- value = 255;
- if (unlikely(value > 255))
- return MDBX_EINVAL;
- env->me_options.spill_max_denominator = (uint8_t)value;
- break;
- case MDBX_opt_spill_min_denominator:
- if (unlikely(value > 255))
- return MDBX_EINVAL;
- env->me_options.spill_min_denominator = (uint8_t)value;
- break;
- case MDBX_opt_spill_parent4child_denominator:
- if (unlikely(value > 255))
- return MDBX_EINVAL;
- env->me_options.spill_parent4child_denominator = (uint8_t)value;
- break;
-
- case MDBX_opt_loose_limit:
- if (value == UINT64_MAX)
- value = 255;
- if (unlikely(value > 255))
- return MDBX_EINVAL;
- env->me_options.dp_loose_limit = (uint8_t)value;
- break;
-
- case MDBX_opt_merge_threshold_16dot16_percent:
- if (value == UINT64_MAX)
- value = 32768;
- if (unlikely(value < 8192 || value > 32768))
- return MDBX_EINVAL;
- env->me_options.merge_threshold_16dot16_percent = (unsigned)value;
- recalculate_merge_threshold(env);
- break;
-
- default:
- return MDBX_EINVAL;
- }
-
- if (should_unlock)
- mdbx_txn_unlock(env);
- return err;
-}
-
-__cold int mdbx_env_get_option(const MDBX_env *env, const MDBX_option_t option,
- uint64_t *pvalue) {
- int err = check_env(env, false);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
- if (unlikely(!pvalue))
- return MDBX_EINVAL;
-
- switch (option) {
- case MDBX_opt_sync_bytes:
- if (unlikely(!(env->me_flags & MDBX_ENV_ACTIVE)))
- return MDBX_EPERM;
- *pvalue = pgno2bytes(
- env, atomic_load32(&env->me_lck->mti_autosync_threshold, mo_Relaxed));
- break;
-
- case MDBX_opt_sync_period:
- if (unlikely(!(env->me_flags & MDBX_ENV_ACTIVE)))
- return MDBX_EPERM;
- *pvalue = mdbx_osal_monotime_to_16dot16(
- atomic_load64(&env->me_lck->mti_autosync_period, mo_Relaxed));
- break;
-
- case MDBX_opt_max_db:
- *pvalue = env->me_maxdbs - CORE_DBS;
- break;
-
- case MDBX_opt_max_readers:
- *pvalue = env->me_maxreaders;
- break;
-
- case MDBX_opt_dp_reserve_limit:
- *pvalue = env->me_options.dp_reserve_limit;
- break;
-
- case MDBX_opt_rp_augment_limit:
- *pvalue = env->me_options.rp_augment_limit;
- break;
-
- case MDBX_opt_txn_dp_limit:
- *pvalue = env->me_options.dp_limit;
- break;
- case MDBX_opt_txn_dp_initial:
- *pvalue = env->me_options.dp_initial;
- break;
-
- case MDBX_opt_spill_max_denominator:
- *pvalue = env->me_options.spill_max_denominator;
- break;
- case MDBX_opt_spill_min_denominator:
- *pvalue = env->me_options.spill_min_denominator;
- break;
- case MDBX_opt_spill_parent4child_denominator:
- *pvalue = env->me_options.spill_parent4child_denominator;
- break;
-
- case MDBX_opt_loose_limit:
- *pvalue = env->me_options.dp_loose_limit;
- break;
-
- case MDBX_opt_merge_threshold_16dot16_percent:
- *pvalue = env->me_options.merge_threshold_16dot16_percent;
- break;
-
- default:
- return MDBX_EINVAL;
- }
-
- return MDBX_SUCCESS;
-}
-
-/*** Attribute support functions for Nexenta **********************************/
-#ifdef MDBX_NEXENTA_ATTRS
-
-static __inline int mdbx_attr_peek(MDBX_val *data, mdbx_attr_t *attrptr) {
- if (unlikely(data->iov_len < sizeof(mdbx_attr_t)))
- return MDBX_INCOMPATIBLE;
-
- if (likely(attrptr != NULL))
- *attrptr = *(mdbx_attr_t *)data->iov_base;
- data->iov_len -= sizeof(mdbx_attr_t);
- data->iov_base =
- likely(data->iov_len > 0) ? ((mdbx_attr_t *)data->iov_base) + 1 : NULL;
-
- return MDBX_SUCCESS;
-}
-
-static __inline int mdbx_attr_poke(MDBX_val *reserved, MDBX_val *data,
- mdbx_attr_t attr, MDBX_put_flags_t flags) {
- mdbx_attr_t *space = reserved->iov_base;
- if (flags & MDBX_RESERVE) {
- if (likely(data != NULL)) {
- data->iov_base = data->iov_len ? space + 1 : NULL;
- }
- } else {
- *space = attr;
- if (likely(data != NULL)) {
- memcpy(space + 1, data->iov_base, data->iov_len);
- }
- }
-
- return MDBX_SUCCESS;
-}
-
-int mdbx_cursor_get_attr(MDBX_cursor *mc, MDBX_val *key, MDBX_val *data,
- mdbx_attr_t *attrptr, MDBX_cursor_op op) {
- int rc = mdbx_cursor_get(mc, key, data, op);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- return mdbx_attr_peek(data, attrptr);
-}
-
-int mdbx_get_attr(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key, MDBX_val *data,
- uint64_t *attrptr) {
- int rc = mdbx_get(txn, dbi, key, data);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- return mdbx_attr_peek(data, attrptr);
-}
-
-int mdbx_put_attr(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key, MDBX_val *data,
- mdbx_attr_t attr, MDBX_put_flags_t flags) {
- MDBX_val reserve;
- reserve.iov_base = NULL;
- reserve.iov_len = (data ? data->iov_len : 0) + sizeof(mdbx_attr_t);
-
- int rc = mdbx_put(txn, dbi, key, &reserve, flags | MDBX_RESERVE);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- return mdbx_attr_poke(&reserve, data, attr, flags);
-}
-
-int mdbx_cursor_put_attr(MDBX_cursor *cursor, MDBX_val *key, MDBX_val *data,
- mdbx_attr_t attr, MDBX_put_flags_t flags) {
- MDBX_val reserve;
- reserve.iov_base = NULL;
- reserve.iov_len = (data ? data->iov_len : 0) + sizeof(mdbx_attr_t);
-
- int rc = mdbx_cursor_put(cursor, key, &reserve, flags | MDBX_RESERVE);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- return mdbx_attr_poke(&reserve, data, attr, flags);
-}
-
-int mdbx_set_attr(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key, MDBX_val *data,
- mdbx_attr_t attr) {
- if (unlikely(!key || !txn))
- return MDBX_EINVAL;
-
- if (unlikely(txn->mt_signature != MDBX_MT_SIGNATURE))
- return MDBX_EBADSIGN;
-
- if (unlikely(!TXN_DBI_EXIST(txn, dbi, DB_USRVALID)))
- return MDBX_BAD_DBI;
-
- if (unlikely(txn->mt_flags & (MDBX_TXN_RDONLY | MDBX_TXN_BLOCKED)))
- return (txn->mt_flags & MDBX_TXN_RDONLY) ? MDBX_EACCESS : MDBX_BAD_TXN;
-
- MDBX_cursor_couple cx;
- MDBX_val old_data;
- int rc = mdbx_cursor_init(&cx.outer, txn, dbi);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- rc = mdbx_cursor_set(&cx.outer, key, &old_data, MDBX_SET, NULL);
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc == MDBX_NOTFOUND && data) {
- cx.outer.mc_next = txn->mt_cursors[dbi];
- txn->mt_cursors[dbi] = &cx.outer;
- rc = mdbx_cursor_put_attr(&cx.outer, key, data, attr, 0);
- txn->mt_cursors[dbi] = cx.outer.mc_next;
- }
- return rc;
- }
-
- mdbx_attr_t old_attr = 0;
- rc = mdbx_attr_peek(&old_data, &old_attr);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
-
- if (old_attr == attr && (!data || (data->iov_len == old_data.iov_len &&
- memcmp(data->iov_base, old_data.iov_base,
- old_data.iov_len) == 0)))
- return MDBX_SUCCESS;
-
- cx.outer.mc_next = txn->mt_cursors[dbi];
- txn->mt_cursors[dbi] = &cx.outer;
- rc = mdbx_cursor_put_attr(&cx.outer, key, data ? data : &old_data, attr,
- MDBX_CURRENT);
- txn->mt_cursors[dbi] = cx.outer.mc_next;
- return rc;
-}
-#endif /* MDBX_NEXENTA_ATTRS */
-
-/******************************************************************************/
-/* *INDENT-OFF* */
-/* clang-format off */
-
-__dll_export
-#ifdef __attribute_used__
- __attribute_used__
-#elif defined(__GNUC__) || __has_attribute(__used__)
- __attribute__((__used__))
-#endif
-#ifdef __attribute_externally_visible__
- __attribute_externally_visible__
-#elif (defined(__GNUC__) && !defined(__clang__)) || \
- __has_attribute(__externally_visible__)
- __attribute__((__externally_visible__))
-#endif
- const struct MDBX_build_info mdbx_build = {
-#ifdef MDBX_BUILD_TIMESTAMP
- MDBX_BUILD_TIMESTAMP
-#else
- "\"" __DATE__ " " __TIME__ "\""
-#endif /* MDBX_BUILD_TIMESTAMP */
-
- ,
-#ifdef MDBX_BUILD_TARGET
- MDBX_BUILD_TARGET
-#else
- #if defined(__ANDROID_API__)
- "Android" MDBX_STRINGIFY(__ANDROID_API__)
- #elif defined(__linux__) || defined(__gnu_linux__)
- "Linux"
- #elif defined(EMSCRIPTEN) || defined(__EMSCRIPTEN__)
- "webassembly"
- #elif defined(__CYGWIN__)
- "CYGWIN"
- #elif defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) \
- || defined(__WINDOWS__)
- "Windows"
- #elif defined(__APPLE__)
- #if (defined(TARGET_OS_IPHONE) && TARGET_OS_IPHONE) \
- || (defined(TARGET_IPHONE_SIMULATOR) && TARGET_IPHONE_SIMULATOR)
- "iOS"
- #else
- "MacOS"
- #endif
- #elif defined(__FreeBSD__)
- "FreeBSD"
- #elif defined(__DragonFly__)
- "DragonFlyBSD"
- #elif defined(__NetBSD__)
- "NetBSD"
- #elif defined(__OpenBSD__)
- "OpenBSD"
- #elif defined(__bsdi__)
- "UnixBSDI"
- #elif defined(__MACH__)
- "MACH"
- #elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC))
- "HPUX"
- #elif defined(_AIX)
- "AIX"
- #elif defined(__sun) && defined(__SVR4)
- "Solaris"
- #elif defined(__BSD__) || defined(BSD)
- "UnixBSD"
- #elif defined(__unix__) || defined(UNIX) || defined(__unix) \
- || defined(__UNIX) || defined(__UNIX__)
- "UNIX"
- #elif defined(_POSIX_VERSION)
- "POSIX" MDBX_STRINGIFY(_POSIX_VERSION)
- #else
- "UnknownOS"
- #endif /* Target OS */
-
- "-"
-
- #if defined(__amd64__)
- "AMD64"
- #elif defined(__ia32__)
- "IA32"
- #elif defined(__e2k__) || defined(__elbrus__)
- "Elbrus"
- #elif defined(__alpha__) || defined(__alpha) || defined(_M_ALPHA)
- "Alpha"
- #elif defined(__aarch64__) || defined(_M_ARM64)
- "ARM64"
- #elif defined(__arm__) || defined(__thumb__) || defined(__TARGET_ARCH_ARM) \
- || defined(__TARGET_ARCH_THUMB) || defined(_ARM) || defined(_M_ARM) \
- || defined(_M_ARMT) || defined(__arm)
- "ARM"
- #elif defined(__mips64) || defined(__mips64__) || (defined(__mips) && (__mips >= 64))
- "MIPS64"
- #elif defined(__mips__) || defined(__mips) || defined(_R4000) || defined(__MIPS__)
- "MIPS"
- #elif defined(__hppa64__) || defined(__HPPA64__) || defined(__hppa64)
- "PARISC64"
- #elif defined(__hppa__) || defined(__HPPA__) || defined(__hppa)
- "PARISC"
- #elif defined(__ia64__) || defined(__ia64) || defined(_IA64) \
- || defined(__IA64__) || defined(_M_IA64) || defined(__itanium__)
- "Itanium"
- #elif defined(__powerpc64__) || defined(__ppc64__) || defined(__ppc64) \
- || defined(__powerpc64) || defined(_ARCH_PPC64)
- "PowerPC64"
- #elif defined(__powerpc__) || defined(__ppc__) || defined(__powerpc) \
- || defined(__ppc) || defined(_ARCH_PPC) || defined(__PPC__) || defined(__POWERPC__)
- "PowerPC"
- #elif defined(__sparc64__) || defined(__sparc64)
- "SPARC64"
- #elif defined(__sparc__) || defined(__sparc)
- "SPARC"
- #elif defined(__s390__) || defined(__s390) || defined(__zarch__) || defined(__zarch)
- "S390"
- #else
- "UnknownARCH"
- #endif
-#endif /* MDBX_BUILD_TARGET */
-
-#ifdef MDBX_BUILD_TYPE
-# if defined(_MSC_VER)
-# pragma message("Configuration-depended MDBX_BUILD_TYPE: " MDBX_BUILD_TYPE)
-# endif
- "-" MDBX_BUILD_TYPE
-#endif /* MDBX_BUILD_TYPE */
- ,
- "MDBX_DEBUG=" MDBX_STRINGIFY(MDBX_DEBUG)
- " MDBX_WORDBITS=" MDBX_STRINGIFY(MDBX_WORDBITS)
- " BYTE_ORDER="
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- "LITTLE_ENDIAN"
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- "BIG_ENDIAN"
-#else
- #error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- " MDBX_ENV_CHECKPID=" MDBX_ENV_CHECKPID_CONFIG
- " MDBX_TXN_CHECKOWNER=" MDBX_TXN_CHECKOWNER_CONFIG
- " MDBX_64BIT_ATOMIC=" MDBX_64BIT_ATOMIC_CONFIG
- " MDBX_64BIT_CAS=" MDBX_64BIT_CAS_CONFIG
- " MDBX_TRUST_RTC=" MDBX_TRUST_RTC_CONFIG
- " MDBX_ENABLE_REFUND=" MDBX_STRINGIFY(MDBX_ENABLE_REFUND)
- " MDBX_ENABLE_MADVISE=" MDBX_STRINGIFY(MDBX_ENABLE_MADVISE)
-#if MDBX_DISABLE_PAGECHECKS
- " MDBX_DISABLE_PAGECHECKS=YES"
-#endif /* MDBX_DISABLE_PAGECHECKS */
-#ifdef __SANITIZE_ADDRESS__
- " SANITIZE_ADDRESS=YES"
-#endif /* __SANITIZE_ADDRESS__ */
-#ifdef MDBX_USE_VALGRIND
- " MDBX_USE_VALGRIND=YES"
-#endif /* MDBX_USE_VALGRIND */
-#if MDBX_FORCE_ASSERTIONS
- " MDBX_FORCE_ASSERTIONS=YES"
-#endif /* MDBX_FORCE_ASSERTIONS */
-#ifdef _GNU_SOURCE
- " _GNU_SOURCE=YES"
-#else
- " _GNU_SOURCE=NO"
-#endif /* _GNU_SOURCE */
-#ifdef __APPLE__
- " MDBX_OSX_SPEED_INSTEADOF_DURABILITY=" MDBX_STRINGIFY(MDBX_OSX_SPEED_INSTEADOF_DURABILITY)
-#endif /* MacOS */
-#if defined(_WIN32) || defined(_WIN64)
- " MDBX_WITHOUT_MSVC_CRT=" MDBX_STRINGIFY(MDBX_WITHOUT_MSVC_CRT)
- " MDBX_BUILD_SHARED_LIBRARY=" MDBX_STRINGIFY(MDBX_BUILD_SHARED_LIBRARY)
-#if !MDBX_BUILD_SHARED_LIBRARY
- " MDBX_MANUAL_MODULE_HANDLER=" MDBX_STRINGIFY(MDBX_MANUAL_MODULE_HANDLER)
-#endif
- " WINVER=" MDBX_STRINGIFY(WINVER)
-#else /* Windows */
- " MDBX_LOCKING=" MDBX_LOCKING_CONFIG
- " MDBX_USE_OFDLOCKS=" MDBX_USE_OFDLOCKS_CONFIG
-#endif /* !Windows */
- " MDBX_CACHELINE_SIZE=" MDBX_STRINGIFY(MDBX_CACHELINE_SIZE)
- " MDBX_CPU_WRITEBACK_INCOHERENT=" MDBX_STRINGIFY(MDBX_CPU_WRITEBACK_INCOHERENT)
- " MDBX_MMAP_INCOHERENT_CPU_CACHE=" MDBX_STRINGIFY(MDBX_MMAP_INCOHERENT_CPU_CACHE)
- " MDBX_MMAP_INCOHERENT_FILE_WRITE=" MDBX_STRINGIFY(MDBX_MMAP_INCOHERENT_FILE_WRITE)
- " MDBX_UNALIGNED_OK=" MDBX_STRINGIFY(MDBX_UNALIGNED_OK)
- " MDBX_PNL_ASCENDING=" MDBX_STRINGIFY(MDBX_PNL_ASCENDING)
- ,
-#ifdef MDBX_BUILD_COMPILER
- MDBX_BUILD_COMPILER
-#else
- #ifdef __INTEL_COMPILER
- "Intel C/C++ " MDBX_STRINGIFY(__INTEL_COMPILER)
- #elif defined(__apple_build_version__)
- "Apple clang " MDBX_STRINGIFY(__apple_build_version__)
- #elif defined(__ibmxl__)
- "IBM clang C " MDBX_STRINGIFY(__ibmxl_version__) "." MDBX_STRINGIFY(__ibmxl_release__)
- "." MDBX_STRINGIFY(__ibmxl_modification__) "." MDBX_STRINGIFY(__ibmxl_ptf_fix_level__)
- #elif defined(__clang__)
- "clang " MDBX_STRINGIFY(__clang_version__)
- #elif defined(__MINGW64__)
- "MINGW-64 " MDBX_STRINGIFY(__MINGW64_MAJOR_VERSION) "." MDBX_STRINGIFY(__MINGW64_MINOR_VERSION)
- #elif defined(__MINGW32__)
- "MINGW-32 " MDBX_STRINGIFY(__MINGW32_MAJOR_VERSION) "." MDBX_STRINGIFY(__MINGW32_MINOR_VERSION)
- #elif defined(__IBMC__)
- "IBM C " MDBX_STRINGIFY(__IBMC__)
- #elif defined(__GNUC__)
- "GNU C/C++ "
- #ifdef __VERSION__
- __VERSION__
- #else
- MDBX_STRINGIFY(__GNUC__) "." MDBX_STRINGIFY(__GNUC_MINOR__) "." MDBX_STRINGIFY(__GNUC_PATCHLEVEL__)
- #endif
- #elif defined(_MSC_VER)
- "MSVC " MDBX_STRINGIFY(_MSC_FULL_VER) "-" MDBX_STRINGIFY(_MSC_BUILD)
- #else
- "Unknown compiler"
- #endif
-#endif /* MDBX_BUILD_COMPILER */
- ,
-#ifdef MDBX_BUILD_FLAGS_CONFIG
- MDBX_BUILD_FLAGS_CONFIG
-#endif /* MDBX_BUILD_FLAGS_CONFIG */
-#ifdef MDBX_BUILD_FLAGS
- MDBX_BUILD_FLAGS
-#endif /* MDBX_BUILD_FLAGS */
-#if !(defined(MDBX_BUILD_FLAGS_CONFIG) || defined(MDBX_BUILD_FLAGS))
- "undefined (please use correct build script)"
-#ifdef _MSC_VER
-#pragma message("warning: Build flags undefined. Please use correct build script")
-#else
-#warning "Build flags undefined. Please use correct build script"
-#endif // _MSC_VER
-#endif
-};
-
-#ifdef __SANITIZE_ADDRESS__
-LIBMDBX_API __attribute__((__weak__)) const char *__asan_default_options() {
- return "symbolize=1:allow_addr2line=1:"
-#if MDBX_DEBUG
- "debug=1:"
- "verbosity=2:"
-#endif /* MDBX_DEBUG */
- "log_threads=1:"
- "report_globals=1:"
- "replace_str=1:replace_intrin=1:"
- "malloc_context_size=9:"
- "detect_leaks=1:"
- "check_printf=1:"
- "detect_deadlocks=1:"
-#ifndef LTO_ENABLED
- "check_initialization_order=1:"
-#endif
- "detect_stack_use_after_return=1:"
- "intercept_tls_get_addr=1:"
- "decorate_proc_maps=1:"
- "abort_on_error=1";
-}
-#endif /* __SANITIZE_ADDRESS__ */
-
-/* *INDENT-ON* */
-/* clang-format on */
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-#if defined(_WIN32) || defined(_WIN64)
-
-#include <winioctl.h>
-
-static int waitstatus2errcode(DWORD result) {
- switch (result) {
- case WAIT_OBJECT_0:
- return MDBX_SUCCESS;
- case WAIT_FAILED:
- return (int)GetLastError();
- case WAIT_ABANDONED:
- return ERROR_ABANDONED_WAIT_0;
- case WAIT_IO_COMPLETION:
- return ERROR_USER_APC;
- case WAIT_TIMEOUT:
- return ERROR_TIMEOUT;
- default:
- return ERROR_UNHANDLED_ERROR;
- }
-}
-
-/* Map a result from an NTAPI call to WIN32 error code. */
-static int ntstatus2errcode(NTSTATUS status) {
- DWORD dummy;
- OVERLAPPED ov;
- memset(&ov, 0, sizeof(ov));
- ov.Internal = status;
- return GetOverlappedResult(NULL, &ov, &dummy, FALSE) ? MDBX_SUCCESS
- : (int)GetLastError();
-}
-
-/* We use native NT APIs to setup the memory map, so that we can
- * let the DB file grow incrementally instead of always preallocating
- * the full size. These APIs are defined in <wdm.h> and <ntifs.h>
- * but those headers are meant for driver-level development and
- * conflict with the regular user-level headers, so we explicitly
- * declare them here. Using these APIs also means we must link to
- * ntdll.dll, which is not linked by default in user code. */
-
-extern NTSTATUS NTAPI NtCreateSection(
- OUT PHANDLE SectionHandle, IN ACCESS_MASK DesiredAccess,
- IN OPTIONAL POBJECT_ATTRIBUTES ObjectAttributes,
- IN OPTIONAL PLARGE_INTEGER MaximumSize, IN ULONG SectionPageProtection,
- IN ULONG AllocationAttributes, IN OPTIONAL HANDLE FileHandle);
-
-typedef struct _SECTION_BASIC_INFORMATION {
- ULONG Unknown;
- ULONG SectionAttributes;
- LARGE_INTEGER SectionSize;
-} SECTION_BASIC_INFORMATION, *PSECTION_BASIC_INFORMATION;
-
-extern NTSTATUS NTAPI NtMapViewOfSection(
- IN HANDLE SectionHandle, IN HANDLE ProcessHandle, IN OUT PVOID *BaseAddress,
- IN ULONG_PTR ZeroBits, IN SIZE_T CommitSize,
- IN OUT OPTIONAL PLARGE_INTEGER SectionOffset, IN OUT PSIZE_T ViewSize,
- IN SECTION_INHERIT InheritDisposition, IN ULONG AllocationType,
- IN ULONG Win32Protect);
-
-extern NTSTATUS NTAPI NtUnmapViewOfSection(IN HANDLE ProcessHandle,
- IN OPTIONAL PVOID BaseAddress);
-
-extern NTSTATUS NTAPI NtClose(HANDLE Handle);
-
-extern NTSTATUS NTAPI NtAllocateVirtualMemory(
- IN HANDLE ProcessHandle, IN OUT PVOID *BaseAddress, IN ULONG_PTR ZeroBits,
- IN OUT PSIZE_T RegionSize, IN ULONG AllocationType, IN ULONG Protect);
-
-extern NTSTATUS NTAPI NtFreeVirtualMemory(IN HANDLE ProcessHandle,
- IN PVOID *BaseAddress,
- IN OUT PSIZE_T RegionSize,
- IN ULONG FreeType);
-
-#ifndef WOF_CURRENT_VERSION
-typedef struct _WOF_EXTERNAL_INFO {
- DWORD Version;
- DWORD Provider;
-} WOF_EXTERNAL_INFO, *PWOF_EXTERNAL_INFO;
-#endif /* WOF_CURRENT_VERSION */
-
-#ifndef WIM_PROVIDER_CURRENT_VERSION
-#define WIM_PROVIDER_HASH_SIZE 20
-
-typedef struct _WIM_PROVIDER_EXTERNAL_INFO {
- DWORD Version;
- DWORD Flags;
- LARGE_INTEGER DataSourceId;
- BYTE ResourceHash[WIM_PROVIDER_HASH_SIZE];
-} WIM_PROVIDER_EXTERNAL_INFO, *PWIM_PROVIDER_EXTERNAL_INFO;
-#endif /* WIM_PROVIDER_CURRENT_VERSION */
-
-#ifndef FILE_PROVIDER_CURRENT_VERSION
-typedef struct _FILE_PROVIDER_EXTERNAL_INFO_V1 {
- ULONG Version;
- ULONG Algorithm;
- ULONG Flags;
-} FILE_PROVIDER_EXTERNAL_INFO_V1, *PFILE_PROVIDER_EXTERNAL_INFO_V1;
-#endif /* FILE_PROVIDER_CURRENT_VERSION */
-
-#ifndef STATUS_OBJECT_NOT_EXTERNALLY_BACKED
-#define STATUS_OBJECT_NOT_EXTERNALLY_BACKED ((NTSTATUS)0xC000046DL)
-#endif
-#ifndef STATUS_INVALID_DEVICE_REQUEST
-#define STATUS_INVALID_DEVICE_REQUEST ((NTSTATUS)0xC0000010L)
-#endif
-#ifndef STATUS_NOT_SUPPORTED
-#define STATUS_NOT_SUPPORTED ((NTSTATUS)0xC00000BBL)
-#endif
-
-#ifndef FILE_DEVICE_FILE_SYSTEM
-#define FILE_DEVICE_FILE_SYSTEM 0x00000009
-#endif
-
-#ifndef FSCTL_GET_EXTERNAL_BACKING
-#define FSCTL_GET_EXTERNAL_BACKING \
- CTL_CODE(FILE_DEVICE_FILE_SYSTEM, 196, METHOD_BUFFERED, FILE_ANY_ACCESS)
-#endif
-
-#ifndef ERROR_NOT_CAPABLE
-#define ERROR_NOT_CAPABLE 775L
-#endif
-
-#endif /* _WIN32 || _WIN64 */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(__UCLIBC__)
-__extern_C void __assert(const char *, const char *, unsigned int, const char *)
-#ifdef __THROW
- __THROW
-#else
- __nothrow
-#endif /* __THROW */
- MDBX_NORETURN;
-#define __assert_fail(assertion, file, line, function) \
- __assert(assertion, file, line, function)
-
-#elif _POSIX_C_SOURCE > 200212 && \
- /* workaround for avoid musl libc wrong prototype */ ( \
- defined(__GLIBC__) || defined(__GNU_LIBRARY__))
-/* Prototype should match libc runtime. ISO POSIX (2003) & LSB 1.x-3.x */
-__extern_C void __assert_fail(const char *assertion, const char *file,
- unsigned line, const char *function)
-#ifdef __THROW
- __THROW
-#else
- __nothrow
-#endif /* __THROW */
- MDBX_NORETURN;
-
-#elif defined(__APPLE__) || defined(__MACH__)
-__extern_C void __assert_rtn(const char *function, const char *file, int line,
- const char *assertion) /* __nothrow */
-#ifdef __dead2
- __dead2
-#else
- MDBX_NORETURN
-#endif /* __dead2 */
-#ifdef __disable_tail_calls
- __disable_tail_calls
-#endif /* __disable_tail_calls */
- ;
-
-#define __assert_fail(assertion, file, line, function) \
- __assert_rtn(function, file, line, assertion)
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-__extern_C void __assert_c99(const char *assection, const char *file, int line,
- const char *function) MDBX_NORETURN;
-#define __assert_fail(assertion, file, line, function) \
- __assert_c99(assertion, file, line, function)
-#elif defined(__OpenBSD__)
-__extern_C __dead void __assert2(const char *file, int line,
- const char *function,
- const char *assertion) /* __nothrow */;
-#define __assert_fail(assertion, file, line, function) \
- __assert2(file, line, function, assertion)
-#elif defined(__NetBSD__)
-__extern_C __dead void __assert13(const char *file, int line,
- const char *function,
- const char *assertion) /* __nothrow */;
-#define __assert_fail(assertion, file, line, function) \
- __assert13(file, line, function, assertion)
-#elif defined(__FreeBSD__) || defined(__BSD__) || defined(__bsdi__) || \
- defined(__DragonFly__)
-__extern_C void __assert(const char *function, const char *file, int line,
- const char *assertion) /* __nothrow */
-#ifdef __dead2
- __dead2
-#else
- MDBX_NORETURN
-#endif /* __dead2 */
-#ifdef __disable_tail_calls
- __disable_tail_calls
-#endif /* __disable_tail_calls */
- ;
-#define __assert_fail(assertion, file, line, function) \
- __assert(function, file, line, assertion)
-
-#endif /* __assert_fail */
-
-#if !defined(__ANDROID_API__) || MDBX_DEBUG
-
-__cold void mdbx_assert_fail(const MDBX_env *env, const char *msg,
- const char *func, int line) {
-#if MDBX_DEBUG
- if (env && env->me_assert_func) {
- env->me_assert_func(env, msg, func, line);
- return;
- }
-#else
- (void)env;
-#endif /* MDBX_DEBUG */
-
- if (mdbx_debug_logger)
- mdbx_debug_log(MDBX_LOG_FATAL, func, line, "assert: %s\n", msg);
- else {
-#if defined(_WIN32) || defined(_WIN64)
- char *message = nullptr;
- const int num = mdbx_asprintf(&message, "\r\nMDBX-ASSERTION: %s, %s:%u",
- msg, func ? func : "unknown", line);
- if (num < 1 || !message)
- message = "<troubles with assertion-message preparation>";
- OutputDebugStringA(message);
- if (IsDebuggerPresent())
- DebugBreak();
-#elif defined(__ANDROID_API__)
- __android_log_assert(msg, "mdbx", "%s:%u", func, line);
-#else
- __assert_fail(msg, "mdbx", line, func);
-#endif
- }
-
-#if defined(_WIN32) || defined(_WIN64)
- FatalExit(ERROR_UNHANDLED_ERROR);
-#else
- abort();
-#endif
-}
-
-#endif /* __ANDROID_API__ || MDBX_DEBUG */
-
-__cold void mdbx_panic(const char *fmt, ...) {
- va_list ap;
- va_start(ap, fmt);
-
- char *message = nullptr;
- const int num = mdbx_vasprintf(&message, fmt, ap);
- va_end(ap);
- const char *const const_message =
- (num < 1 || !message) ? "<troubles with panic-message preparation>"
- : message;
-
-#if defined(_WIN32) || defined(_WIN64)
- OutputDebugStringA("\r\nMDBX-PANIC: ");
- OutputDebugStringA(const_message);
- if (IsDebuggerPresent())
- DebugBreak();
- FatalExit(ERROR_UNHANDLED_ERROR);
-#else
-#if defined(__ANDROID_API__)
- __android_log_assert("panic", "mdbx", "%s", const_message);
-#else
- __assert_fail(const_message, "mdbx", 0, "panic");
-#endif /* __ANDROID_API__ */
- abort();
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef mdbx_vasprintf
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt,
- va_list ap) {
- va_list ones;
- va_copy(ones, ap);
- int needed = vsnprintf(nullptr, 0, fmt, ap);
-
- if (unlikely(needed < 0 || needed >= INT_MAX)) {
- *strp = nullptr;
- va_end(ones);
- return needed;
- }
-
- *strp = mdbx_malloc(needed + 1);
- if (unlikely(*strp == nullptr)) {
- va_end(ones);
-#if defined(_WIN32) || defined(_WIN64)
- SetLastError(MDBX_ENOMEM);
-#else
- errno = MDBX_ENOMEM;
-#endif
- return -1;
- }
-
- int actual = vsnprintf(*strp, needed + 1, fmt, ones);
- va_end(ones);
-
- assert(actual == needed);
- if (unlikely(actual < 0)) {
- mdbx_free(*strp);
- *strp = nullptr;
- }
- return actual;
-}
-#endif /* mdbx_vasprintf */
-
-#ifndef mdbx_asprintf
-MDBX_INTERNAL_FUNC int mdbx_asprintf(char **strp, const char *fmt, ...) {
- va_list ap;
- va_start(ap, fmt);
- int rc = mdbx_vasprintf(strp, fmt, ap);
- va_end(ap);
- return rc;
-}
-#endif /* mdbx_asprintf */
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result) {
- assert(is_powerof2(alignment) && alignment >= sizeof(void *));
-#if defined(_WIN32) || defined(_WIN64)
- (void)alignment;
- *result = VirtualAlloc(NULL, bytes, MEM_COMMIT | MEM_RESERVE, PAGE_READWRITE);
- return *result ? MDBX_SUCCESS : MDBX_ENOMEM /* ERROR_OUTOFMEMORY */;
-#elif defined(_ISOC11_SOURCE)
- *result = aligned_alloc(alignment, ceil_powerof2(bytes, alignment));
- return *result ? MDBX_SUCCESS : errno;
-#elif _POSIX_VERSION >= 200112L && \
- (!defined(__ANDROID_API__) || __ANDROID_API__ >= 17)
- *result = nullptr;
- return posix_memalign(result, alignment, bytes);
-#elif __GLIBC_PREREQ(2, 16) || __STDC_VERSION__ >= 201112L
- *result = memalign(alignment, bytes);
- return *result ? MDBX_SUCCESS : errno;
-#else
-#error FIXME
-#endif
-}
-#endif /* mdbx_memalign_alloc */
-
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr) {
-#if defined(_WIN32) || defined(_WIN64)
- VirtualFree(ptr, 0, MEM_RELEASE);
-#else
- mdbx_free(ptr);
-#endif
-}
-#endif /* mdbx_memalign_free */
-
-#ifndef mdbx_strdup
-char *mdbx_strdup(const char *str) {
- if (!str)
- return NULL;
- size_t bytes = strlen(str) + 1;
- char *dup = mdbx_malloc(bytes);
- if (dup)
- memcpy(dup, str, bytes);
- return dup;
-}
-#endif /* mdbx_strdup */
-
-/*----------------------------------------------------------------------------*/
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair) {
- int rc;
- memset(condpair, 0, sizeof(mdbx_condpair_t));
-#if defined(_WIN32) || defined(_WIN64)
- if ((condpair->mutex = CreateMutexW(NULL, FALSE, NULL)) == NULL) {
- rc = (int)GetLastError();
- goto bailout_mutex;
- }
- if ((condpair->event[0] = CreateEventW(NULL, FALSE, FALSE, NULL)) == NULL) {
- rc = (int)GetLastError();
- goto bailout_event;
- }
- if ((condpair->event[1] = CreateEventW(NULL, FALSE, FALSE, NULL)) != NULL)
- return MDBX_SUCCESS;
-
- rc = (int)GetLastError();
- (void)CloseHandle(condpair->event[0]);
-bailout_event:
- (void)CloseHandle(condpair->mutex);
-#else
- rc = pthread_mutex_init(&condpair->mutex, NULL);
- if (unlikely(rc != 0))
- goto bailout_mutex;
- rc = pthread_cond_init(&condpair->cond[0], NULL);
- if (unlikely(rc != 0))
- goto bailout_cond;
- rc = pthread_cond_init(&condpair->cond[1], NULL);
- if (likely(rc == 0))
- return MDBX_SUCCESS;
-
- (void)pthread_cond_destroy(&condpair->cond[0]);
-bailout_cond:
- (void)pthread_mutex_destroy(&condpair->mutex);
-#endif
-bailout_mutex:
- memset(condpair, 0, sizeof(mdbx_condpair_t));
- return rc;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair) {
-#if defined(_WIN32) || defined(_WIN64)
- int rc = CloseHandle(condpair->mutex) ? MDBX_SUCCESS : (int)GetLastError();
- rc = CloseHandle(condpair->event[0]) ? rc : (int)GetLastError();
- rc = CloseHandle(condpair->event[1]) ? rc : (int)GetLastError();
-#else
- int err, rc = pthread_mutex_destroy(&condpair->mutex);
- rc = (err = pthread_cond_destroy(&condpair->cond[0])) ? err : rc;
- rc = (err = pthread_cond_destroy(&condpair->cond[1])) ? err : rc;
-#endif
- memset(condpair, 0, sizeof(mdbx_condpair_t));
- return rc;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD code = WaitForSingleObject(condpair->mutex, INFINITE);
- return waitstatus2errcode(code);
-#else
- return mdbx_pthread_mutex_lock(&condpair->mutex);
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair) {
-#if defined(_WIN32) || defined(_WIN64)
- return ReleaseMutex(condpair->mutex) ? MDBX_SUCCESS : (int)GetLastError();
-#else
- return pthread_mutex_unlock(&condpair->mutex);
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part) {
-#if defined(_WIN32) || defined(_WIN64)
- return SetEvent(condpair->event[part]) ? MDBX_SUCCESS : (int)GetLastError();
-#else
- return pthread_cond_signal(&condpair->cond[part]);
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair,
- bool part) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD code = SignalObjectAndWait(condpair->mutex, condpair->event[part],
- INFINITE, FALSE);
- if (code == WAIT_OBJECT_0) {
- code = WaitForSingleObject(condpair->mutex, INFINITE);
- if (code == WAIT_OBJECT_0)
- return MDBX_SUCCESS;
- }
- return waitstatus2errcode(code);
-#else
- return pthread_cond_wait(&condpair->cond[part], &condpair->mutex);
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex) {
-#if defined(_WIN32) || defined(_WIN64)
- InitializeCriticalSection(fastmutex);
- return MDBX_SUCCESS;
-#else
- return pthread_mutex_init(fastmutex, NULL);
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex) {
-#if defined(_WIN32) || defined(_WIN64)
- DeleteCriticalSection(fastmutex);
- return MDBX_SUCCESS;
-#else
- return pthread_mutex_destroy(fastmutex);
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex) {
-#if defined(_WIN32) || defined(_WIN64)
- __try {
- EnterCriticalSection(fastmutex);
- } __except (
- (GetExceptionCode() ==
- 0xC0000194 /* STATUS_POSSIBLE_DEADLOCK / EXCEPTION_POSSIBLE_DEADLOCK */)
- ? EXCEPTION_EXECUTE_HANDLER
- : EXCEPTION_CONTINUE_SEARCH) {
- return ERROR_POSSIBLE_DEADLOCK;
- }
- return MDBX_SUCCESS;
-#else
- return mdbx_pthread_mutex_lock(fastmutex);
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex) {
-#if defined(_WIN32) || defined(_WIN64)
- LeaveCriticalSection(fastmutex);
- return MDBX_SUCCESS;
-#else
- return pthread_mutex_unlock(fastmutex);
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname) {
-#if defined(_WIN32) || defined(_WIN64)
- const size_t wlen = mbstowcs(nullptr, pathname, INT_MAX);
- if (wlen < 1 || wlen > /* MAX_PATH */ INT16_MAX)
- return ERROR_INVALID_NAME;
- wchar_t *const pathnameW = _alloca((wlen + 1) * sizeof(wchar_t));
- if (wlen != mbstowcs(pathnameW, pathname, wlen + 1))
- return ERROR_INVALID_NAME;
- return DeleteFileW(pathnameW) ? MDBX_SUCCESS : (int)GetLastError();
-#else
- return unlink(pathname) ? errno : MDBX_SUCCESS;
-#endif
-}
-
-#if !(defined(_WIN32) || defined(_WIN64))
-static bool is_valid_fd(int fd) { return !(isatty(fd) < 0 && errno == EBADF); }
-#endif /*! Windows */
-
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname) {
-#if defined(_WIN32) || defined(_WIN64)
- const size_t wlen = mbstowcs(nullptr, pathname, INT_MAX);
- if (wlen < 1 || wlen > /* MAX_PATH */ INT16_MAX)
- return ERROR_INVALID_NAME;
- wchar_t *const pathnameW = _alloca((wlen + 1) * sizeof(wchar_t));
- if (wlen != mbstowcs(pathnameW, pathname, wlen + 1))
- return ERROR_INVALID_NAME;
- return RemoveDirectoryW(pathnameW) ? MDBX_SUCCESS : (int)GetLastError();
-#else
- return rmdir(pathname) ? errno : MDBX_SUCCESS;
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits) {
- *fd = INVALID_HANDLE_VALUE;
-
-#if defined(_WIN32) || defined(_WIN64)
- const size_t wlen = mbstowcs(nullptr, pathname, INT_MAX);
- if (wlen < 1 || wlen > /* MAX_PATH */ INT16_MAX)
- return ERROR_INVALID_NAME;
- wchar_t *const pathnameW = _alloca((wlen + 1) * sizeof(wchar_t));
- if (wlen != mbstowcs(pathnameW, pathname, wlen + 1))
- return ERROR_INVALID_NAME;
-
- DWORD CreationDisposition = unix_mode_bits ? OPEN_ALWAYS : OPEN_EXISTING;
- DWORD FlagsAndAttributes =
- FILE_FLAG_POSIX_SEMANTICS | FILE_ATTRIBUTE_NOT_CONTENT_INDEXED;
- DWORD DesiredAccess = FILE_READ_ATTRIBUTES;
- DWORD ShareMode = (env->me_flags & MDBX_EXCLUSIVE)
- ? 0
- : (FILE_SHARE_READ | FILE_SHARE_WRITE);
-
- switch (purpose) {
- default:
- return ERROR_INVALID_PARAMETER;
- case MDBX_OPEN_LCK:
- CreationDisposition = OPEN_ALWAYS;
- DesiredAccess |= GENERIC_READ | GENERIC_WRITE;
- FlagsAndAttributes |= FILE_ATTRIBUTE_HIDDEN | FILE_ATTRIBUTE_TEMPORARY;
- break;
- case MDBX_OPEN_DXB_READ:
- CreationDisposition = OPEN_EXISTING;
- DesiredAccess |= GENERIC_READ;
- ShareMode |= FILE_SHARE_READ;
- break;
- case MDBX_OPEN_DXB_LAZY:
- DesiredAccess |= GENERIC_READ | GENERIC_WRITE;
- break;
- case MDBX_OPEN_DXB_DSYNC:
- CreationDisposition = OPEN_EXISTING;
- DesiredAccess |= GENERIC_WRITE;
- FlagsAndAttributes |= FILE_FLAG_WRITE_THROUGH;
- break;
- case MDBX_OPEN_COPY:
- CreationDisposition = CREATE_NEW;
- ShareMode = 0;
- DesiredAccess |= GENERIC_WRITE;
- FlagsAndAttributes |=
- (env->me_psize < env->me_os_psize) ? 0 : FILE_FLAG_NO_BUFFERING;
- break;
- case MDBX_OPEN_DELETE:
- CreationDisposition = OPEN_EXISTING;
- ShareMode |= FILE_SHARE_DELETE;
- DesiredAccess =
- FILE_READ_ATTRIBUTES | FILE_WRITE_ATTRIBUTES | DELETE | SYNCHRONIZE;
- break;
- }
-
- *fd = CreateFileW(pathnameW, DesiredAccess, ShareMode, NULL,
- CreationDisposition, FlagsAndAttributes, NULL);
- if (*fd == INVALID_HANDLE_VALUE)
- return (int)GetLastError();
-
- BY_HANDLE_FILE_INFORMATION info;
- if (!GetFileInformationByHandle(*fd, &info)) {
- int err = (int)GetLastError();
- CloseHandle(*fd);
- *fd = INVALID_HANDLE_VALUE;
- return err;
- }
- const DWORD AttributesDiff =
- (info.dwFileAttributes ^ FlagsAndAttributes) &
- (FILE_ATTRIBUTE_HIDDEN | FILE_ATTRIBUTE_NOT_CONTENT_INDEXED |
- FILE_ATTRIBUTE_TEMPORARY | FILE_ATTRIBUTE_COMPRESSED);
- if (AttributesDiff)
- (void)SetFileAttributesW(pathnameW, info.dwFileAttributes ^ AttributesDiff);
-
-#else
- int flags = unix_mode_bits ? O_CREAT : 0;
- switch (purpose) {
- default:
- return EINVAL;
- case MDBX_OPEN_LCK:
- flags |= O_RDWR;
- break;
- case MDBX_OPEN_DXB_READ:
- flags = O_RDONLY;
- break;
- case MDBX_OPEN_DXB_LAZY:
- flags |= O_RDWR;
- break;
- case MDBX_OPEN_COPY:
- flags = O_CREAT | O_WRONLY | O_EXCL;
- break;
- case MDBX_OPEN_DXB_DSYNC:
- flags |= O_WRONLY;
-#if defined(O_DSYNC)
- flags |= O_DSYNC;
-#elif defined(O_SYNC)
- flags |= O_SYNC;
-#elif defined(O_FSYNC)
- flags |= O_FSYNC;
-#endif
- break;
- case MDBX_OPEN_DELETE:
- flags = O_RDWR;
- break;
- }
-
- const bool direct_nocache_for_copy =
- env->me_psize >= env->me_os_psize && purpose == MDBX_OPEN_COPY;
- if (direct_nocache_for_copy) {
-#if defined(O_DIRECT)
- flags |= O_DIRECT;
-#endif /* O_DIRECT */
-#if defined(O_NOCACHE)
- flags |= O_NOCACHE;
-#endif /* O_NOCACHE */
- }
-
-#ifdef O_CLOEXEC
- flags |= O_CLOEXEC;
-#endif /* O_CLOEXEC */
-
- /* Safeguard for todo4recovery://erased_by_github/libmdbx/issues/144 */
-#if STDIN_FILENO == 0 && STDOUT_FILENO == 1 && STDERR_FILENO == 2
- int stub_fd0 = -1, stub_fd1 = -1, stub_fd2 = -1;
- static const char dev_null[] = "/dev/null";
- if (!is_valid_fd(STDIN_FILENO)) {
- mdbx_warning("STD%s_FILENO/%d is invalid, open %s for temporary stub", "IN",
- STDIN_FILENO, dev_null);
- stub_fd0 = open(dev_null, O_RDONLY | O_NOCTTY);
- }
- if (!is_valid_fd(STDOUT_FILENO)) {
- mdbx_warning("STD%s_FILENO/%d is invalid, open %s for temporary stub",
- "OUT", STDOUT_FILENO, dev_null);
- stub_fd1 = open(dev_null, O_WRONLY | O_NOCTTY);
- }
- if (!is_valid_fd(STDERR_FILENO)) {
- mdbx_warning("STD%s_FILENO/%d is invalid, open %s for temporary stub",
- "ERR", STDERR_FILENO, dev_null);
- stub_fd2 = open(dev_null, O_WRONLY | O_NOCTTY);
- }
-#else
-#error "Unexpected or unsupported UNIX or POSIX system"
-#endif /* STDIN_FILENO == 0 && STDERR_FILENO == 2 */
-
- *fd = open(pathname, flags, unix_mode_bits);
-#if defined(O_DIRECT)
- if (*fd < 0 && (flags & O_DIRECT) &&
- (errno == EINVAL || errno == EAFNOSUPPORT)) {
- flags &= ~(O_DIRECT | O_EXCL);
- *fd = open(pathname, flags, unix_mode_bits);
- }
-#endif /* O_DIRECT */
-
- /* Safeguard for todo4recovery://erased_by_github/libmdbx/issues/144 */
-#if STDIN_FILENO == 0 && STDOUT_FILENO == 1 && STDERR_FILENO == 2
- if (*fd == STDIN_FILENO) {
- mdbx_warning("Got STD%s_FILENO/%d, avoid using it by dup(fd)", "IN",
- STDIN_FILENO);
- assert(stub_fd0 == -1);
- *fd = dup(stub_fd0 = *fd);
- }
- if (*fd == STDOUT_FILENO) {
- mdbx_warning("Got STD%s_FILENO/%d, avoid using it by dup(fd)", "OUT",
- STDOUT_FILENO);
- assert(stub_fd1 == -1);
- *fd = dup(stub_fd1 = *fd);
- }
- if (*fd == STDERR_FILENO) {
- mdbx_warning("Got STD%s_FILENO/%d, avoid using it by dup(fd)", "ERR",
- STDERR_FILENO);
- assert(stub_fd2 == -1);
- *fd = dup(stub_fd2 = *fd);
- }
- if (stub_fd0 != -1)
- close(stub_fd0);
- if (stub_fd1 != -1)
- close(stub_fd1);
- if (stub_fd2 != -1)
- close(stub_fd2);
- if (*fd >= STDIN_FILENO && *fd <= STDERR_FILENO) {
- mdbx_error(
- "Rejecting the use of a FD in the range "
- "STDIN_FILENO/%d..STDERR_FILENO/%d to prevent database corruption",
- STDIN_FILENO, STDERR_FILENO);
- close(*fd);
- return EBADF;
- }
-#else
-#error "Unexpected or unsupported UNIX or POSIX system"
-#endif /* STDIN_FILENO == 0 && STDERR_FILENO == 2 */
-
- if (*fd < 0)
- return errno;
-
-#if defined(FD_CLOEXEC) && !defined(O_CLOEXEC)
- const int fd_flags = fcntl(*fd, F_GETFD);
- if (fd_flags != -1)
- (void)fcntl(*fd, F_SETFD, fd_flags | FD_CLOEXEC);
-#endif /* FD_CLOEXEC && !O_CLOEXEC */
-
- if (direct_nocache_for_copy) {
-#if defined(F_NOCACHE) && !defined(O_NOCACHE)
- (void)fcntl(*fd, F_NOCACHE, 1);
-#endif /* F_NOCACHE */
- }
-
-#endif
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd) {
-#if defined(_WIN32) || defined(_WIN64)
- return CloseHandle(fd) ? MDBX_SUCCESS : (int)GetLastError();
-#else
- assert(fd > STDERR_FILENO);
- return (close(fd) == 0) ? MDBX_SUCCESS : errno;
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t bytes,
- uint64_t offset) {
- if (bytes > MAX_WRITE)
- return MDBX_EINVAL;
-#if defined(_WIN32) || defined(_WIN64)
- OVERLAPPED ov;
- ov.hEvent = 0;
- ov.Offset = (DWORD)offset;
- ov.OffsetHigh = HIGH_DWORD(offset);
-
- DWORD read = 0;
- if (unlikely(!ReadFile(fd, buf, (DWORD)bytes, &read, &ov))) {
- int rc = (int)GetLastError();
- return (rc == MDBX_SUCCESS) ? /* paranoia */ ERROR_READ_FAULT : rc;
- }
-#else
- STATIC_ASSERT_MSG(sizeof(off_t) >= sizeof(size_t),
- "libmdbx requires 64-bit file I/O on 64-bit systems");
- intptr_t read = pread(fd, buf, bytes, offset);
- if (read < 0) {
- int rc = errno;
- return (rc == MDBX_SUCCESS) ? /* paranoia */ MDBX_EIO : rc;
- }
-#endif
- return (bytes == (size_t)read) ? MDBX_SUCCESS : MDBX_ENODATA;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t bytes, uint64_t offset) {
- while (true) {
-#if defined(_WIN32) || defined(_WIN64)
- OVERLAPPED ov;
- ov.hEvent = 0;
- ov.Offset = (DWORD)offset;
- ov.OffsetHigh = HIGH_DWORD(offset);
-
- DWORD written;
- if (unlikely(!WriteFile(
- fd, buf, likely(bytes <= MAX_WRITE) ? (DWORD)bytes : MAX_WRITE,
- &written, &ov)))
- return (int)GetLastError();
- if (likely(bytes == written))
- return MDBX_SUCCESS;
-#else
- STATIC_ASSERT_MSG(sizeof(off_t) >= sizeof(size_t),
- "libmdbx requires 64-bit file I/O on 64-bit systems");
- const intptr_t written =
- pwrite(fd, buf, likely(bytes <= MAX_WRITE) ? bytes : MAX_WRITE, offset);
- if (likely(bytes == (size_t)written))
- return MDBX_SUCCESS;
- if (written < 0) {
- const int rc = errno;
- if (rc != EINTR)
- return rc;
- continue;
- }
-#endif
- bytes -= written;
- offset += written;
- buf = (char *)buf + written;
- }
-}
-
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t bytes) {
- while (true) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD written;
- if (unlikely(!WriteFile(
- fd, buf, likely(bytes <= MAX_WRITE) ? (DWORD)bytes : MAX_WRITE,
- &written, nullptr)))
- return (int)GetLastError();
- if (likely(bytes == written))
- return MDBX_SUCCESS;
-#else
- STATIC_ASSERT_MSG(sizeof(off_t) >= sizeof(size_t),
- "libmdbx requires 64-bit file I/O on 64-bit systems");
- const intptr_t written =
- write(fd, buf, likely(bytes <= MAX_WRITE) ? bytes : MAX_WRITE);
- if (likely(bytes == (size_t)written))
- return MDBX_SUCCESS;
- if (written < 0) {
- const int rc = errno;
- if (rc != EINTR)
- return rc;
- continue;
- }
-#endif
- bytes -= written;
- buf = (char *)buf + written;
- }
-}
-
-int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov, int iovcnt,
- uint64_t offset, size_t expected_written) {
-#if defined(_WIN32) || defined(_WIN64) || defined(__APPLE__) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ < 24)
- size_t written = 0;
- for (int i = 0; i < iovcnt; ++i) {
- int rc = mdbx_pwrite(fd, iov[i].iov_base, iov[i].iov_len, offset);
- if (unlikely(rc != MDBX_SUCCESS))
- return rc;
- written += iov[i].iov_len;
- offset += iov[i].iov_len;
- }
- return (expected_written == written) ? MDBX_SUCCESS
- : MDBX_EIO /* ERROR_WRITE_FAULT */;
-#else
- int rc;
- intptr_t written;
- do {
- STATIC_ASSERT_MSG(sizeof(off_t) >= sizeof(size_t),
- "libmdbx requires 64-bit file I/O on 64-bit systems");
- written = pwritev(fd, iov, iovcnt, offset);
- if (likely(expected_written == (size_t)written))
- return MDBX_SUCCESS;
- rc = errno;
- } while (rc == EINTR);
- return (written < 0) ? rc : MDBX_EIO /* Use which error code? */;
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- enum mdbx_syncmode_bits mode_bits) {
-#if defined(_WIN32) || defined(_WIN64)
- if ((mode_bits & (MDBX_SYNC_DATA | MDBX_SYNC_IODQ)) && !FlushFileBuffers(fd))
- return (int)GetLastError();
- return MDBX_SUCCESS;
-#else
-
-#if defined(__APPLE__) && \
- MDBX_OSX_SPEED_INSTEADOF_DURABILITY == MDBX_OSX_WANNA_DURABILITY
- if (mode_bits & MDBX_SYNC_IODQ)
- return likely(fcntl(fd, F_FULLFSYNC) != -1) ? MDBX_SUCCESS : errno;
-#endif /* MacOS */
-
- /* LY: This approach is always safe and without appreciable performance
- * degradation, even on a kernel with fdatasync's bug.
- *
- * For more info about of a corresponding fdatasync() bug
- * see http://www.spinics.net/lists/linux-ext4/msg33714.html */
- while (1) {
- switch (mode_bits & (MDBX_SYNC_DATA | MDBX_SYNC_SIZE)) {
- case MDBX_SYNC_NONE:
- return MDBX_SUCCESS /* nothing to do */;
-#if defined(_POSIX_SYNCHRONIZED_IO) && _POSIX_SYNCHRONIZED_IO > 0
- case MDBX_SYNC_DATA:
- if (fdatasync(fd) == 0)
- return MDBX_SUCCESS;
- break /* error */;
-#if defined(__linux__) || defined(__gnu_linux__)
- case MDBX_SYNC_SIZE:
- if (mdbx_linux_kernel_version >= 0x03060000)
- return MDBX_SUCCESS;
- __fallthrough /* fall through */;
-#endif /* Linux */
-#endif /* _POSIX_SYNCHRONIZED_IO > 0 */
- default:
- if (fsync(fd) == 0)
- return MDBX_SUCCESS;
- }
-
- int rc = errno;
- if (rc != EINTR)
- return rc;
- }
-#endif
-}
-
-int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length) {
-#if defined(_WIN32) || defined(_WIN64)
- BY_HANDLE_FILE_INFORMATION info;
- if (!GetFileInformationByHandle(fd, &info))
- return (int)GetLastError();
- *length = info.nFileSizeLow | (uint64_t)info.nFileSizeHigh << 32;
-#else
- struct stat st;
-
- STATIC_ASSERT_MSG(sizeof(off_t) <= sizeof(uint64_t),
- "libmdbx requires 64-bit file I/O on 64-bit systems");
- if (fstat(fd, &st))
- return errno;
-
- *length = st.st_size;
-#endif
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd) {
-#if defined(_WIN32) || defined(_WIN64)
- switch (GetFileType(fd)) {
- case FILE_TYPE_DISK:
- return MDBX_RESULT_FALSE;
- case FILE_TYPE_CHAR:
- case FILE_TYPE_PIPE:
- return MDBX_RESULT_TRUE;
- default:
- return (int)GetLastError();
- }
-#else
- struct stat info;
- if (fstat(fd, &info))
- return errno;
- switch (info.st_mode & S_IFMT) {
- case S_IFBLK:
- case S_IFREG:
- return MDBX_RESULT_FALSE;
- case S_IFCHR:
- case S_IFIFO:
- case S_IFSOCK:
- return MDBX_RESULT_TRUE;
- case S_IFDIR:
- case S_IFLNK:
- default:
- return MDBX_INCOMPATIBLE;
- }
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length) {
-#if defined(_WIN32) || defined(_WIN64)
- if (mdbx_SetFileInformationByHandle) {
- FILE_END_OF_FILE_INFO EndOfFileInfo;
- EndOfFileInfo.EndOfFile.QuadPart = length;
- return mdbx_SetFileInformationByHandle(fd, FileEndOfFileInfo,
- &EndOfFileInfo,
- sizeof(FILE_END_OF_FILE_INFO))
- ? MDBX_SUCCESS
- : (int)GetLastError();
- } else {
- LARGE_INTEGER li;
- li.QuadPart = length;
- return (SetFilePointerEx(fd, li, NULL, FILE_BEGIN) && SetEndOfFile(fd))
- ? MDBX_SUCCESS
- : (int)GetLastError();
- }
-#else
- STATIC_ASSERT_MSG(sizeof(off_t) >= sizeof(size_t),
- "libmdbx requires 64-bit file I/O on 64-bit systems");
- return ftruncate(fd, length) == 0 ? MDBX_SUCCESS : errno;
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos) {
-#if defined(_WIN32) || defined(_WIN64)
- LARGE_INTEGER li;
- li.QuadPart = pos;
- return SetFilePointerEx(fd, li, NULL, FILE_BEGIN) ? MDBX_SUCCESS
- : (int)GetLastError();
-#else
- STATIC_ASSERT_MSG(sizeof(off_t) >= sizeof(size_t),
- "libmdbx requires 64-bit file I/O on 64-bit systems");
- return (lseek(fd, pos, SEEK_SET) < 0) ? errno : MDBX_SUCCESS;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg) {
-#if defined(_WIN32) || defined(_WIN64)
- *thread = CreateThread(NULL, 0, start_routine, arg, 0, NULL);
- return *thread ? MDBX_SUCCESS : (int)GetLastError();
-#else
- return pthread_create(thread, NULL, start_routine, arg);
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD code = WaitForSingleObject(thread, INFINITE);
- return waitstatus2errcode(code);
-#else
- void *unused_retval = &unused_retval;
- return pthread_join(thread, &unused_retval);
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits) {
- uint8_t *ptr = (uint8_t *)map->address + offset;
-#if defined(_WIN32) || defined(_WIN64)
- if (!FlushViewOfFile(ptr, length))
- return (int)GetLastError();
-#else
-#if defined(__linux__) || defined(__gnu_linux__)
- if (mode_bits == MDBX_SYNC_NONE && mdbx_linux_kernel_version > 0x02061300)
- /* Since Linux 2.6.19, MS_ASYNC is in fact a no-op. The kernel properly
- * tracks dirty pages and flushes them to storage as necessary. */
- return MDBX_SUCCESS;
-#endif /* Linux */
- if (msync(ptr, length, (mode_bits & MDBX_SYNC_DATA) ? MS_SYNC : MS_ASYNC))
- return errno;
- mode_bits &= ~MDBX_SYNC_DATA;
-#endif
- return mdbx_fsync(map->fd, mode_bits);
-}
-
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err) {
-#if defined(_WIN32) || defined(_WIN64)
- (void)pathname;
- (void)err;
- if (!mdbx_GetVolumeInformationByHandleW)
- return MDBX_ENOSYS;
- DWORD unused, flags;
- if (!mdbx_GetVolumeInformationByHandleW(handle, nullptr, 0, nullptr, &unused,
- &flags, nullptr, 0))
- return (int)GetLastError();
- if ((flags & FILE_READ_ONLY_VOLUME) == 0)
- return MDBX_EACCESS;
-#else
- struct statvfs info;
- if (err != MDBX_ENOFILE) {
- if (statvfs(pathname, &info))
- return errno;
- if ((info.f_flag & ST_RDONLY) == 0)
- return err;
- }
- if (fstatvfs(handle, &info))
- return errno;
- if ((info.f_flag & ST_RDONLY) == 0)
- return (err == MDBX_ENOFILE) ? MDBX_EACCESS : err;
-#endif /* !Windows */
- return MDBX_SUCCESS;
-}
-
-static int mdbx_check_fs_local(mdbx_filehandle_t handle, int flags) {
-#if defined(_WIN32) || defined(_WIN64)
- if (mdbx_RunningUnderWine() && !(flags & MDBX_EXCLUSIVE))
- return ERROR_NOT_CAPABLE /* workaround for Wine */;
-
- if (GetFileType(handle) != FILE_TYPE_DISK)
- return ERROR_FILE_OFFLINE;
-
- if (mdbx_GetFileInformationByHandleEx) {
- FILE_REMOTE_PROTOCOL_INFO RemoteProtocolInfo;
- if (mdbx_GetFileInformationByHandleEx(handle, FileRemoteProtocolInfo,
- &RemoteProtocolInfo,
- sizeof(RemoteProtocolInfo))) {
- if ((RemoteProtocolInfo.Flags & REMOTE_PROTOCOL_INFO_FLAG_OFFLINE) &&
- !(flags & MDBX_RDONLY))
- return ERROR_FILE_OFFLINE;
- if (!(RemoteProtocolInfo.Flags & REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK) &&
- !(flags & MDBX_EXCLUSIVE))
- return ERROR_REMOTE_STORAGE_MEDIA_ERROR;
- }
- }
-
- if (mdbx_NtFsControlFile) {
- NTSTATUS rc;
- struct {
- WOF_EXTERNAL_INFO wof_info;
- union {
- WIM_PROVIDER_EXTERNAL_INFO wim_info;
- FILE_PROVIDER_EXTERNAL_INFO_V1 file_info;
- };
- size_t reserved_for_microsoft_madness[42];
- } GetExternalBacking_OutputBuffer;
- IO_STATUS_BLOCK StatusBlock;
- rc = mdbx_NtFsControlFile(handle, NULL, NULL, NULL, &StatusBlock,
- FSCTL_GET_EXTERNAL_BACKING, NULL, 0,
- &GetExternalBacking_OutputBuffer,
- sizeof(GetExternalBacking_OutputBuffer));
- if (NT_SUCCESS(rc)) {
- if (!(flags & MDBX_EXCLUSIVE))
- return ERROR_REMOTE_STORAGE_MEDIA_ERROR;
- } else if (rc != STATUS_OBJECT_NOT_EXTERNALLY_BACKED &&
- rc != STATUS_INVALID_DEVICE_REQUEST &&
- rc != STATUS_NOT_SUPPORTED)
- return ntstatus2errcode(rc);
- }
-
- if (mdbx_GetVolumeInformationByHandleW && mdbx_GetFinalPathNameByHandleW) {
- WCHAR *PathBuffer = mdbx_malloc(sizeof(WCHAR) * INT16_MAX);
- if (!PathBuffer)
- return MDBX_ENOMEM;
-
- int rc = MDBX_SUCCESS;
- DWORD VolumeSerialNumber, FileSystemFlags;
- if (!mdbx_GetVolumeInformationByHandleW(handle, PathBuffer, INT16_MAX,
- &VolumeSerialNumber, NULL,
- &FileSystemFlags, NULL, 0)) {
- rc = (int)GetLastError();
- goto bailout;
- }
-
- if ((flags & MDBX_RDONLY) == 0) {
- if (FileSystemFlags &
- (FILE_SEQUENTIAL_WRITE_ONCE | FILE_READ_ONLY_VOLUME |
- FILE_VOLUME_IS_COMPRESSED)) {
- rc = ERROR_REMOTE_STORAGE_MEDIA_ERROR;
- goto bailout;
- }
- }
-
- if (!mdbx_GetFinalPathNameByHandleW(handle, PathBuffer, INT16_MAX,
- FILE_NAME_NORMALIZED |
- VOLUME_NAME_NT)) {
- rc = (int)GetLastError();
- goto bailout;
- }
-
- if (_wcsnicmp(PathBuffer, L"\\Device\\Mup\\", 12) == 0) {
- if (!(flags & MDBX_EXCLUSIVE)) {
- rc = ERROR_REMOTE_STORAGE_MEDIA_ERROR;
- goto bailout;
- }
- } else if (mdbx_GetFinalPathNameByHandleW(handle, PathBuffer, INT16_MAX,
- FILE_NAME_NORMALIZED |
- VOLUME_NAME_DOS)) {
- UINT DriveType = GetDriveTypeW(PathBuffer);
- if (DriveType == DRIVE_NO_ROOT_DIR &&
- _wcsnicmp(PathBuffer, L"\\\\?\\", 4) == 0 &&
- _wcsnicmp(PathBuffer + 5, L":\\", 2) == 0) {
- PathBuffer[7] = 0;
- DriveType = GetDriveTypeW(PathBuffer + 4);
- }
- switch (DriveType) {
- case DRIVE_CDROM:
- if (flags & MDBX_RDONLY)
- break;
- // fall through
- case DRIVE_UNKNOWN:
- case DRIVE_NO_ROOT_DIR:
- case DRIVE_REMOTE:
- default:
- if (!(flags & MDBX_EXCLUSIVE))
- rc = ERROR_REMOTE_STORAGE_MEDIA_ERROR;
- // fall through
- case DRIVE_REMOVABLE:
- case DRIVE_FIXED:
- case DRIVE_RAMDISK:
- break;
- }
- }
- bailout:
- mdbx_free(PathBuffer);
- return rc;
- }
-
-#else
-
- struct statvfs statvfs_info;
- if (fstatvfs(handle, &statvfs_info))
- return errno;
-#if defined(ST_LOCAL) || defined(ST_EXPORTED)
- const unsigned long st_flags = statvfs_info.f_flag;
-#endif /* ST_LOCAL || ST_EXPORTED */
-
-#if defined(__NetBSD__)
- const unsigned type = 0;
- const char *const name = statvfs_info.f_fstypename;
- const size_t name_len = VFS_NAMELEN;
-#elif defined(_AIX) || defined(__OS400__)
- const char *const name = statvfs_info.f_basetype;
- const size_t name_len = sizeof(statvfs_info.f_basetype);
- struct stat st;
- if (fstat(handle, &st))
- return errno;
- const unsigned type = st.st_vfstype;
- if ((st.st_flag & FS_REMOTE) != 0 && !(flags & MDBX_EXCLUSIVE))
- return MDBX_EREMOTE;
-#elif defined(FSTYPSZ) || defined(_FSTYPSZ)
- const unsigned type = 0;
- const char *const name = statvfs_info.f_basetype;
- const size_t name_len = sizeof(statvfs_info.f_basetype);
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(ST_FSTYPSZ) || defined(_ST_FSTYPSZ)
- const unsigned type = 0;
- struct stat st;
- if (fstat(handle, &st))
- return errno;
- const char *const name = st.st_fstype;
- const size_t name_len = strlen(name);
-#else
- struct statfs statfs_info;
- if (fstatfs(handle, &statfs_info))
- return errno;
-#if defined(__OpenBSD__)
- const unsigned type = 0;
-#else
- const unsigned type = statfs_info.f_type;
-#endif
-#if defined(MNT_LOCAL) || defined(MNT_EXPORTED)
- const unsigned long mnt_flags = statfs_info.f_flags;
-#endif /* MNT_LOCAL || MNT_EXPORTED */
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__) || defined(MFSNAMELEN) || \
- defined(MFSTYPENAMELEN) || defined(VFS_NAMELEN)
- const char *const name = statfs_info.f_fstypename;
- const size_t name_len = sizeof(statfs_info.f_fstypename);
-#elif defined(__ANDROID_API__) && __ANDROID_API__ < 21
- const char *const name = "";
- const unsigned name_len = 0;
-#else
-
- const char *name = "";
- unsigned name_len = 0;
-
- struct stat st;
- if (fstat(handle, &st))
- return errno;
-
- char pathbuf[PATH_MAX];
- FILE *mounted = nullptr;
-#if defined(__linux__) || defined(__gnu_linux__)
- mounted = setmntent("/proc/mounts", "r");
-#endif /* Linux */
- if (!mounted)
- mounted = setmntent("/etc/mtab", "r");
- if (mounted) {
- const struct mntent *ent;
-#if defined(_BSD_SOURCE) || defined(_SVID_SOURCE) || defined(__BIONIC__) || \
- (defined(_DEFAULT_SOURCE) && __GLIBC_PREREQ(2, 19))
- struct mntent entbuf;
- const bool should_copy = false;
- while (nullptr !=
- (ent = getmntent_r(mounted, &entbuf, pathbuf, sizeof(pathbuf))))
-#else
- const bool should_copy = true;
- while (nullptr != (ent = getmntent(mounted)))
-#endif
- {
- struct stat mnt;
- if (!stat(ent->mnt_dir, &mnt) && mnt.st_dev == st.st_dev) {
- if (should_copy) {
- name =
- strncpy(pathbuf, ent->mnt_fsname, name_len = sizeof(pathbuf) - 1);
- pathbuf[name_len] = 0;
- } else {
- name = ent->mnt_fsname;
- name_len = strlen(name);
- }
- break;
- }
- }
- endmntent(mounted);
- }
-#endif /* !xBSD && !Android/Bionic */
-#endif
-
- if (name_len) {
- if (((name_len > 2 && strncasecmp("nfs", name, 3) == 0) ||
- strncasecmp("cifs", name, name_len) == 0 ||
- strncasecmp("ncpfs", name, name_len) == 0 ||
- strncasecmp("smbfs", name, name_len) == 0 ||
- strcasecmp("9P" /* WSL2 */, name) == 0 ||
- ((name_len > 3 && strncasecmp("fuse", name, 4) == 0) &&
- strncasecmp("fuseblk", name, name_len) != 0)) &&
- !(flags & MDBX_EXCLUSIVE))
- return MDBX_EREMOTE;
- if (strcasecmp("ftp", name) == 0 || strcasecmp("http", name) == 0 ||
- strcasecmp("sshfs", name) == 0)
- return MDBX_EREMOTE;
- }
-
-#ifdef ST_LOCAL
- if ((st_flags & ST_LOCAL) == 0 && !(flags & MDBX_EXCLUSIVE))
- return MDBX_EREMOTE;
-#elif defined(MNT_LOCAL)
- if ((mnt_flags & MNT_LOCAL) == 0 && !(flags & MDBX_EXCLUSIVE))
- return MDBX_EREMOTE;
-#endif /* ST/MNT_LOCAL */
-
-#ifdef ST_EXPORTED
- if ((st_flags & ST_EXPORTED) != 0 && !(flags & MDBX_RDONLY))
- return MDBX_EREMOTE;
-#elif defined(MNT_EXPORTED)
- if ((mnt_flags & MNT_EXPORTED) != 0 && !(flags & MDBX_RDONLY))
- return MDBX_EREMOTE;
-#endif /* ST/MNT_EXPORTED */
-
- switch (type) {
- case 0xFF534D42 /* CIFS_MAGIC_NUMBER */:
- case 0x6969 /* NFS_SUPER_MAGIC */:
- case 0x564c /* NCP_SUPER_MAGIC */:
- case 0x517B /* SMB_SUPER_MAGIC */:
-#if defined(__digital__) || defined(__osf__) || defined(__osf)
- case 0x0E /* Tru64 NFS */:
-#endif
-#ifdef ST_FST_NFS
- case ST_FST_NFS:
-#endif
- if ((flags & MDBX_EXCLUSIVE) == 0)
- return MDBX_EREMOTE;
- case 0:
- default:
- break;
- }
-#endif /* Unix */
-
- return MDBX_SUCCESS;
-}
-
-static int check_mmap_limit(const size_t limit) {
- const bool should_check =
-#if defined(__SANITIZE_ADDRESS__)
- true;
-#else
- RUNNING_ON_VALGRIND;
-#endif /* __SANITIZE_ADDRESS__ */
-
- if (should_check) {
- intptr_t pagesize, total_ram_pages, avail_ram_pages;
- int err =
- mdbx_get_sysraminfo(&pagesize, &total_ram_pages, &avail_ram_pages);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- const int log2page = log2n_powerof2(pagesize);
- if ((limit >> (log2page + 7)) > (size_t)total_ram_pages ||
- (limit >> (log2page + 6)) > (size_t)avail_ram_pages) {
- mdbx_error(
- "%s (%zu pages) is too large for available (%zu pages) or total "
- "(%zu pages) system RAM",
- "database upper size limit", limit >> log2page, avail_ram_pages,
- total_ram_pages);
- return MDBX_TOO_LARGE;
- }
- }
-
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t size, const size_t limit,
- const unsigned options) {
- assert(size <= limit);
- map->limit = 0;
- map->current = 0;
- map->address = nullptr;
- map->filesize = 0;
-#if defined(_WIN32) || defined(_WIN64)
- map->section = NULL;
-#endif /* Windows */
-
- int err = mdbx_check_fs_local(map->fd, flags);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- err = check_mmap_limit(limit);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- if ((flags & MDBX_RDONLY) == 0 && (options & MMAP_OPTION_TRUNCATE) != 0) {
- err = mdbx_ftruncate(map->fd, size);
- if (err != MDBX_SUCCESS)
- return err;
- map->filesize = size;
-#if !(defined(_WIN32) || defined(_WIN64))
- map->current = size;
-#endif /* !Windows */
- } else {
- err = mdbx_filesize(map->fd, &map->filesize);
- if (err != MDBX_SUCCESS)
- return err;
-#if !(defined(_WIN32) || defined(_WIN64))
- map->current = (map->filesize > limit) ? limit : (size_t)map->filesize;
-#endif /* !Windows */
- }
-
-#if defined(_WIN32) || defined(_WIN64)
- LARGE_INTEGER SectionSize;
- SectionSize.QuadPart = size;
- err = NtCreateSection(
- &map->section,
- /* DesiredAccess */
- (flags & MDBX_WRITEMAP)
- ? SECTION_QUERY | SECTION_MAP_READ | SECTION_EXTEND_SIZE |
- SECTION_MAP_WRITE
- : SECTION_QUERY | SECTION_MAP_READ | SECTION_EXTEND_SIZE,
- /* ObjectAttributes */ NULL, /* MaximumSize (InitialSize) */ &SectionSize,
- /* SectionPageProtection */
- (flags & MDBX_RDONLY) ? PAGE_READONLY : PAGE_READWRITE,
- /* AllocationAttributes */ SEC_RESERVE, map->fd);
- if (!NT_SUCCESS(err))
- return ntstatus2errcode(err);
-
- SIZE_T ViewSize = (flags & MDBX_RDONLY) ? 0
- : mdbx_RunningUnderWine() ? size
- : limit;
- err = NtMapViewOfSection(
- map->section, GetCurrentProcess(), &map->address,
- /* ZeroBits */ 0,
- /* CommitSize */ 0,
- /* SectionOffset */ NULL, &ViewSize,
- /* InheritDisposition */ ViewUnmap,
- /* AllocationType */ (flags & MDBX_RDONLY) ? 0 : MEM_RESERVE,
- /* Win32Protect */
- (flags & MDBX_WRITEMAP) ? PAGE_READWRITE : PAGE_READONLY);
- if (!NT_SUCCESS(err)) {
- NtClose(map->section);
- map->section = 0;
- map->address = nullptr;
- return ntstatus2errcode(err);
- }
- assert(map->address != MAP_FAILED);
-
- map->current = (size_t)SectionSize.QuadPart;
- map->limit = ViewSize;
-
-#else /* Windows */
-
-#ifndef MAP_TRYFIXED
-#define MAP_TRYFIXED 0
-#endif
-
-#ifndef MAP_HASSEMAPHORE
-#define MAP_HASSEMAPHORE 0
-#endif
-
-#ifndef MAP_CONCEAL
-#define MAP_CONCEAL 0
-#endif
-
-#ifndef MAP_NOSYNC
-#define MAP_NOSYNC 0
-#endif
-
-#ifndef MAP_FIXED_NOREPLACE
-#define MAP_FIXED_NOREPLACE 0
-#endif
-
-#ifndef MAP_NORESERVE
-#define MAP_NORESERVE 0
-#endif
-
- map->address = mmap(
- NULL, limit, (flags & MDBX_WRITEMAP) ? PROT_READ | PROT_WRITE : PROT_READ,
- MAP_SHARED | MAP_FILE | MAP_NORESERVE |
- (F_ISSET(flags, MDBX_UTTERLY_NOSYNC) ? MAP_NOSYNC : 0) |
- ((options & MMAP_OPTION_SEMAPHORE) ? MAP_HASSEMAPHORE | MAP_NOSYNC
- : MAP_CONCEAL),
- map->fd, 0);
-
- if (unlikely(map->address == MAP_FAILED)) {
- map->limit = 0;
- map->current = 0;
- map->address = nullptr;
- return errno;
- }
- map->limit = limit;
-
-#if MDBX_ENABLE_MADVISE
-#ifdef MADV_DONTFORK
- if (unlikely(madvise(map->address, map->limit, MADV_DONTFORK) != 0))
- return errno;
-#endif /* MADV_DONTFORK */
-#ifdef MADV_NOHUGEPAGE
- (void)madvise(map->address, map->limit, MADV_NOHUGEPAGE);
-#endif /* MADV_NOHUGEPAGE */
-#endif /* MDBX_ENABLE_MADVISE */
-
-#endif /* ! Windows */
-
- VALGRIND_MAKE_MEM_DEFINED(map->address, map->current);
- MDBX_ASAN_UNPOISON_MEMORY_REGION(map->address, map->current);
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map) {
- VALGRIND_MAKE_MEM_NOACCESS(map->address, map->current);
- /* Unpoisoning is required for ASAN to avoid false-positive diagnostic
- * when this memory will re-used by malloc or another mmapping.
- * See todo4recovery://erased_by_github/libmdbx/pull/93#issuecomment-613687203
- */
- MDBX_ASAN_UNPOISON_MEMORY_REGION(map->address,
- (map->filesize && map->filesize < map->limit)
- ? map->filesize
- : map->limit);
-#if defined(_WIN32) || defined(_WIN64)
- if (map->section)
- NtClose(map->section);
- NTSTATUS rc = NtUnmapViewOfSection(GetCurrentProcess(), map->address);
- if (!NT_SUCCESS(rc))
- ntstatus2errcode(rc);
-#else
- if (unlikely(munmap(map->address, map->limit)))
- return errno;
-#endif /* ! Windows */
-
- map->limit = 0;
- map->current = 0;
- map->address = nullptr;
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit) {
- assert(size <= limit);
-#if defined(_WIN32) || defined(_WIN64)
- assert(size != map->current || limit != map->limit || size < map->filesize);
-
- NTSTATUS status;
- LARGE_INTEGER SectionSize;
- int err, rc = MDBX_SUCCESS;
-
- if (!(flags & MDBX_RDONLY) && limit == map->limit && size > map->current &&
- /* workaround for Wine */ mdbx_NtExtendSection) {
- /* growth rw-section */
- SectionSize.QuadPart = size;
- status = mdbx_NtExtendSection(map->section, &SectionSize);
- if (!NT_SUCCESS(status))
- return ntstatus2errcode(status);
- map->current = size;
- if (map->filesize < size)
- map->filesize = size;
- return MDBX_SUCCESS;
- }
-
- if (limit > map->limit) {
- err = check_mmap_limit(limit);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- /* check ability of address space for growth before unmap */
- PVOID BaseAddress = (PBYTE)map->address + map->limit;
- SIZE_T RegionSize = limit - map->limit;
- status = NtAllocateVirtualMemory(GetCurrentProcess(), &BaseAddress, 0,
- &RegionSize, MEM_RESERVE, PAGE_NOACCESS);
- if (status == (NTSTATUS) /* STATUS_CONFLICTING_ADDRESSES */ 0xC0000018)
- return MDBX_UNABLE_EXTEND_MAPSIZE;
- if (!NT_SUCCESS(status))
- return ntstatus2errcode(status);
-
- status = NtFreeVirtualMemory(GetCurrentProcess(), &BaseAddress, &RegionSize,
- MEM_RELEASE);
- if (!NT_SUCCESS(status))
- return ntstatus2errcode(status);
- }
-
- /* Windows unable:
- * - shrink a mapped file;
- * - change size of mapped view;
- * - extend read-only mapping;
- * Therefore we should unmap/map entire section. */
- if ((flags & MDBX_MRESIZE_MAY_UNMAP) == 0)
- return MDBX_EPERM;
-
- /* Unpoisoning is required for ASAN to avoid false-positive diagnostic
- * when this memory will re-used by malloc or another mmapping.
- * See todo4recovery://erased_by_github/libmdbx/pull/93#issuecomment-613687203
- */
- MDBX_ASAN_UNPOISON_MEMORY_REGION(map->address, map->limit);
- status = NtUnmapViewOfSection(GetCurrentProcess(), map->address);
- if (!NT_SUCCESS(status))
- return ntstatus2errcode(status);
- status = NtClose(map->section);
- map->section = NULL;
- PVOID ReservedAddress = NULL;
- SIZE_T ReservedSize = limit;
-
- if (!NT_SUCCESS(status)) {
- bailout_ntstatus:
- err = ntstatus2errcode(status);
- bailout:
- map->address = NULL;
- map->current = map->limit = 0;
- if (ReservedAddress) {
- ReservedSize = 0;
- status = NtFreeVirtualMemory(GetCurrentProcess(), &ReservedAddress,
- &ReservedSize, MEM_RELEASE);
- assert(NT_SUCCESS(status));
- (void)status;
- }
- return err;
- }
-
-retry_file_and_section:
- /* resizing of the file may take a while,
- * therefore we reserve address space to avoid occupy it by other threads */
- ReservedAddress = map->address;
- status = NtAllocateVirtualMemory(GetCurrentProcess(), &ReservedAddress, 0,
- &ReservedSize, MEM_RESERVE, PAGE_NOACCESS);
- if (!NT_SUCCESS(status)) {
- ReservedAddress = NULL;
- if (status != (NTSTATUS) /* STATUS_CONFLICTING_ADDRESSES */ 0xC0000018)
- goto bailout_ntstatus /* no way to recovery */;
-
- if (flags & MDBX_MRESIZE_MAY_MOVE)
- /* the base address could be changed */
- map->address = NULL;
- }
-
- err = mdbx_filesize(map->fd, &map->filesize);
- if (err != MDBX_SUCCESS)
- goto bailout;
-
- if ((flags & MDBX_RDONLY) == 0 && map->filesize != size) {
- err = mdbx_ftruncate(map->fd, size);
- if (err == MDBX_SUCCESS)
- map->filesize = size;
- /* ignore error, because Windows unable shrink file
- * that already mapped (by another process) */
- }
-
- SectionSize.QuadPart = size;
- status = NtCreateSection(
- &map->section,
- /* DesiredAccess */
- (flags & MDBX_WRITEMAP)
- ? SECTION_QUERY | SECTION_MAP_READ | SECTION_EXTEND_SIZE |
- SECTION_MAP_WRITE
- : SECTION_QUERY | SECTION_MAP_READ | SECTION_EXTEND_SIZE,
- /* ObjectAttributes */ NULL,
- /* MaximumSize (InitialSize) */ &SectionSize,
- /* SectionPageProtection */
- (flags & MDBX_RDONLY) ? PAGE_READONLY : PAGE_READWRITE,
- /* AllocationAttributes */ SEC_RESERVE, map->fd);
-
- if (!NT_SUCCESS(status))
- goto bailout_ntstatus;
-
- if (ReservedAddress) {
- /* release reserved address space */
- ReservedSize = 0;
- status = NtFreeVirtualMemory(GetCurrentProcess(), &ReservedAddress,
- &ReservedSize, MEM_RELEASE);
- ReservedAddress = NULL;
- if (!NT_SUCCESS(status))
- goto bailout_ntstatus;
- }
-
-retry_mapview:;
- SIZE_T ViewSize = (flags & MDBX_RDONLY) ? size : limit;
- status = NtMapViewOfSection(
- map->section, GetCurrentProcess(), &map->address,
- /* ZeroBits */ 0,
- /* CommitSize */ 0,
- /* SectionOffset */ NULL, &ViewSize,
- /* InheritDisposition */ ViewUnmap,
- /* AllocationType */ (flags & MDBX_RDONLY) ? 0 : MEM_RESERVE,
- /* Win32Protect */
- (flags & MDBX_WRITEMAP) ? PAGE_READWRITE : PAGE_READONLY);
-
- if (!NT_SUCCESS(status)) {
- if (status == (NTSTATUS) /* STATUS_CONFLICTING_ADDRESSES */ 0xC0000018 &&
- map->address && (flags & MDBX_MRESIZE_MAY_MOVE) != 0) {
- /* try remap at another base address */
- map->address = NULL;
- goto retry_mapview;
- }
- NtClose(map->section);
- map->section = NULL;
-
- if (map->address && (size != map->current || limit != map->limit)) {
- /* try remap with previously size and limit,
- * but will return MDBX_UNABLE_EXTEND_MAPSIZE on success */
- rc = (limit > map->limit) ? MDBX_UNABLE_EXTEND_MAPSIZE : MDBX_EPERM;
- size = map->current;
- ReservedSize = limit = map->limit;
- goto retry_file_and_section;
- }
-
- /* no way to recovery */
- goto bailout_ntstatus;
- }
- assert(map->address != MAP_FAILED);
-
- map->current = (size_t)SectionSize.QuadPart;
- map->limit = ViewSize;
-
-#else /* Windows */
-
- map->filesize = 0;
- int rc = mdbx_filesize(map->fd, &map->filesize);
- if (rc != MDBX_SUCCESS)
- return rc;
-
- if (flags & MDBX_RDONLY) {
- map->current = (map->filesize > limit) ? limit : (size_t)map->filesize;
- if (map->current != size)
- rc = (size > map->current) ? MDBX_UNABLE_EXTEND_MAPSIZE : MDBX_EPERM;
- } else {
- if (map->filesize != size) {
- rc = mdbx_ftruncate(map->fd, size);
- if (rc != MDBX_SUCCESS)
- return rc;
- map->filesize = size;
- }
-
- if (map->current > size) {
- /* Clearing asan's bitmask for the region which released in shrinking,
- * since:
- * - after the shrinking we will get an exception when accessing
- * this region and (therefore) do not need the help of ASAN.
- * - this allows us to clear the mask only within the file size
- * when closing the mapping. */
- MDBX_ASAN_UNPOISON_MEMORY_REGION(
- (char *)map->address + size,
- ((map->current < map->limit) ? map->current : map->limit) - size);
- }
- map->current = size;
- }
-
- if (limit == map->limit)
- return rc;
-
- if (limit < map->limit) {
- /* unmap an excess at end of mapping. */
- // coverity[offset_free : FALSE]
- if (unlikely(munmap(map->dxb + limit, map->limit - limit)))
- return errno;
- map->limit = limit;
- return rc;
- }
-
- int err = check_mmap_limit(limit);
- if (unlikely(err != MDBX_SUCCESS))
- return err;
-
- assert(limit > map->limit);
- uint8_t *ptr = MAP_FAILED;
-
-#if defined(MREMAP_MAYMOVE)
- ptr = mremap(map->address, map->limit, limit,
- (flags & MDBX_MRESIZE_MAY_MOVE) ? MREMAP_MAYMOVE : 0);
- if (ptr == MAP_FAILED) {
- err = errno;
- switch (err) {
- default:
- return err;
- case EAGAIN:
- case ENOMEM:
- return MDBX_UNABLE_EXTEND_MAPSIZE;
- case EFAULT /* MADV_DODUMP / MADV_DONTDUMP are mixed for mmap-range */:
- break;
- }
- }
-#endif /* MREMAP_MAYMOVE */
-
- const unsigned mmap_flags =
- MAP_CONCEAL | MAP_SHARED | MAP_FILE | MAP_NORESERVE |
- (F_ISSET(flags, MDBX_UTTERLY_NOSYNC) ? MAP_NOSYNC : 0);
- const unsigned mmap_prot =
- (flags & MDBX_WRITEMAP) ? PROT_READ | PROT_WRITE : PROT_READ;
-
- if (ptr == MAP_FAILED) {
- /* Try to mmap additional space beyond the end of mapping. */
- ptr = mmap(map->dxb + map->limit, limit - map->limit, mmap_prot,
- mmap_flags | MAP_FIXED_NOREPLACE, map->fd, map->limit);
- if (ptr == map->dxb + map->limit)
- ptr = map->dxb;
- else if (ptr != MAP_FAILED) {
- /* the desired address is busy, unmap unsuitable one */
- if (unlikely(munmap(ptr, limit - map->limit)))
- return errno;
- ptr = MAP_FAILED;
- } else {
- err = errno;
- switch (err) {
- default:
- return err;
- case EAGAIN:
- case ENOMEM:
- return MDBX_UNABLE_EXTEND_MAPSIZE;
- case EEXIST: /* address busy */
- case EINVAL: /* kernel don't support MAP_FIXED_NOREPLACE */
- break;
- }
- }
- }
-
- if (ptr == MAP_FAILED) {
- /* unmap and map again whole region */
- if ((flags & MDBX_MRESIZE_MAY_UNMAP) == 0) {
- /* TODO: Perhaps here it is worth to implement suspend/resume threads
- * and perform unmap/map as like for Windows. */
- return MDBX_UNABLE_EXTEND_MAPSIZE;
- }
-
- if (unlikely(munmap(map->address, map->limit)))
- return errno;
-
- // coverity[pass_freed_arg : FALSE]
- ptr = mmap(map->address, limit, mmap_prot,
- (flags & MDBX_MRESIZE_MAY_MOVE)
- ? mmap_flags
- : mmap_flags | (MAP_FIXED_NOREPLACE ? MAP_FIXED_NOREPLACE
- : MAP_FIXED),
- map->fd, 0);
- if (MAP_FIXED_NOREPLACE != 0 && MAP_FIXED_NOREPLACE != MAP_FIXED &&
- unlikely(ptr == MAP_FAILED) && !(flags & MDBX_MRESIZE_MAY_MOVE) &&
- errno == /* kernel don't support MAP_FIXED_NOREPLACE */ EINVAL)
- // coverity[pass_freed_arg : FALSE]
- ptr = mmap(map->address, limit, mmap_prot, mmap_flags | MAP_FIXED,
- map->fd, 0);
-
- if (unlikely(ptr == MAP_FAILED)) {
- /* try to restore prev mapping */
- // coverity[pass_freed_arg : FALSE]
- ptr = mmap(map->address, map->limit, mmap_prot,
- (flags & MDBX_MRESIZE_MAY_MOVE)
- ? mmap_flags
- : mmap_flags | (MAP_FIXED_NOREPLACE ? MAP_FIXED_NOREPLACE
- : MAP_FIXED),
- map->fd, 0);
- if (MAP_FIXED_NOREPLACE != 0 && MAP_FIXED_NOREPLACE != MAP_FIXED &&
- unlikely(ptr == MAP_FAILED) && !(flags & MDBX_MRESIZE_MAY_MOVE) &&
- errno == /* kernel don't support MAP_FIXED_NOREPLACE */ EINVAL)
- // coverity[pass_freed_arg : FALSE]
- ptr = mmap(map->address, map->limit, mmap_prot, mmap_flags | MAP_FIXED,
- map->fd, 0);
- if (unlikely(ptr == MAP_FAILED)) {
- VALGRIND_MAKE_MEM_NOACCESS(map->address, map->current);
- /* Unpoisoning is required for ASAN to avoid false-positive diagnostic
- * when this memory will re-used by malloc or another mmapping.
- * See
- * todo4recovery://erased_by_github/libmdbx/pull/93#issuecomment-613687203
- */
- MDBX_ASAN_UNPOISON_MEMORY_REGION(
- map->address,
- (map->current < map->limit) ? map->current : map->limit);
- map->limit = 0;
- map->current = 0;
- map->address = nullptr;
- return errno;
- }
- rc = MDBX_UNABLE_EXTEND_MAPSIZE;
- limit = map->limit;
- }
- }
-
- assert(ptr && ptr != MAP_FAILED);
- if (map->address != ptr) {
- VALGRIND_MAKE_MEM_NOACCESS(map->address, map->current);
- /* Unpoisoning is required for ASAN to avoid false-positive diagnostic
- * when this memory will re-used by malloc or another mmapping.
- * See
- * todo4recovery://erased_by_github/libmdbx/pull/93#issuecomment-613687203
- */
- MDBX_ASAN_UNPOISON_MEMORY_REGION(
- map->address, (map->current < map->limit) ? map->current : map->limit);
-
- VALGRIND_MAKE_MEM_DEFINED(ptr, map->current);
- MDBX_ASAN_UNPOISON_MEMORY_REGION(ptr, map->current);
- map->address = ptr;
- }
- map->limit = limit;
-
-#if MDBX_ENABLE_MADVISE
-#ifdef MADV_DONTFORK
- if (unlikely(madvise(map->address, map->limit, MADV_DONTFORK) != 0))
- return errno;
-#endif /* MADV_DONTFORK */
-#ifdef MADV_NOHUGEPAGE
- (void)madvise(map->address, map->limit, MADV_NOHUGEPAGE);
-#endif /* MADV_NOHUGEPAGE */
-#endif /* MDBX_ENABLE_MADVISE */
-
-#endif /* POSIX / Windows */
-
- return rc;
-}
-
-/*----------------------------------------------------------------------------*/
-
-__cold MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny) {
- for (;;) {
-#if defined(_M_IX86) || defined(_M_X64) || defined(__i386__) || \
- defined(__x86_64__)
- const unsigned salt = 277u * (unsigned)__rdtsc();
-#elif (defined(_WIN32) || defined(_WIN64)) && MDBX_WITHOUT_MSVC_CRT
- static ULONG state;
- const unsigned salt = (unsigned)RtlRandomEx(&state);
-#else
- const unsigned salt = rand();
-#endif
-
- const unsigned coin = salt % (tiny ? 29u : 43u);
- if (coin < 43 / 3)
- break;
-#if defined(_WIN32) || defined(_WIN64)
- SwitchToThread();
- if (coin > 43 * 2 / 3)
- Sleep(1);
-#else
- sched_yield();
- if (coin > 43 * 2 / 3)
- usleep(coin);
-#endif
- }
-}
-
-#if defined(_WIN32) || defined(_WIN64)
-#elif defined(__APPLE__) || defined(__MACH__)
-#include <mach/mach_time.h>
-#elif defined(__linux__) || defined(__gnu_linux__)
-__cold static clockid_t choice_monoclock(void) {
- struct timespec probe;
-#if defined(CLOCK_BOOTTIME)
- if (clock_gettime(CLOCK_BOOTTIME, &probe) == 0)
- return CLOCK_BOOTTIME;
-#elif defined(CLOCK_MONOTONIC_RAW)
- if (clock_gettime(CLOCK_MONOTONIC_RAW, &probe) == 0)
- return CLOCK_MONOTONIC_RAW;
-#elif defined(CLOCK_MONOTONIC_COARSE)
- if (clock_gettime(CLOCK_MONOTONIC_COARSE, &probe) == 0)
- return CLOCK_MONOTONIC_COARSE;
-#endif
- return CLOCK_MONOTONIC;
-}
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_WIN32) || defined(_WIN64)
-static LARGE_INTEGER performance_frequency;
-#elif defined(__APPLE__) || defined(__MACH__)
-static uint64_t ratio_16dot16_to_monotine;
-#endif
-
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16) {
-#if defined(_WIN32) || defined(_WIN64)
- if (unlikely(performance_frequency.QuadPart == 0))
- QueryPerformanceFrequency(&performance_frequency);
- const uint64_t ratio = performance_frequency.QuadPart;
-#elif defined(__APPLE__) || defined(__MACH__)
- if (unlikely(ratio_16dot16_to_monotine == 0)) {
- mach_timebase_info_data_t ti;
- mach_timebase_info(&ti);
- ratio_16dot16_to_monotine = UINT64_C(1000000000) * ti.denom / ti.numer;
- }
- const uint64_t ratio = ratio_16dot16_to_monotine;
-#else
- const uint64_t ratio = UINT64_C(1000000000);
-#endif
- const uint64_t ret = (ratio * seconds_16dot16 + 32768) >> 16;
- return likely(ret || seconds_16dot16 == 0) ? ret : /* fix underflow */ 1;
-}
-
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime) {
- static uint64_t limit;
- if (unlikely(monotime > limit)) {
- if (limit != 0)
- return UINT32_MAX;
- limit = mdbx_osal_16dot16_to_monotime(UINT32_MAX - 1);
- if (monotime > limit)
- return UINT32_MAX;
- }
- const uint32_t ret =
-#if defined(_WIN32) || defined(_WIN64)
- (uint32_t)((monotime << 16) / performance_frequency.QuadPart);
-#elif defined(__APPLE__) || defined(__MACH__)
- (uint32_t)((monotime << 16) / ratio_16dot16_to_monotine);
-#else
- (uint32_t)(monotime * 128 / 1953125);
-#endif
- return likely(ret || monotime == 0) ? ret : /* fix underflow */ 1;
-}
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void) {
-#if defined(_WIN32) || defined(_WIN64)
- LARGE_INTEGER counter;
- counter.QuadPart = 0;
- QueryPerformanceCounter(&counter);
- return counter.QuadPart;
-#elif defined(__APPLE__) || defined(__MACH__)
- return mach_absolute_time();
-#else
-
-#if defined(__linux__) || defined(__gnu_linux__)
- static clockid_t posix_clockid = -1;
- if (unlikely(posix_clockid < 0))
- posix_clockid = choice_monoclock();
-#elif defined(CLOCK_MONOTONIC)
-#define posix_clockid CLOCK_MONOTONIC
-#else
-#define posix_clockid CLOCK_REALTIME
-#endif
-
- struct timespec ts;
- if (unlikely(clock_gettime(posix_clockid, &ts) != 0)) {
- ts.tv_nsec = 0;
- ts.tv_sec = 0;
- }
- return ts.tv_sec * UINT64_C(1000000000) + ts.tv_nsec;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-
-static void bootid_shake(bin128_t *p) {
- /* Bob Jenkins's PRNG: https://burtleburtle.net/bob/rand/smallprng.html */
- const uint32_t e = p->a - (p->b << 23 | p->b >> 9);
- p->a = p->b ^ (p->c << 16 | p->c >> 16);
- p->b = p->c + (p->d << 11 | p->d >> 21);
- p->c = p->d + e;
- p->d = e + p->a;
-}
-
-static void bootid_collect(bin128_t *p, const void *s, size_t n) {
- p->y += UINT64_C(64526882297375213);
- bootid_shake(p);
- for (size_t i = 0; i < n; ++i) {
- bootid_shake(p);
- p->y ^= UINT64_C(48797879452804441) * ((const uint8_t *)s)[i];
- bootid_shake(p);
- p->y += 14621231;
- }
- bootid_shake(p);
-
- /* minor non-linear tomfoolery */
- const unsigned z = p->x % 61;
- p->y = p->y << z | p->y >> (64 - z);
- bootid_shake(p);
- bootid_shake(p);
- const unsigned q = p->x % 59;
- p->y = p->y << q | p->y >> (64 - q);
- bootid_shake(p);
- bootid_shake(p);
- bootid_shake(p);
-}
-
-#if defined(_WIN32) || defined(_WIN64)
-
-static uint64_t windows_systemtime_ms() {
- FILETIME ft;
- GetSystemTimeAsFileTime(&ft);
- return ((uint64_t)ft.dwHighDateTime << 32 | ft.dwLowDateTime) / 10000ul;
-}
-
-static uint64_t windows_bootime(void) {
- unsigned confirmed = 0;
- uint64_t boottime = 0;
- uint64_t up0 = mdbx_GetTickCount64();
- uint64_t st0 = windows_systemtime_ms();
- for (uint64_t fuse = st0; up0 && st0 < fuse + 1000 * 1000u / 42;) {
- YieldProcessor();
- const uint64_t up1 = mdbx_GetTickCount64();
- const uint64_t st1 = windows_systemtime_ms();
- if (st1 > fuse && st1 == st0 && up1 == up0) {
- uint64_t diff = st1 - up1;
- if (boottime == diff) {
- if (++confirmed > 4)
- return boottime;
- } else {
- confirmed = 0;
- boottime = diff;
- }
- fuse = st1;
- Sleep(1);
- }
- st0 = st1;
- up0 = up1;
- }
- return 0;
-}
-
-static LSTATUS mdbx_RegGetValue(HKEY hKey, LPCSTR lpSubKey, LPCSTR lpValue,
- PVOID pvData, LPDWORD pcbData) {
- LSTATUS rc;
- if (!mdbx_RegGetValueA) {
- /* an old Windows 2000/XP */
- HKEY hSubKey;
- rc = RegOpenKeyA(hKey, lpSubKey, &hSubKey);
- if (rc == ERROR_SUCCESS) {
- rc = RegQueryValueExA(hSubKey, lpValue, NULL, NULL, pvData, pcbData);
- RegCloseKey(hSubKey);
- }
- return rc;
- }
-
- rc = mdbx_RegGetValueA(hKey, lpSubKey, lpValue, RRF_RT_ANY, NULL, pvData,
- pcbData);
- if (rc != ERROR_FILE_NOT_FOUND)
- return rc;
-
- rc = mdbx_RegGetValueA(hKey, lpSubKey, lpValue,
- RRF_RT_ANY | 0x00010000 /* RRF_SUBKEY_WOW6464KEY */,
- NULL, pvData, pcbData);
- if (rc != ERROR_FILE_NOT_FOUND)
- return rc;
- return mdbx_RegGetValueA(hKey, lpSubKey, lpValue,
- RRF_RT_ANY | 0x00020000 /* RRF_SUBKEY_WOW6432KEY */,
- NULL, pvData, pcbData);
-}
-#endif
-
-__cold MDBX_MAYBE_UNUSED static bool
-bootid_parse_uuid(bin128_t *s, const void *p, const size_t n) {
- if (n > 31) {
- unsigned bits = 0;
- for (unsigned i = 0; i < n; ++i) /* try parse an UUID in text form */ {
- uint8_t c = ((const uint8_t *)p)[i];
- if (c >= '0' && c <= '9')
- c -= '0';
- else if (c >= 'a' && c <= 'f')
- c -= 'a' - 10;
- else if (c >= 'A' && c <= 'F')
- c -= 'A' - 10;
- else
- continue;
- assert(c <= 15);
- c ^= s->y >> 60;
- s->y = s->y << 4 | s->x >> 60;
- s->x = s->x << 4 | c;
- bits += 4;
- }
- if (bits > 42 * 3)
- /* UUID parsed successfully */
- return true;
- }
-
- if (n > 15) /* is enough handle it as a binary? */ {
- if (n == sizeof(bin128_t)) {
- bin128_t aligned;
- memcpy(&aligned, p, sizeof(bin128_t));
- s->x += aligned.x;
- s->y += aligned.y;
- } else
- bootid_collect(s, p, n);
- return true;
- }
-
- if (n)
- bootid_collect(s, p, n);
- return false;
-}
-
-__cold MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void) {
- bin128_t bin = {{0, 0}};
- bool got_machineid = false, got_boottime = false, got_bootseq = false;
-
-#if defined(__linux__) || defined(__gnu_linux__)
- {
- const int fd =
- open("/proc/sys/kernel/random/boot_id", O_RDONLY | O_NOFOLLOW);
- if (fd != -1) {
- struct statfs fs;
- char buf[42];
- const ssize_t len =
- (fstatfs(fd, &fs) == 0 && fs.f_type == /* procfs */ 0x9FA0)
- ? read(fd, buf, sizeof(buf))
- : -1;
- const int err = close(fd);
- assert(err == 0);
- (void)err;
- if (len > 0 && bootid_parse_uuid(&bin, buf, len))
- return bin;
- }
- }
-#endif /* Linux */
-
-#if defined(__APPLE__) || defined(__MACH__)
- {
- char buf[42];
- size_t len = sizeof(buf);
- if (!sysctlbyname("kern.bootsessionuuid", buf, &len, nullptr, 0) &&
- bootid_parse_uuid(&bin, buf, len))
- return bin;
-
-#if defined(__MAC_OS_X_VERSION_MIN_REQUIRED) && \
- __MAC_OS_X_VERSION_MIN_REQUIRED > 1050
- uuid_t uuid;
- struct timespec wait = {0, 1000000000u / 42};
- if (!gethostuuid(uuid, &wait) &&
- bootid_parse_uuid(&bin, uuid, sizeof(uuid)))
- got_machineid = true;
-#endif /* > 10.5 */
-
- struct timeval boottime;
- len = sizeof(boottime);
- if (!sysctlbyname("kern.boottime", &boottime, &len, nullptr, 0) &&
- len == sizeof(boottime) && boottime.tv_sec)
- got_boottime = true;
- }
-#endif /* Apple/Darwin */
-
-#if defined(_WIN32) || defined(_WIN64)
- {
- union buf {
- DWORD BootId;
- DWORD BaseTime;
- SYSTEM_TIMEOFDAY_INFORMATION SysTimeOfDayInfo;
- struct {
- LARGE_INTEGER BootTime;
- LARGE_INTEGER CurrentTime;
- LARGE_INTEGER TimeZoneBias;
- ULONG TimeZoneId;
- ULONG Reserved;
- ULONGLONG BootTimeBias;
- ULONGLONG SleepTimeBias;
- } SysTimeOfDayInfoHacked;
- wchar_t MachineGuid[42];
- char DigitalProductId[248];
- } buf;
-
- static const char HKLM_MicrosoftCryptography[] =
- "SOFTWARE\\Microsoft\\Cryptography";
- DWORD len = sizeof(buf);
- /* Windows is madness and must die */
- if (mdbx_RegGetValue(HKEY_LOCAL_MACHINE, HKLM_MicrosoftCryptography,
- "MachineGuid", &buf.MachineGuid,
- &len) == ERROR_SUCCESS &&
- len < sizeof(buf))
- got_machineid = bootid_parse_uuid(&bin, &buf.MachineGuid, len);
-
- if (!got_machineid) {
- /* again, Windows is madness */
- static const char HKLM_WindowsNT[] =
- "SOFTWARE\\Microsoft\\Windows NT\\CurrentVersion";
- static const char HKLM_WindowsNT_DPK[] =
- "SOFTWARE\\Microsoft\\Windows "
- "NT\\CurrentVersion\\DefaultProductKey";
- static const char HKLM_WindowsNT_DPK2[] =
- "SOFTWARE\\Microsoft\\Windows "
- "NT\\CurrentVersion\\DefaultProductKey2";
-
- len = sizeof(buf);
- if (mdbx_RegGetValue(HKEY_LOCAL_MACHINE, HKLM_WindowsNT,
- "DigitalProductId", &buf.DigitalProductId,
- &len) == ERROR_SUCCESS &&
- len > 42 && len < sizeof(buf)) {
- bootid_collect(&bin, &buf.DigitalProductId, len);
- got_machineid = true;
- }
- len = sizeof(buf);
- if (mdbx_RegGetValue(HKEY_LOCAL_MACHINE, HKLM_WindowsNT_DPK,
- "DigitalProductId", &buf.DigitalProductId,
- &len) == ERROR_SUCCESS &&
- len > 42 && len < sizeof(buf)) {
- bootid_collect(&bin, &buf.DigitalProductId, len);
- got_machineid = true;
- }
- len = sizeof(buf);
- if (mdbx_RegGetValue(HKEY_LOCAL_MACHINE, HKLM_WindowsNT_DPK2,
- "DigitalProductId", &buf.DigitalProductId,
- &len) == ERROR_SUCCESS &&
- len > 42 && len < sizeof(buf)) {
- bootid_collect(&bin, &buf.DigitalProductId, len);
- got_machineid = true;
- }
- }
-
- static const char HKLM_PrefetcherParams[] =
- "SYSTEM\\CurrentControlSet\\Control\\Session Manager\\Memory "
- "Management\\PrefetchParameters";
- len = sizeof(buf);
- if (mdbx_RegGetValue(HKEY_LOCAL_MACHINE, HKLM_PrefetcherParams, "BootId",
- &buf.BootId, &len) == ERROR_SUCCESS &&
- len > 1 && len < sizeof(buf)) {
- bootid_collect(&bin, &buf.BootId, len);
- got_bootseq = true;
- }
-
- len = sizeof(buf);
- if (mdbx_RegGetValue(HKEY_LOCAL_MACHINE, HKLM_PrefetcherParams, "BaseTime",
- &buf.BaseTime, &len) == ERROR_SUCCESS &&
- len >= sizeof(buf.BaseTime) && buf.BaseTime) {
- bootid_collect(&bin, &buf.BaseTime, len);
- got_boottime = true;
- }
-
- /* BootTime from SYSTEM_TIMEOFDAY_INFORMATION */
- NTSTATUS status = NtQuerySystemInformation(
- 0x03 /* SystemTmeOfDayInformation */, &buf.SysTimeOfDayInfo,
- sizeof(buf.SysTimeOfDayInfo), &len);
- if (NT_SUCCESS(status) &&
- len >= offsetof(union buf, SysTimeOfDayInfoHacked.BootTimeBias) +
- sizeof(buf.SysTimeOfDayInfoHacked.BootTimeBias) &&
- buf.SysTimeOfDayInfoHacked.BootTime.QuadPart) {
- const uint64_t UnbiasedBootTime =
- buf.SysTimeOfDayInfoHacked.BootTime.QuadPart -
- buf.SysTimeOfDayInfoHacked.BootTimeBias;
- if (UnbiasedBootTime) {
- bootid_collect(&bin, &UnbiasedBootTime, sizeof(UnbiasedBootTime));
- got_boottime = true;
- }
- }
-
- if (!got_boottime) {
- uint64_t boottime = windows_bootime();
- if (boottime) {
- bootid_collect(&bin, &boottime, sizeof(boottime));
- got_boottime = true;
- }
- }
- }
-#endif /* Windows */
-
-#if defined(CTL_HW) && defined(HW_UUID)
- if (!got_machineid) {
- static const int mib[] = {CTL_HW, HW_UUID};
- char buf[42];
- size_t len = sizeof(buf);
- if (sysctl(
-#ifdef SYSCTL_LEGACY_NONCONST_MIB
- (int *)
-#endif
- mib,
- ARRAY_LENGTH(mib), &buf, &len, NULL, 0) == 0)
- got_machineid = bootid_parse_uuid(&bin, buf, len);
- }
-#endif /* CTL_HW && HW_UUID */
-
-#if defined(CTL_KERN) && defined(KERN_HOSTUUID)
- if (!got_machineid) {
- static const int mib[] = {CTL_KERN, KERN_HOSTUUID};
- char buf[42];
- size_t len = sizeof(buf);
- if (sysctl(
-#ifdef SYSCTL_LEGACY_NONCONST_MIB
- (int *)
-#endif
- mib,
- ARRAY_LENGTH(mib), &buf, &len, NULL, 0) == 0)
- got_machineid = bootid_parse_uuid(&bin, buf, len);
- }
-#endif /* CTL_KERN && KERN_HOSTUUID */
-
-#if defined(__NetBSD__)
- if (!got_machineid) {
- char buf[42];
- size_t len = sizeof(buf);
- if (sysctlbyname("machdep.dmi.system-uuid", buf, &len, NULL, 0) == 0)
- got_machineid = bootid_parse_uuid(&bin, buf, len);
- }
-#endif /* __NetBSD__ */
-
-#if _XOPEN_SOURCE_EXTENDED
- if (!got_machineid) {
- const int hostid = gethostid();
- if (hostid > 0) {
- bootid_collect(&bin, &hostid, sizeof(hostid));
- got_machineid = true;
- }
- }
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
- if (!got_machineid) {
- lack:
- bin.x = bin.y = 0;
- return bin;
- }
-
- /*--------------------------------------------------------------------------*/
-
-#if defined(CTL_KERN) && defined(KERN_BOOTTIME)
- if (!got_boottime) {
- static const int mib[] = {CTL_KERN, KERN_BOOTTIME};
- struct timeval boottime;
- size_t len = sizeof(boottime);
- if (sysctl(
-#ifdef SYSCTL_LEGACY_NONCONST_MIB
- (int *)
-#endif
- mib,
- ARRAY_LENGTH(mib), &boottime, &len, NULL, 0) == 0 &&
- len == sizeof(boottime) && boottime.tv_sec) {
- bootid_collect(&bin, &boottime, len);
- got_boottime = true;
- }
- }
-#endif /* CTL_KERN && KERN_BOOTTIME */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
- if (!got_boottime) {
- kstat_ctl_t *kc = kstat_open();
- if (kc) {
- kstat_t *kp = kstat_lookup(kc, "unix", 0, "system_misc");
- if (kp && kstat_read(kc, kp, 0) != -1) {
- kstat_named_t *kn = (kstat_named_t *)kstat_data_lookup(kp, "boot_time");
- if (kn) {
- switch (kn->data_type) {
- case KSTAT_DATA_INT32:
- case KSTAT_DATA_UINT32:
- bootid_collect(&bin, &kn->value, sizeof(int32_t));
- got_boottime = true;
- case KSTAT_DATA_INT64:
- case KSTAT_DATA_UINT64:
- bootid_collect(&bin, &kn->value, sizeof(int64_t));
- got_boottime = true;
- }
- }
- }
- kstat_close(kc);
- }
- }
-#endif /* SunOS / Solaris */
-
-#if _XOPEN_SOURCE_EXTENDED && defined(BOOT_TIME)
- if (!got_boottime) {
- setutxent();
- const struct utmpx id = {.ut_type = BOOT_TIME};
- const struct utmpx *entry = getutxid(&id);
- if (entry) {
- bootid_collect(&bin, entry, sizeof(*entry));
- got_boottime = true;
- while (unlikely((entry = getutxid(&id)) != nullptr)) {
- /* have multiple reboot records, assuming we can distinguish next
- * bootsession even if RTC is wrong or absent */
- bootid_collect(&bin, entry, sizeof(*entry));
- got_bootseq = true;
- }
- }
- endutxent();
- }
-#endif /* _XOPEN_SOURCE_EXTENDED && BOOT_TIME */
-
- if (!got_bootseq) {
- if (!got_boottime || !MDBX_TRUST_RTC)
- goto lack;
-
-#if defined(_WIN32) || defined(_WIN64)
- FILETIME now;
- GetSystemTimeAsFileTime(&now);
- if (0x1CCCCCC > now.dwHighDateTime)
-#else
- struct timespec mono, real;
- if (clock_gettime(CLOCK_MONOTONIC, &mono) ||
- clock_gettime(CLOCK_REALTIME, &real) ||
- /* wrong time, RTC is mad or absent */
- 1555555555l > real.tv_sec ||
- /* seems no adjustment by RTC/NTP, i.e. a fake time */
- real.tv_sec < mono.tv_sec || 1234567890l > real.tv_sec - mono.tv_sec ||
- (real.tv_sec - mono.tv_sec) % 900u == 0)
-#endif
- goto lack;
- }
-
- return bin;
-}
-
-__cold int mdbx_get_sysraminfo(intptr_t *page_size, intptr_t *total_pages,
- intptr_t *avail_pages) {
- if (!page_size && !total_pages && !avail_pages)
- return MDBX_EINVAL;
- if (total_pages)
- *total_pages = -1;
- if (avail_pages)
- *avail_pages = -1;
-
- const intptr_t pagesize = mdbx_syspagesize();
- if (page_size)
- *page_size = pagesize;
- if (unlikely(pagesize < MIN_PAGESIZE || !is_powerof2(pagesize)))
- return MDBX_INCOMPATIBLE;
-
- MDBX_MAYBE_UNUSED const int log2page = log2n_powerof2(pagesize);
- assert(pagesize == (INT64_C(1) << log2page));
- (void)log2page;
-
-#if defined(_WIN32) || defined(_WIN64)
- MEMORYSTATUSEX info;
- memset(&info, 0, sizeof(info));
- info.dwLength = sizeof(info);
- if (!GlobalMemoryStatusEx(&info))
- return (int)GetLastError();
-#endif
-
- if (total_pages) {
-#if defined(_WIN32) || defined(_WIN64)
- const intptr_t total_ram_pages = (intptr_t)(info.ullTotalPhys >> log2page);
-#elif defined(_SC_PHYS_PAGES)
- const intptr_t total_ram_pages = sysconf(_SC_PHYS_PAGES);
- if (total_ram_pages == -1)
- return errno;
-#elif defined(_SC_AIX_REALMEM)
- const intptr_t total_ram_Kb = sysconf(_SC_AIX_REALMEM);
- if (total_ram_Kb == -1)
- return errno;
- const intptr_t total_ram_pages = (total_ram_Kb << 10) >> log2page;
-#elif defined(HW_USERMEM) || defined(HW_PHYSMEM64) || defined(HW_MEMSIZE) || \
- defined(HW_PHYSMEM)
- size_t ram, len = sizeof(ram);
- static const int mib[] = {
- CTL_HW,
-#if defined(HW_USERMEM)
- HW_USERMEM
-#elif defined(HW_PHYSMEM64)
- HW_PHYSMEM64
-#elif defined(HW_MEMSIZE)
- HW_MEMSIZE
-#else
- HW_PHYSMEM
-#endif
- };
- if (sysctl(
-#ifdef SYSCTL_LEGACY_NONCONST_MIB
- (int *)
-#endif
- mib,
- ARRAY_LENGTH(mib), &ram, &len, NULL, 0) != 0)
- return errno;
- if (len != sizeof(ram))
- return MDBX_ENOSYS;
- const intptr_t total_ram_pages = (intptr_t)(ram >> log2page);
-#else
-#error "FIXME: Get User-accessible or physical RAM"
-#endif
- *total_pages = total_ram_pages;
- if (total_ram_pages < 1)
- return MDBX_ENOSYS;
- }
-
- if (avail_pages) {
-#if defined(_WIN32) || defined(_WIN64)
- const intptr_t avail_ram_pages = (intptr_t)(info.ullAvailPhys >> log2page);
-#elif defined(_SC_AVPHYS_PAGES)
- const intptr_t avail_ram_pages = sysconf(_SC_AVPHYS_PAGES);
- if (avail_ram_pages == -1)
- return errno;
-#elif defined(__MACH__)
- mach_msg_type_number_t count = HOST_VM_INFO_COUNT;
- vm_statistics_data_t vmstat;
- mach_port_t mport = mach_host_self();
- kern_return_t kerr = host_statistics(mach_host_self(), HOST_VM_INFO,
- (host_info_t)&vmstat, &count);
- mach_port_deallocate(mach_task_self(), mport);
- if (unlikely(kerr != KERN_SUCCESS))
- return MDBX_ENOSYS;
- const intptr_t avail_ram_pages = vmstat.free_count;
-#elif defined(VM_TOTAL) || defined(VM_METER)
- struct vmtotal info;
- size_t len = sizeof(info);
- static const int mib[] = {
- CTL_VM,
-#if defined(VM_TOTAL)
- VM_TOTAL
-#elif defined(VM_METER)
- VM_METER
-#endif
- };
- if (sysctl(
-#ifdef SYSCTL_LEGACY_NONCONST_MIB
- (int *)
-#endif
- mib,
- ARRAY_LENGTH(mib), &info, &len, NULL, 0) != 0)
- return errno;
- if (len != sizeof(info))
- return MDBX_ENOSYS;
- const intptr_t avail_ram_pages = info.t_free;
-#else
-#error "FIXME: Get Available RAM"
-#endif
- *avail_pages = avail_ram_pages;
- if (avail_ram_pages < 1)
- return MDBX_ENOSYS;
- }
-
- return MDBX_SUCCESS;
-}
-/* This is CMake-template for libmdbx's version.c
- ******************************************************************************/
-
-
-#if MDBX_VERSION_MAJOR != 0 || \
- MDBX_VERSION_MINOR != 11
-#error "API version mismatch! Had `git fetch --tags` done?"
-#endif
-
-static const char sourcery[] = MDBX_STRINGIFY(MDBX_BUILD_SOURCERY);
-
-__dll_export
-#ifdef __attribute_used__
- __attribute_used__
-#elif defined(__GNUC__) || __has_attribute(__used__)
- __attribute__((__used__))
-#endif
-#ifdef __attribute_externally_visible__
- __attribute_externally_visible__
-#elif (defined(__GNUC__) && !defined(__clang__)) || \
- __has_attribute(__externally_visible__)
- __attribute__((__externally_visible__))
-#endif
- const struct MDBX_version_info mdbx_version = {
- 0,
- 11,
- 7,
- 0,
- {"2022-04-22T23:28:56+03:00", "6fdefd6844bd6d212ea6dec99892ae1fac981521", "40ec559c8c1e0b9794cdc0cd12c155d05e7e8547",
- "v0.11.7-0-g40ec559c"},
- sourcery};
-
-__dll_export
-#ifdef __attribute_used__
- __attribute_used__
-#elif defined(__GNUC__) || __has_attribute(__used__)
- __attribute__((__used__))
-#endif
-#ifdef __attribute_externally_visible__
- __attribute_externally_visible__
-#elif (defined(__GNUC__) && !defined(__clang__)) || \
- __has_attribute(__externally_visible__)
- __attribute__((__externally_visible__))
-#endif
- const char *const mdbx_sourcery_anchor = sourcery;
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-#if defined(_WIN32) || defined(_WIN64) /* Windows LCK-implementation */
-
-/* PREAMBLE FOR WINDOWS:
- *
- * We are not concerned for performance here.
- * If you are running Windows a performance could NOT be the goal.
- * Otherwise please use Linux. */
-
-
-static void mdbx_winnt_import(void);
-
-#if MDBX_BUILD_SHARED_LIBRARY
-#if MDBX_WITHOUT_MSVC_CRT && defined(NDEBUG)
-/* DEBUG/CHECKED builds still require MSVC's CRT for runtime checks.
- *
- * Define dll's entry point only for Release build when NDEBUG is defined and
- * MDBX_WITHOUT_MSVC_CRT=ON. if the entry point isn't defined then MSVC's will
- * automatically use DllMainCRTStartup() from CRT library, which also
- * automatically call DllMain() from our mdbx.dll */
-#pragma comment(linker, "/ENTRY:DllMain")
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-BOOL APIENTRY DllMain(HANDLE module, DWORD reason, LPVOID reserved)
-#else
-#if !MDBX_MANUAL_MODULE_HANDLER
-static
-#endif /* !MDBX_MANUAL_MODULE_HANDLER */
- void NTAPI
- mdbx_module_handler(PVOID module, DWORD reason, PVOID reserved)
-#endif /* MDBX_BUILD_SHARED_LIBRARY */
-{
- (void)reserved;
- switch (reason) {
- case DLL_PROCESS_ATTACH:
- mdbx_winnt_import();
- mdbx_rthc_global_init();
- break;
- case DLL_PROCESS_DETACH:
- mdbx_rthc_global_dtor();
- break;
-
- case DLL_THREAD_ATTACH:
- break;
- case DLL_THREAD_DETACH:
- mdbx_rthc_thread_dtor(module);
- break;
- }
-#if MDBX_BUILD_SHARED_LIBRARY
- return TRUE;
-#endif
-}
-
-#if !MDBX_BUILD_SHARED_LIBRARY && !MDBX_MANUAL_MODULE_HANDLER
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(_MSC_VER)
-# pragma const_seg(push)
-# pragma data_seg(push)
-
-# ifndef _M_IX86
- /* kick a linker to create the TLS directory if not already done */
-# pragma comment(linker, "/INCLUDE:_tls_used")
- /* Force some symbol references. */
-# pragma comment(linker, "/INCLUDE:mdbx_tls_anchor")
- /* specific const-segment for WIN64 */
-# pragma const_seg(".CRT$XLB")
- const
-# else
- /* kick a linker to create the TLS directory if not already done */
-# pragma comment(linker, "/INCLUDE:__tls_used")
- /* Force some symbol references. */
-# pragma comment(linker, "/INCLUDE:_mdbx_tls_anchor")
- /* specific data-segment for WIN32 */
-# pragma data_seg(".CRT$XLB")
-# endif
-
- __declspec(allocate(".CRT$XLB")) PIMAGE_TLS_CALLBACK mdbx_tls_anchor = mdbx_module_handler;
-# pragma data_seg(pop)
-# pragma const_seg(pop)
-
-#elif defined(__GNUC__)
-# ifndef _M_IX86
- const
-# endif
- PIMAGE_TLS_CALLBACK mdbx_tls_anchor __attribute__((__section__(".CRT$XLB"), used)) = mdbx_module_handler;
-#else
-# error FIXME
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* !MDBX_BUILD_SHARED_LIBRARY && !MDBX_MANUAL_MODULE_HANDLER */
-
-/*----------------------------------------------------------------------------*/
-
-#define LCK_SHARED 0
-#define LCK_EXCLUSIVE LOCKFILE_EXCLUSIVE_LOCK
-#define LCK_WAITFOR 0
-#define LCK_DONTWAIT LOCKFILE_FAIL_IMMEDIATELY
-
-static __inline BOOL flock(mdbx_filehandle_t fd, DWORD flags, uint64_t offset,
- size_t bytes) {
- OVERLAPPED ov;
- ov.hEvent = 0;
- ov.Offset = (DWORD)offset;
- ov.OffsetHigh = HIGH_DWORD(offset);
- return LockFileEx(fd, flags, 0, (DWORD)bytes, HIGH_DWORD(bytes), &ov);
-}
-
-static __inline BOOL funlock(mdbx_filehandle_t fd, uint64_t offset,
- size_t bytes) {
- return UnlockFile(fd, (DWORD)offset, HIGH_DWORD(offset), (DWORD)bytes,
- HIGH_DWORD(bytes));
-}
-
-/*----------------------------------------------------------------------------*/
-/* global `write` lock for write-txt processing,
- * exclusive locking both meta-pages) */
-
-#define LCK_MAXLEN (1u + ((~(size_t)0) >> 1))
-#define LCK_META_OFFSET 0
-#define LCK_META_LEN (MAX_PAGESIZE * NUM_METAS)
-#define LCK_BODY_OFFSET LCK_META_LEN
-#define LCK_BODY_LEN (LCK_MAXLEN - LCK_BODY_OFFSET)
-#define LCK_BODY LCK_BODY_OFFSET, LCK_BODY_LEN
-#define LCK_WHOLE 0, LCK_MAXLEN
-
-int mdbx_txn_lock(MDBX_env *env, bool dontwait) {
- if (dontwait) {
- if (!TryEnterCriticalSection(&env->me_windowsbug_lock))
- return MDBX_BUSY;
- } else {
- __try {
- EnterCriticalSection(&env->me_windowsbug_lock);
- }
- __except ((GetExceptionCode() ==
- 0xC0000194 /* STATUS_POSSIBLE_DEADLOCK / EXCEPTION_POSSIBLE_DEADLOCK */)
- ? EXCEPTION_EXECUTE_HANDLER
- : EXCEPTION_CONTINUE_SEARCH) {
- return ERROR_POSSIBLE_DEADLOCK;
- }
- }
-
- if ((env->me_flags & MDBX_EXCLUSIVE) ||
- flock(env->me_lazy_fd,
- dontwait ? (LCK_EXCLUSIVE | LCK_DONTWAIT)
- : (LCK_EXCLUSIVE | LCK_WAITFOR),
- LCK_BODY))
- return MDBX_SUCCESS;
- int rc = (int)GetLastError();
- LeaveCriticalSection(&env->me_windowsbug_lock);
- return (!dontwait || rc != ERROR_LOCK_VIOLATION) ? rc : MDBX_BUSY;
-}
-
-void mdbx_txn_unlock(MDBX_env *env) {
- int rc = (env->me_flags & MDBX_EXCLUSIVE)
- ? TRUE
- : funlock(env->me_lazy_fd, LCK_BODY);
- LeaveCriticalSection(&env->me_windowsbug_lock);
- if (!rc)
- mdbx_panic("%s failed: err %u", __func__, (int)GetLastError());
-}
-
-/*----------------------------------------------------------------------------*/
-/* global `read` lock for readers registration,
- * exclusive locking `mti_numreaders` (second) cacheline */
-
-#define LCK_LO_OFFSET 0
-#define LCK_LO_LEN offsetof(MDBX_lockinfo, mti_numreaders)
-#define LCK_UP_OFFSET LCK_LO_LEN
-#define LCK_UP_LEN (sizeof(MDBX_lockinfo) - LCK_UP_OFFSET)
-#define LCK_LOWER LCK_LO_OFFSET, LCK_LO_LEN
-#define LCK_UPPER LCK_UP_OFFSET, LCK_UP_LEN
-
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env) {
- mdbx_srwlock_AcquireShared(&env->me_remap_guard);
- if (env->me_lfd == INVALID_HANDLE_VALUE)
- return MDBX_SUCCESS; /* readonly database in readonly filesystem */
-
- /* transition from S-? (used) to S-E (locked),
- * e.g. exclusive lock upper-part */
- if ((env->me_flags & MDBX_EXCLUSIVE) ||
- flock(env->me_lfd, LCK_EXCLUSIVE | LCK_WAITFOR, LCK_UPPER))
- return MDBX_SUCCESS;
-
- int rc = (int)GetLastError();
- mdbx_srwlock_ReleaseShared(&env->me_remap_guard);
- return rc;
-}
-
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env) {
- if (env->me_lfd != INVALID_HANDLE_VALUE) {
- /* transition from S-E (locked) to S-? (used), e.g. unlock upper-part */
- if ((env->me_flags & MDBX_EXCLUSIVE) == 0 &&
- !funlock(env->me_lfd, LCK_UPPER))
- mdbx_panic("%s failed: err %u", __func__, (int)GetLastError());
- }
- mdbx_srwlock_ReleaseShared(&env->me_remap_guard);
-}
-
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait) {
- return flock(fd,
- wait ? LCK_EXCLUSIVE | LCK_WAITFOR
- : LCK_EXCLUSIVE | LCK_DONTWAIT,
- 0, LCK_MAXLEN)
- ? MDBX_SUCCESS
- : (int)GetLastError();
-}
-
-static int suspend_and_append(mdbx_handle_array_t **array,
- const DWORD ThreadId) {
- const unsigned limit = (*array)->limit;
- if ((*array)->count == limit) {
- void *ptr = mdbx_realloc(
- (limit > ARRAY_LENGTH((*array)->handles))
- ? *array
- : /* don't free initial array on the stack */ NULL,
- sizeof(mdbx_handle_array_t) +
- sizeof(HANDLE) * (limit * 2 - ARRAY_LENGTH((*array)->handles)));
- if (!ptr)
- return MDBX_ENOMEM;
- if (limit == ARRAY_LENGTH((*array)->handles))
- memcpy(ptr, *array, sizeof(mdbx_handle_array_t));
- *array = (mdbx_handle_array_t *)ptr;
- (*array)->limit = limit * 2;
- }
-
- HANDLE hThread = OpenThread(THREAD_SUSPEND_RESUME | THREAD_QUERY_INFORMATION,
- FALSE, ThreadId);
- if (hThread == NULL)
- return (int)GetLastError();
-
- if (SuspendThread(hThread) == (DWORD)-1) {
- int err = (int)GetLastError();
- DWORD ExitCode;
- if (err == /* workaround for Win10 UCRT bug */ ERROR_ACCESS_DENIED ||
- !GetExitCodeThread(hThread, &ExitCode) || ExitCode != STILL_ACTIVE)
- err = MDBX_SUCCESS;
- CloseHandle(hThread);
- return err;
- }
-
- (*array)->handles[(*array)->count++] = hThread;
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array) {
- mdbx_assert(env, (env->me_flags & MDBX_NOTLS) == 0);
- const uintptr_t CurrentTid = GetCurrentThreadId();
- int rc;
- if (env->me_lck_mmap.lck) {
- /* Scan LCK for threads of the current process */
- const MDBX_reader *const begin = env->me_lck_mmap.lck->mti_readers;
- const MDBX_reader *const end =
- begin +
- atomic_load32(&env->me_lck_mmap.lck->mti_numreaders, mo_AcquireRelease);
- const uintptr_t WriteTxnOwner = env->me_txn0 ? env->me_txn0->mt_owner : 0;
- for (const MDBX_reader *reader = begin; reader < end; ++reader) {
- if (reader->mr_pid.weak != env->me_pid || !reader->mr_tid.weak) {
- skip_lck:
- continue;
- }
- if (reader->mr_tid.weak == CurrentTid ||
- reader->mr_tid.weak == WriteTxnOwner)
- goto skip_lck;
-
- rc = suspend_and_append(array, (mdbx_tid_t)reader->mr_tid.weak);
- if (rc != MDBX_SUCCESS) {
- bailout_lck:
- (void)mdbx_resume_threads_after_remap(*array);
- return rc;
- }
- }
- if (WriteTxnOwner && WriteTxnOwner != CurrentTid) {
- rc = suspend_and_append(array, (mdbx_tid_t)WriteTxnOwner);
- if (rc != MDBX_SUCCESS)
- goto bailout_lck;
- }
- } else {
- /* Without LCK (i.e. read-only mode).
- * Walk through a snapshot of all running threads */
- mdbx_assert(env, env->me_flags & (MDBX_EXCLUSIVE | MDBX_RDONLY));
- const HANDLE hSnapshot = CreateToolhelp32Snapshot(TH32CS_SNAPTHREAD, 0);
- if (hSnapshot == INVALID_HANDLE_VALUE)
- return (int)GetLastError();
-
- THREADENTRY32 entry;
- entry.dwSize = sizeof(THREADENTRY32);
-
- if (!Thread32First(hSnapshot, &entry)) {
- rc = (int)GetLastError();
- bailout_toolhelp:
- CloseHandle(hSnapshot);
- (void)mdbx_resume_threads_after_remap(*array);
- return rc;
- }
-
- do {
- if (entry.th32OwnerProcessID != env->me_pid ||
- entry.th32ThreadID == CurrentTid)
- continue;
-
- rc = suspend_and_append(array, entry.th32ThreadID);
- if (rc != MDBX_SUCCESS)
- goto bailout_toolhelp;
-
- } while (Thread32Next(hSnapshot, &entry));
-
- rc = (int)GetLastError();
- if (rc != ERROR_NO_MORE_FILES)
- goto bailout_toolhelp;
- CloseHandle(hSnapshot);
- }
-
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array) {
- int rc = MDBX_SUCCESS;
- for (unsigned i = 0; i < array->count; ++i) {
- const HANDLE hThread = array->handles[i];
- if (ResumeThread(hThread) == (DWORD)-1) {
- const int err = (int)GetLastError();
- DWORD ExitCode;
- if (err != /* workaround for Win10 UCRT bug */ ERROR_ACCESS_DENIED &&
- GetExitCodeThread(hThread, &ExitCode) && ExitCode == STILL_ACTIVE)
- rc = err;
- }
- CloseHandle(hThread);
- }
- return rc;
-}
-
-/*----------------------------------------------------------------------------*/
-/* global `initial` lock for lockfile initialization,
- * exclusive/shared locking first cacheline */
-
-/* Briefly description of locking schema/algorithm:
- * - Windows does not support upgrading or downgrading for file locking.
- * - Therefore upgrading/downgrading is emulated by shared and exclusive
- * locking of upper and lower halves.
- * - In other words, we have FSM with possible 9 states,
- * i.e. free/shared/exclusive x free/shared/exclusive == 9.
- * Only 6 states of FSM are used, which 2 of ones are transitive.
- *
- * States:
- * ?-? = free, i.e. unlocked
- * S-? = used, i.e. shared lock
- * E-? = exclusive-read, i.e. operational exclusive
- * ?-S
- * ?-E = middle (transitive state)
- * S-S
- * S-E = locked (transitive state)
- * E-S
- * E-E = exclusive-write, i.e. exclusive due (re)initialization
- *
- * The mdbx_lck_seize() moves the locking-FSM from the initial free/unlocked
- * state to the "exclusive write" (and returns MDBX_RESULT_TRUE) if possible,
- * or to the "used" (and returns MDBX_RESULT_FALSE).
- *
- * The mdbx_lck_downgrade() moves the locking-FSM from "exclusive write"
- * state to the "used" (i.e. shared) state.
- *
- * The mdbx_lck_upgrade() moves the locking-FSM from "used" (i.e. shared)
- * state to the "exclusive write" state.
- */
-
-static void lck_unlock(MDBX_env *env) {
- int err;
-
- if (env->me_lfd != INVALID_HANDLE_VALUE) {
- /* double `unlock` for robustly remove overlapped shared/exclusive locks */
- while (funlock(env->me_lfd, LCK_LOWER))
- ;
- err = (int)GetLastError();
- assert(err == ERROR_NOT_LOCKED ||
- (mdbx_RunningUnderWine() && err == ERROR_LOCK_VIOLATION));
- (void)err;
- SetLastError(ERROR_SUCCESS);
-
- while (funlock(env->me_lfd, LCK_UPPER))
- ;
- err = (int)GetLastError();
- assert(err == ERROR_NOT_LOCKED ||
- (mdbx_RunningUnderWine() && err == ERROR_LOCK_VIOLATION));
- (void)err;
- SetLastError(ERROR_SUCCESS);
- }
-
- if (env->me_lazy_fd != INVALID_HANDLE_VALUE) {
- /* explicitly unlock to avoid latency for other processes (windows kernel
- * releases such locks via deferred queues) */
- while (funlock(env->me_lazy_fd, LCK_BODY))
- ;
- err = (int)GetLastError();
- assert(err == ERROR_NOT_LOCKED ||
- (mdbx_RunningUnderWine() && err == ERROR_LOCK_VIOLATION));
- (void)err;
- SetLastError(ERROR_SUCCESS);
-
- while (funlock(env->me_lazy_fd, LCK_WHOLE))
- ;
- err = (int)GetLastError();
- assert(err == ERROR_NOT_LOCKED ||
- (mdbx_RunningUnderWine() && err == ERROR_LOCK_VIOLATION));
- (void)err;
- SetLastError(ERROR_SUCCESS);
- }
-}
-
-/* Seize state as 'exclusive-write' (E-E and returns MDBX_RESULT_TRUE)
- * or as 'used' (S-? and returns MDBX_RESULT_FALSE).
- * Otherwise returns an error. */
-static int internal_seize_lck(HANDLE lfd) {
- int rc;
- assert(lfd != INVALID_HANDLE_VALUE);
-
- /* 1) now on ?-? (free), get ?-E (middle) */
- mdbx_jitter4testing(false);
- if (!flock(lfd, LCK_EXCLUSIVE | LCK_WAITFOR, LCK_UPPER)) {
- rc = (int)GetLastError() /* 2) something went wrong, give up */;
- mdbx_error("%s, err %u", "?-?(free) >> ?-E(middle)", rc);
- return rc;
- }
-
- /* 3) now on ?-E (middle), try E-E (exclusive-write) */
- mdbx_jitter4testing(false);
- if (flock(lfd, LCK_EXCLUSIVE | LCK_DONTWAIT, LCK_LOWER))
- return MDBX_RESULT_TRUE /* 4) got E-E (exclusive-write), done */;
-
- /* 5) still on ?-E (middle) */
- rc = (int)GetLastError();
- mdbx_jitter4testing(false);
- if (rc != ERROR_SHARING_VIOLATION && rc != ERROR_LOCK_VIOLATION) {
- /* 6) something went wrong, give up */
- if (!funlock(lfd, LCK_UPPER))
- mdbx_panic("%s(%s) failed: err %u", __func__, "?-E(middle) >> ?-?(free)",
- (int)GetLastError());
- return rc;
- }
-
- /* 7) still on ?-E (middle), try S-E (locked) */
- mdbx_jitter4testing(false);
- rc = flock(lfd, LCK_SHARED | LCK_DONTWAIT, LCK_LOWER) ? MDBX_RESULT_FALSE
- : (int)GetLastError();
-
- mdbx_jitter4testing(false);
- if (rc != MDBX_RESULT_FALSE)
- mdbx_error("%s, err %u", "?-E(middle) >> S-E(locked)", rc);
-
- /* 8) now on S-E (locked) or still on ?-E (middle),
- * transition to S-? (used) or ?-? (free) */
- if (!funlock(lfd, LCK_UPPER))
- mdbx_panic("%s(%s) failed: err %u", __func__,
- "X-E(locked/middle) >> X-?(used/free)", (int)GetLastError());
-
- /* 9) now on S-? (used, DONE) or ?-? (free, FAILURE) */
- return rc;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env) {
- int rc;
-
- assert(env->me_lazy_fd != INVALID_HANDLE_VALUE);
- if (env->me_flags & MDBX_EXCLUSIVE)
- return MDBX_RESULT_TRUE /* nope since files were must be opened
- non-shareable */
- ;
-
- if (env->me_lfd == INVALID_HANDLE_VALUE) {
- /* LY: without-lck mode (e.g. on read-only filesystem) */
- mdbx_jitter4testing(false);
- if (!flock(env->me_lazy_fd, LCK_SHARED | LCK_DONTWAIT, LCK_WHOLE)) {
- rc = (int)GetLastError();
- mdbx_error("%s, err %u", "without-lck", rc);
- return rc;
- }
- return MDBX_RESULT_FALSE;
- }
-
- rc = internal_seize_lck(env->me_lfd);
- mdbx_jitter4testing(false);
- if (rc == MDBX_RESULT_TRUE && (env->me_flags & MDBX_RDONLY) == 0) {
- /* Check that another process don't operates in without-lck mode.
- * Doing such check by exclusive locking the body-part of db. Should be
- * noted:
- * - we need an exclusive lock for do so;
- * - we can't lock meta-pages, otherwise other process could get an error
- * while opening db in valid (non-conflict) mode. */
- if (!flock(env->me_lazy_fd, LCK_EXCLUSIVE | LCK_DONTWAIT, LCK_BODY)) {
- rc = (int)GetLastError();
- mdbx_error("%s, err %u", "lock-against-without-lck", rc);
- mdbx_jitter4testing(false);
- lck_unlock(env);
- } else {
- mdbx_jitter4testing(false);
- if (!funlock(env->me_lazy_fd, LCK_BODY))
- mdbx_panic("%s(%s) failed: err %u", __func__,
- "unlock-against-without-lck", (int)GetLastError());
- }
- }
-
- return rc;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env) {
- /* Transite from exclusive-write state (E-E) to used (S-?) */
- assert(env->me_lazy_fd != INVALID_HANDLE_VALUE);
- assert(env->me_lfd != INVALID_HANDLE_VALUE);
-
- if (env->me_flags & MDBX_EXCLUSIVE)
- return MDBX_SUCCESS /* nope since files were must be opened non-shareable */
- ;
- /* 1) now at E-E (exclusive-write), transition to ?_E (middle) */
- if (!funlock(env->me_lfd, LCK_LOWER))
- mdbx_panic("%s(%s) failed: err %u", __func__,
- "E-E(exclusive-write) >> ?-E(middle)", (int)GetLastError());
-
- /* 2) now at ?-E (middle), transition to S-E (locked) */
- if (!flock(env->me_lfd, LCK_SHARED | LCK_DONTWAIT, LCK_LOWER)) {
- int rc = (int)GetLastError() /* 3) something went wrong, give up */;
- mdbx_error("%s, err %u", "?-E(middle) >> S-E(locked)", rc);
- return rc;
- }
-
- /* 4) got S-E (locked), continue transition to S-? (used) */
- if (!funlock(env->me_lfd, LCK_UPPER))
- mdbx_panic("%s(%s) failed: err %u", __func__, "S-E(locked) >> S-?(used)",
- (int)GetLastError());
-
- return MDBX_SUCCESS /* 5) now at S-? (used), done */;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_lck_upgrade(MDBX_env *env) {
- /* Transite from used state (S-?) to exclusive-write (E-E) */
- assert(env->me_lfd != INVALID_HANDLE_VALUE);
-
- if (env->me_flags & MDBX_EXCLUSIVE)
- return MDBX_SUCCESS /* nope since files were must be opened non-shareable */
- ;
-
- int rc;
- /* 1) now on S-? (used), try S-E (locked) */
- mdbx_jitter4testing(false);
- if (!flock(env->me_lfd, LCK_EXCLUSIVE | LCK_DONTWAIT, LCK_UPPER)) {
- rc = (int)GetLastError() /* 2) something went wrong, give up */;
- mdbx_verbose("%s, err %u", "S-?(used) >> S-E(locked)", rc);
- return rc;
- }
-
- /* 3) now on S-E (locked), transition to ?-E (middle) */
- if (!funlock(env->me_lfd, LCK_LOWER))
- mdbx_panic("%s(%s) failed: err %u", __func__, "S-E(locked) >> ?-E(middle)",
- (int)GetLastError());
-
- /* 4) now on ?-E (middle), try E-E (exclusive-write) */
- mdbx_jitter4testing(false);
- if (!flock(env->me_lfd, LCK_EXCLUSIVE | LCK_DONTWAIT, LCK_LOWER)) {
- rc = (int)GetLastError() /* 5) something went wrong, give up */;
- mdbx_verbose("%s, err %u", "?-E(middle) >> E-E(exclusive-write)", rc);
- return rc;
- }
-
- return MDBX_SUCCESS /* 6) now at E-E (exclusive-write), done */;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag) {
- (void)env;
- (void)inprocess_neighbor;
- (void)global_uniqueness_flag;
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor) {
- /* LY: should unmap before releasing the locks to avoid race condition and
- * STATUS_USER_MAPPED_FILE/ERROR_USER_MAPPED_FILE */
- if (env->me_map)
- mdbx_munmap(&env->me_dxb_mmap);
- if (env->me_lck_mmap.lck) {
- const bool synced = env->me_lck_mmap.lck->mti_unsynced_pages.weak == 0;
- mdbx_munmap(&env->me_lck_mmap);
- if (synced && !inprocess_neighbor && env->me_lfd != INVALID_HANDLE_VALUE &&
- mdbx_lck_upgrade(env) == MDBX_SUCCESS)
- /* this will fail if LCK is used/mmapped by other process(es) */
- mdbx_ftruncate(env->me_lfd, 0);
- }
- lck_unlock(env);
- return MDBX_SUCCESS;
-}
-
-/*----------------------------------------------------------------------------*/
-/* reader checking (by pid) */
-
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env) {
- (void)env;
- return MDBX_SUCCESS;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env) {
- (void)env;
- return MDBX_SUCCESS;
-}
-
-/* Checks reader by pid.
- *
- * Returns:
- * MDBX_RESULT_TRUE, if pid is live (unable to acquire lock)
- * MDBX_RESULT_FALSE, if pid is dead (lock acquired)
- * or otherwise the errcode. */
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid) {
- (void)env;
- HANDLE hProcess = OpenProcess(SYNCHRONIZE, FALSE, pid);
- int rc;
- if (likely(hProcess)) {
- rc = WaitForSingleObject(hProcess, 0);
- if (unlikely(rc == (int)WAIT_FAILED))
- rc = (int)GetLastError();
- CloseHandle(hProcess);
- } else {
- rc = (int)GetLastError();
- }
-
- switch (rc) {
- case ERROR_INVALID_PARAMETER:
- /* pid seems invalid */
- return MDBX_RESULT_FALSE;
- case WAIT_OBJECT_0:
- /* process just exited */
- return MDBX_RESULT_FALSE;
- case ERROR_ACCESS_DENIED:
- /* The ERROR_ACCESS_DENIED would be returned for CSRSS-processes, etc.
- * assume pid exists */
- return MDBX_RESULT_TRUE;
- case WAIT_TIMEOUT:
- /* pid running */
- return MDBX_RESULT_TRUE;
- default:
- /* failure */
- return rc;
- }
-}
-
-//----------------------------------------------------------------------------
-// Stub for slim read-write lock
-// Copyright (C) 1995-2002 Brad Wilson
-
-static void WINAPI stub_srwlock_Init(MDBX_srwlock *srwl) {
- srwl->readerCount = srwl->writerCount = 0;
-}
-
-static void WINAPI stub_srwlock_AcquireShared(MDBX_srwlock *srwl) {
- while (true) {
- assert(srwl->writerCount >= 0 && srwl->readerCount >= 0);
-
- // If there's a writer already, spin without unnecessarily
- // interlocking the CPUs
- if (srwl->writerCount != 0) {
- YieldProcessor();
- continue;
- }
-
- // Add to the readers list
- _InterlockedIncrement(&srwl->readerCount);
-
- // Check for writers again (we may have been preempted). If
- // there are no writers writing or waiting, then we're done.
- if (srwl->writerCount == 0)
- break;
-
- // Remove from the readers list, spin, try again
- _InterlockedDecrement(&srwl->readerCount);
- YieldProcessor();
- }
-}
-
-static void WINAPI stub_srwlock_ReleaseShared(MDBX_srwlock *srwl) {
- assert(srwl->readerCount > 0);
- _InterlockedDecrement(&srwl->readerCount);
-}
-
-static void WINAPI stub_srwlock_AcquireExclusive(MDBX_srwlock *srwl) {
- while (true) {
- assert(srwl->writerCount >= 0 && srwl->readerCount >= 0);
-
- // If there's a writer already, spin without unnecessarily
- // interlocking the CPUs
- if (srwl->writerCount != 0) {
- YieldProcessor();
- continue;
- }
-
- // See if we can become the writer (expensive, because it inter-
- // locks the CPUs, so writing should be an infrequent process)
- if (_InterlockedExchange(&srwl->writerCount, 1) == 0)
- break;
- }
-
- // Now we're the writer, but there may be outstanding readers.
- // Spin until there aren't any more; new readers will wait now
- // that we're the writer.
- while (srwl->readerCount != 0) {
- assert(srwl->writerCount >= 0 && srwl->readerCount >= 0);
- YieldProcessor();
- }
-}
-
-static void WINAPI stub_srwlock_ReleaseExclusive(MDBX_srwlock *srwl) {
- assert(srwl->writerCount == 1 && srwl->readerCount >= 0);
- srwl->writerCount = 0;
-}
-
-static uint64_t WINAPI stub_GetTickCount64(void) {
- LARGE_INTEGER Counter, Frequency;
- return (QueryPerformanceFrequency(&Frequency) &&
- QueryPerformanceCounter(&Counter))
- ? Counter.QuadPart * 1000ul / Frequency.QuadPart
- : 0;
-}
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef xMDBX_ALLOY
-MDBX_srwlock_function mdbx_srwlock_Init, mdbx_srwlock_AcquireShared,
- mdbx_srwlock_ReleaseShared, mdbx_srwlock_AcquireExclusive,
- mdbx_srwlock_ReleaseExclusive;
-
-MDBX_NtExtendSection mdbx_NtExtendSection;
-MDBX_GetFileInformationByHandleEx mdbx_GetFileInformationByHandleEx;
-MDBX_GetVolumeInformationByHandleW mdbx_GetVolumeInformationByHandleW;
-MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-MDBX_SetFileInformationByHandle mdbx_SetFileInformationByHandle;
-MDBX_NtFsControlFile mdbx_NtFsControlFile;
-MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-MDBX_GetTickCount64 mdbx_GetTickCount64;
-MDBX_RegGetValueA mdbx_RegGetValueA;
-#endif /* xMDBX_ALLOY */
-
-#if __GNUC_PREREQ(8, 0)
-#pragma GCC diagnostic push
-#pragma GCC diagnostic ignored "-Wcast-function-type"
-#endif /* GCC/MINGW */
-
-static void mdbx_winnt_import(void) {
- const HINSTANCE hNtdll = GetModuleHandleA("ntdll.dll");
-
-#define GET_PROC_ADDR(dll, ENTRY) \
- mdbx_##ENTRY = (MDBX_##ENTRY)GetProcAddress(dll, #ENTRY)
-
- if (GetProcAddress(hNtdll, "wine_get_version")) {
- assert(mdbx_RunningUnderWine());
- } else {
- GET_PROC_ADDR(hNtdll, NtFsControlFile);
- GET_PROC_ADDR(hNtdll, NtExtendSection);
- assert(!mdbx_RunningUnderWine());
- }
-
- const HINSTANCE hKernel32dll = GetModuleHandleA("kernel32.dll");
- GET_PROC_ADDR(hKernel32dll, GetFileInformationByHandleEx);
- GET_PROC_ADDR(hKernel32dll, GetTickCount64);
- if (!mdbx_GetTickCount64)
- mdbx_GetTickCount64 = stub_GetTickCount64;
- if (!mdbx_RunningUnderWine()) {
- GET_PROC_ADDR(hKernel32dll, SetFileInformationByHandle);
- GET_PROC_ADDR(hKernel32dll, GetVolumeInformationByHandleW);
- GET_PROC_ADDR(hKernel32dll, GetFinalPathNameByHandleW);
- GET_PROC_ADDR(hKernel32dll, PrefetchVirtualMemory);
- }
-
- const HINSTANCE hAdvapi32dll = GetModuleHandleA("advapi32.dll");
- GET_PROC_ADDR(hAdvapi32dll, RegGetValueA);
-#undef GET_PROC_ADDR
-
- const MDBX_srwlock_function init =
- (MDBX_srwlock_function)GetProcAddress(hKernel32dll, "InitializeSRWLock");
- if (init != NULL) {
- mdbx_srwlock_Init = init;
- mdbx_srwlock_AcquireShared = (MDBX_srwlock_function)GetProcAddress(
- hKernel32dll, "AcquireSRWLockShared");
- mdbx_srwlock_ReleaseShared = (MDBX_srwlock_function)GetProcAddress(
- hKernel32dll, "ReleaseSRWLockShared");
- mdbx_srwlock_AcquireExclusive = (MDBX_srwlock_function)GetProcAddress(
- hKernel32dll, "AcquireSRWLockExclusive");
- mdbx_srwlock_ReleaseExclusive = (MDBX_srwlock_function)GetProcAddress(
- hKernel32dll, "ReleaseSRWLockExclusive");
- } else {
- mdbx_srwlock_Init = stub_srwlock_Init;
- mdbx_srwlock_AcquireShared = stub_srwlock_AcquireShared;
- mdbx_srwlock_ReleaseShared = stub_srwlock_ReleaseShared;
- mdbx_srwlock_AcquireExclusive = stub_srwlock_AcquireExclusive;
- mdbx_srwlock_ReleaseExclusive = stub_srwlock_ReleaseExclusive;
- }
-}
-
-#if __GNUC_PREREQ(8, 0)
-#pragma GCC diagnostic pop
-#endif /* GCC/MINGW */
-
-#endif /* Windows LCK-implementation */
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-#if !(defined(_WIN32) || defined(_WIN64)) /* !Windows LCK-implementation */
-
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
-#include <sys/sem.h>
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
-/*----------------------------------------------------------------------------*/
-/* global constructor/destructor */
-
-#if defined(__linux__) || defined(__gnu_linux__)
-
-#include <sys/utsname.h>
-
-#ifndef xMDBX_ALLOY
-uint32_t mdbx_linux_kernel_version;
-bool mdbx_RunningOnWSL1;
-#endif /* xMDBX_ALLOY */
-
-__cold static uint8_t probe_for_WSL(const char *tag) {
- const char *const WSL = strstr(tag, "WSL");
- if (WSL && WSL[3] >= '2' && WSL[3] <= '9')
- return WSL[3] - '0';
- const char *const wsl = strstr(tag, "wsl");
- if (wsl && wsl[3] >= '2' && wsl[3] <= '9')
- return wsl[3] - '0';
- if (WSL || wsl || strcasestr(tag, "Microsoft"))
- /* Expecting no new kernel within WSL1, either it will explicitly
- * marked by an appropriate WSL-version hint. */
- return (mdbx_linux_kernel_version < /* 4.19.x */ 0x04130000) ? 1 : 2;
- return 0;
-}
-
-#endif /* Linux */
-
-__cold static __attribute__((__constructor__)) void
-mdbx_global_constructor(void) {
-#if defined(__linux__) || defined(__gnu_linux__)
- struct utsname buffer;
- if (uname(&buffer) == 0) {
- int i = 0;
- char *p = buffer.release;
- while (*p && i < 4) {
- if (*p >= '0' && *p <= '9') {
- long number = strtol(p, &p, 10);
- if (number > 0) {
- if (number > 255)
- number = 255;
- mdbx_linux_kernel_version += number << (24 - i * 8);
- }
- ++i;
- } else {
- ++p;
- }
- }
- /* "Official" way of detecting WSL1 but not WSL2
- * https://github.com/Microsoft/WSL/issues/423#issuecomment-221627364
- *
- * WARNING: False negative detection of WSL1 will result in DATA LOSS!
- * So, the REQUIREMENTS for this code:
- * 1. MUST detect WSL1 without false-negatives.
- * 2. DESIRABLE detect WSL2 but without the risk of violating the first. */
- mdbx_RunningOnWSL1 = probe_for_WSL(buffer.version) == 1 ||
- probe_for_WSL(buffer.sysname) == 1 ||
- probe_for_WSL(buffer.release) == 1;
- }
-#endif /* Linux */
-
- mdbx_rthc_global_init();
-}
-
-__cold static __attribute__((__destructor__)) void
-mdbx_global_destructor(void) {
- mdbx_rthc_global_dtor();
-}
-
-/*----------------------------------------------------------------------------*/
-/* lck */
-
-/* ОпиÑание реализации блокировок Ð´Ð»Ñ POSIX & Linux:
- *
- * lck-файл отображаетÑÑ Ð² памÑÑ‚ÑŒ, в нём организуетÑÑ Ñ‚Ð°Ð±Ð»Ð¸Ñ†Ð° читателей и
- * размещаютÑÑ ÑовмеÑтно иÑпользуемые posix-мьютекÑÑ‹ (futex). ПоÑредÑтвом
- * Ñтих мьютекÑов (Ñм struct MDBX_lockinfo) реализуютÑÑ:
- * - Блокировка таблицы читателей Ð´Ð»Ñ Ñ€ÐµÐ³Ð¸Ñтрации,
- * т.е. функции mdbx_rdt_lock() и mdbx_rdt_unlock().
- * - Блокировка БД Ð´Ð»Ñ Ð¿Ð¸ÑˆÑƒÑ‰Ð¸Ñ… транзакций,
- * т.е. функции mdbx_txn_lock() и mdbx_txn_unlock().
- *
- * ОÑтальной функционал реализуетÑÑ Ð¾Ñ‚Ð´ÐµÐ»ÑŒÐ½Ð¾ поÑредÑтвом файловых блокировок:
- * - Первоначальный захват БД в режиме exclusive/shared и поÑледующий перевод
- * в операционный режим, функции mdbx_lck_seize() и mdbx_lck_downgrade().
- * - Проверка приÑутÑтвие процеÑÑов-читателей,
- * т.е. функции mdbx_rpid_set(), mdbx_rpid_clear() и mdbx_rpid_check().
- *
- * Ð”Ð»Ñ Ð±Ð»Ð¾ÐºÐ¸Ñ€Ð¾Ð²ÐºÐ¸ файлов иÑпользуетÑÑ fcntl(F_SETLK), так как:
- * - lockf() оперирует только ÑкÑклюзивной блокировкой и требует
- * Ð¾Ñ‚ÐºÑ€Ñ‹Ñ‚Ð¸Ñ Ñ„Ð°Ð¹Ð»Ð° в RW-режиме.
- * - flock() не гарантирует атомарноÑти при Ñмене блокировок
- * и оперирует только вÑем файлом целиком.
- * - Ð”Ð»Ñ ÐºÐ¾Ð½Ñ‚Ñ€Ð¾Ð»Ñ Ð¿Ñ€Ð¾Ñ†ÐµÑÑов-читателей иÑпользуютÑÑ Ð¾Ð´Ð½Ð¾Ð±Ð°Ð¹Ñ‚Ð¾Ð²Ñ‹Ðµ
- * range-блокировки lck-файла поÑредÑтвом fcntl(F_SETLK). При Ñтом
- * в качеÑтве позиции иÑпользуетÑÑ pid процеÑÑа-читателÑ.
- * - Ð”Ð»Ñ Ð¿ÐµÑ€Ð²Ð¾Ð½Ð°Ñ‡Ð°Ð»ÑŒÐ½Ð¾Ð³Ð¾ захвата и shared/exclusive выполнÑетÑÑ Ð±Ð»Ð¾ÐºÐ¸Ñ€Ð¾Ð²ÐºÐ°
- * оÑновного файла БД и при уÑпехе lck-файла.
- *
- * ----------------------------------------------------------------------------
- * УДЕРЖИВÐЕМЫЕ БЛОКИРОВКИ Ð’ ЗÐВИСИМОСТИ ОТ РЕЖИМРИ СОСТОЯÐИЯ
- *
- * ÐкÑклюзивный режим без lck-файла:
- * = заблокирован веÑÑŒ dxb-файл поÑредÑтвом F_RDLCK или F_WRLCK,
- * в завиÑимоÑти от MDBX_RDONLY.
- *
- * Ðе-операционный режим на Ð²Ñ€ÐµÐ¼Ñ Ð¿ÐµÑ€Ðµ-инициализации и разрушении lck-файла:
- * = F_WRLCK блокировка первого байта lck-файла, другие процеÑÑÑ‹ ждут её
- * ÑнÑÑ‚Ð¸Ñ Ð¿Ñ€Ð¸ получении F_RDLCK через F_SETLKW.
- * - блокировки dxb-файла могут менÑÑ‚ÑŒÑÑ Ð´Ð¾ ÑнÑтие ÑкÑклюзивной блокировки
- * lck-файла:
- * + Ð´Ð»Ñ ÐЕ-ÑкÑклюзивного режима блокировка pid-байта в dxb-файле
- * поÑредÑтвом F_RDLCK или F_WRLCK, в завиÑимоÑти от MDBX_RDONLY.
- * + Ð´Ð»Ñ ÐКСКЛЮЗИВÐОГО режима блокировка pid-байта вÑего dxb-файла
- * поÑредÑтвом F_RDLCK или F_WRLCK, в завиÑимоÑти от MDBX_RDONLY.
- *
- * ОПЕРÐЦИОÐÐЫЙ режим Ñ lck-файлом:
- * = F_RDLCK блокировка первого байта lck-файла, другие процеÑÑÑ‹ не могут
- * получить F_WRLCK и таким образом видÑÑ‚ что БД иÑпользуетÑÑ.
- * + F_WRLCK блокировка pid-байта в clk-файле поÑле первой транзакции чтениÑ.
- * + Ð´Ð»Ñ ÐЕ-ÑкÑклюзивного режима блокировка pid-байта в dxb-файле
- * поÑредÑтвом F_RDLCK или F_WRLCK, в завиÑимоÑти от MDBX_RDONLY.
- * + Ð´Ð»Ñ ÐКСКЛЮЗИВÐОГО режима блокировка pid-байта вÑего dxb-файла
- * поÑредÑтвом F_RDLCK или F_WRLCK, в завиÑимоÑти от MDBX_RDONLY.
- */
-
-#if MDBX_USE_OFDLOCKS
-static int op_setlk, op_setlkw, op_getlk;
-__cold static void choice_fcntl() {
- assert(!op_setlk && !op_setlkw && !op_getlk);
- if ((mdbx_runtime_flags & MDBX_DBG_LEGACY_MULTIOPEN) == 0
-#if defined(__linux__) || defined(__gnu_linux__)
- && mdbx_linux_kernel_version >
- 0x030f0000 /* OFD locks are available since 3.15, but engages here
- only for 3.16 and later kernels (i.e. LTS) because
- of reliability reasons */
-#endif /* linux */
- ) {
- op_setlk = F_OFD_SETLK;
- op_setlkw = F_OFD_SETLKW;
- op_getlk = F_OFD_GETLK;
- return;
- }
- op_setlk = F_SETLK;
- op_setlkw = F_SETLKW;
- op_getlk = F_GETLK;
-}
-#else
-#define op_setlk F_SETLK
-#define op_setlkw F_SETLKW
-#define op_getlk F_GETLK
-#endif /* MDBX_USE_OFDLOCKS */
-
-#ifndef OFF_T_MAX
-#define OFF_T_MAX \
- (((sizeof(off_t) > 4) ? INT64_MAX : INT32_MAX) & ~(size_t)0xffff)
-#endif
-
-static int lck_op(const mdbx_filehandle_t fd, int cmd, const int lck,
- const off_t offset, off_t len) {
- STATIC_ASSERT(sizeof(off_t) >= sizeof(void *) &&
- sizeof(off_t) >= sizeof(size_t));
-#ifdef __ANDROID_API__
- STATIC_ASSERT_MSG((sizeof(off_t) * 8 == MDBX_WORDBITS),
- "The bitness of system `off_t` type is mismatch. Please "
- "fix build and/or NDK configuration.");
-#endif /* Android */
- mdbx_jitter4testing(true);
- assert(offset >= 0 && len > 0);
- assert((uint64_t)offset < (uint64_t)INT64_MAX &&
- (uint64_t)len < (uint64_t)INT64_MAX &&
- (uint64_t)(offset + len) > (uint64_t)offset);
-
- assert((uint64_t)offset < (uint64_t)OFF_T_MAX &&
- (uint64_t)len <= (uint64_t)OFF_T_MAX &&
- (uint64_t)(offset + len) <= (uint64_t)OFF_T_MAX);
-
- assert((uint64_t)((off_t)((uint64_t)offset + (uint64_t)len)) ==
- ((uint64_t)offset + (uint64_t)len));
- for (;;) {
- struct flock lock_op;
- STATIC_ASSERT(sizeof(off_t) <= sizeof(lock_op.l_start) &&
- sizeof(off_t) <= sizeof(lock_op.l_len) &&
- OFF_T_MAX == (off_t)OFF_T_MAX);
- memset(&lock_op, 0, sizeof(lock_op));
- lock_op.l_type = lck;
- lock_op.l_whence = SEEK_SET;
- lock_op.l_start = offset;
- lock_op.l_len = len;
- int rc = fcntl(fd, cmd, &lock_op);
- mdbx_jitter4testing(true);
- if (rc != -1) {
- if (cmd == op_getlk) {
- /* Checks reader by pid. Returns:
- * MDBX_RESULT_TRUE - if pid is live (reader holds a lock).
- * MDBX_RESULT_FALSE - if pid is dead (a lock could be placed). */
- return (lock_op.l_type == F_UNLCK) ? MDBX_RESULT_FALSE
- : MDBX_RESULT_TRUE;
- }
- return MDBX_SUCCESS;
- }
- rc = errno;
-#if MDBX_USE_OFDLOCKS
- if (rc == EINVAL &&
- (cmd == F_OFD_SETLK || cmd == F_OFD_SETLKW || cmd == F_OFD_GETLK)) {
- /* fallback to non-OFD locks */
- if (cmd == F_OFD_SETLK)
- cmd = F_SETLK;
- else if (cmd == F_OFD_SETLKW)
- cmd = F_SETLKW;
- else
- cmd = F_GETLK;
- op_setlk = F_SETLK;
- op_setlkw = F_SETLKW;
- op_getlk = F_GETLK;
- continue;
- }
-#endif /* MDBX_USE_OFDLOCKS */
- if (rc != EINTR || cmd == op_setlkw) {
- assert(MDBX_IS_ERROR(rc));
- return rc;
- }
- }
-}
-
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait) {
-#if MDBX_USE_OFDLOCKS
- if (unlikely(op_setlk == 0))
- choice_fcntl();
-#endif /* MDBX_USE_OFDLOCKS */
- return lck_op(fd, wait ? op_setlkw : op_setlk, F_WRLCK, 0, OFF_T_MAX);
-}
-
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env) {
- assert(env->me_lfd != INVALID_HANDLE_VALUE);
- assert(env->me_pid > 0);
- if (unlikely(mdbx_getpid() != env->me_pid))
- return MDBX_PANIC;
- return lck_op(env->me_lfd, op_setlk, F_WRLCK, env->me_pid, 1);
-}
-
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env) {
- assert(env->me_lfd != INVALID_HANDLE_VALUE);
- assert(env->me_pid > 0);
- return lck_op(env->me_lfd, op_setlk, F_UNLCK, env->me_pid, 1);
-}
-
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid) {
- assert(env->me_lfd != INVALID_HANDLE_VALUE);
- assert(pid > 0);
- return lck_op(env->me_lfd, op_getlk, F_WRLCK, pid, 1);
-}
-
-/*---------------------------------------------------------------------------*/
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc) {
-#if MDBX_LOCKING == MDBX_LOCKING_POSIX1988
- return sem_init(ipc, false, 1) ? errno : 0;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
- return pthread_mutex_init(ipc, nullptr);
-#else
-#error "FIXME"
-#endif
-}
-
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc) {
-#if MDBX_LOCKING == MDBX_LOCKING_POSIX1988
- return sem_destroy(ipc) ? errno : 0;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
- return pthread_mutex_destroy(ipc);
-#else
-#error "FIXME"
-#endif
-}
-#endif /* MDBX_LOCKING > MDBX_LOCKING_SYSV */
-
-static int check_fstat(MDBX_env *env) {
- struct stat st;
-
- int rc = MDBX_SUCCESS;
- if (fstat(env->me_lazy_fd, &st)) {
- rc = errno;
- mdbx_error("fstat(%s), err %d", "DXB", rc);
- return rc;
- }
-
- if (!S_ISREG(st.st_mode) || st.st_nlink < 1) {
-#ifdef EBADFD
- rc = EBADFD;
-#else
- rc = EPERM;
-#endif
- mdbx_error("%s %s, err %d", "DXB",
- (st.st_nlink < 1) ? "file was removed" : "not a regular file",
- rc);
- return rc;
- }
-
- if (st.st_size < (off_t)(MDBX_MIN_PAGESIZE * NUM_METAS)) {
- mdbx_verbose("dxb-file is too short (%u), exclusive-lock needed",
- (unsigned)st.st_size);
- rc = MDBX_RESULT_TRUE;
- }
-
- //----------------------------------------------------------------------------
-
- if (fstat(env->me_lfd, &st)) {
- rc = errno;
- mdbx_error("fstat(%s), err %d", "LCK", rc);
- return rc;
- }
-
- if (!S_ISREG(st.st_mode) || st.st_nlink < 1) {
-#ifdef EBADFD
- rc = EBADFD;
-#else
- rc = EPERM;
-#endif
- mdbx_error("%s %s, err %d", "LCK",
- (st.st_nlink < 1) ? "file was removed" : "not a regular file",
- rc);
- return rc;
- }
-
- /* Checking file size for detect the situation when we got the shared lock
- * immediately after mdbx_lck_destroy(). */
- if (st.st_size < (off_t)(sizeof(MDBX_lockinfo) + sizeof(MDBX_reader))) {
- mdbx_verbose("lck-file is too short (%u), exclusive-lock needed",
- (unsigned)st.st_size);
- rc = MDBX_RESULT_TRUE;
- }
-
- return rc;
-}
-
-__cold MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env) {
- assert(env->me_lazy_fd != INVALID_HANDLE_VALUE);
- if (unlikely(mdbx_getpid() != env->me_pid))
- return MDBX_PANIC;
-#if MDBX_USE_OFDLOCKS
- if (unlikely(op_setlk == 0))
- choice_fcntl();
-#endif /* MDBX_USE_OFDLOCKS */
-
- int rc = MDBX_SUCCESS;
-#if defined(__linux__) || defined(__gnu_linux__)
- if (unlikely(mdbx_RunningOnWSL1)) {
- rc = ENOLCK /* No record locks available */;
- mdbx_error("%s, err %u",
- "WSL1 (Windows Subsystem for Linux) is mad and trouble-full, "
- "injecting failure to avoid data loss",
- rc);
- return rc;
- }
-#endif /* Linux */
-
- if (env->me_lfd == INVALID_HANDLE_VALUE) {
- /* LY: without-lck mode (e.g. exclusive or on read-only filesystem) */
- rc =
- lck_op(env->me_lazy_fd, op_setlk,
- (env->me_flags & MDBX_RDONLY) ? F_RDLCK : F_WRLCK, 0, OFF_T_MAX);
- if (rc != MDBX_SUCCESS) {
- mdbx_error("%s, err %u", "without-lck", rc);
- mdbx_assert(env, MDBX_IS_ERROR(rc));
- return rc;
- }
- return MDBX_RESULT_TRUE /* Done: return with exclusive locking. */;
- }
-#if defined(_POSIX_PRIORITY_SCHEDULING) && _POSIX_PRIORITY_SCHEDULING > 0
- sched_yield();
-#endif
-
-retry:
- if (rc == MDBX_RESULT_TRUE) {
- rc = lck_op(env->me_lfd, op_setlk, F_UNLCK, 0, 1);
- if (rc != MDBX_SUCCESS) {
- mdbx_error("%s, err %u", "unlock-before-retry", rc);
- mdbx_assert(env, MDBX_IS_ERROR(rc));
- return rc;
- }
- }
-
- /* Firstly try to get exclusive locking. */
- rc = lck_op(env->me_lfd, op_setlk, F_WRLCK, 0, 1);
- if (rc == MDBX_SUCCESS) {
- rc = check_fstat(env);
- if (MDBX_IS_ERROR(rc))
- return rc;
-
- continue_dxb_exclusive:
- rc =
- lck_op(env->me_lazy_fd, op_setlk,
- (env->me_flags & MDBX_RDONLY) ? F_RDLCK : F_WRLCK, 0, OFF_T_MAX);
- if (rc == MDBX_SUCCESS)
- return MDBX_RESULT_TRUE /* Done: return with exclusive locking. */;
-
- int err = check_fstat(env);
- if (MDBX_IS_ERROR(err))
- return err;
-
- /* the cause may be a collision with POSIX's file-lock recovery. */
- if (!(rc == EAGAIN || rc == EACCES || rc == EBUSY || rc == EWOULDBLOCK ||
- rc == EDEADLK)) {
- mdbx_error("%s, err %u", "dxb-exclusive", rc);
- mdbx_assert(env, MDBX_IS_ERROR(rc));
- return rc;
- }
-
- /* Fallback to lck-shared */
- } else if (!(rc == EAGAIN || rc == EACCES || rc == EBUSY ||
- rc == EWOULDBLOCK || rc == EDEADLK)) {
- mdbx_error("%s, err %u", "try-exclusive", rc);
- mdbx_assert(env, MDBX_IS_ERROR(rc));
- return rc;
- }
-
- /* Here could be one of two:
- * - mdbx_lck_destroy() from the another process was hold the lock
- * during a destruction.
- * - either mdbx_lck_seize() from the another process was got the exclusive
- * lock and doing initialization.
- * For distinguish these cases will use size of the lck-file later. */
-
- /* Wait for lck-shared now. */
- /* Here may be await during transient processes, for instance until another
- * competing process doesn't call lck_downgrade(). */
- rc = lck_op(env->me_lfd, op_setlkw, F_RDLCK, 0, 1);
- if (rc != MDBX_SUCCESS) {
- mdbx_error("%s, err %u", "try-shared", rc);
- mdbx_assert(env, MDBX_IS_ERROR(rc));
- return rc;
- }
-
- rc = check_fstat(env);
- if (rc == MDBX_RESULT_TRUE)
- goto retry;
- if (rc != MDBX_SUCCESS) {
- mdbx_error("%s, err %u", "lck_fstat", rc);
- return rc;
- }
-
- /* got shared, retry exclusive */
- rc = lck_op(env->me_lfd, op_setlk, F_WRLCK, 0, 1);
- if (rc == MDBX_SUCCESS)
- goto continue_dxb_exclusive;
-
- if (!(rc == EAGAIN || rc == EACCES || rc == EBUSY || rc == EWOULDBLOCK ||
- rc == EDEADLK)) {
- mdbx_error("%s, err %u", "try-exclusive", rc);
- mdbx_assert(env, MDBX_IS_ERROR(rc));
- return rc;
- }
-
- /* Lock against another process operating in without-lck or exclusive mode. */
- rc =
- lck_op(env->me_lazy_fd, op_setlk,
- (env->me_flags & MDBX_RDONLY) ? F_RDLCK : F_WRLCK, env->me_pid, 1);
- if (rc != MDBX_SUCCESS) {
- mdbx_error("%s, err %u", "lock-against-without-lck", rc);
- mdbx_assert(env, MDBX_IS_ERROR(rc));
- return rc;
- }
-
- /* Done: return with shared locking. */
- return MDBX_RESULT_FALSE;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env) {
- assert(env->me_lfd != INVALID_HANDLE_VALUE);
- if (unlikely(mdbx_getpid() != env->me_pid))
- return MDBX_PANIC;
-
- int rc = MDBX_SUCCESS;
- if ((env->me_flags & MDBX_EXCLUSIVE) == 0) {
- rc = lck_op(env->me_lazy_fd, op_setlk, F_UNLCK, 0, env->me_pid);
- if (rc == MDBX_SUCCESS)
- rc = lck_op(env->me_lazy_fd, op_setlk, F_UNLCK, env->me_pid + 1,
- OFF_T_MAX - env->me_pid - 1);
- }
- if (rc == MDBX_SUCCESS)
- rc = lck_op(env->me_lfd, op_setlk, F_RDLCK, 0, 1);
- if (unlikely(rc != 0)) {
- mdbx_error("%s, err %u", "lck", rc);
- assert(MDBX_IS_ERROR(rc));
- }
- return rc;
-}
-
-__cold MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor) {
- if (unlikely(mdbx_getpid() != env->me_pid))
- return MDBX_PANIC;
-
- int rc = MDBX_SUCCESS;
- struct stat lck_info;
- MDBX_lockinfo *lck = env->me_lck_mmap.lck;
- if (env->me_lfd != INVALID_HANDLE_VALUE && !inprocess_neighbor && lck &&
- /* try get exclusive access */
- lck_op(env->me_lfd, op_setlk, F_WRLCK, 0, OFF_T_MAX) == 0 &&
- /* if LCK was not removed */
- fstat(env->me_lfd, &lck_info) == 0 && lck_info.st_nlink > 0 &&
- lck_op(env->me_lazy_fd, op_setlk,
- (env->me_flags & MDBX_RDONLY) ? F_RDLCK : F_WRLCK, 0,
- OFF_T_MAX) == 0) {
-
- mdbx_verbose("%p got exclusive, drown locks", (void *)env);
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- if (env->me_sysv_ipc.semid != -1)
- rc = semctl(env->me_sysv_ipc.semid, 2, IPC_RMID) ? errno : 0;
-#else
- rc = mdbx_ipclock_destroy(&lck->mti_rlock);
- if (rc == 0)
- rc = mdbx_ipclock_destroy(&lck->mti_wlock);
-#endif /* MDBX_LOCKING */
-
- mdbx_assert(env, rc == 0);
- if (rc == 0) {
- const bool synced = lck->mti_unsynced_pages.weak == 0;
- mdbx_munmap(&env->me_lck_mmap);
- if (synced)
- rc = ftruncate(env->me_lfd, 0) ? errno : 0;
- }
-
- mdbx_jitter4testing(false);
- }
-
- /* 1) POSIX's fcntl() locks (i.e. when op_setlk == F_SETLK) should be restored
- * after file was closed.
- *
- * 2) File locks would be released (by kernel) while the file-descriptors will
- * be closed. But to avoid false-positive EACCESS and EDEADLK from the kernel,
- * locks should be released here explicitly with properly order. */
-
- /* close dxb and restore lock */
- if (env->me_dsync_fd != INVALID_HANDLE_VALUE) {
- if (unlikely(close(env->me_dsync_fd) != 0) && rc == MDBX_SUCCESS)
- rc = errno;
- env->me_dsync_fd = INVALID_HANDLE_VALUE;
- }
- if (env->me_lazy_fd != INVALID_HANDLE_VALUE) {
- if (unlikely(close(env->me_lazy_fd) != 0) && rc == MDBX_SUCCESS)
- rc = errno;
- env->me_lazy_fd = INVALID_HANDLE_VALUE;
- if (op_setlk == F_SETLK && inprocess_neighbor && rc == MDBX_SUCCESS) {
- /* restore file-lock */
- rc = lck_op(
- inprocess_neighbor->me_lazy_fd, F_SETLKW,
- (inprocess_neighbor->me_flags & MDBX_RDONLY) ? F_RDLCK : F_WRLCK,
- (inprocess_neighbor->me_flags & MDBX_EXCLUSIVE)
- ? 0
- : inprocess_neighbor->me_pid,
- (inprocess_neighbor->me_flags & MDBX_EXCLUSIVE) ? OFF_T_MAX : 1);
- }
- }
-
- /* close clk and restore locks */
- if (env->me_lfd != INVALID_HANDLE_VALUE) {
- if (unlikely(close(env->me_lfd) != 0) && rc == MDBX_SUCCESS)
- rc = errno;
- env->me_lfd = INVALID_HANDLE_VALUE;
- if (op_setlk == F_SETLK && inprocess_neighbor && rc == MDBX_SUCCESS) {
- /* restore file-locks */
- rc = lck_op(inprocess_neighbor->me_lfd, F_SETLKW, F_RDLCK, 0, 1);
- if (rc == MDBX_SUCCESS && inprocess_neighbor->me_live_reader)
- rc = mdbx_rpid_set(inprocess_neighbor);
- }
- }
-
- if (inprocess_neighbor && rc != MDBX_SUCCESS)
- inprocess_neighbor->me_flags |= MDBX_FATAL_ERROR;
- return rc;
-}
-
-/*---------------------------------------------------------------------------*/
-
-__cold MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag) {
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- int semid = -1;
- /* don't initialize semaphores twice */
- (void)inprocess_neighbor;
- if (global_uniqueness_flag == MDBX_RESULT_TRUE) {
- struct stat st;
- if (fstat(env->me_lazy_fd, &st))
- return errno;
- sysv_retry_create:
- semid = semget(env->me_sysv_ipc.key, 2,
- IPC_CREAT | IPC_EXCL |
- (st.st_mode & (S_IRWXU | S_IRWXG | S_IRWXO)));
- if (unlikely(semid == -1)) {
- int err = errno;
- if (err != EEXIST)
- return err;
-
- /* remove and re-create semaphore set */
- semid = semget(env->me_sysv_ipc.key, 2, 0);
- if (semid == -1) {
- err = errno;
- if (err != ENOENT)
- return err;
- goto sysv_retry_create;
- }
- if (semctl(semid, 2, IPC_RMID)) {
- err = errno;
- if (err != EIDRM)
- return err;
- }
- goto sysv_retry_create;
- }
-
- unsigned short val_array[2] = {1, 1};
- if (semctl(semid, 2, SETALL, val_array))
- return errno;
- } else {
- semid = semget(env->me_sysv_ipc.key, 2, 0);
- if (semid == -1)
- return errno;
-
- /* check read & write access */
- struct semid_ds data[2];
- if (semctl(semid, 2, IPC_STAT, data) || semctl(semid, 2, IPC_SET, data))
- return errno;
- }
-
- env->me_sysv_ipc.semid = semid;
- return MDBX_SUCCESS;
-
-#elif MDBX_LOCKING == MDBX_LOCKING_FUTEX
- (void)inprocess_neighbor;
- if (global_uniqueness_flag != MDBX_RESULT_TRUE)
- return MDBX_SUCCESS;
-#error "FIXME: Not implemented"
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-
- /* don't initialize semaphores twice */
- (void)inprocess_neighbor;
- if (global_uniqueness_flag == MDBX_RESULT_TRUE) {
- if (sem_init(&env->me_lck_mmap.lck->mti_rlock, true, 1))
- return errno;
- if (sem_init(&env->me_lck_mmap.lck->mti_wlock, true, 1))
- return errno;
- }
- return MDBX_SUCCESS;
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
- if (inprocess_neighbor)
- return MDBX_SUCCESS /* don't need any initialization for mutexes
- if LCK already opened/used inside current process */
- ;
-
- /* FIXME: Unfortunately, there is no other reliable way but to long testing
- * on each platform. On the other hand, behavior like FreeBSD is incorrect
- * and we can expect it to be rare. Moreover, even on FreeBSD without
- * additional in-process initialization, the probability of an problem
- * occurring is vanishingly small, and the symptom is a return of EINVAL
- * while locking a mutex. In other words, in the worst case, the problem
- * results in an EINVAL error at the start of the transaction, but NOT data
- * loss, nor database corruption, nor other fatal troubles. Thus, the code
- * below I am inclined to think the workaround for erroneous platforms (like
- * FreeBSD), rather than a defect of libmdbx. */
-#if defined(__FreeBSD__)
- /* seems that shared mutexes on FreeBSD required in-process initialization */
- (void)global_uniqueness_flag;
-#else
- /* shared mutexes on many other platforms (including Darwin and Linux's
- * futexes) doesn't need any addition in-process initialization */
- if (global_uniqueness_flag != MDBX_RESULT_TRUE)
- return MDBX_SUCCESS;
-#endif
-
- pthread_mutexattr_t ma;
- int rc = pthread_mutexattr_init(&ma);
- if (rc)
- return rc;
-
- rc = pthread_mutexattr_setpshared(&ma, PTHREAD_PROCESS_SHARED);
- if (rc)
- goto bailout;
-
-#if MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#if defined(PTHREAD_MUTEX_ROBUST) || defined(pthread_mutexattr_setrobust)
- rc = pthread_mutexattr_setrobust(&ma, PTHREAD_MUTEX_ROBUST);
-#elif defined(PTHREAD_MUTEX_ROBUST_NP) || \
- defined(pthread_mutexattr_setrobust_np)
- rc = pthread_mutexattr_setrobust_np(&ma, PTHREAD_MUTEX_ROBUST_NP);
-#elif _POSIX_THREAD_PROCESS_SHARED < 200809L
- rc = pthread_mutexattr_setrobust_np(&ma, PTHREAD_MUTEX_ROBUST_NP);
-#else
- rc = pthread_mutexattr_setrobust(&ma, PTHREAD_MUTEX_ROBUST);
-#endif
- if (rc)
- goto bailout;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_POSIX2008 */
-
-#if defined(_POSIX_THREAD_PRIO_INHERIT) && _POSIX_THREAD_PRIO_INHERIT >= 0 && \
- !defined(MDBX_SAFE4QEMU)
- rc = pthread_mutexattr_setprotocol(&ma, PTHREAD_PRIO_INHERIT);
- if (rc == ENOTSUP)
- rc = pthread_mutexattr_setprotocol(&ma, PTHREAD_PRIO_NONE);
- if (rc && rc != ENOTSUP)
- goto bailout;
-#endif /* PTHREAD_PRIO_INHERIT */
-
- rc = pthread_mutexattr_settype(&ma, PTHREAD_MUTEX_ERRORCHECK);
- if (rc && rc != ENOTSUP)
- goto bailout;
-
- rc = pthread_mutex_init(&env->me_lck_mmap.lck->mti_rlock, &ma);
- if (rc)
- goto bailout;
- rc = pthread_mutex_init(&env->me_lck_mmap.lck->mti_wlock, &ma);
-
-bailout:
- pthread_mutexattr_destroy(&ma);
- return rc;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING > 0 */
-}
-
-__cold static int mdbx_ipclock_failed(MDBX_env *env, mdbx_ipclock_t *ipc,
- const int err) {
- int rc = err;
-#if MDBX_LOCKING == MDBX_LOCKING_POSIX2008 || MDBX_LOCKING == MDBX_LOCKING_SYSV
- if (err == EOWNERDEAD) {
- /* We own the mutex. Clean up after dead previous owner. */
-
- const bool rlocked = ipc == &env->me_lck->mti_rlock;
- rc = MDBX_SUCCESS;
- if (!rlocked) {
- if (unlikely(env->me_txn)) {
- /* env is hosed if the dead thread was ours */
- env->me_flags |= MDBX_FATAL_ERROR;
- env->me_txn = NULL;
- rc = MDBX_PANIC;
- }
- }
- mdbx_warning("%clock owner died, %s", (rlocked ? 'r' : 'w'),
- (rc ? "this process' env is hosed" : "recovering"));
-
- int check_rc = mdbx_cleanup_dead_readers(env, rlocked, NULL);
- check_rc = (check_rc == MDBX_SUCCESS) ? MDBX_RESULT_TRUE : check_rc;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- rc = (rc == MDBX_SUCCESS) ? check_rc : rc;
-#else
-#if defined(PTHREAD_MUTEX_ROBUST) || defined(pthread_mutex_consistent)
- int mreco_rc = pthread_mutex_consistent(ipc);
-#elif defined(PTHREAD_MUTEX_ROBUST_NP) || defined(pthread_mutex_consistent_np)
- int mreco_rc = pthread_mutex_consistent_np(ipc);
-#elif _POSIX_THREAD_PROCESS_SHARED < 200809L
- int mreco_rc = pthread_mutex_consistent_np(ipc);
-#else
- int mreco_rc = pthread_mutex_consistent(ipc);
-#endif
- check_rc = (mreco_rc == 0) ? check_rc : mreco_rc;
-
- if (unlikely(mreco_rc))
- mdbx_error("lock recovery failed, %s", mdbx_strerror(mreco_rc));
-
- rc = (rc == MDBX_SUCCESS) ? check_rc : rc;
- if (MDBX_IS_ERROR(rc))
- pthread_mutex_unlock(ipc);
-#endif /* MDBX_LOCKING == MDBX_LOCKING_POSIX2008 */
- return rc;
- }
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001
- (void)ipc;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
- (void)ipc;
-#elif MDBX_LOCKING == MDBX_LOCKING_FUTEX
-#ifdef _MSC_VER
-#pragma message("warning: TODO")
-#else
-#warning "TODO"
-#endif
- (void)ipc;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
- mdbx_error("mutex (un)lock failed, %s", mdbx_strerror(err));
- if (rc != EDEADLK)
- env->me_flags |= MDBX_FATAL_ERROR;
- return rc;
-}
-
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void) {
- /* avoid 32-bit Bionic bug/hang with 32-pit TID */
- if (sizeof(pthread_mutex_t) < sizeof(pid_t) + sizeof(unsigned)) {
- pid_t tid = gettid();
- if (unlikely(tid > 0xffff)) {
- mdbx_fatal("Raise the ENOSYS(%d) error to avoid hang due "
- "the 32-bit Bionic/Android bug with tid/thread_id 0x%08x(%i) "
- "that don’t fit in 16 bits, see "
- "https://android.googlesource.com/platform/bionic/+/master/"
- "docs/32-bit-abi.md#is-too-small-for-large-pids",
- ENOSYS, tid, tid);
- return ENOSYS;
- }
- }
- return 0;
-}
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-static int mdbx_ipclock_lock(MDBX_env *env, mdbx_ipclock_t *ipc,
- const bool dont_wait) {
-#if MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
- int rc = mdbx_check_tid4bionic();
- if (likely(rc == 0))
- rc = dont_wait ? pthread_mutex_trylock(ipc) : pthread_mutex_lock(ipc);
- rc = (rc == EBUSY && dont_wait) ? MDBX_BUSY : rc;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
- int rc = MDBX_SUCCESS;
- if (dont_wait) {
- if (sem_trywait(ipc)) {
- rc = errno;
- if (rc == EAGAIN)
- rc = MDBX_BUSY;
- }
- } else if (sem_wait(ipc))
- rc = errno;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
- struct sembuf op = {.sem_num = (ipc != &env->me_lck->mti_wlock),
- .sem_op = -1,
- .sem_flg = dont_wait ? IPC_NOWAIT | SEM_UNDO : SEM_UNDO};
- int rc;
- if (semop(env->me_sysv_ipc.semid, &op, 1)) {
- rc = errno;
- if (dont_wait && rc == EAGAIN)
- rc = MDBX_BUSY;
- } else {
- rc = *ipc ? EOWNERDEAD : MDBX_SUCCESS;
- *ipc = env->me_pid;
- }
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
- if (unlikely(rc != MDBX_SUCCESS && rc != MDBX_BUSY))
- rc = mdbx_ipclock_failed(env, ipc, rc);
- return rc;
-}
-
-static int mdbx_ipclock_unlock(MDBX_env *env, mdbx_ipclock_t *ipc) {
-#if MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
- int rc = pthread_mutex_unlock(ipc);
- (void)env;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
- int rc = sem_post(ipc) ? errno : MDBX_SUCCESS;
- (void)env;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
- if (unlikely(*ipc != (pid_t)env->me_pid))
- return EPERM;
- *ipc = 0;
- struct sembuf op = {.sem_num = (ipc != &env->me_lck->mti_wlock),
- .sem_op = 1,
- .sem_flg = SEM_UNDO};
- int rc = semop(env->me_sysv_ipc.semid, &op, 1) ? errno : MDBX_SUCCESS;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
- return rc;
-}
-
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env) {
- mdbx_trace("%s", ">>");
- mdbx_jitter4testing(true);
- int rc = mdbx_ipclock_lock(env, &env->me_lck->mti_rlock, false);
- mdbx_trace("<< rc %d", rc);
- return rc;
-}
-
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env) {
- mdbx_trace("%s", ">>");
- int rc = mdbx_ipclock_unlock(env, &env->me_lck->mti_rlock);
- mdbx_trace("<< rc %d", rc);
- if (unlikely(rc != MDBX_SUCCESS))
- mdbx_panic("%s() failed: err %d\n", __func__, rc);
- mdbx_jitter4testing(true);
-}
-
-int mdbx_txn_lock(MDBX_env *env, bool dont_wait) {
- mdbx_trace("%swait %s", dont_wait ? "dont-" : "", ">>");
- mdbx_jitter4testing(true);
- int rc = mdbx_ipclock_lock(env, &env->me_lck->mti_wlock, dont_wait);
- mdbx_trace("<< rc %d", rc);
- return MDBX_IS_ERROR(rc) ? rc : MDBX_SUCCESS;
-}
-
-void mdbx_txn_unlock(MDBX_env *env) {
- mdbx_trace("%s", ">>");
- int rc = mdbx_ipclock_unlock(env, &env->me_lck->mti_wlock);
- mdbx_trace("<< rc %d", rc);
- if (unlikely(rc != MDBX_SUCCESS))
- mdbx_panic("%s() failed: err %d\n", __func__, rc);
- mdbx_jitter4testing(true);
-}
-
-#else
-#ifdef _MSC_VER
-#pragma warning(disable : 4206) /* nonstandard extension used: translation \
- unit is empty */
-#endif /* _MSC_VER (warnings) */
-#endif /* !Windows LCK-implementation */
diff --git a/3rdparty/libmdbx/mdbx.c++ b/3rdparty/libmdbx/mdbx.c++
deleted file mode 100644
index b890b48b0c5654447382248b69c3c5d62f9239f0..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx.c++
+++ /dev/null
@@ -1,5202 +0,0 @@
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#define xMDBX_ALLOY 1
-#define MDBX_BUILD_SOURCERY c510b93d78b5848764f976ea67f06d278a7b99f141b4c98ceb095285f834f31c_v0_11_7_0_g40ec559c
-#ifdef MDBX_CONFIG_H
-#include MDBX_CONFIG_H
-#endif
-
-#define LIBMDBX_INTERNALS
-#ifdef xMDBX_TOOLS
-#define MDBX_DEPRECATED
-#endif /* xMDBX_TOOLS */
-
-#ifdef xMDBX_ALLOY
-/* Amalgamated build */
-#define MDBX_INTERNAL_FUNC static
-#define MDBX_INTERNAL_VAR static
-#else
-/* Non-amalgamated build */
-#define MDBX_INTERNAL_FUNC
-#define MDBX_INTERNAL_VAR extern
-#endif /* xMDBX_ALLOY */
-
-/** Disables using GNU/Linux libc extensions.
- * \ingroup build_option
- * \note This option couldn't be moved to the options.h since dependant
- * control macros/defined should be prepared before include the options.h */
-#ifndef MDBX_DISABLE_GNU_SOURCE
-#define MDBX_DISABLE_GNU_SOURCE 0
-#endif
-#if MDBX_DISABLE_GNU_SOURCE
-#undef _GNU_SOURCE
-#elif (defined(__linux__) || defined(__gnu_linux__)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif /* MDBX_DISABLE_GNU_SOURCE */
-
-/*----------------------------------------------------------------------------*/
-
-/* Should be defined before any includes */
-#if !defined(_FILE_OFFSET_BITS) && !defined(__ANDROID_API__) && \
- !defined(ANDROID)
-#define _FILE_OFFSET_BITS 64
-#endif
-
-#ifdef __APPLE__
-#define _DARWIN_C_SOURCE
-#endif
-
-#ifdef _MSC_VER
-#if _MSC_FULL_VER < 190024234
-/* Actually libmdbx was not tested with compilers older than 19.00.24234 (Visual
- * Studio 2015 Update 3). But you could remove this #error and try to continue
- * at your own risk. In such case please don't rise up an issues related ONLY to
- * old compilers.
- *
- * NOTE:
- * Unfortunately, there are several different builds of "Visual Studio" that
- * are called "Visual Studio 2015 Update 3".
- *
- * The 190024234 is used here because it is minimal version of Visual Studio
- * that was used for build and testing libmdbx in recent years. Soon this
- * value will be increased to 19.0.24241.7, since build and testing using
- * "Visual Studio 2015" will be performed only at https://ci.appveyor.com.
- *
- * Please ask Microsoft (but not us) for information about version differences
- * and how to and where you can obtain the latest "Visual Studio 2015" build
- * with all fixes.
- */
-#error \
- "At least \"Microsoft C/C++ Compiler\" version 19.00.24234 (Visual Studio 2015 Update 3) is required."
-#endif
-#ifndef _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#if _MSC_VER > 1913
-#pragma warning(disable : 5045) /* Compiler will insert Spectre mitigation... \
- */
-#endif
-#if _MSC_VER > 1914
-#pragma warning( \
- disable : 5105) /* winbase.h(9531): warning C5105: macro expansion \
- producing 'defined' has undefined behavior */
-#endif
-#pragma warning(disable : 4710) /* 'xyz': function not inlined */
-#pragma warning(disable : 4711) /* function 'xyz' selected for automatic \
- inline expansion */
-#pragma warning( \
- disable : 4201) /* nonstandard extension used : nameless struct / union */
-#pragma warning(disable : 4702) /* unreachable code */
-#pragma warning(disable : 4706) /* assignment within conditional expression */
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4324) /* 'xyz': structure was padded due to \
- alignment specifier */
-#pragma warning(disable : 4310) /* cast truncates constant value */
-#pragma warning( \
- disable : 4820) /* bytes padding added after data member for alignment */
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4366) /* the result of the unary '&' operator may be \
- unaligned */
-#pragma warning(disable : 4200) /* nonstandard extension used: zero-sized \
- array in struct/union */
-#pragma warning(disable : 4204) /* nonstandard extension used: non-constant \
- aggregate initializer */
-#pragma warning( \
- disable : 4505) /* unreferenced local function has been removed */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(__GNUC__) && __GNUC__ < 9
-#pragma GCC diagnostic ignored "-Wattributes"
-#endif /* GCC < 9 */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-#include "mdbx.h++"
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#ifndef __GNUC_PREREQ
-# if defined(__GNUC__) && defined(__GNUC_MINOR__)
-# define __GNUC_PREREQ(maj, min) \
- ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GNUC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GNUC_PREREQ */
-
-#ifndef __CLANG_PREREQ
-# ifdef __clang__
-# define __CLANG_PREREQ(maj,min) \
- ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min))
-# else
-# define __CLANG_PREREQ(maj,min) (0)
-# endif
-#endif /* __CLANG_PREREQ */
-
-#ifndef __GLIBC_PREREQ
-# if defined(__GLIBC__) && defined(__GLIBC_MINOR__)
-# define __GLIBC_PREREQ(maj, min) \
- ((__GLIBC__ << 16) + __GLIBC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GLIBC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GLIBC_PREREQ */
-
-#ifndef __has_warning
-# define __has_warning(x) (0)
-#endif
-
-#ifndef __has_include
-# define __has_include(x) (0)
-#endif
-
-#if __has_feature(thread_sanitizer)
-# define __SANITIZE_THREAD__ 1
-#endif
-
-#if __has_feature(address_sanitizer)
-# define __SANITIZE_ADDRESS__ 1
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __extern_C
-# ifdef __cplusplus
-# define __extern_C extern "C"
-# else
-# define __extern_C
-# endif
-#endif /* __extern_C */
-
-#if !defined(nullptr) && !defined(__cplusplus) || (__cplusplus < 201103L && !defined(_MSC_VER))
-# define nullptr NULL
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __always_inline
-# if defined(__GNUC__) || __has_attribute(__always_inline__)
-# define __always_inline __inline __attribute__((__always_inline__))
-# elif defined(_MSC_VER)
-# define __always_inline __forceinline
-# else
-# define __always_inline
-# endif
-#endif /* __always_inline */
-
-#ifndef __noinline
-# if defined(__GNUC__) || __has_attribute(__noinline__)
-# define __noinline __attribute__((__noinline__))
-# elif defined(_MSC_VER)
-# define __noinline __declspec(noinline)
-# else
-# define __noinline
-# endif
-#endif /* __noinline */
-
-#ifndef __must_check_result
-# if defined(__GNUC__) || __has_attribute(__warn_unused_result__)
-# define __must_check_result __attribute__((__warn_unused_result__))
-# else
-# define __must_check_result
-# endif
-#endif /* __must_check_result */
-
-#if !defined(__noop) && !defined(_MSC_VER)
-# define __noop(...) do {} while(0)
-#endif /* __noop */
-
-#if defined(__fallthrough) && \
- (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__))
-#undef __fallthrough
-#endif /* __fallthrough workaround for MinGW */
-
-#ifndef __fallthrough
-# if defined(__cplusplus) && (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || __cplusplus >= 201703L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(8, 0) && defined(__cplusplus) && __cplusplus >= 201103L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(7, 0) && \
- (!defined(__LCC__) || (__LCC__ == 124 && __LCC_MINOR__ >= 12) || \
- (__LCC__ == 125 && __LCC_MINOR__ >= 5) || (__LCC__ >= 126))
-# define __fallthrough __attribute__((__fallthrough__))
-# elif defined(__clang__) && defined(__cplusplus) && __cplusplus >= 201103L &&\
- __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
-# define __fallthrough [[clang::fallthrough]]
-# else
-# define __fallthrough
-# endif
-#endif /* __fallthrough */
-
-#ifndef __unreachable
-# if __GNUC_PREREQ(4,5) || __has_builtin(__builtin_unreachable)
-# define __unreachable() __builtin_unreachable()
-# elif defined(_MSC_VER)
-# define __unreachable() __assume(0)
-# else
-# define __unreachable() __noop()
-# endif
-#endif /* __unreachable */
-
-#ifndef __prefetch
-# if defined(__GNUC__) || defined(__clang__) || __has_builtin(__builtin_prefetch)
-# define __prefetch(ptr) __builtin_prefetch(ptr)
-# else
-# define __prefetch(ptr) __noop(ptr)
-# endif
-#endif /* __prefetch */
-
-#ifndef __nothrow
-# if defined(__cplusplus)
-# if __cplusplus < 201703L
-# define __nothrow throw()
-# else
-# define __nothrow noexcept(true)
-# endif /* __cplusplus */
-# elif defined(__GNUC__) || __has_attribute(__nothrow__)
-# define __nothrow __attribute__((__nothrow__))
-# elif defined(_MSC_VER) && defined(__cplusplus)
-# define __nothrow __declspec(nothrow)
-# else
-# define __nothrow
-# endif
-#endif /* __nothrow */
-
-#ifndef __hidden
-# if defined(__GNUC__) || __has_attribute(__visibility__)
-# define __hidden __attribute__((__visibility__("hidden")))
-# else
-# define __hidden
-# endif
-#endif /* __hidden */
-
-#ifndef __optimize
-# if defined(__OPTIMIZE__)
-# if (defined(__GNUC__) && !defined(__clang__)) || __has_attribute(__optimize__)
-# define __optimize(ops) __attribute__((__optimize__(ops)))
-# else
-# define __optimize(ops)
-# endif
-# else
-# define __optimize(ops)
-# endif
-#endif /* __optimize */
-
-#ifndef __hot
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __hot __attribute__((__hot__)) __optimize(3)
-# elif defined(__clang__) && !__has_attribute(__hot_) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put frequently used functions in separate section */
-# define __hot __attribute__((__section__("text.hot"))) __optimize("O3")
-# elif defined(__GNUC__) || __has_attribute(__hot__)
-# define __hot __attribute__((__hot__)) __optimize("O3")
-# else
-# define __hot __optimize("O3")
-# endif
-# else
-# define __hot
-# endif
-#endif /* __hot */
-
-#ifndef __cold
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __cold __attribute__((__cold__)) __optimize(1)
-# elif defined(__clang__) && !__has_attribute(cold) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put infrequently used functions in separate section */
-# define __cold __attribute__((__section__("text.unlikely"))) __optimize("Os")
-# elif defined(__GNUC__) || __has_attribute(cold)
-# define __cold __attribute__((__cold__)) __optimize("Os")
-# else
-# define __cold __optimize("Os")
-# endif
-# else
-# define __cold
-# endif
-#endif /* __cold */
-
-#ifndef __flatten
-# if defined(__OPTIMIZE__) && (defined(__GNUC__) || __has_attribute(__flatten__))
-# define __flatten __attribute__((__flatten__))
-# else
-# define __flatten
-# endif
-#endif /* __flatten */
-
-#ifndef likely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define likely(cond) __builtin_expect(!!(cond), 1)
-# else
-# define likely(x) (!!(x))
-# endif
-#endif /* likely */
-
-#ifndef unlikely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define unlikely(cond) __builtin_expect(!!(cond), 0)
-# else
-# define unlikely(x) (!!(x))
-# endif
-#endif /* unlikely */
-
-#ifndef __anonymous_struct_extension__
-# if defined(__GNUC__)
-# define __anonymous_struct_extension__ __extension__
-# else
-# define __anonymous_struct_extension__
-# endif
-#endif /* __anonymous_struct_extension__ */
-
-#ifndef __Wpedantic_format_voidptr
- MDBX_MAYBE_UNUSED MDBX_PURE_FUNCTION static __inline const void*
- __Wpedantic_format_voidptr(const void* ptr) {return ptr;}
-# define __Wpedantic_format_voidptr(ARG) __Wpedantic_format_voidptr(ARG)
-#endif /* __Wpedantic_format_voidptr */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(MDBX_USE_VALGRIND)
-# include <valgrind/memcheck.h>
-# ifndef VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE
- /* LY: available since Valgrind 3.10 */
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# endif
-#elif !defined(RUNNING_ON_VALGRIND)
-# define VALGRIND_CREATE_MEMPOOL(h,r,z)
-# define VALGRIND_DESTROY_MEMPOOL(h)
-# define VALGRIND_MEMPOOL_TRIM(h,a,s)
-# define VALGRIND_MEMPOOL_ALLOC(h,a,s)
-# define VALGRIND_MEMPOOL_FREE(h,a)
-# define VALGRIND_MEMPOOL_CHANGE(h,a,b,s)
-# define VALGRIND_MAKE_MEM_NOACCESS(a,s)
-# define VALGRIND_MAKE_MEM_DEFINED(a,s)
-# define VALGRIND_MAKE_MEM_UNDEFINED(a,s)
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_CHECK_MEM_IS_ADDRESSABLE(a,s) (0)
-# define VALGRIND_CHECK_MEM_IS_DEFINED(a,s) (0)
-# define RUNNING_ON_VALGRIND (0)
-#endif /* MDBX_USE_VALGRIND */
-
-#ifdef __SANITIZE_ADDRESS__
-# include <sanitizer/asan_interface.h>
-#elif !defined(ASAN_POISON_MEMORY_REGION)
-# define ASAN_POISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-# define ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-#endif /* __SANITIZE_ADDRESS__ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef ARRAY_LENGTH
-# ifdef __cplusplus
- template <typename T, size_t N>
- char (&__ArraySizeHelper(T (&array)[N]))[N];
-# define ARRAY_LENGTH(array) (sizeof(::__ArraySizeHelper(array)))
-# else
-# define ARRAY_LENGTH(array) (sizeof(array) / sizeof(array[0]))
-# endif
-#endif /* ARRAY_LENGTH */
-
-#ifndef ARRAY_END
-# define ARRAY_END(array) (&array[ARRAY_LENGTH(array)])
-#endif /* ARRAY_END */
-
-#define CONCAT(a,b) a##b
-#define XCONCAT(a,b) CONCAT(a,b)
-
-#ifndef offsetof
-# define offsetof(type, member) __builtin_offsetof(type, member)
-#endif /* offsetof */
-
-#ifndef container_of
-# define container_of(ptr, type, member) \
- ((type *)((char *)(ptr) - offsetof(type, member)))
-#endif /* container_of */
-
-#define MDBX_TETRAD(a, b, c, d) \
- ((uint32_t)(a) << 24 | (uint32_t)(b) << 16 | (uint32_t)(c) << 8 | (d))
-
-#define MDBX_STRING_TETRAD(str) MDBX_TETRAD(str[0], str[1], str[2], str[3])
-
-#define FIXME "FIXME: " __FILE__ ", " MDBX_STRINGIFY(__LINE__)
-
-#ifndef STATIC_ASSERT_MSG
-# if defined(static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) static_assert(expr, msg)
-# elif defined(_STATIC_ASSERT)
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif defined(_MSC_VER)
-# include <crtdbg.h>
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) \
- || __has_feature(c_static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) _Static_assert(expr, msg)
-# else
-# define STATIC_ASSERT_MSG(expr, msg) switch (0) {case 0:case (expr):;}
-# endif
-#endif /* STATIC_ASSERT */
-
-#ifndef STATIC_ASSERT
-# define STATIC_ASSERT(expr) STATIC_ASSERT_MSG(expr, #expr)
-#endif
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__GNUC__) && !__GNUC_PREREQ(4, 2)
-/* Actually libmdbx was not tested with compilers older than GCC 4.2.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required GCC >= 4.2"
-#endif
-
-#if defined(__clang__) && !__CLANG_PREREQ(3, 8)
-/* Actually libmdbx was not tested with CLANG older than 3.8.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required CLANG >= 3.8"
-#endif
-
-#if defined(__GLIBC__) && !__GLIBC_PREREQ(2, 12)
-/* Actually libmdbx was not tested with something older than glibc 2.12.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old systems.
- */
-#warning "libmdbx was only tested with GLIBC >= 2.12."
-#endif
-
-#ifdef __SANITIZE_THREAD__
-#warning \
- "libmdbx don't compatible with ThreadSanitizer, you will get a lot of false-positive issues."
-#endif /* __SANITIZE_THREAD__ */
-
-#if __has_warning("-Wnested-anon-types")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wnested-anon-types"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wnested-anon-types"
-#else
-#pragma warning disable "nested-anon-types"
-#endif
-#endif /* -Wnested-anon-types */
-
-#if __has_warning("-Wconstant-logical-operand")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wconstant-logical-operand"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wconstant-logical-operand"
-#else
-#pragma warning disable "constant-logical-operand"
-#endif
-#endif /* -Wconstant-logical-operand */
-
-#if defined(__LCC__) && (__LCC__ <= 121)
-/* bug #2798 */
-#pragma diag_suppress alignment_reduction_ignored
-#elif defined(__ICC)
-#pragma warning(disable : 3453 1366)
-#elif __has_warning("-Walignment-reduction-ignored")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Walignment-reduction-ignored"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Walignment-reduction-ignored"
-#else
-#pragma warning disable "alignment-reduction-ignored"
-#endif
-#endif /* -Walignment-reduction-ignored */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(_WIN32) || defined(_WIN64)
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if !defined(_NO_CRT_STDIO_INLINE) && MDBX_BUILD_SHARED_LIBRARY && \
- !defined(xMDBX_TOOLS) && MDBX_WITHOUT_MSVC_CRT
-#define _NO_CRT_STDIO_INLINE
-#endif
-#elif !defined(_POSIX_C_SOURCE)
-#define _POSIX_C_SOURCE 200809L
-#endif /* Windows */
-
-/*----------------------------------------------------------------------------*/
-/* C99 includes */
-#include <inttypes.h>
-#include <stddef.h>
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <assert.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-/* C11' alignas() */
-#if __has_include(<stdalign.h>)
-#include <stdalign.h>
-#endif
-#if defined(alignas) || defined(__cplusplus)
-#define MDBX_ALIGNAS(N) alignas(N)
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
-#define MDBX_ALIGNAS(N) _Alignas(N)
-#elif defined(_MSC_VER)
-#define MDBX_ALIGNAS(N) __declspec(align(N))
-#elif __has_attribute(__aligned__) || defined(__GNUC__)
-#define MDBX_ALIGNAS(N) __attribute__((__aligned__(N)))
-#else
-#error "FIXME: Required alignas() or equivalent."
-#endif /* MDBX_ALIGNAS */
-
-/*----------------------------------------------------------------------------*/
-/* Systems includes */
-
-#ifdef __APPLE__
-#include <TargetConditionals.h>
-#endif /* Apple OSX & iOS */
-
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__)
-#include <sys/cdefs.h>
-#include <sys/mount.h>
-#include <sys/sysctl.h>
-#include <sys/types.h>
-#if defined(__FreeBSD__) || defined(__DragonFly__)
-#include <vm/vm_param.h>
-#elif defined(__OpenBSD__) || defined(__NetBSD__)
-#include <uvm/uvm_param.h>
-#else
-#define SYSCTL_LEGACY_NONCONST_MIB
-#endif
-#ifndef __MACH__
-#include <sys/vmmeter.h>
-#endif
-#else
-#include <malloc.h>
-#if !(defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(_WIN32) || defined(_WIN64))
-#include <mntent.h>
-#endif /* !Solaris */
-#endif /* !xBSD */
-
-#if defined(__FreeBSD__) || __has_include(<malloc_np.h>)
-#include <malloc_np.h>
-#endif
-
-#if defined(__APPLE__) || defined(__MACH__) || __has_include(<malloc/malloc.h>)
-#include <malloc/malloc.h>
-#endif /* MacOS */
-
-#if defined(__MACH__)
-#include <mach/host_info.h>
-#include <mach/mach_host.h>
-#include <mach/mach_port.h>
-#include <uuid/uuid.h>
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-#include <sched.h>
-#include <sys/sendfile.h>
-#include <sys/statfs.h>
-#endif /* Linux */
-
-#ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 0
-#endif
-
-#ifndef _XOPEN_SOURCE_EXTENDED
-#define _XOPEN_SOURCE_EXTENDED 0
-#else
-#include <utmpx.h>
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#include <kstat.h>
-#include <sys/mnttab.h>
-/* On Solaris, it's easier to add a missing prototype rather than find a
- * combination of #defines that break nothing. */
-__extern_C key_t ftok(const char *, int);
-#endif /* SunOS/Solaris */
-
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef _WIN32_WINNT
-#define _WIN32_WINNT 0x0601 /* Windows 7 */
-#elif _WIN32_WINNT < 0x0500
-#error At least 'Windows 2000' API is required for libmdbx.
-#endif /* _WIN32_WINNT */
-#if (defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* MinGW */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif /* WIN32_LEAN_AND_MEAN */
-#include <excpt.h>
-#include <tlhelp32.h>
-#include <windows.h>
-#include <winnt.h>
-#include <winternl.h>
-#define HAVE_SYS_STAT_H
-#define HAVE_SYS_TYPES_H
-typedef HANDLE mdbx_thread_t;
-typedef unsigned mdbx_thread_key_t;
-#define MAP_FAILED NULL
-#define HIGH_DWORD(v) ((DWORD)((sizeof(v) > 4) ? ((uint64_t)(v) >> 32) : 0))
-#define THREAD_CALL WINAPI
-#define THREAD_RESULT DWORD
-typedef struct {
- HANDLE mutex;
- HANDLE event[2];
-} mdbx_condpair_t;
-typedef CRITICAL_SECTION mdbx_fastmutex_t;
-
-#if !defined(_MSC_VER) && !defined(__try)
-/* *INDENT-OFF* */
-/* clang-format off */
-#define __try
-#define __except(COND) if(false)
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* stub for MSVC's __try/__except */
-
-#if MDBX_WITHOUT_MSVC_CRT
-
-#ifndef mdbx_malloc
-static inline void *mdbx_malloc(size_t bytes) {
- return HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_malloc */
-
-#ifndef mdbx_calloc
-static inline void *mdbx_calloc(size_t nelem, size_t size) {
- return HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, nelem * size);
-}
-#endif /* mdbx_calloc */
-
-#ifndef mdbx_realloc
-static inline void *mdbx_realloc(void *ptr, size_t bytes) {
- return ptr ? HeapReAlloc(GetProcessHeap(), 0, ptr, bytes)
- : HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_realloc */
-
-#ifndef mdbx_free
-static inline void mdbx_free(void *ptr) { HeapFree(GetProcessHeap(), 0, ptr); }
-#endif /* mdbx_free */
-
-#else /* MDBX_WITHOUT_MSVC_CRT */
-
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup _strdup
-
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-#ifndef snprintf
-#define snprintf _snprintf /* ntdll */
-#endif
-
-#ifndef vsnprintf
-#define vsnprintf _vsnprintf /* ntdll */
-#endif
-
-#else /*----------------------------------------------------------------------*/
-
-#include <unistd.h>
-#if !defined(_POSIX_MAPPED_FILES) || _POSIX_MAPPED_FILES < 1
-#error "libmdbx requires the _POSIX_MAPPED_FILES feature"
-#endif /* _POSIX_MAPPED_FILES */
-
-#include <pthread.h>
-#include <semaphore.h>
-#include <signal.h>
-#include <sys/file.h>
-#include <sys/ipc.h>
-#include <sys/mman.h>
-#include <sys/param.h>
-#include <sys/stat.h>
-#include <sys/statvfs.h>
-#include <sys/uio.h>
-typedef pthread_t mdbx_thread_t;
-typedef pthread_key_t mdbx_thread_key_t;
-#define INVALID_HANDLE_VALUE (-1)
-#define THREAD_CALL
-#define THREAD_RESULT void *
-typedef struct {
- pthread_mutex_t mutex;
- pthread_cond_t cond[2];
-} mdbx_condpair_t;
-typedef pthread_mutex_t mdbx_fastmutex_t;
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup strdup
-#endif /* Platform */
-
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
-/* malloc_usable_size() already provided */
-#elif defined(__APPLE__)
-#define malloc_usable_size(ptr) malloc_size(ptr)
-#elif defined(_MSC_VER) && !MDBX_WITHOUT_MSVC_CRT
-#define malloc_usable_size(ptr) _msize(ptr)
-#endif /* malloc_usable_size */
-
-#ifdef __ANDROID_API__
-#include <android/log.h>
-#if __ANDROID_API__ >= 21
-#include <sys/sendfile.h>
-#endif
-#endif /* Android */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(HAVE_SYS_STAT_H) || __has_include(<sys/stat.h>)
-#include <sys/stat.h>
-#endif
-#if defined(HAVE_SYS_TYPES_H) || __has_include(<sys/types.h>)
-#include <sys/types.h>
-#endif
-#if defined(HAVE_SYS_FILE_H) || __has_include(<sys/file.h>)
-#include <sys/file.h>
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-
-#ifndef SSIZE_MAX
-#define SSIZE_MAX INTPTR_MAX
-#endif
-
-#if !defined(MADV_DODUMP) && defined(MADV_CORE)
-#define MADV_DODUMP MADV_CORE
-#endif /* MADV_CORE -> MADV_DODUMP */
-
-#if !defined(MADV_DONTDUMP) && defined(MADV_NOCORE)
-#define MADV_DONTDUMP MADV_NOCORE
-#endif /* MADV_NOCORE -> MADV_DONTDUMP */
-
-#if defined(i386) || defined(__386) || defined(__i386) || defined(__i386__) || \
- defined(i486) || defined(__i486) || defined(__i486__) || \
- defined(i586) | defined(__i586) || defined(__i586__) || defined(i686) || \
- defined(__i686) || defined(__i686__) || defined(_M_IX86) || \
- defined(_X86_) || defined(__THW_INTEL__) || defined(__I86__) || \
- defined(__INTEL__) || defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64__) || defined(__amd64) || defined(_M_X64) || \
- defined(_M_AMD64) || defined(__IA32__) || defined(__INTEL__)
-#ifndef __ia32__
-/* LY: define neutral __ia32__ for x86 and x86-64 */
-#define __ia32__ 1
-#endif /* __ia32__ */
-#if !defined(__amd64__) && (defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64) || defined(_M_X64))
-/* LY: define trusty __amd64__ for all AMD64/x86-64 arch */
-#define __amd64__ 1
-#endif /* __amd64__ */
-#endif /* all x86 */
-
-#if (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
-#error \
- "Sanity checking failed: Two's complement, reasonably sized integer types"
-#endif
-
-#if UINTPTR_MAX > 0xffffFFFFul || ULONG_MAX > 0xffffFFFFul
-#define MDBX_WORDBITS 64
-#else
-#define MDBX_WORDBITS 32
-#endif /* MDBX_WORDBITS */
-
-#if defined(__ANDROID_API__) || defined(ANDROID)
-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "_FILE_OFFSET_BITS != MDBX_WORDBITS" (_FILE_OFFSET_BITS != MDBX_WORDBITS)
-#elif defined(__FILE_OFFSET_BITS) && __FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "__FILE_OFFSET_BITS != MDBX_WORDBITS" (__FILE_OFFSET_BITS != MDBX_WORDBITS)
-#endif
-#endif /* Android */
-
-/*----------------------------------------------------------------------------*/
-/* Compiler's includes for builtins/intrinsics */
-
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
-#include <intrin.h>
-#elif __GNUC_PREREQ(4, 4) || defined(__clang__)
-#if defined(__ia32__) || defined(__e2k__)
-#include <x86intrin.h>
-#endif /* __ia32__ */
-#if defined(__ia32__)
-#include <cpuid.h>
-#endif /* __ia32__ */
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
-#include <mbarrier.h>
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
-#include <machine/sys/inline.h>
-#elif defined(__IBMC__) && defined(__powerpc)
-#include <atomic.h>
-#elif defined(_AIX)
-#include <builtins.h>
-#include <sys/atomic_op.h>
-#elif (defined(__osf__) && defined(__DECC)) || defined(__alpha)
-#include <c_asm.h>
-#include <machine/builtins.h>
-#elif defined(__MWERKS__)
-/* CodeWarrior - troubles ? */
-#pragma gcc_extensions
-#elif defined(__SNC__)
-/* Sony PS3 - troubles ? */
-#elif defined(__hppa__) || defined(__hppa)
-#include <machine/inline.h>
-#else
-#error Unsupported C compiler, please use GNU C 4.4 or newer
-#endif /* Compiler */
-
-/*----------------------------------------------------------------------------*/
-/* Byteorder */
-
-#if !defined(__BYTE_ORDER__) || !defined(__ORDER_LITTLE_ENDIAN__) || \
- !defined(__ORDER_BIG_ENDIAN__)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(__GLIBC__) || defined(__GNU_LIBRARY__) || defined(__ANDROID_API__) || \
- defined(HAVE_ENDIAN_H) || __has_include(<endian.h>)
-#include <endian.h>
-#elif defined(__APPLE__) || defined(__MACH__) || defined(__OpenBSD__) || \
- defined(HAVE_MACHINE_ENDIAN_H) || __has_include(<machine/endian.h>)
-#include <machine/endian.h>
-#elif defined(HAVE_SYS_ISA_DEFS_H) || __has_include(<sys/isa_defs.h>)
-#include <sys/isa_defs.h>
-#elif (defined(HAVE_SYS_TYPES_H) && defined(HAVE_SYS_ENDIAN_H)) || \
- (__has_include(<sys/types.h>) && __has_include(<sys/endian.h>))
-#include <sys/endian.h>
-#include <sys/types.h>
-#elif defined(__bsdi__) || defined(__DragonFly__) || defined(__FreeBSD__) || \
- defined(__NetBSD__) || \
- defined(HAVE_SYS_PARAM_H) || __has_include(<sys/param.h>)
-#include <sys/param.h>
-#endif /* OS */
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-
-#else
-#error __BYTE_ORDER__ should be defined.
-#endif /* Arch */
-
-#endif
-#endif /* __BYTE_ORDER__ || __ORDER_LITTLE_ENDIAN__ || __ORDER_BIG_ENDIAN__ */
-
-/* Get the size of a memory page for the system.
- * This is the basic size that the platform's memory manager uses, and is
- * fundamental to the use of memory-mapped files. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline size_t
-mdbx_syspagesize(void) {
-#if defined(_WIN32) || defined(_WIN64)
- SYSTEM_INFO si;
- GetSystemInfo(&si);
- return si.dwPageSize;
-#else
- return sysconf(_SC_PAGE_SIZE);
-#endif
-}
-
-typedef struct mdbx_mmap_param {
- union {
- void *address;
- uint8_t *dxb;
- struct MDBX_lockinfo *lck;
- };
- mdbx_filehandle_t fd;
- size_t limit; /* mapping length, but NOT a size of file nor DB */
- size_t current; /* mapped region size, i.e. the size of file and DB */
- uint64_t filesize /* in-process cache of a file size */;
-#if defined(_WIN32) || defined(_WIN64)
- HANDLE section; /* memory-mapped section handle */
-#endif
-} mdbx_mmap_t;
-
-typedef union bin128 {
- __anonymous_struct_extension__ struct { uint64_t x, y; };
- __anonymous_struct_extension__ struct { uint32_t a, b, c, d; };
-} bin128_t;
-
-#if defined(_WIN32) || defined(_WIN64)
-typedef union MDBX_srwlock {
- __anonymous_struct_extension__ struct {
- long volatile readerCount;
- long volatile writerCount;
- };
- RTL_SRWLOCK native;
-} MDBX_srwlock;
-#endif /* Windows */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny);
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny);
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-#if defined(__cplusplus) && !defined(__STDC_NO_ATOMICS__) && (__has_include(<cstdatomic>) || __has_extension(cxx_atomic))
-#include <cstdatomic>
-#define MDBX_HAVE_C11ATOMICS
-#elif !defined(__cplusplus) && \
- (__STDC_VERSION__ >= 201112L || __has_extension(c_atomic)) && \
- !defined(__STDC_NO_ATOMICS__) && \
- (__GNUC_PREREQ(4, 9) || __CLANG_PREREQ(3, 8) || \
- !(defined(__GNUC__) || defined(__clang__)))
-#include <stdatomic.h>
-#define MDBX_HAVE_C11ATOMICS
-#elif defined(__GNUC__) || defined(__clang__)
-#elif defined(_MSC_VER)
-#pragma warning(disable : 4163) /* 'xyz': not available as an intrinsic */
-#pragma warning(disable : 4133) /* 'function': incompatible types - from \
- 'size_t' to 'LONGLONG' */
-#pragma warning(disable : 4244) /* 'return': conversion from 'LONGLONG' to \
- 'std::size_t', possible loss of data */
-#pragma warning(disable : 4267) /* 'function': conversion from 'size_t' to \
- 'long', possible loss of data */
-#pragma intrinsic(_InterlockedExchangeAdd, _InterlockedCompareExchange)
-#pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedCompareExchange64)
-#elif defined(__APPLE__)
-#include <libkern/OSAtomic.h>
-#else
-#error FIXME atomic-ops
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Memory/Compiler barriers, cache coherence */
-
-#if __has_include(<sys/cachectl.h>)
-#include <sys/cachectl.h>
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS should have explicit cache control */
-#include <sys/cachectl.h>
-#endif
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_compiler_barrier(void) {
-#if defined(__clang__) || defined(__GNUC__)
- __asm__ __volatile__("" ::: "memory");
-#elif defined(_MSC_VER)
- _ReadWriteBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
- __memory_barrier();
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __compiler_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_sched_fence(/* LY: no-arg meaning 'all expect ALU', e.g. 0x3D3D */);
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __fence();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_memory_barrier(void) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(memory_order_seq_cst);
-#elif defined(__ATOMIC_SEQ_CST)
-#ifdef __clang__
- __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
- __atomic_thread_fence(__ATOMIC_SEQ_CST);
-#endif
-#elif defined(__clang__) || defined(__GNUC__)
- __sync_synchronize();
-#elif defined(_WIN32) || defined(_WIN64)
- MemoryBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
-#if defined(__ia32__)
- _mm_mfence();
-#else
- __mf();
-#endif
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __machine_rw_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_mf();
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __lwsync();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* libc compatibility stuff */
-
-#if (!defined(__GLIBC__) && __GLIBC_PREREQ(2, 1)) && \
- (defined(_GNU_SOURCE) || defined(_BSD_SOURCE))
-#define mdbx_asprintf asprintf
-#define mdbx_vasprintf vasprintf
-#else
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC
- MDBX_PRINTF_ARGS(2, 3) int mdbx_asprintf(char **strp, const char *fmt, ...);
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt, va_list ap);
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* OS abstraction layer stuff */
-
-/* max bytes to write in one call */
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_WRITE UINT32_C(0x01000000)
-#else
-#define MAX_WRITE UINT32_C(0x3fff0000)
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-MDBX_INTERNAL_VAR uint32_t mdbx_linux_kernel_version;
-MDBX_INTERNAL_VAR bool mdbx_RunningOnWSL1 /* Windows Subsystem 1 for Linux */;
-#endif /* Linux */
-
-#ifndef mdbx_strdup
-LIBMDBX_API char *mdbx_strdup(const char *str);
-#endif
-
-MDBX_MAYBE_UNUSED static __inline int mdbx_get_errno(void) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD rc = GetLastError();
-#else
- int rc = errno;
-#endif
- return rc;
-}
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result);
-#endif
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr);
-#endif
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair, bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair);
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex);
-
-MDBX_INTERNAL_FUNC int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov,
- int iovcnt, uint64_t offset,
- size_t expected_written);
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t count,
- uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t count, uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t count);
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg);
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread);
-
-enum mdbx_syncmode_bits {
- MDBX_SYNC_NONE = 0,
- MDBX_SYNC_DATA = 1,
- MDBX_SYNC_SIZE = 2,
- MDBX_SYNC_IODQ = 4
-};
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- const enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length);
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos);
-MDBX_INTERNAL_FUNC int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length);
-
-enum mdbx_openfile_purpose {
- MDBX_OPEN_DXB_READ = 0,
- MDBX_OPEN_DXB_LAZY = 1,
- MDBX_OPEN_DXB_DSYNC = 2,
- MDBX_OPEN_LCK = 3,
- MDBX_OPEN_COPY = 4,
- MDBX_OPEN_DELETE = 5
-};
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait);
-
-#define MMAP_OPTION_TRUNCATE 1
-#define MMAP_OPTION_SEMAPHORE 2
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t must, const size_t limit,
- const unsigned options);
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map);
-#define MDBX_MRESIZE_MAY_MOVE 0x00000100
-#define MDBX_MRESIZE_MAY_UNMAP 0x00000200
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit);
-#if defined(_WIN32) || defined(_WIN64)
-typedef struct {
- unsigned limit, count;
- HANDLE handles[31];
-} mdbx_handle_array_t;
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array);
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array);
-#endif /* Windows */
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err);
-
-MDBX_MAYBE_UNUSED static __inline uint32_t mdbx_getpid(void) {
- STATIC_ASSERT(sizeof(mdbx_pid_t) <= sizeof(uint32_t));
-#if defined(_WIN32) || defined(_WIN64)
- return GetCurrentProcessId();
-#else
- STATIC_ASSERT(sizeof(pid_t) <= sizeof(uint32_t));
- return getpid();
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline uintptr_t mdbx_thread_self(void) {
- mdbx_tid_t thunk;
- STATIC_ASSERT(sizeof(uintptr_t) >= sizeof(thunk));
-#if defined(_WIN32) || defined(_WIN64)
- thunk = GetCurrentThreadId();
-#else
- thunk = pthread_self();
-#endif
- return (uintptr_t)thunk;
-}
-
-#if !defined(_WIN32) && !defined(_WIN64)
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void);
-#else
-static __inline int mdbx_check_tid4bionic(void) { return 0; }
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-MDBX_MAYBE_UNUSED static __inline int
-mdbx_pthread_mutex_lock(pthread_mutex_t *mutex) {
- int err = mdbx_check_tid4bionic();
- return unlikely(err) ? err : pthread_mutex_lock(mutex);
-}
-#endif /* !Windows */
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void);
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16);
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime);
-
-MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void);
-/*----------------------------------------------------------------------------*/
-/* lck stuff */
-
-/// \brief Initialization of synchronization primitives linked with MDBX_env
-/// instance both in LCK-file and within the current process.
-/// \param
-/// global_uniqueness_flag = true - denotes that there are no other processes
-/// working with DB and LCK-file. Thus the function MUST initialize
-/// shared synchronization objects in memory-mapped LCK-file.
-/// global_uniqueness_flag = false - denotes that at least one process is
-/// already working with DB and LCK-file, including the case when DB
-/// has already been opened in the current process. Thus the function
-/// MUST NOT initialize shared synchronization objects in memory-mapped
-/// LCK-file that are already in use.
-/// \return Error code or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag);
-
-/// \brief Disconnects from shared interprocess objects and destructs
-/// synchronization objects linked with MDBX_env instance
-/// within the current process.
-/// \param
-/// inprocess_neighbor = NULL - if the current process does not have other
-/// instances of MDBX_env linked with the DB being closed.
-/// Thus the function MUST check for other processes working with DB or
-/// LCK-file, and keep or destroy shared synchronization objects in
-/// memory-mapped LCK-file depending on the result.
-/// inprocess_neighbor = not-NULL - pointer to another instance of MDBX_env
-/// (anyone of there is several) working with DB or LCK-file within the
-/// current process. Thus the function MUST NOT try to acquire exclusive
-/// lock and/or try to destruct shared synchronization objects linked with
-/// DB or LCK-file. Moreover, the implementation MUST ensure correct work
-/// of other instances of MDBX_env within the current process, e.g.
-/// restore POSIX-fcntl locks after the closing of file descriptors.
-/// \return Error code (MDBX_PANIC) or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor);
-
-/// \brief Connects to shared interprocess locking objects and tries to acquire
-/// the maximum lock level (shared if exclusive is not available)
-/// Depending on implementation or/and platform (Windows) this function may
-/// acquire the non-OS super-level lock (e.g. for shared synchronization
-/// objects initialization), which will be downgraded to OS-exclusive or
-/// shared via explicit calling of mdbx_lck_downgrade().
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if an exclusive lock was acquired and thus
-/// the current process is the first and only after the last use of DB.
-/// MDBX_RESULT_FALSE (0) - if a shared lock was acquired and thus
-/// DB has already been opened and now is used by other processes.
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env);
-
-/// \brief Downgrades the level of initially acquired lock to
-/// operational level specified by argument. The reson for such downgrade:
-/// - unblocking of other processes that are waiting for access, i.e.
-/// if (env->me_flags & MDBX_EXCLUSIVE) != 0, then other processes
-/// should be made aware that access is unavailable rather than
-/// wait for it.
-/// - freeing locks that interfere file operation (especially for Windows)
-/// (env->me_flags & MDBX_EXCLUSIVE) == 0 - downgrade to shared lock.
-/// (env->me_flags & MDBX_EXCLUSIVE) != 0 - downgrade to exclusive
-/// operational lock.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env);
-
-/// \brief Locks LCK-file or/and table of readers for (de)registering.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env);
-
-/// \brief Unlocks LCK-file or/and table of readers after (de)registering.
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env);
-
-/// \brief Acquires lock for DB change (on writing transaction start)
-/// Reading transactions will not be blocked.
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-/// \return Error code or zero on success
-LIBMDBX_API int mdbx_txn_lock(MDBX_env *env, bool dont_wait);
-
-/// \brief Releases lock once DB changes is made (after writing transaction
-/// has finished).
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-LIBMDBX_API void mdbx_txn_unlock(MDBX_env *env);
-
-/// \brief Sets alive-flag of reader presence (indicative lock) for PID of
-/// the current process. The function does no more than needed for
-/// the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env);
-
-/// \brief Resets alive-flag of reader presence (indicative lock)
-/// for PID of the current process. The function does no more than needed
-/// for the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env);
-
-/// \brief Checks for reading process status with the given pid with help of
-/// alive-flag of presence (indicative lock) or using another way.
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if the reader process with the given PID is alive
-/// and working with DB (indicative lock is present).
-/// MDBX_RESULT_FALSE (0) - if the reader process with the given PID is absent
-/// or not working with DB (indicative lock is not present).
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid);
-
-#if defined(_WIN32) || defined(_WIN64)
-
-typedef void(WINAPI *MDBX_srwlock_function)(MDBX_srwlock *);
-MDBX_INTERNAL_VAR MDBX_srwlock_function mdbx_srwlock_Init,
- mdbx_srwlock_AcquireShared, mdbx_srwlock_ReleaseShared,
- mdbx_srwlock_AcquireExclusive, mdbx_srwlock_ReleaseExclusive;
-
-#if _WIN32_WINNT < 0x0600 /* prior to Windows Vista */
-typedef enum _FILE_INFO_BY_HANDLE_CLASS {
- FileBasicInfo,
- FileStandardInfo,
- FileNameInfo,
- FileRenameInfo,
- FileDispositionInfo,
- FileAllocationInfo,
- FileEndOfFileInfo,
- FileStreamInfo,
- FileCompressionInfo,
- FileAttributeTagInfo,
- FileIdBothDirectoryInfo,
- FileIdBothDirectoryRestartInfo,
- FileIoPriorityHintInfo,
- FileRemoteProtocolInfo,
- MaximumFileInfoByHandleClass
-} FILE_INFO_BY_HANDLE_CLASS,
- *PFILE_INFO_BY_HANDLE_CLASS;
-
-typedef struct _FILE_END_OF_FILE_INFO {
- LARGE_INTEGER EndOfFile;
-} FILE_END_OF_FILE_INFO, *PFILE_END_OF_FILE_INFO;
-
-#define REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK 0x00000001
-#define REMOTE_PROTOCOL_INFO_FLAG_OFFLINE 0x00000002
-
-typedef struct _FILE_REMOTE_PROTOCOL_INFO {
- USHORT StructureVersion;
- USHORT StructureSize;
- DWORD Protocol;
- USHORT ProtocolMajorVersion;
- USHORT ProtocolMinorVersion;
- USHORT ProtocolRevision;
- USHORT Reserved;
- DWORD Flags;
- struct {
- DWORD Reserved[8];
- } GenericReserved;
- struct {
- DWORD Reserved[16];
- } ProtocolSpecificReserved;
-} FILE_REMOTE_PROTOCOL_INFO, *PFILE_REMOTE_PROTOCOL_INFO;
-
-#endif /* _WIN32_WINNT < 0x0600 (prior to Windows Vista) */
-
-typedef BOOL(WINAPI *MDBX_GetFileInformationByHandleEx)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_GetFileInformationByHandleEx
- mdbx_GetFileInformationByHandleEx;
-
-typedef BOOL(WINAPI *MDBX_GetVolumeInformationByHandleW)(
- _In_ HANDLE hFile, _Out_opt_ LPWSTR lpVolumeNameBuffer,
- _In_ DWORD nVolumeNameSize, _Out_opt_ LPDWORD lpVolumeSerialNumber,
- _Out_opt_ LPDWORD lpMaximumComponentLength,
- _Out_opt_ LPDWORD lpFileSystemFlags,
- _Out_opt_ LPWSTR lpFileSystemNameBuffer, _In_ DWORD nFileSystemNameSize);
-MDBX_INTERNAL_VAR MDBX_GetVolumeInformationByHandleW
- mdbx_GetVolumeInformationByHandleW;
-
-typedef DWORD(WINAPI *MDBX_GetFinalPathNameByHandleW)(_In_ HANDLE hFile,
- _Out_ LPWSTR lpszFilePath,
- _In_ DWORD cchFilePath,
- _In_ DWORD dwFlags);
-MDBX_INTERNAL_VAR MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-
-typedef BOOL(WINAPI *MDBX_SetFileInformationByHandle)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_SetFileInformationByHandle
- mdbx_SetFileInformationByHandle;
-
-typedef NTSTATUS(NTAPI *MDBX_NtFsControlFile)(
- IN HANDLE FileHandle, IN OUT HANDLE Event,
- IN OUT PVOID /* PIO_APC_ROUTINE */ ApcRoutine, IN OUT PVOID ApcContext,
- OUT PIO_STATUS_BLOCK IoStatusBlock, IN ULONG FsControlCode,
- IN OUT PVOID InputBuffer, IN ULONG InputBufferLength,
- OUT OPTIONAL PVOID OutputBuffer, IN ULONG OutputBufferLength);
-MDBX_INTERNAL_VAR MDBX_NtFsControlFile mdbx_NtFsControlFile;
-
-typedef uint64_t(WINAPI *MDBX_GetTickCount64)(void);
-MDBX_INTERNAL_VAR MDBX_GetTickCount64 mdbx_GetTickCount64;
-
-#if !defined(_WIN32_WINNT_WIN8) || _WIN32_WINNT < _WIN32_WINNT_WIN8
-typedef struct _WIN32_MEMORY_RANGE_ENTRY {
- PVOID VirtualAddress;
- SIZE_T NumberOfBytes;
-} WIN32_MEMORY_RANGE_ENTRY, *PWIN32_MEMORY_RANGE_ENTRY;
-#endif /* Windows 8.x */
-
-typedef BOOL(WINAPI *MDBX_PrefetchVirtualMemory)(
- HANDLE hProcess, ULONG_PTR NumberOfEntries,
- PWIN32_MEMORY_RANGE_ENTRY VirtualAddresses, ULONG Flags);
-MDBX_INTERNAL_VAR MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-
-typedef enum _SECTION_INHERIT { ViewShare = 1, ViewUnmap = 2 } SECTION_INHERIT;
-
-typedef NTSTATUS(NTAPI *MDBX_NtExtendSection)(IN HANDLE SectionHandle,
- IN PLARGE_INTEGER NewSectionSize);
-MDBX_INTERNAL_VAR MDBX_NtExtendSection mdbx_NtExtendSection;
-
-static __inline bool mdbx_RunningUnderWine(void) {
- return !mdbx_NtExtendSection;
-}
-
-typedef LSTATUS(WINAPI *MDBX_RegGetValueA)(HKEY hkey, LPCSTR lpSubKey,
- LPCSTR lpValue, DWORD dwFlags,
- LPDWORD pdwType, PVOID pvData,
- LPDWORD pcbData);
-MDBX_INTERNAL_VAR MDBX_RegGetValueA mdbx_RegGetValueA;
-
-NTSYSAPI ULONG RtlRandomEx(PULONG Seed);
-
-#endif /* Windows */
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_MSC_VER) && _MSC_VER >= 1900
-/* LY: MSVC 2015/2017/2019 has buggy/inconsistent PRIuPTR/PRIxPTR macros
- * for internal format-args checker. */
-#undef PRIuPTR
-#undef PRIiPTR
-#undef PRIdPTR
-#undef PRIxPTR
-#define PRIuPTR "Iu"
-#define PRIiPTR "Ii"
-#define PRIdPTR "Id"
-#define PRIxPTR "Ix"
-#define PRIuSIZE "zu"
-#define PRIiSIZE "zi"
-#define PRIdSIZE "zd"
-#define PRIxSIZE "zx"
-#endif /* fix PRI*PTR for _MSC_VER */
-
-#ifndef PRIuSIZE
-#define PRIuSIZE PRIuPTR
-#define PRIiSIZE PRIiPTR
-#define PRIdSIZE PRIdPTR
-#define PRIxSIZE PRIxPTR
-#endif /* PRI*SIZE macros for MSVC */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-#define mdbx_sourcery_anchor XCONCAT(mdbx_sourcery_, MDBX_BUILD_SOURCERY)
-#if defined(xMDBX_TOOLS)
-extern LIBMDBX_API const char *const mdbx_sourcery_anchor;
-#endif
-
-/*******************************************************************************
- *******************************************************************************
- *******************************************************************************
- *
- *
- * #### ##### ##### # #### # # ####
- * # # # # # # # # ## # #
- * # # # # # # # # # # # ####
- * # # ##### # # # # # # # #
- * # # # # # # # # ## # #
- * #### # # # #### # # ####
- *
- *
- */
-
-/** \defgroup build_option Build options
- * The libmdbx build options.
- @{ */
-
-/** Using fcntl(F_FULLFSYNC) with 5-10 times slowdown */
-#define MDBX_OSX_WANNA_DURABILITY 0
-/** Using fsync() with chance of data lost on power failure */
-#define MDBX_OSX_WANNA_SPEED 1
-
-#ifndef MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-/** Choices \ref MDBX_OSX_WANNA_DURABILITY or \ref MDBX_OSX_WANNA_SPEED
- * for OSX & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY MDBX_OSX_WANNA_DURABILITY
-#endif /* MDBX_OSX_SPEED_INSTEADOF_DURABILITY */
-
-/** Controls checking PID against reuse DB environment after the fork() */
-#ifndef MDBX_ENV_CHECKPID
-#if defined(MADV_DONTFORK) || defined(_WIN32) || defined(_WIN64)
-/* PID check could be omitted:
- * - on Linux when madvise(MADV_DONTFORK) is available, i.e. after the fork()
- * mapped pages will not be available for child process.
- * - in Windows where fork() not available. */
-#define MDBX_ENV_CHECKPID 0
-#else
-#define MDBX_ENV_CHECKPID 1
-#endif
-#define MDBX_ENV_CHECKPID_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#else
-#define MDBX_ENV_CHECKPID_CONFIG MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#endif /* MDBX_ENV_CHECKPID */
-
-/** Controls checking transaction owner thread against misuse transactions from
- * other threads. */
-#ifndef MDBX_TXN_CHECKOWNER
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_TXN_CHECKOWNER_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#else
-#define MDBX_TXN_CHECKOWNER_CONFIG MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#endif /* MDBX_TXN_CHECKOWNER */
-
-/** Does a system have battery-backed Real-Time Clock or just a fake. */
-#ifndef MDBX_TRUST_RTC
-#if defined(__linux__) || defined(__gnu_linux__) || defined(__NetBSD__) || \
- defined(__OpenBSD__)
-#define MDBX_TRUST_RTC 0 /* a lot of embedded systems have a fake RTC */
-#else
-#define MDBX_TRUST_RTC 1
-#endif
-#define MDBX_TRUST_RTC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#else
-#define MDBX_TRUST_RTC_CONFIG MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#endif /* MDBX_TRUST_RTC */
-
-/** Controls online database auto-compactification during write-transactions. */
-#ifndef MDBX_ENABLE_REFUND
-#define MDBX_ENABLE_REFUND 1
-#elif !(MDBX_ENABLE_REFUND == 0 || MDBX_ENABLE_REFUND == 1)
-#error MDBX_ENABLE_REFUND must be defined as 0 or 1
-#endif /* MDBX_ENABLE_REFUND */
-
-/** Controls gathering statistics for page operations. */
-#ifndef MDBX_ENABLE_PGOP_STAT
-#define MDBX_ENABLE_PGOP_STAT 1
-#elif !(MDBX_ENABLE_PGOP_STAT == 0 || MDBX_ENABLE_PGOP_STAT == 1)
-#error MDBX_ENABLE_PGOP_STAT must be defined as 0 or 1
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-/** Controls use of POSIX madvise() hints and friends. */
-#ifndef MDBX_ENABLE_MADVISE
-#define MDBX_ENABLE_MADVISE 1
-#elif !(MDBX_ENABLE_MADVISE == 0 || MDBX_ENABLE_MADVISE == 1)
-#error MDBX_ENABLE_MADVISE must be defined as 0 or 1
-#endif /* MDBX_ENABLE_MADVISE */
-
-/** Disable some checks to reduce an overhead and detection probability of
- * database corruption to a values closer to the LMDB. */
-#ifndef MDBX_DISABLE_PAGECHECKS
-#define MDBX_DISABLE_PAGECHECKS 0
-#elif !(MDBX_DISABLE_PAGECHECKS == 0 || MDBX_DISABLE_PAGECHECKS == 1)
-#error MDBX_DISABLE_PAGECHECKS must be defined as 0 or 1
-#endif /* MDBX_DISABLE_PAGECHECKS */
-
-#ifndef MDBX_PNL_PREALLOC_FOR_RADIXSORT
-#define MDBX_PNL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_PNL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_PNL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_PNL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
-
-#ifndef MDBX_DPL_PREALLOC_FOR_RADIXSORT
-#define MDBX_DPL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_DPL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_DPL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_DPL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
-
-/* Basically, this build-option is for TODO. Guess it should be replaced
- * with MDBX_ENABLE_WRITEMAP_SPILLING with the three variants:
- * 0/OFF = Don't track dirty pages at all and don't spilling ones.
- * This should be by-default on Linux and may-be other systems
- * (not sure: Darwin/OSX, FreeBSD, Windows 10) where kernel provides
- * properly LRU tracking and async writing on-demand.
- * 1/ON = Lite tracking of dirty pages but with LRU labels and explicit
- * spilling with msync(MS_ASYNC). */
-#ifndef MDBX_FAKE_SPILL_WRITEMAP
-#if defined(__linux__) || defined(__gnu_linux__)
-#define MDBX_FAKE_SPILL_WRITEMAP 1 /* msync(MS_ASYNC) is no-op on Linux */
-#else
-#define MDBX_FAKE_SPILL_WRITEMAP 0
-#endif
-#elif !(MDBX_FAKE_SPILL_WRITEMAP == 0 || MDBX_FAKE_SPILL_WRITEMAP == 1)
-#error MDBX_FAKE_SPILL_WRITEMAP must be defined as 0 or 1
-#endif /* MDBX_FAKE_SPILL_WRITEMAP */
-
-/** Controls sort order of internal page number lists.
- * This mostly experimental/advanced option with not for regular MDBX users.
- * \warning The database format depend on this option and libmdbx builded with
- * different option value are incompatible. */
-#ifndef MDBX_PNL_ASCENDING
-#define MDBX_PNL_ASCENDING 0
-#elif !(MDBX_PNL_ASCENDING == 0 || MDBX_PNL_ASCENDING == 1)
-#error MDBX_PNL_ASCENDING must be defined as 0 or 1
-#endif /* MDBX_PNL_ASCENDING */
-
-/** Avoid dependence from MSVC CRT and use ntdll.dll instead. */
-#ifndef MDBX_WITHOUT_MSVC_CRT
-#define MDBX_WITHOUT_MSVC_CRT 1
-#elif !(MDBX_WITHOUT_MSVC_CRT == 0 || MDBX_WITHOUT_MSVC_CRT == 1)
-#error MDBX_WITHOUT_MSVC_CRT must be defined as 0 or 1
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-/** Size of buffer used during copying a environment/database file. */
-#ifndef MDBX_ENVCOPY_WRITEBUF
-#define MDBX_ENVCOPY_WRITEBUF 1048576u
-#elif MDBX_ENVCOPY_WRITEBUF < 65536u || MDBX_ENVCOPY_WRITEBUF > 1073741824u || \
- MDBX_ENVCOPY_WRITEBUF % 65536u
-#error MDBX_ENVCOPY_WRITEBUF must be defined in range 65536..1073741824 and be multiple of 65536
-#endif /* MDBX_ENVCOPY_WRITEBUF */
-
-/** Forces assertion checking */
-#ifndef MDBX_FORCE_ASSERTIONS
-#define MDBX_FORCE_ASSERTIONS 0
-#elif !(MDBX_FORCE_ASSERTIONS == 0 || MDBX_FORCE_ASSERTIONS == 1)
-#error MDBX_FORCE_ASSERTIONS must be defined as 0 or 1
-#endif /* MDBX_FORCE_ASSERTIONS */
-
-/** Presumed malloc size overhead for each allocation
- * to adjust allocations to be more aligned. */
-#ifndef MDBX_ASSUME_MALLOC_OVERHEAD
-#ifdef __SIZEOF_POINTER__
-#define MDBX_ASSUME_MALLOC_OVERHEAD (__SIZEOF_POINTER__ * 2u)
-#else
-#define MDBX_ASSUME_MALLOC_OVERHEAD (sizeof(void *) * 2u)
-#endif
-#elif MDBX_ASSUME_MALLOC_OVERHEAD < 0 || MDBX_ASSUME_MALLOC_OVERHEAD > 64 || \
- MDBX_ASSUME_MALLOC_OVERHEAD % 4
-#error MDBX_ASSUME_MALLOC_OVERHEAD must be defined in range 0..64 and be multiple of 4
-#endif /* MDBX_ASSUME_MALLOC_OVERHEAD */
-
-/** In case the MDBX_DEBUG is undefined set it corresponding to NDEBUG */
-#ifndef MDBX_DEBUG
-#ifdef NDEBUG
-#define MDBX_DEBUG 0
-#else
-#define MDBX_DEBUG 1
-#endif
-#endif /* MDBX_DEBUG */
-
-/** If defined then enables integration with Valgrind,
- * a memory analyzing tool. */
-#ifndef MDBX_USE_VALGRIND
-#endif /* MDBX_USE_VALGRIND */
-
-/** If defined then enables use C11 atomics,
- * otherwise detects ones availability automatically. */
-#ifndef MDBX_HAVE_C11ATOMICS
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-//------------------------------------------------------------------------------
-
-/** Win32 File Locking API for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_WIN32FILES -1
-
-/** SystemV IPC semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_SYSV 5
-
-/** POSIX-1 Shared anonymous semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX1988 1988
-
-/** POSIX-2001 Shared Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2001 2001
-
-/** POSIX-2008 Robust Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2008 2008
-
-/** BeOS Benaphores, aka Futexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_BENAPHORE 1995
-
-/** Advanced: Choices the locking implementation (autodetection by default). */
-#if defined(_WIN32) || defined(_WIN64)
-#define MDBX_LOCKING MDBX_LOCKING_WIN32FILES
-#else
-#ifndef MDBX_LOCKING
-#if defined(_POSIX_THREAD_PROCESS_SHARED) && \
- _POSIX_THREAD_PROCESS_SHARED >= 200112L && !defined(__FreeBSD__)
-
-/* Some platforms define the EOWNERDEAD error code even though they
- * don't support Robust Mutexes. If doubt compile with -MDBX_LOCKING=2001. */
-#if defined(EOWNERDEAD) && _POSIX_THREAD_PROCESS_SHARED >= 200809L && \
- ((defined(_POSIX_THREAD_ROBUST_PRIO_INHERIT) && \
- _POSIX_THREAD_ROBUST_PRIO_INHERIT > 0) || \
- (defined(_POSIX_THREAD_ROBUST_PRIO_PROTECT) && \
- _POSIX_THREAD_ROBUST_PRIO_PROTECT > 0) || \
- defined(PTHREAD_MUTEX_ROBUST) || defined(PTHREAD_MUTEX_ROBUST_NP)) && \
- (!defined(__GLIBC__) || \
- __GLIBC_PREREQ(2, 10) /* troubles with Robust mutexes before 2.10 */)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2008
-#else
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2001
-#endif
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX1988
-#else
-#define MDBX_LOCKING MDBX_LOCKING_SYSV
-#endif
-#define MDBX_LOCKING_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_LOCKING)
-#else
-#define MDBX_LOCKING_CONFIG MDBX_STRINGIFY(MDBX_LOCKING)
-#endif /* MDBX_LOCKING */
-#endif /* !Windows */
-
-/** Advanced: Using POSIX OFD-locks (autodetection by default). */
-#ifndef MDBX_USE_OFDLOCKS
-#if defined(F_OFD_SETLK) && defined(F_OFD_SETLKW) && defined(F_OFD_GETLK) && \
- !defined(MDBX_SAFE4QEMU) && \
- !defined(__sun) /* OFD-lock are broken on Solaris */
-#define MDBX_USE_OFDLOCKS 1
-#else
-#define MDBX_USE_OFDLOCKS 0
-#endif
-#define MDBX_USE_OFDLOCKS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#else
-#define MDBX_USE_OFDLOCKS_CONFIG MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#endif /* MDBX_USE_OFDLOCKS */
-
-/** Advanced: Using sendfile() syscall (autodetection by default). */
-#ifndef MDBX_USE_SENDFILE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 21)
-#define MDBX_USE_SENDFILE 1
-#else
-#define MDBX_USE_SENDFILE 0
-#endif
-#endif /* MDBX_USE_SENDFILE */
-
-/** Advanced: Using copy_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_COPYFILERANGE
-#if __GLIBC_PREREQ(2, 27) && defined(_GNU_SOURCE)
-#define MDBX_USE_COPYFILERANGE 1
-#else
-#define MDBX_USE_COPYFILERANGE 0
-#endif
-#endif /* MDBX_USE_COPYFILERANGE */
-
-/** Advanced: Using sync_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_SYNCFILERANGE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- defined(SYNC_FILE_RANGE_WRITE) && !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 26)
-#define MDBX_USE_SYNCFILERANGE 1
-#else
-#define MDBX_USE_SYNCFILERANGE 0
-#endif
-#endif /* MDBX_USE_SYNCFILERANGE */
-
-//------------------------------------------------------------------------------
-
-#ifndef MDBX_CPU_WRITEBACK_INCOHERENT
-#if defined(__ia32__) || defined(__e2k__) || defined(__hppa) || \
- defined(__hppa__) || defined(DOXYGEN)
-#define MDBX_CPU_WRITEBACK_INCOHERENT 0
-#else
-#define MDBX_CPU_WRITEBACK_INCOHERENT 1
-#endif
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-#ifndef MDBX_MMAP_INCOHERENT_FILE_WRITE
-#ifdef __OpenBSD__
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 1
-#else
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#ifndef MDBX_MMAP_INCOHERENT_CPU_CACHE
-#if defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS has cache coherency issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 1
-#else
-/* LY: assume no relevant mmap/dcache issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#ifndef MDBX_64BIT_ATOMIC
-#if MDBX_WORDBITS >= 64 || defined(DOXYGEN)
-#define MDBX_64BIT_ATOMIC 1
-#else
-#define MDBX_64BIT_ATOMIC 0
-#endif
-#define MDBX_64BIT_ATOMIC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#else
-#define MDBX_64BIT_ATOMIC_CONFIG MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#endif /* MDBX_64BIT_ATOMIC */
-
-#ifndef MDBX_64BIT_CAS
-#if defined(ATOMIC_LLONG_LOCK_FREE)
-#if ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__GCC_ATOMIC_LLONG_LOCK_FREE)
-#if __GCC_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__CLANG_ATOMIC_LLONG_LOCK_FREE)
-#if __CLANG_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(_MSC_VER) || defined(__APPLE__) || defined(DOXYGEN)
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS MDBX_64BIT_ATOMIC
-#endif
-#define MDBX_64BIT_CAS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#else
-#define MDBX_64BIT_CAS_CONFIG MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#endif /* MDBX_64BIT_CAS */
-
-#ifndef MDBX_UNALIGNED_OK
-#if defined(__ALIGNED__) || defined(__SANITIZE_UNDEFINED__)
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#elif defined(__ARM_FEATURE_UNALIGNED)
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#elif __CLANG_PREREQ(5, 0) || __GNUC_PREREQ(5, 0)
-/* expecting an optimization will well done, also this
- * hushes false-positives from UBSAN (undefined behaviour sanitizer) */
-#define MDBX_UNALIGNED_OK 0
-#elif defined(__e2k__) || defined(__elbrus__)
-#if __iset__ > 4
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#endif
-#elif defined(__ia32__)
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#endif
-#elif MDBX_UNALIGNED_OK == 1
-#undef MDBX_UNALIGNED_OK
-#define MDBX_UNALIGNED_OK 32 /* any unaligned access allowed */
-#endif /* MDBX_UNALIGNED_OK */
-
-#ifndef MDBX_CACHELINE_SIZE
-#if defined(SYSTEM_CACHE_ALIGNMENT_SIZE)
-#define MDBX_CACHELINE_SIZE SYSTEM_CACHE_ALIGNMENT_SIZE
-#elif defined(__ia64__) || defined(__ia64) || defined(_M_IA64)
-#define MDBX_CACHELINE_SIZE 128
-#else
-#define MDBX_CACHELINE_SIZE 64
-#endif
-#endif /* MDBX_CACHELINE_SIZE */
-
-/** @} end of build options */
-/*******************************************************************************
- *******************************************************************************
- ******************************************************************************/
-
-#ifdef DOXYGEN
-/* !!! Actually this is a fake definitions !!!
- * !!! for documentation generation by Doxygen !!! */
-
-/** Controls enabling of debugging features.
- *
- * - `MDBX_DEBUG = 0` (by default) Disables any debugging features at all,
- * including logging and assertion controls.
- * Logging level and corresponding debug flags changing
- * by \ref mdbx_setup_debug() will not have effect.
- * - `MDBX_DEBUG > 0` Enables code for the debugging features (logging,
- * assertions checking and internal audit).
- * Simultaneously sets the default logging level
- * to the `MDBX_DEBUG` value.
- * Also enables \ref MDBX_DBG_AUDIT if `MDBX_DEBUG >= 2`.
- *
- * \ingroup build_option */
-#define MDBX_DEBUG 0...7
-
-/** Disables using of GNU libc extensions. */
-#define MDBX_DISABLE_GNU_SOURCE 0 or 1
-
-#endif /* DOXYGEN */
-
-/* Undefine the NDEBUG if debugging is enforced by MDBX_DEBUG */
-#if MDBX_DEBUG
-#undef NDEBUG
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-enum MDBX_memory_order {
- mo_Relaxed,
- mo_AcquireRelease,
- mo_SequentialConsistency
-};
-
-typedef union {
- volatile uint32_t weak;
-#ifdef MDBX_HAVE_C11ATOMICS
- volatile _Atomic uint32_t c11a;
-#endif /* MDBX_HAVE_C11ATOMICS */
-} MDBX_atomic_uint32_t;
-
-typedef union {
- volatile uint64_t weak;
-#if defined(MDBX_HAVE_C11ATOMICS) && (MDBX_64BIT_CAS || MDBX_64BIT_ATOMIC)
- volatile _Atomic uint64_t c11a;
-#endif
-#if !defined(MDBX_HAVE_C11ATOMICS) || !MDBX_64BIT_CAS || !MDBX_64BIT_ATOMIC
- __anonymous_struct_extension__ struct {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- MDBX_atomic_uint32_t low, high;
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- MDBX_atomic_uint32_t high, low;
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- };
-#endif
-} MDBX_atomic_uint64_t;
-
-#ifdef MDBX_HAVE_C11ATOMICS
-
-/* Crutches for C11 atomic compiler's bugs */
-#if defined(__e2k__) && defined(__LCC__) && __LCC__ < /* FIXME */ 127
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->weak)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->weak)
-#elif defined(__clang__) && __clang__ < 8
-#define MDBX_c11a_ro(type, ptr) ((volatile _Atomic(type) *)&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#else
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#endif /* Crutches for C11 atomic compiler's bugs */
-
-static __always_inline memory_order mo_c11_store(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_release;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-
-static __always_inline memory_order mo_c11_load(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_acquire;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED static __always_inline void
-mdbx_memory_fence(enum MDBX_memory_order order, bool write) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(write ? mo_c11_store(order) : mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_compiler_barrier();
- if (write &&
- order > (MDBX_CPU_WRITEBACK_INCOHERENT ? mo_Relaxed : mo_AcquireRelease))
- mdbx_memory_barrier();
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_store32(MDBX_atomic_uint32_t *p, const uint32_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint32_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_load32(const MDBX_atomic_uint32_t *p, enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint32_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint32_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint32_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-/* Basic constants and types */
-
-/* A stamp that identifies a file as an MDBX file.
- * There's nothing special about this value other than that it is easily
- * recognizable, and it will reflect any byte order mismatches. */
-#define MDBX_MAGIC UINT64_C(/* 56-bit prime */ 0x59659DBDEF4C11)
-
-/* FROZEN: The version number for a database's datafile format. */
-#define MDBX_DATA_VERSION 3
-/* The version number for a database's lockfile format. */
-#define MDBX_LOCK_VERSION 4
-
-/* handle for the DB used to track free pages. */
-#define FREE_DBI 0
-/* handle for the default DB. */
-#define MAIN_DBI 1
-/* Number of DBs in metapage (free and main) - also hardcoded elsewhere */
-#define CORE_DBS 2
-
-/* Number of meta pages - also hardcoded elsewhere */
-#define NUM_METAS 3
-
-/* A page number in the database.
- *
- * MDBX uses 32 bit for page numbers. This limits database
- * size up to 2^44 bytes, in case of 4K pages. */
-typedef uint32_t pgno_t;
-typedef MDBX_atomic_uint32_t atomic_pgno_t;
-#define PRIaPGNO PRIu32
-#define MAX_PAGENO UINT32_C(0x7FFFffff)
-#define MIN_PAGENO NUM_METAS
-
-#define SAFE64_INVALID_THRESHOLD UINT64_C(0xffffFFFF00000000)
-
-/* A transaction ID. */
-typedef uint64_t txnid_t;
-typedef MDBX_atomic_uint64_t atomic_txnid_t;
-#define PRIaTXN PRIi64
-#define MIN_TXNID UINT64_C(1)
-#define MAX_TXNID (SAFE64_INVALID_THRESHOLD - 1)
-#define INITIAL_TXNID (MIN_TXNID + NUM_METAS - 1)
-#define INVALID_TXNID UINT64_MAX
-/* LY: for testing non-atomic 64-bit txnid on 32-bit arches.
- * #define xMDBX_TXNID_STEP (UINT32_MAX / 3) */
-#ifndef xMDBX_TXNID_STEP
-#if MDBX_64BIT_CAS
-#define xMDBX_TXNID_STEP 1u
-#else
-#define xMDBX_TXNID_STEP 2u
-#endif
-#endif /* xMDBX_TXNID_STEP */
-
-/* Used for offsets within a single page.
- * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
- * this is plenty. */
-typedef uint16_t indx_t;
-
-#define MEGABYTE ((size_t)1 << 20)
-
-/*----------------------------------------------------------------------------*/
-/* Core structures for database and shared memory (i.e. format definition) */
-#pragma pack(push, 4)
-
-/* Information about a single database in the environment. */
-typedef struct MDBX_db {
- uint16_t md_flags; /* see mdbx_dbi_open */
- uint16_t md_depth; /* depth of this tree */
- uint32_t md_xsize; /* key-size for MDBX_DUPFIXED (LEAF2 pages) */
- pgno_t md_root; /* the root page of this tree */
- pgno_t md_branch_pages; /* number of internal pages */
- pgno_t md_leaf_pages; /* number of leaf pages */
- pgno_t md_overflow_pages; /* number of overflow pages */
- uint64_t md_seq; /* table sequence counter */
- uint64_t md_entries; /* number of data items */
- uint64_t md_mod_txnid; /* txnid of last committed modification */
-} MDBX_db;
-
-/* database size-related parameters */
-typedef struct MDBX_geo {
- uint16_t grow_pv; /* datafile growth step as a 16-bit packed (exponential
- quantized) value */
- uint16_t shrink_pv; /* datafile shrink threshold as a 16-bit packed
- (exponential quantized) value */
- pgno_t lower; /* minimal size of datafile in pages */
- pgno_t upper; /* maximal size of datafile in pages */
- pgno_t now; /* current size of datafile in pages */
- pgno_t next; /* first unused page in the datafile,
- but actually the file may be shorter. */
-} MDBX_geo;
-
-/* Meta page content.
- * A meta page is the start point for accessing a database snapshot.
- * Pages 0-1 are meta pages. Transaction N writes meta page (N % 2). */
-typedef struct MDBX_meta {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with MDBX_DATA_VERSION. */
- uint32_t mm_magic_and_version[2];
-
- /* txnid that committed this page, the first of a two-phase-update pair */
- uint32_t mm_txnid_a[2];
-
- uint16_t mm_extra_flags; /* extra DB flags, zero (nothing) for now */
- uint8_t mm_validator_id; /* ID of checksum and page validation method,
- * zero (nothing) for now */
- uint8_t mm_extra_pagehdr; /* extra bytes in the page header,
- * zero (nothing) for now */
-
- MDBX_geo mm_geo; /* database size-related parameters */
-
- MDBX_db mm_dbs[CORE_DBS]; /* first is free space, 2nd is main db */
- /* The size of pages used in this DB */
-#define mm_psize mm_dbs[FREE_DBI].md_xsize
- MDBX_canary mm_canary;
-
-#define MDBX_DATASIGN_NONE 0u
-#define MDBX_DATASIGN_WEAK 1u
-#define SIGN_IS_STEADY(sign) ((sign) > MDBX_DATASIGN_WEAK)
-#define META_IS_STEADY(meta) \
- SIGN_IS_STEADY(unaligned_peek_u64_volatile(4, (meta)->mm_datasync_sign))
- uint32_t mm_datasync_sign[2];
-
- /* txnid that committed this page, the second of a two-phase-update pair */
- uint32_t mm_txnid_b[2];
-
- /* Number of non-meta pages which were put in GC after COW. May be 0 in case
- * DB was previously handled by libmdbx without corresponding feature.
- * This value in couple with mr_snapshot_pages_retired allows fast estimation
- * of "how much reader is restraining GC recycling". */
- uint32_t mm_pages_retired[2];
-
- /* The analogue /proc/sys/kernel/random/boot_id or similar to determine
- * whether the system was rebooted after the last use of the database files.
- * If there was no reboot, but there is no need to rollback to the last
- * steady sync point. Zeros mean that no relevant information is available
- * from the system. */
- bin128_t mm_bootid;
-
-} MDBX_meta;
-
-#pragma pack(1)
-
-/* Common header for all page types. The page type depends on mp_flags.
- *
- * P_BRANCH and P_LEAF pages have unsorted 'MDBX_node's at the end, with
- * sorted mp_ptrs[] entries referring to them. Exception: P_LEAF2 pages
- * omit mp_ptrs and pack sorted MDBX_DUPFIXED values after the page header.
- *
- * P_OVERFLOW records occupy one or more contiguous pages where only the
- * first has a page header. They hold the real data of F_BIGDATA nodes.
- *
- * P_SUBP sub-pages are small leaf "pages" with duplicate data.
- * A node with flag F_DUPDATA but not F_SUBDATA contains a sub-page.
- * (Duplicate data can also go in sub-databases, which use normal pages.)
- *
- * P_META pages contain MDBX_meta, the start point of an MDBX snapshot.
- *
- * Each non-metapage up to MDBX_meta.mm_last_pg is reachable exactly once
- * in the snapshot: Either used by a database or listed in a GC record. */
-typedef struct MDBX_page {
- union {
-#define IS_FROZEN(txn, p) ((p)->mp_txnid < (txn)->mt_txnid)
-#define IS_SPILLED(txn, p) ((p)->mp_txnid == (txn)->mt_txnid)
-#define IS_SHADOWED(txn, p) ((p)->mp_txnid > (txn)->mt_txnid)
-#define IS_VALID(txn, p) ((p)->mp_txnid <= (txn)->mt_front)
-#define IS_MODIFIABLE(txn, p) ((p)->mp_txnid == (txn)->mt_front)
- uint64_t mp_txnid;
- struct MDBX_page *mp_next; /* for in-memory list of freed pages */
- };
- uint16_t mp_leaf2_ksize; /* key size if this is a LEAF2 page */
-#define P_BRANCH 0x01 /* branch page */
-#define P_LEAF 0x02 /* leaf page */
-#define P_OVERFLOW 0x04 /* overflow page */
-#define P_META 0x08 /* meta page */
-#define P_BAD 0x10 /* explicit flag for invalid/bad page */
-#define P_LEAF2 0x20 /* for MDBX_DUPFIXED records */
-#define P_SUBP 0x40 /* for MDBX_DUPSORT sub-pages */
-#define P_SPILLED 0x2000 /* spilled in parent txn */
-#define P_LOOSE 0x4000 /* page was dirtied then freed, can be reused */
-#define P_FROZEN 0x8000 /* used for retire page with known status */
-#define P_ILL_BITS (~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_SPILLED))
- uint16_t mp_flags;
- union {
- uint32_t mp_pages; /* number of overflow pages */
- __anonymous_struct_extension__ struct {
- indx_t mp_lower; /* lower bound of free space */
- indx_t mp_upper; /* upper bound of free space */
- };
- };
- pgno_t mp_pgno; /* page number */
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- indx_t mp_ptrs[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_page;
-
-/* Size of the page header, excluding dynamic data at the end */
-#define PAGEHDRSZ ((unsigned)offsetof(MDBX_page, mp_ptrs))
-
-#pragma pack(pop)
-
-#if MDBX_ENABLE_PGOP_STAT
-/* Statistics of page operations overall of all (running, completed and aborted)
- * transactions */
-typedef struct {
- MDBX_atomic_uint64_t newly; /* Quantity of a new pages added */
- MDBX_atomic_uint64_t cow; /* Quantity of pages copied for update */
- MDBX_atomic_uint64_t clone; /* Quantity of parent's dirty pages clones
- for nested transactions */
- MDBX_atomic_uint64_t split; /* Page splits */
- MDBX_atomic_uint64_t merge; /* Page merges */
- MDBX_atomic_uint64_t spill; /* Quantity of spilled dirty pages */
- MDBX_atomic_uint64_t unspill; /* Quantity of unspilled/reloaded pages */
- MDBX_atomic_uint64_t
- wops; /* Number of explicit write operations (not a pages) to a disk */
-} MDBX_pgop_stat_t;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-#if MDBX_LOCKING == MDBX_LOCKING_WIN32FILES
-#define MDBX_CLOCK_SIGN UINT32_C(0xF10C)
-typedef void mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
-
-#define MDBX_CLOCK_SIGN UINT32_C(0xF18D)
-typedef mdbx_pid_t mdbx_ipclock_t;
-#ifndef EOWNERDEAD
-#define EOWNERDEAD MDBX_RESULT_TRUE
-#endif
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#define MDBX_CLOCK_SIGN UINT32_C(0x8017)
-typedef pthread_mutex_t mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-#define MDBX_CLOCK_SIGN UINT32_C(0xFC29)
-typedef sem_t mdbx_ipclock_t;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV && !defined(__cplusplus)
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc);
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc);
-#endif /* MDBX_LOCKING */
-
-/* Reader Lock Table
- *
- * Readers don't acquire any locks for their data access. Instead, they
- * simply record their transaction ID in the reader table. The reader
- * mutex is needed just to find an empty slot in the reader table. The
- * slot's address is saved in thread-specific data so that subsequent
- * read transactions started by the same thread need no further locking to
- * proceed.
- *
- * If MDBX_NOTLS is set, the slot address is not saved in thread-specific data.
- * No reader table is used if the database is on a read-only filesystem.
- *
- * Since the database uses multi-version concurrency control, readers don't
- * actually need any locking. This table is used to keep track of which
- * readers are using data from which old transactions, so that we'll know
- * when a particular old transaction is no longer in use. Old transactions
- * that have discarded any data pages can then have those pages reclaimed
- * for use by a later write transaction.
- *
- * The lock table is constructed such that reader slots are aligned with the
- * processor's cache line size. Any slot is only ever used by one thread.
- * This alignment guarantees that there will be no contention or cache
- * thrashing as threads update their own slot info, and also eliminates
- * any need for locking when accessing a slot.
- *
- * A writer thread will scan every slot in the table to determine the oldest
- * outstanding reader transaction. Any freed pages older than this will be
- * reclaimed by the writer. The writer doesn't use any locks when scanning
- * this table. This means that there's no guarantee that the writer will
- * see the most up-to-date reader info, but that's not required for correct
- * operation - all we need is to know the upper bound on the oldest reader,
- * we don't care at all about the newest reader. So the only consequence of
- * reading stale information here is that old pages might hang around a
- * while longer before being reclaimed. That's actually good anyway, because
- * the longer we delay reclaiming old pages, the more likely it is that a
- * string of contiguous pages can be found after coalescing old pages from
- * many old transactions together. */
-
-/* The actual reader record, with cacheline padding. */
-typedef struct MDBX_reader {
- /* Current Transaction ID when this transaction began, or (txnid_t)-1.
- * Multiple readers that start at the same time will probably have the
- * same ID here. Again, it's not important to exclude them from
- * anything; all we need to know is which version of the DB they
- * started from so we can avoid overwriting any data used in that
- * particular version. */
- MDBX_atomic_uint64_t /* txnid_t */ mr_txnid;
-
- /* The information we store in a single slot of the reader table.
- * In addition to a transaction ID, we also record the process and
- * thread ID that owns a slot, so that we can detect stale information,
- * e.g. threads or processes that went away without cleaning up.
- *
- * NOTE: We currently don't check for stale records.
- * We simply re-init the table when we know that we're the only process
- * opening the lock file. */
-
- /* The thread ID of the thread owning this txn. */
- MDBX_atomic_uint64_t mr_tid;
-
- /* The process ID of the process owning this reader txn. */
- MDBX_atomic_uint32_t mr_pid;
-
- /* The number of pages used in the reader's MVCC snapshot,
- * i.e. the value of meta->mm_geo.next and txn->mt_next_pgno */
- atomic_pgno_t mr_snapshot_pages_used;
- /* Number of retired pages at the time this reader starts transaction. So,
- * at any time the difference mm_pages_retired - mr_snapshot_pages_retired
- * will give the number of pages which this reader restraining from reuse. */
- MDBX_atomic_uint64_t mr_snapshot_pages_retired;
-} MDBX_reader;
-
-/* The header for the reader table (a memory-mapped lock file). */
-typedef struct MDBX_lockinfo {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with with MDBX_LOCK_VERSION. */
- uint64_t mti_magic_and_version;
-
- /* Format of this lock file. Must be set to MDBX_LOCK_FORMAT. */
- uint32_t mti_os_and_format;
-
- /* Flags which environment was opened. */
- MDBX_atomic_uint32_t mti_envmode;
-
- /* Threshold of un-synced-with-disk pages for auto-sync feature,
- * zero means no-threshold, i.e. auto-sync is disabled. */
- atomic_pgno_t mti_autosync_threshold;
-
- /* Low 32-bit of txnid with which meta-pages was synced,
- * i.e. for sync-polling in the MDBX_NOMETASYNC mode. */
- MDBX_atomic_uint32_t mti_meta_sync_txnid;
-
- /* Period for timed auto-sync feature, i.e. at the every steady checkpoint
- * the mti_unsynced_timeout sets to the current_time + mti_autosync_period.
- * The time value is represented in a suitable system-dependent form, for
- * example clock_gettime(CLOCK_BOOTTIME) or clock_gettime(CLOCK_MONOTONIC).
- * Zero means timed auto-sync is disabled. */
- MDBX_atomic_uint64_t mti_autosync_period;
-
- /* Marker to distinguish uniqueness of DB/CLK. */
- MDBX_atomic_uint64_t mti_bait_uniqueness;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
-#if MDBX_ENABLE_PGOP_STAT
- /* Statistics of costly ops of all (running, completed and aborted)
- * transactions */
- MDBX_pgop_stat_t mti_pgop_stat;
-#endif /* MDBX_ENABLE_PGOP_STAT*/
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Write transaction lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_wlock;
-#endif /* MDBX_LOCKING > 0 */
-
- atomic_txnid_t mti_oldest_reader;
-
- /* Timestamp of the last steady sync. Value is represented in a suitable
- * system-dependent form, for example clock_gettime(CLOCK_BOOTTIME) or
- * clock_gettime(CLOCK_MONOTONIC). */
- MDBX_atomic_uint64_t mti_sync_timestamp;
-
- /* Number un-synced-with-disk pages for auto-sync feature. */
- atomic_pgno_t mti_unsynced_pages;
-
- /* Number of page which was discarded last time by madvise(MADV_FREE). */
- atomic_pgno_t mti_discarded_tail;
-
- /* Timestamp of the last readers check. */
- MDBX_atomic_uint64_t mti_reader_check_timestamp;
-
- /* Shared anchor for tracking readahead edge and enabled/disabled status. */
- pgno_t mti_readahead_anchor;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Readeaders registration lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_rlock;
-#endif /* MDBX_LOCKING > 0 */
-
- /* The number of slots that have been used in the reader table.
- * This always records the maximum count, it is not decremented
- * when readers release their slots. */
- MDBX_atomic_uint32_t mti_numreaders;
- MDBX_atomic_uint32_t mti_readers_refresh_flag;
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
- MDBX_reader mti_readers[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_lockinfo;
-
-/* Lockfile format signature: version, features and field layout */
-#define MDBX_LOCK_FORMAT \
- (MDBX_CLOCK_SIGN * 27733 + (unsigned)sizeof(MDBX_reader) * 13 + \
- (unsigned)offsetof(MDBX_reader, mr_snapshot_pages_used) * 251 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_oldest_reader) * 83 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_numreaders) * 37 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_readers) * 29)
-
-#define MDBX_DATA_MAGIC \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + MDBX_DATA_VERSION)
-
-#define MDBX_DATA_MAGIC_LEGACY_COMPAT \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + 2)
-
-#define MDBX_DATA_MAGIC_LEGACY_DEVEL ((MDBX_MAGIC << 8) + 255)
-
-#define MDBX_LOCK_MAGIC ((MDBX_MAGIC << 8) + MDBX_LOCK_VERSION)
-
-/* The maximum size of a database page.
- *
- * It is 64K, but value-PAGEHDRSZ must fit in MDBX_page.mp_upper.
- *
- * MDBX will use database pages < OS pages if needed.
- * That causes more I/O in write transactions: The OS must
- * know (read) the whole page before writing a partial page.
- *
- * Note that we don't currently support Huge pages. On Linux,
- * regular data files cannot use Huge pages, and in general
- * Huge pages aren't actually pageable. We rely on the OS
- * demand-pager to read our data and page it out when memory
- * pressure from other processes is high. So until OSs have
- * actual paging support for Huge pages, they're not viable. */
-#define MAX_PAGESIZE MDBX_MAX_PAGESIZE
-#define MIN_PAGESIZE MDBX_MIN_PAGESIZE
-
-#define MIN_MAPSIZE (MIN_PAGESIZE * MIN_PAGENO)
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_MAPSIZE32 UINT32_C(0x38000000)
-#else
-#define MAX_MAPSIZE32 UINT32_C(0x7f000000)
-#endif
-#define MAX_MAPSIZE64 ((MAX_PAGENO + 1) * (uint64_t)MAX_PAGESIZE)
-
-#if MDBX_WORDBITS >= 64
-#define MAX_MAPSIZE MAX_MAPSIZE64
-#define MDBX_PGL_LIMIT ((size_t)MAX_PAGENO)
-#else
-#define MAX_MAPSIZE MAX_MAPSIZE32
-#define MDBX_PGL_LIMIT (MAX_MAPSIZE32 / MIN_PAGESIZE)
-#endif /* MDBX_WORDBITS */
-
-#define MDBX_READERS_LIMIT 32767
-#define MDBX_RADIXSORT_THRESHOLD 333
-
-/*----------------------------------------------------------------------------*/
-
-/* An PNL is an Page Number List, a sorted array of IDs.
- * The first element of the array is a counter for how many actual page-numbers
- * are in the list. By default PNLs are sorted in descending order, this allow
- * cut off a page with lowest pgno (at the tail) just truncating the list. The
- * sort order of PNLs is controlled by the MDBX_PNL_ASCENDING build option. */
-typedef pgno_t *MDBX_PNL;
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_ORDERED(first, last) ((first) < (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) >= (last))
-#else
-#define MDBX_PNL_ORDERED(first, last) ((first) > (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) <= (last))
-#endif
-
-/* List of txnid, only for MDBX_txn.tw.lifo_reclaimed */
-typedef txnid_t *MDBX_TXL;
-
-/* An Dirty-Page list item is an pgno/pointer pair. */
-typedef struct MDBX_dp {
- MDBX_page *ptr;
- pgno_t pgno;
- union {
- unsigned extra;
- __anonymous_struct_extension__ struct {
- unsigned multi : 1;
- unsigned lru : 31;
- };
- };
-} MDBX_dp;
-
-/* An DPL (dirty-page list) is a sorted array of MDBX_DPs. */
-typedef struct MDBX_dpl {
- unsigned sorted;
- unsigned length;
- unsigned detent; /* allocated size excluding the MDBX_DPL_RESERVE_GAP */
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_dp items[] /* dynamic size with holes at zero and after the last */;
-#endif
-} MDBX_dpl;
-
-/* PNL sizes */
-#define MDBX_PNL_GRANULATE 1024
-#define MDBX_PNL_INITIAL \
- (MDBX_PNL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(pgno_t))
-
-#define MDBX_TXL_GRANULATE 32
-#define MDBX_TXL_INITIAL \
- (MDBX_TXL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-#define MDBX_TXL_MAX \
- ((1u << 17) - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-
-#define MDBX_PNL_ALLOCLEN(pl) ((pl)[-1])
-#define MDBX_PNL_SIZE(pl) ((pl)[0])
-#define MDBX_PNL_FIRST(pl) ((pl)[1])
-#define MDBX_PNL_LAST(pl) ((pl)[MDBX_PNL_SIZE(pl)])
-#define MDBX_PNL_BEGIN(pl) (&(pl)[1])
-#define MDBX_PNL_END(pl) (&(pl)[MDBX_PNL_SIZE(pl) + 1])
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_FIRST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_LAST(pl)
-#else
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_LAST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_FIRST(pl)
-#endif
-
-#define MDBX_PNL_SIZEOF(pl) ((MDBX_PNL_SIZE(pl) + 1) * sizeof(pgno_t))
-#define MDBX_PNL_IS_EMPTY(pl) (MDBX_PNL_SIZE(pl) == 0)
-
-/*----------------------------------------------------------------------------*/
-/* Internal structures */
-
-/* Auxiliary DB info.
- * The information here is mostly static/read-only. There is
- * only a single copy of this record in the environment. */
-typedef struct MDBX_dbx {
- MDBX_val md_name; /* name of the database */
- MDBX_cmp_func *md_cmp; /* function for comparing keys */
- MDBX_cmp_func *md_dcmp; /* function for comparing data items */
- size_t md_klen_min, md_klen_max; /* min/max key length for the database */
- size_t md_vlen_min,
- md_vlen_max; /* min/max value/data length for the database */
-} MDBX_dbx;
-
-/* A database transaction.
- * Every operation requires a transaction handle. */
-struct MDBX_txn {
-#define MDBX_MT_SIGNATURE UINT32_C(0x93D53A31)
- uint32_t mt_signature;
-
- /* Transaction Flags */
- /* mdbx_txn_begin() flags */
-#define MDBX_TXN_RO_BEGIN_FLAGS (MDBX_TXN_RDONLY | MDBX_TXN_RDONLY_PREPARE)
-#define MDBX_TXN_RW_BEGIN_FLAGS \
- (MDBX_TXN_NOMETASYNC | MDBX_TXN_NOSYNC | MDBX_TXN_TRY)
- /* Additional flag for mdbx_sync_locked() */
-#define MDBX_SHRINK_ALLOWED UINT32_C(0x40000000)
-
- /* internal txn flags */
-#define MDBX_TXN_FINISHED 0x01 /* txn is finished or never began */
-#define MDBX_TXN_ERROR 0x02 /* txn is unusable after an error */
-#define MDBX_TXN_DIRTY 0x04 /* must write, even if dirty list is empty */
-#define MDBX_TXN_SPILLS 0x08 /* txn or a parent has spilled pages */
-#define MDBX_TXN_HAS_CHILD 0x10 /* txn has an MDBX_txn.mt_child */
- /* most operations on the txn are currently illegal */
-#define MDBX_TXN_BLOCKED \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_HAS_CHILD)
-
-#define TXN_FLAGS \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_DIRTY | MDBX_TXN_SPILLS | \
- MDBX_TXN_HAS_CHILD)
-
-#if (TXN_FLAGS & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS)) || \
- ((MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS | TXN_FLAGS) & \
- MDBX_SHRINK_ALLOWED)
-#error "Oops, some flags overlapped or wrong"
-#endif
- uint32_t mt_flags;
-
- MDBX_txn *mt_parent; /* parent of a nested txn */
- /* Nested txn under this txn, set together with flag MDBX_TXN_HAS_CHILD */
- MDBX_txn *mt_child;
- MDBX_geo mt_geo;
- /* next unallocated page */
-#define mt_next_pgno mt_geo.next
- /* corresponding to the current size of datafile */
-#define mt_end_pgno mt_geo.now
-
- /* The ID of this transaction. IDs are integers incrementing from 1.
- * Only committed write transactions increment the ID. If a transaction
- * aborts, the ID may be re-used by the next writer. */
- txnid_t mt_txnid;
- txnid_t mt_front;
-
- MDBX_env *mt_env; /* the DB environment */
- /* Array of records for each DB known in the environment. */
- MDBX_dbx *mt_dbxs;
- /* Array of MDBX_db records for each known DB */
- MDBX_db *mt_dbs;
- /* Array of sequence numbers for each DB handle */
- unsigned *mt_dbiseqs;
-
- /* Transaction DBI Flags */
-#define DBI_DIRTY MDBX_DBI_DIRTY /* DB was written in this txn */
-#define DBI_STALE MDBX_DBI_STALE /* Named-DB record is older than txnID */
-#define DBI_FRESH MDBX_DBI_FRESH /* Named-DB handle opened in this txn */
-#define DBI_CREAT MDBX_DBI_CREAT /* Named-DB handle created in this txn */
-#define DBI_VALID 0x10 /* DB handle is valid, see also DB_VALID */
-#define DBI_USRVALID 0x20 /* As DB_VALID, but not set for FREE_DBI */
-#define DBI_AUDITED 0x40 /* Internal flag for accounting during audit */
- /* Array of flags for each DB */
- uint8_t *mt_dbistate;
- /* Number of DB records in use, or 0 when the txn is finished.
- * This number only ever increments until the txn finishes; we
- * don't decrement it when individual DB handles are closed. */
- MDBX_dbi mt_numdbs;
- size_t mt_owner; /* thread ID that owns this transaction */
- MDBX_canary mt_canary;
- void *mt_userctx; /* User-settable context */
- MDBX_cursor **mt_cursors;
-
- union {
- struct {
- /* For read txns: This thread/txn's reader table slot, or NULL. */
- MDBX_reader *reader;
- } to;
- struct {
- /* In write txns, array of cursors for each DB */
- pgno_t *reclaimed_pglist; /* Reclaimed GC pages */
- txnid_t last_reclaimed; /* ID of last used record */
-#if MDBX_ENABLE_REFUND
- pgno_t loose_refund_wl /* FIXME: describe */;
-#endif /* MDBX_ENABLE_REFUND */
- /* dirtylist room: Dirty array size - dirty pages visible to this txn.
- * Includes ancestor txns' dirty pages not hidden by other txns'
- * dirty/spilled pages. Thus commit(nested txn) has room to merge
- * dirtylist into mt_parent after freeing hidden mt_parent pages. */
- unsigned dirtyroom;
- /* a sequence to spilling dirty page with LRU policy */
- unsigned dirtylru;
- /* For write txns: Modified pages. Sorted when not MDBX_WRITEMAP. */
- MDBX_dpl *dirtylist;
- /* The list of reclaimed txns from GC */
- MDBX_TXL lifo_reclaimed;
- /* The list of pages that became unused during this transaction. */
- MDBX_PNL retired_pages;
- /* The list of loose pages that became unused and may be reused
- * in this transaction, linked through `mp_next`. */
- MDBX_page *loose_pages;
- /* Number of loose pages (tw.loose_pages) */
- unsigned loose_count;
- /* The sorted list of dirty pages we temporarily wrote to disk
- * because the dirty list was full. page numbers in here are
- * shifted left by 1, deleted slots have the LSB set. */
- MDBX_PNL spill_pages;
- unsigned spill_least_removed;
- } tw;
- };
-};
-
-#if MDBX_WORDBITS >= 64
-#define CURSOR_STACK 32
-#else
-#define CURSOR_STACK 24
-#endif
-
-struct MDBX_xcursor;
-
-/* Cursors are used for all DB operations.
- * A cursor holds a path of (page pointer, key index) from the DB
- * root to a position in the DB, plus other state. MDBX_DUPSORT
- * cursors include an xcursor to the current data item. Write txns
- * track their cursors and keep them up to date when data moves.
- * Exception: An xcursor's pointer to a P_SUBP page can be stale.
- * (A node with F_DUPDATA but no F_SUBDATA contains a subpage). */
-struct MDBX_cursor {
-#define MDBX_MC_LIVE UINT32_C(0xFE05D5B1)
-#define MDBX_MC_READY4CLOSE UINT32_C(0x2817A047)
-#define MDBX_MC_WAIT4EOT UINT32_C(0x90E297A7)
- uint32_t mc_signature;
- /* The database handle this cursor operates on */
- MDBX_dbi mc_dbi;
- /* Next cursor on this DB in this txn */
- MDBX_cursor *mc_next;
- /* Backup of the original cursor if this cursor is a shadow */
- MDBX_cursor *mc_backup;
- /* Context used for databases with MDBX_DUPSORT, otherwise NULL */
- struct MDBX_xcursor *mc_xcursor;
- /* The transaction that owns this cursor */
- MDBX_txn *mc_txn;
- /* The database record for this cursor */
- MDBX_db *mc_db;
- /* The database auxiliary record for this cursor */
- MDBX_dbx *mc_dbx;
- /* The mt_dbistate for this database */
- uint8_t *mc_dbistate;
- unsigned mc_snum; /* number of pushed pages */
- unsigned mc_top; /* index of top page, normally mc_snum-1 */
-
- /* Cursor state flags. */
-#define C_INITIALIZED 0x01 /* cursor has been initialized and is valid */
-#define C_EOF 0x02 /* No more data */
-#define C_SUB 0x04 /* Cursor is a sub-cursor */
-#define C_DEL 0x08 /* last op was a cursor_del */
-#define C_UNTRACK 0x10 /* Un-track cursor when closing */
-#define C_RECLAIMING 0x20 /* GC lookup is prohibited */
-#define C_GCFREEZE 0x40 /* reclaimed_pglist must not be updated */
-
- /* Cursor checking flags. */
-#define C_COPYING 0x100 /* skip key-value length check (copying simplify) */
-#define C_UPDATING 0x200 /* update/rebalance pending */
-#define C_RETIRING 0x400 /* refs to child pages may be invalid */
-#define C_SKIPORD 0x800 /* don't check keys ordering */
-
- unsigned mc_flags; /* see mdbx_cursor */
- MDBX_page *mc_pg[CURSOR_STACK]; /* stack of pushed pages */
- indx_t mc_ki[CURSOR_STACK]; /* stack of page indices */
-};
-
-/* Context for sorted-dup records.
- * We could have gone to a fully recursive design, with arbitrarily
- * deep nesting of sub-databases. But for now we only handle these
- * levels - main DB, optional sub-DB, sorted-duplicate DB. */
-typedef struct MDBX_xcursor {
- /* A sub-cursor for traversing the Dup DB */
- MDBX_cursor mx_cursor;
- /* The database record for this Dup DB */
- MDBX_db mx_db;
- /* The auxiliary DB record for this Dup DB */
- MDBX_dbx mx_dbx;
-} MDBX_xcursor;
-
-typedef struct MDBX_cursor_couple {
- MDBX_cursor outer;
- void *mc_userctx; /* User-settable context */
- MDBX_xcursor inner;
-} MDBX_cursor_couple;
-
-/* The database environment. */
-struct MDBX_env {
- /* ----------------------------------------------------- mostly static part */
-#define MDBX_ME_SIGNATURE UINT32_C(0x9A899641)
- MDBX_atomic_uint32_t me_signature;
- /* Failed to update the meta page. Probably an I/O error. */
-#define MDBX_FATAL_ERROR UINT32_C(0x80000000)
- /* Some fields are initialized. */
-#define MDBX_ENV_ACTIVE UINT32_C(0x20000000)
- /* me_txkey is set */
-#define MDBX_ENV_TXKEY UINT32_C(0x10000000)
- /* Legacy MDBX_MAPASYNC (prior v0.9) */
-#define MDBX_DEPRECATED_MAPASYNC UINT32_C(0x100000)
-#define ENV_INTERNAL_FLAGS (MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE | MDBX_ENV_TXKEY)
- uint32_t me_flags;
- mdbx_mmap_t me_dxb_mmap; /* The main data file */
-#define me_map me_dxb_mmap.dxb
-#define me_lazy_fd me_dxb_mmap.fd
- mdbx_filehandle_t me_dsync_fd;
- mdbx_mmap_t me_lck_mmap; /* The lock file */
-#define me_lfd me_lck_mmap.fd
- struct MDBX_lockinfo *me_lck;
-
- unsigned me_psize; /* DB page size, initialized from me_os_psize */
- unsigned me_leaf_nodemax; /* max size of a leaf-node */
- uint8_t me_psize2log; /* log2 of DB page size */
- int8_t me_stuck_meta; /* recovery-only: target meta page or less that zero */
- uint16_t me_merge_threshold,
- me_merge_threshold_gc; /* pages emptier than this are candidates for
- merging */
- unsigned me_os_psize; /* OS page size, from mdbx_syspagesize() */
- unsigned me_maxreaders; /* size of the reader table */
- MDBX_dbi me_maxdbs; /* size of the DB table */
- uint32_t me_pid; /* process ID of this env */
- mdbx_thread_key_t me_txkey; /* thread-key for readers */
- char *me_pathname; /* path to the DB files */
- void *me_pbuf; /* scratch area for DUPSORT put() */
- MDBX_txn *me_txn0; /* preallocated write transaction */
-
- MDBX_dbx *me_dbxs; /* array of static DB info */
- uint16_t *me_dbflags; /* array of flags from MDBX_db.md_flags */
- unsigned *me_dbiseqs; /* array of dbi sequence numbers */
- unsigned
- me_maxgc_ov1page; /* Number of pgno_t fit in a single overflow page */
- uint32_t me_live_reader; /* have liveness lock in reader table */
- void *me_userctx; /* User-settable context */
- MDBX_hsr_func *me_hsr_callback; /* Callback for kicking laggard readers */
-
- struct {
- unsigned dp_reserve_limit;
- unsigned rp_augment_limit;
- unsigned dp_limit;
- unsigned dp_initial;
- uint8_t dp_loose_limit;
- uint8_t spill_max_denominator;
- uint8_t spill_min_denominator;
- uint8_t spill_parent4child_denominator;
- unsigned merge_threshold_16dot16_percent;
- union {
- unsigned all;
- /* tracks options with non-auto values but tuned by user */
- struct {
- unsigned dp_limit : 1;
- } non_auto;
- } flags;
- } me_options;
-
- /* struct me_dbgeo used for accepting db-geo params from user for the new
- * database creation, i.e. when mdbx_env_set_geometry() was called before
- * mdbx_env_open(). */
- struct {
- size_t lower; /* minimal size of datafile */
- size_t upper; /* maximal size of datafile */
- size_t now; /* current size of datafile */
- size_t grow; /* step to grow datafile */
- size_t shrink; /* threshold to shrink datafile */
- } me_dbgeo;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- union {
- key_t key;
- int semid;
- } me_sysv_ipc;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
- MDBX_env *me_lcklist_next;
-
- /* --------------------------------------------------- mostly volatile part */
-
- MDBX_txn *me_txn; /* current write transaction */
- mdbx_fastmutex_t me_dbi_lock;
- MDBX_dbi me_numdbs; /* number of DBs opened */
-
- MDBX_page *me_dp_reserve; /* list of malloc'ed blocks for re-use */
- unsigned me_dp_reserve_len;
- /* PNL of pages that became unused in a write txn */
- MDBX_PNL me_retired_pages;
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_srwlock me_remap_guard;
- /* Workaround for LockFileEx and WriteFile multithread bug */
- CRITICAL_SECTION me_windowsbug_lock;
-#else
- mdbx_fastmutex_t me_remap_guard;
-#endif
-
- /* -------------------------------------------------------------- debugging */
-
-#if MDBX_DEBUG
- MDBX_assert_func *me_assert_func; /* Callback for assertion failures */
-#endif
-#ifdef MDBX_USE_VALGRIND
- int me_valgrind_handle;
-#endif
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- pgno_t me_poison_edge;
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-#ifndef xMDBX_DEBUG_SPILLING
-#define xMDBX_DEBUG_SPILLING 0
-#endif
-#if xMDBX_DEBUG_SPILLING == 2
- unsigned debug_dirtied_est, debug_dirtied_act;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- /* ------------------------------------------------- stub for lck-less mode */
- MDBX_atomic_uint64_t
- x_lckless_stub[(sizeof(MDBX_lockinfo) + MDBX_CACHELINE_SIZE - 1) /
- sizeof(MDBX_atomic_uint64_t)];
-};
-
-#ifndef __cplusplus
-/*----------------------------------------------------------------------------*/
-/* Debug and Logging stuff */
-
-#define MDBX_RUNTIME_FLAGS_INIT \
- ((MDBX_DEBUG) > 0) * MDBX_DBG_ASSERT + ((MDBX_DEBUG) > 1) * MDBX_DBG_AUDIT
-
-extern uint8_t mdbx_runtime_flags;
-extern uint8_t mdbx_loglevel;
-extern MDBX_debug_func *mdbx_debug_logger;
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny) {
-#if MDBX_DEBUG
- if (MDBX_DBG_JITTER & mdbx_runtime_flags)
- mdbx_osal_jitter(tiny);
-#else
- (void)tiny;
-#endif
-}
-
-MDBX_INTERNAL_FUNC void MDBX_PRINTF_ARGS(4, 5)
- mdbx_debug_log(int level, const char *function, int line, const char *fmt,
- ...) MDBX_PRINTF_ARGS(4, 5);
-MDBX_INTERNAL_FUNC void mdbx_debug_log_va(int level, const char *function,
- int line, const char *fmt,
- va_list args);
-
-#if MDBX_DEBUG
-#define mdbx_log_enabled(msg) unlikely(msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() unlikely((mdbx_runtime_flags & MDBX_DBG_AUDIT))
-#else /* MDBX_DEBUG */
-#define mdbx_log_enabled(msg) (msg < MDBX_LOG_VERBOSE && msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() (0)
-#endif /* MDBX_DEBUG */
-
-#if MDBX_FORCE_ASSERTIONS
-#define mdbx_assert_enabled() (1)
-#elif MDBX_DEBUG
-#define mdbx_assert_enabled() likely((mdbx_runtime_flags & MDBX_DBG_ASSERT))
-#else
-#define mdbx_assert_enabled() (0)
-#endif /* assertions */
-
-#if !MDBX_DEBUG && defined(__ANDROID_API__)
-#define mdbx_assert_fail(env, msg, func, line) \
- __android_log_assert(msg, "mdbx", "%s:%u", func, line)
-#else
-void mdbx_assert_fail(const MDBX_env *env, const char *msg, const char *func,
- int line);
-#endif
-
-#define mdbx_debug_extra(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, __func__, __LINE__, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug_extra_print(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, NULL, 0, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_trace(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_TRACE)) \
- mdbx_debug_log(MDBX_LOG_TRACE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_DEBUG)) \
- mdbx_debug_log(MDBX_LOG_DEBUG, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_verbose(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_VERBOSE)) \
- mdbx_debug_log(MDBX_LOG_VERBOSE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_notice(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_NOTICE)) \
- mdbx_debug_log(MDBX_LOG_NOTICE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_warning(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_WARN)) \
- mdbx_debug_log(MDBX_LOG_WARN, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_error(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) \
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_fatal(fmt, ...) \
- mdbx_debug_log(MDBX_LOG_FATAL, __func__, __LINE__, fmt "\n", __VA_ARGS__);
-
-#define mdbx_ensure_msg(env, expr, msg) \
- do { \
- if (unlikely(!(expr))) \
- mdbx_assert_fail(env, msg, __func__, __LINE__); \
- } while (0)
-
-#define mdbx_ensure(env, expr) mdbx_ensure_msg(env, expr, #expr)
-
-/* assert(3) variant in environment context */
-#define mdbx_assert(env, expr) \
- do { \
- if (mdbx_assert_enabled()) \
- mdbx_ensure(env, expr); \
- } while (0)
-
-/* assert(3) variant in cursor context */
-#define mdbx_cassert(mc, expr) mdbx_assert((mc)->mc_txn->mt_env, expr)
-
-/* assert(3) variant in transaction context */
-#define mdbx_tassert(txn, expr) mdbx_assert((txn)->mt_env, expr)
-
-#ifndef xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-#undef assert
-#define assert(expr) mdbx_assert(NULL, expr)
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Cache coherence and mmap invalidation */
-
-#if MDBX_CPU_WRITEBACK_INCOHERENT
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_memory_barrier()
-#else
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_compiler_barrier()
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-MDBX_MAYBE_UNUSED static __inline void
-mdbx_flush_incoherent_mmap(void *addr, size_t nbytes, const intptr_t pagesize) {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- char *const begin = (char *)(-pagesize & (intptr_t)addr);
- char *const end =
- (char *)(-pagesize & (intptr_t)((char *)addr + nbytes + pagesize - 1));
- int err = msync(begin, end - begin, MS_SYNC | MS_INVALIDATE) ? errno : 0;
- mdbx_assert(nullptr, err == 0);
- (void)err;
-#else
- (void)pagesize;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#if MDBX_MMAP_INCOHERENT_CPU_CACHE
-#ifdef DCACHE
- /* MIPS has cache coherency issues.
- * Note: for any nbytes >= on-chip cache size, entire is flushed. */
- cacheflush(addr, nbytes, DCACHE);
-#else
-#error "Oops, cacheflush() not available"
-#endif /* DCACHE */
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#if !MDBX_MMAP_INCOHERENT_FILE_WRITE && !MDBX_MMAP_INCOHERENT_CPU_CACHE
- (void)addr;
- (void)nbytes;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* Internal prototypes */
-
-MDBX_INTERNAL_FUNC int mdbx_cleanup_dead_readers(MDBX_env *env, int rlocked,
- int *dead);
-MDBX_INTERNAL_FUNC int mdbx_rthc_alloc(mdbx_thread_key_t *key,
- MDBX_reader *begin, MDBX_reader *end);
-MDBX_INTERNAL_FUNC void mdbx_rthc_remove(const mdbx_thread_key_t key);
-
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_init(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_dtor(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_thread_dtor(void *ptr);
-
-#endif /* !__cplusplus */
-
-#define MDBX_IS_ERROR(rc) \
- ((rc) != MDBX_RESULT_TRUE && (rc) != MDBX_RESULT_FALSE)
-
-/* Internal error codes, not exposed outside libmdbx */
-#define MDBX_NO_ROOT (MDBX_LAST_ADDED_ERRCODE + 10)
-
-/* Debugging output value of a cursor DBI: Negative in a sub-cursor. */
-#define DDBI(mc) \
- (((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
-
-/* Key size which fits in a DKBUF (debug key buffer). */
-#define DKBUF_MAX 511
-#define DKBUF char _kbuf[DKBUF_MAX * 4 + 2]
-#define DKEY(x) mdbx_dump_val(x, _kbuf, DKBUF_MAX * 2 + 1)
-#define DVAL(x) mdbx_dump_val(x, _kbuf + DKBUF_MAX * 2 + 1, DKBUF_MAX * 2 + 1)
-
-#if MDBX_DEBUG
-#define DKBUF_DEBUG DKBUF
-#define DKEY_DEBUG(x) DKEY(x)
-#define DVAL_DEBUG(x) DVAL(x)
-#else
-#define DKBUF_DEBUG ((void)(0))
-#define DKEY_DEBUG(x) ("-")
-#define DVAL_DEBUG(x) ("-")
-#endif
-
-/* An invalid page number.
- * Mainly used to denote an empty tree. */
-#define P_INVALID (~(pgno_t)0)
-
-/* Test if the flags f are set in a flag word w. */
-#define F_ISSET(w, f) (((w) & (f)) == (f))
-
-/* Round n up to an even number. */
-#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
-
-/* Default size of memory map.
- * This is certainly too small for any actual applications. Apps should
- * always set the size explicitly using mdbx_env_set_geometry(). */
-#define DEFAULT_MAPSIZE MEGABYTE
-
-/* Number of slots in the reader table.
- * This value was chosen somewhat arbitrarily. The 61 is a prime number,
- * and such readers plus a couple mutexes fit into single 4KB page.
- * Applications should set the table size using mdbx_env_set_maxreaders(). */
-#define DEFAULT_READERS 61
-
-/* Test if a page is a leaf page */
-#define IS_LEAF(p) (((p)->mp_flags & P_LEAF) != 0)
-/* Test if a page is a LEAF2 page */
-#define IS_LEAF2(p) unlikely(((p)->mp_flags & P_LEAF2) != 0)
-/* Test if a page is a branch page */
-#define IS_BRANCH(p) (((p)->mp_flags & P_BRANCH) != 0)
-/* Test if a page is an overflow page */
-#define IS_OVERFLOW(p) unlikely(((p)->mp_flags & P_OVERFLOW) != 0)
-/* Test if a page is a sub page */
-#define IS_SUBP(p) (((p)->mp_flags & P_SUBP) != 0)
-
-#define PAGETYPE(p) ((p)->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW))
-
-/* Header for a single key/data pair within a page.
- * Used in pages of type P_BRANCH and P_LEAF without P_LEAF2.
- * We guarantee 2-byte alignment for 'MDBX_node's.
- *
- * Leaf node flags describe node contents. F_BIGDATA says the node's
- * data part is the page number of an overflow page with actual data.
- * F_DUPDATA and F_SUBDATA can be combined giving duplicate data in
- * a sub-page/sub-database, and named databases (just F_SUBDATA). */
-typedef struct MDBX_node {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- union {
- uint32_t mn_dsize;
- uint32_t mn_pgno32;
- };
- uint8_t mn_flags; /* see mdbx_node flags */
- uint8_t mn_extra;
- uint16_t mn_ksize; /* key size */
-#else
- uint16_t mn_ksize; /* key size */
- uint8_t mn_extra;
- uint8_t mn_flags; /* see mdbx_node flags */
- union {
- uint32_t mn_pgno32;
- uint32_t mn_dsize;
- };
-#endif /* __BYTE_ORDER__ */
-
- /* mdbx_node Flags */
-#define F_BIGDATA 0x01 /* data put on overflow page */
-#define F_SUBDATA 0x02 /* data is a sub-database */
-#define F_DUPDATA 0x04 /* data has duplicates */
-
- /* valid flags for mdbx_node_add() */
-#define NODE_ADD_FLAGS (F_DUPDATA | F_SUBDATA | MDBX_RESERVE | MDBX_APPEND)
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- uint8_t mn_data[] /* key and data are appended here */;
-#endif /* C99 */
-} MDBX_node;
-
-#define DB_PERSISTENT_FLAGS \
- (MDBX_REVERSEKEY | MDBX_DUPSORT | MDBX_INTEGERKEY | MDBX_DUPFIXED | \
- MDBX_INTEGERDUP | MDBX_REVERSEDUP)
-
-/* mdbx_dbi_open() flags */
-#define DB_USABLE_FLAGS (DB_PERSISTENT_FLAGS | MDBX_CREATE | MDBX_DB_ACCEDE)
-
-#define DB_VALID 0x8000 /* DB handle is valid, for me_dbflags */
-#define DB_INTERNAL_FLAGS DB_VALID
-
-#if DB_INTERNAL_FLAGS & DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-#if DB_PERSISTENT_FLAGS & ~DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-
-/* max number of pages to commit in one writev() call */
-#define MDBX_COMMIT_PAGES 64
-#if defined(IOV_MAX) && IOV_MAX < MDBX_COMMIT_PAGES /* sysconf(_SC_IOV_MAX) */
-#undef MDBX_COMMIT_PAGES
-#define MDBX_COMMIT_PAGES IOV_MAX
-#endif
-
-/*
- * /
- * | -1, a < b
- * CMP2INT(a,b) = < 0, a == b
- * | 1, a > b
- * \
- */
-#ifndef __e2k__
-/* LY: fast enough on most systems */
-#define CMP2INT(a, b) (((b) > (a)) ? -1 : (a) > (b))
-#else
-/* LY: more parallelable on VLIW Elbrus */
-#define CMP2INT(a, b) (((a) > (b)) - ((b) > (a)))
-#endif
-
-/* Do not spill pages to disk if txn is getting full, may fail instead */
-#define MDBX_NOSPILL 0x8000
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-int64pgno(int64_t i64) {
- if (likely(i64 >= (int64_t)MIN_PAGENO && i64 <= (int64_t)MAX_PAGENO + 1))
- return (pgno_t)i64;
- return (i64 < (int64_t)MIN_PAGENO) ? MIN_PAGENO : MAX_PAGENO;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_add(size_t base, size_t augend) {
- assert(base <= MAX_PAGENO + 1 && augend < MAX_PAGENO);
- return int64pgno(base + augend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_sub(size_t base, size_t subtrahend) {
- assert(base >= MIN_PAGENO && base <= MAX_PAGENO + 1 &&
- subtrahend < MAX_PAGENO);
- return int64pgno(base - subtrahend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline bool
-is_powerof2(size_t x) {
- return (x & (x - 1)) == 0;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-floor_powerof2(size_t value, size_t granularity) {
- assert(is_powerof2(granularity));
- return value & ~(granularity - 1);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-ceil_powerof2(size_t value, size_t granularity) {
- return floor_powerof2(value + granularity - 1, granularity);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static unsigned
-log2n_powerof2(size_t value) {
- assert(value > 0 && value < INT32_MAX && is_powerof2(value));
- assert((value & -(int32_t)value) == value);
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_ctzl)
- return __builtin_ctzl(value);
-#elif defined(_MSC_VER)
- unsigned long index;
- _BitScanForward(&index, (unsigned long)value);
- return index;
-#else
- static const uint8_t debruijn_ctz32[32] = {
- 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
- 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9};
- return debruijn_ctz32[(uint32_t)(value * 0x077CB531u) >> 27];
-#endif
-}
-
-/* Only a subset of the mdbx_env flags can be changed
- * at runtime. Changing other flags requires closing the
- * environment and re-opening it with the new flags. */
-#define ENV_CHANGEABLE_FLAGS \
- (MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC | MDBX_DEPRECATED_MAPASYNC | \
- MDBX_NOMEMINIT | MDBX_COALESCE | MDBX_PAGEPERTURB | MDBX_ACCEDE)
-#define ENV_CHANGELESS_FLAGS \
- (MDBX_NOSUBDIR | MDBX_RDONLY | MDBX_WRITEMAP | MDBX_NOTLS | MDBX_NORDAHEAD | \
- MDBX_LIFORECLAIM | MDBX_EXCLUSIVE)
-#define ENV_USABLE_FLAGS (ENV_CHANGEABLE_FLAGS | ENV_CHANGELESS_FLAGS)
-
-#if !defined(__cplusplus) || CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_MAYBE_UNUSED static void static_checks(void) {
- STATIC_ASSERT_MSG(INT16_MAX - CORE_DBS == MDBX_MAX_DBI,
- "Oops, MDBX_MAX_DBI or CORE_DBS?");
- STATIC_ASSERT_MSG((unsigned)(MDBX_DB_ACCEDE | MDBX_CREATE) ==
- ((DB_USABLE_FLAGS | DB_INTERNAL_FLAGS) &
- (ENV_USABLE_FLAGS | ENV_INTERNAL_FLAGS)),
- "Oops, some flags overlapped or wrong");
- STATIC_ASSERT_MSG((ENV_INTERNAL_FLAGS & ENV_USABLE_FLAGS) == 0,
- "Oops, some flags overlapped or wrong");
-}
-#endif /* Disabled for MSVC 19.0 (VisualStudio 2015) */
-
-#ifdef __cplusplus
-}
-#endif
-
-#define MDBX_ASAN_POISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("POISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_POISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#define MDBX_ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("UNPOISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_UNPOISON_MEMORY_REGION(addr, size); \
- } while (0)
-//
-// Copyright (c) 2020-2022, Leonid Yuriev <leo@yuriev.ru>.
-// SPDX-License-Identifier: Apache-2.0
-//
-// Non-inline part of the libmdbx C++ API
-//
-
-#if defined(_MSC_VER) && !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-
-
-#include <array>
-#include <atomic>
-#include <cctype> // for isxdigit(), etc
-#include <system_error>
-
-namespace {
-
-#if 0 /* Unused for now */
-
-class trouble_location {
-
-#ifndef TROUBLE_PROVIDE_LINENO
-#define TROUBLE_PROVIDE_LINENO 1
-#endif
-
-#ifndef TROUBLE_PROVIDE_CONDITION
-#define TROUBLE_PROVIDE_CONDITION 1
-#endif
-
-#ifndef TROUBLE_PROVIDE_FUNCTION
-#define TROUBLE_PROVIDE_FUNCTION 1
-#endif
-
-#ifndef TROUBLE_PROVIDE_FILENAME
-#define TROUBLE_PROVIDE_FILENAME 1
-#endif
-
-#if TROUBLE_PROVIDE_LINENO
- const unsigned line_;
-#endif
-#if TROUBLE_PROVIDE_CONDITION
- const char *const condition_;
-#endif
-#if TROUBLE_PROVIDE_FUNCTION
- const char *const function_;
-#endif
-#if TROUBLE_PROVIDE_FILENAME
- const char *const filename_;
-#endif
-
-public:
- MDBX_CXX11_CONSTEXPR trouble_location(unsigned line, const char *condition,
- const char *function, const char *filename)
- :
-#if TROUBLE_PROVIDE_LINENO
- line_(line)
-#endif
-#if TROUBLE_PROVIDE_CONDITION
- ,
- condition_(condition)
-#endif
-#if TROUBLE_PROVIDE_FUNCTION
- ,
- function_(function)
-#endif
-#if TROUBLE_PROVIDE_FILENAME
- ,
- filename_(filename)
-#endif
- {
-#if !TROUBLE_PROVIDE_LINENO
- (void)line;
-#endif
-#if !TROUBLE_PROVIDE_CONDITION
- (void)condition;
-#endif
-#if !TROUBLE_PROVIDE_FUNCTION
- (void)function;
-#endif
-#if !TROUBLE_PROVIDE_FILENAME
- (void)filename;
-#endif
- }
-
- trouble_location(const trouble_location &&) = delete;
-
- unsigned line() const {
-#if TROUBLE_PROVIDE_LINENO
- return line_;
-#else
- return 0;
-#endif
- }
-
- const char *condition() const {
-#if TROUBLE_PROVIDE_CONDITION
- return condition_;
-#else
- return "";
-#endif
- }
-
- const char *function() const {
-#if TROUBLE_PROVIDE_FUNCTION
- return function_;
-#else
- return "";
-#endif
- }
-
- const char *filename() const {
-#if TROUBLE_PROVIDE_FILENAME
- return filename_;
-#else
- return "";
-#endif
- }
-};
-
-//------------------------------------------------------------------------------
-
-__cold std::string format_va(const char *fmt, va_list ap) {
- va_list ones;
- va_copy(ones, ap);
-#ifdef _MSC_VER
- int needed = _vscprintf(fmt, ap);
-#else
- int needed = vsnprintf(nullptr, 0, fmt, ap);
-#endif
- assert(needed >= 0);
- std::string result;
- result.reserve(size_t(needed + 1));
- result.resize(size_t(needed), '\0');
- assert(int(result.capacity()) > needed);
- int actual = vsnprintf(const_cast<char *>(result.data()), result.capacity(),
- fmt, ones);
- assert(actual == needed);
- (void)actual;
- va_end(ones);
- return result;
-}
-
-__cold std::string format(const char *fmt, ...) {
- va_list ap;
- va_start(ap, fmt);
- std::string result = format_va(fmt, ap);
- va_end(ap);
- return result;
-}
-
-class bug : public std::runtime_error {
- const trouble_location &location_;
-
-public:
- bug(const trouble_location &) noexcept;
- /* temporary workaround for "private field 'FOO' is not used" from CLANG
- * and for "function 'BAR' was declared but never referenced" from LCC. */
-#ifndef __LCC__
- const trouble_location &location() const noexcept { return location_; }
-#endif
- virtual ~bug() noexcept;
-};
-
-__cold bug::bug(const trouble_location &location) noexcept
- : std::runtime_error(format("mdbx.bug: %s.%s at %s:%u", location.function(),
- location.condition(), location.filename(),
- location.line())),
- location_(location) {}
-
-__cold bug::~bug() noexcept {}
-
-[[noreturn]] __cold void raise_bug(const trouble_location &what_and_where) {
- throw bug(what_and_where);
-}
-
-#define RAISE_BUG(line, condition, function, file) \
- do { \
- static MDBX_CXX11_CONSTEXPR_VAR trouble_location bug(line, condition, \
- function, file); \
- raise_bug(bug); \
- } while (0)
-
-#define ENSURE(condition) \
- do \
- if (MDBX_UNLIKELY(!(condition))) \
- MDBX_CXX20_UNLIKELY RAISE_BUG(__LINE__, #condition, __func__, __FILE__); \
- while (0)
-
-#define NOT_IMPLEMENTED() \
- RAISE_BUG(__LINE__, "not_implemented", __func__, __FILE__);
-
-#endif /* Unused*/
-
-//------------------------------------------------------------------------------
-
-template <typename PATH> struct path_to_pchar {
- const std::string str;
- path_to_pchar(const PATH &path) : str(path.generic_string()) {}
- operator const char *() const { return str.c_str(); }
-};
-
-template <typename PATH>
-MDBX_MAYBE_UNUSED PATH pchar_to_path(const char *c_str) {
- return PATH(c_str);
-}
-
-template <> struct path_to_pchar<std::string> {
- const char *const ptr;
- path_to_pchar(const std::string &path) : ptr(path.c_str()) {}
- operator const char *() const { return ptr; }
-};
-
-#if defined(_WIN32) || defined(_WIN64)
-
-#ifndef WC_ERR_INVALID_CHARS
-static const DWORD WC_ERR_INVALID_CHARS =
- (6 /* Windows Vista */ <= /* MajorVersion */ LOBYTE(LOWORD(GetVersion())))
- ? 0x00000080
- : 0;
-#endif /* WC_ERR_INVALID_CHARS */
-
-template <> struct path_to_pchar<std::wstring> {
- std::string str;
- path_to_pchar(const std::wstring &path) {
- if (!path.empty()) {
- const int chars =
- WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, path.data(),
- int(path.size()), nullptr, 0, nullptr, nullptr);
- if (chars == 0)
- mdbx::error::throw_exception(GetLastError());
- str.append(chars, '\0');
- WideCharToMultiByte(CP_UTF8, WC_ERR_INVALID_CHARS, path.data(),
- int(path.size()), const_cast<char *>(str.data()),
- chars, nullptr, nullptr);
- }
- }
- operator const char *() const { return str.c_str(); }
-};
-
-template <>
-MDBX_MAYBE_UNUSED std::wstring pchar_to_path<std::wstring>(const char *c_str) {
- std::wstring wstr;
- if (c_str && *c_str) {
- const int chars = MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, c_str,
- int(strlen(c_str)), nullptr, 0);
- if (chars == 0)
- mdbx::error::throw_exception(GetLastError());
- wstr.append(chars, '\0');
- MultiByteToWideChar(CP_UTF8, MB_ERR_INVALID_CHARS, c_str,
- int(strlen(c_str)), const_cast<wchar_t *>(wstr.data()),
- chars);
- }
- return wstr;
-}
-
-#endif /* Windows */
-
-} // namespace
-
-//------------------------------------------------------------------------------
-
-namespace mdbx {
-
-[[noreturn]] __cold void throw_max_length_exceeded() {
- throw std::length_error(
- "mdbx:: Exceeded the maximal length of data/slice/buffer.");
-}
-
-[[noreturn]] __cold void throw_too_small_target_buffer() {
- throw std::length_error("mdbx:: The target buffer is too small.");
-}
-
-[[noreturn]] __cold void throw_out_range() {
- throw std::out_of_range("mdbx:: Slice or buffer method was called with "
- "an argument that exceeds the length.");
-}
-
-[[noreturn]] __cold void throw_allocators_mismatch() {
- throw std::logic_error(
- "mdbx:: An allocators mismatch, so an object could not be transferred "
- "into an incompatible memory allocation scheme.");
-}
-
-__cold exception::exception(const ::mdbx::error &error) noexcept
- : base(error.what()), error_(error) {}
-
-__cold exception::~exception() noexcept {}
-
-static std::atomic_int fatal_countdown;
-
-__cold fatal::fatal(const ::mdbx::error &error) noexcept : base(error) {
- ++fatal_countdown;
-}
-
-__cold fatal::~fatal() noexcept {
- if (--fatal_countdown == 0)
- std::terminate();
-}
-
-#define DEFINE_EXCEPTION(NAME) \
- __cold NAME::NAME(const ::mdbx::error &rc) : exception(rc) {} \
- __cold NAME::~NAME() noexcept {}
-
-DEFINE_EXCEPTION(bad_map_id)
-DEFINE_EXCEPTION(bad_transaction)
-DEFINE_EXCEPTION(bad_value_size)
-DEFINE_EXCEPTION(db_corrupted)
-DEFINE_EXCEPTION(db_full)
-DEFINE_EXCEPTION(db_invalid)
-DEFINE_EXCEPTION(db_too_large)
-DEFINE_EXCEPTION(db_unable_extend)
-DEFINE_EXCEPTION(db_version_mismatch)
-DEFINE_EXCEPTION(db_wanna_write_for_recovery)
-DEFINE_EXCEPTION(incompatible_operation)
-DEFINE_EXCEPTION(internal_page_full)
-DEFINE_EXCEPTION(internal_problem)
-DEFINE_EXCEPTION(key_exists)
-DEFINE_EXCEPTION(key_mismatch)
-DEFINE_EXCEPTION(max_maps_reached)
-DEFINE_EXCEPTION(max_readers_reached)
-DEFINE_EXCEPTION(multivalue)
-DEFINE_EXCEPTION(no_data)
-DEFINE_EXCEPTION(not_found)
-DEFINE_EXCEPTION(operation_not_permitted)
-DEFINE_EXCEPTION(permission_denied_or_not_writeable)
-DEFINE_EXCEPTION(reader_slot_busy)
-DEFINE_EXCEPTION(remote_media)
-DEFINE_EXCEPTION(something_busy)
-DEFINE_EXCEPTION(thread_mismatch)
-DEFINE_EXCEPTION(transaction_full)
-DEFINE_EXCEPTION(transaction_overlapping)
-
-#undef DEFINE_EXCEPTION
-
-__cold const char *error::what() const noexcept {
- if (is_mdbx_error())
- return mdbx_liberr2str(code());
-
- switch (code()) {
-#define ERROR_CASE(CODE) \
- case CODE: \
- return MDBX_STRINGIFY(CODE)
- ERROR_CASE(MDBX_ENODATA);
- ERROR_CASE(MDBX_EINVAL);
- ERROR_CASE(MDBX_EACCESS);
- ERROR_CASE(MDBX_ENOMEM);
- ERROR_CASE(MDBX_EROFS);
- ERROR_CASE(MDBX_ENOSYS);
- ERROR_CASE(MDBX_EIO);
- ERROR_CASE(MDBX_EPERM);
- ERROR_CASE(MDBX_EINTR);
- ERROR_CASE(MDBX_ENOFILE);
- ERROR_CASE(MDBX_EREMOTE);
-#undef ERROR_CASE
- default:
- return "SYSTEM";
- }
-}
-
-__cold std::string error::message() const {
- char buf[1024];
- const char *msg = ::mdbx_strerror_r(code(), buf, sizeof(buf));
- return std::string(msg ? msg : "unknown");
-}
-
-[[noreturn]] __cold void error::panic(const char *context,
- const char *func) const noexcept {
- assert(code() != MDBX_SUCCESS);
- ::mdbx_panic("mdbx::%s.%s(): \"%s\" (%d)", context, func, what(), code());
- std::terminate();
-}
-
-__cold void error::throw_exception() const {
- switch (code()) {
- case MDBX_EINVAL:
- throw std::invalid_argument("mdbx");
- case MDBX_ENOMEM:
- throw std::bad_alloc();
- case MDBX_SUCCESS:
- static_assert(MDBX_SUCCESS == MDBX_RESULT_FALSE, "WTF?");
- throw std::logic_error("MDBX_SUCCESS (MDBX_RESULT_FALSE)");
- case MDBX_RESULT_TRUE:
- throw std::logic_error("MDBX_RESULT_TRUE");
-#define CASE_EXCEPTION(NAME, CODE) \
- case CODE: \
- throw NAME(code())
- CASE_EXCEPTION(bad_map_id, MDBX_BAD_DBI);
- CASE_EXCEPTION(bad_transaction, MDBX_BAD_TXN);
- CASE_EXCEPTION(bad_value_size, MDBX_BAD_VALSIZE);
- CASE_EXCEPTION(db_corrupted, MDBX_CORRUPTED);
- CASE_EXCEPTION(db_corrupted, MDBX_CURSOR_FULL); /* branch-pages loop */
- CASE_EXCEPTION(db_corrupted, MDBX_PAGE_NOTFOUND);
- CASE_EXCEPTION(db_full, MDBX_MAP_FULL);
- CASE_EXCEPTION(db_invalid, MDBX_INVALID);
- CASE_EXCEPTION(db_too_large, MDBX_TOO_LARGE);
- CASE_EXCEPTION(db_unable_extend, MDBX_UNABLE_EXTEND_MAPSIZE);
- CASE_EXCEPTION(db_version_mismatch, MDBX_VERSION_MISMATCH);
- CASE_EXCEPTION(db_wanna_write_for_recovery, MDBX_WANNA_RECOVERY);
- CASE_EXCEPTION(fatal, MDBX_EBADSIGN);
- CASE_EXCEPTION(fatal, MDBX_PANIC);
- CASE_EXCEPTION(incompatible_operation, MDBX_INCOMPATIBLE);
- CASE_EXCEPTION(internal_page_full, MDBX_PAGE_FULL);
- CASE_EXCEPTION(internal_problem, MDBX_PROBLEM);
- CASE_EXCEPTION(key_mismatch, MDBX_EKEYMISMATCH);
- CASE_EXCEPTION(max_maps_reached, MDBX_DBS_FULL);
- CASE_EXCEPTION(max_readers_reached, MDBX_READERS_FULL);
- CASE_EXCEPTION(multivalue, MDBX_EMULTIVAL);
- CASE_EXCEPTION(no_data, MDBX_ENODATA);
- CASE_EXCEPTION(not_found, MDBX_NOTFOUND);
- CASE_EXCEPTION(operation_not_permitted, MDBX_EPERM);
- CASE_EXCEPTION(permission_denied_or_not_writeable, MDBX_EACCESS);
- CASE_EXCEPTION(reader_slot_busy, MDBX_BAD_RSLOT);
- CASE_EXCEPTION(remote_media, MDBX_EREMOTE);
- CASE_EXCEPTION(something_busy, MDBX_BUSY);
- CASE_EXCEPTION(thread_mismatch, MDBX_THREAD_MISMATCH);
- CASE_EXCEPTION(transaction_full, MDBX_TXN_FULL);
- CASE_EXCEPTION(transaction_overlapping, MDBX_TXN_OVERLAPPING);
-#undef CASE_EXCEPTION
- default:
- if (is_mdbx_error())
- throw exception(*this);
- throw std::system_error(std::error_code(code(), std::system_category()));
- }
-}
-
-//------------------------------------------------------------------------------
-
-bool slice::is_printable(bool disable_utf8) const noexcept {
- enum : byte {
- LS = 4, // shift for UTF8 sequence length
- P_ = 1 << LS, // printable ASCII flag
- N_ = 0, // non-printable ASCII
- second_range_mask = P_ - 1, // mask for range flag
- r80_BF = 0, // flag for UTF8 2nd byte range
- rA0_BF = 1, // flag for UTF8 2nd byte range
- r80_9F = 2, // flag for UTF8 2nd byte range
- r90_BF = 3, // flag for UTF8 2nd byte range
- r80_8F = 4, // flag for UTF8 2nd byte range
-
- // valid utf-8 byte sequences
- // http://www.unicode.org/versions/Unicode6.0.0/ch03.pdf - page 94
- // Code | Bytes | | |
- // Points | 1st | 2nd | 3rd |4th
- // --------------------|--------|--------|--------|---
- C2 = 2 << LS | r80_BF, // U+000080..U+0007FF | C2..DF | 80..BF | |
- E0 = 3 << LS | rA0_BF, // U+000800..U+000FFF | E0 | A0..BF | 80..BF |
- E1 = 3 << LS | r80_BF, // U+001000..U+00CFFF | E1..EC | 80..BF | 80..BF |
- ED = 3 << LS | r80_9F, // U+00D000..U+00D7FF | ED | 80..9F | 80..BF |
- EE = 3 << LS | r80_BF, // U+00E000..U+00FFFF | EE..EF | 80..BF | 80..BF |
- F0 = 4 << LS | r90_BF, // U+010000..U+03FFFF | F0 | 90..BF | 80..BF |...
- F1 = 4 << LS | r80_BF, // U+040000..U+0FFFFF | F1..F3 | 80..BF | 80..BF |...
- F4 = 4 << LS | r80_BF, // U+100000..U+10FFFF | F4 | 80..8F | 80..BF |...
- };
-
- static const byte range_from[] = {0x80, 0xA0, 0x80, 0x90, 0x80};
- static const byte range_to[] = {0xBF, 0xBF, 0x9F, 0xBF, 0x8F};
-
- static const byte map[256] = {
- // 1 2 3 4 5 6 7 8 9 a b c d e f
- N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, // 00
- N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, N_, // 10
- P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, // 20
- P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, // 30
- P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, // 40
- P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, // 50
- P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, // 60
- P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, N_, // 70
- N_, N_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, N_, P_, N_, // 80
- N_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, N_, P_, P_, // 90
- P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, // a0
- P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, // b0
- P_, P_, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, // c0
- C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, C2, // df
- E0, E1, E1, E1, E1, E1, E1, E1, E1, E1, E1, E1, E1, ED, EE, EE, // e0
- F0, F1, F1, F1, F4, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_, P_ // f0
- };
-
- if (MDBX_UNLIKELY(length() < 1))
- MDBX_CXX20_UNLIKELY return false;
-
- auto src = byte_ptr();
- const auto end = src + length();
- if (MDBX_UNLIKELY(disable_utf8)) {
- do
- if (MDBX_UNLIKELY((P_ & map[*src]) == 0))
- MDBX_CXX20_UNLIKELY return false;
- while (++src < end);
- return true;
- }
-
- do {
- const auto bits = map[*src];
- const auto second_from = range_from[bits & second_range_mask];
- const auto second_to = range_to[bits & second_range_mask];
- switch (bits >> LS) {
- default:
- MDBX_CXX20_UNLIKELY return false;
- case 1:
- src += 1;
- continue;
- case 2:
- if (MDBX_UNLIKELY(src + 1 >= end))
- MDBX_CXX20_UNLIKELY return false;
- if (MDBX_UNLIKELY(src[1] < second_from || src[1] > second_to))
- MDBX_CXX20_UNLIKELY return false;
- src += 2;
- continue;
- case 3:
- if (MDBX_UNLIKELY(src + 3 >= end))
- MDBX_CXX20_UNLIKELY return false;
- if (MDBX_UNLIKELY(src[1] < second_from || src[1] > second_to))
- MDBX_CXX20_UNLIKELY return false;
- if (MDBX_UNLIKELY(src[2] < 0x80 || src[2] > 0xBF))
- MDBX_CXX20_UNLIKELY return false;
- src += 3;
- continue;
- case 4:
- if (MDBX_UNLIKELY(src + 4 >= end))
- MDBX_CXX20_UNLIKELY return false;
- if (MDBX_UNLIKELY(src[1] < second_from || src[1] > second_to))
- MDBX_CXX20_UNLIKELY return false;
- if (MDBX_UNLIKELY(src[2] < 0x80 || src[2] > 0xBF))
- MDBX_CXX20_UNLIKELY return false;
- if (MDBX_UNLIKELY(src[3] < 0x80 || src[3] > 0xBF))
- MDBX_CXX20_UNLIKELY return false;
- src += 4;
- continue;
- }
- } while (src < end);
-
- return true;
-}
-
-//------------------------------------------------------------------------------
-
-char *to_hex::write_bytes(char *__restrict const dest, size_t dest_size) const {
- if (MDBX_UNLIKELY(envisage_result_length() > dest_size))
- MDBX_CXX20_UNLIKELY throw_too_small_target_buffer();
-
- auto ptr = dest;
- auto src = source.byte_ptr();
- const char alphabase = (uppercase ? 'A' : 'a') - 10;
- auto line = ptr;
- for (const auto end = source.end_byte_ptr(); src != end; ++src) {
- if (wrap_width && size_t(ptr - line) >= wrap_width) {
- *ptr = '\n';
- line = ++ptr;
- }
- const int8_t hi = *src >> 4;
- const int8_t lo = *src & 15;
- ptr[0] = char(alphabase + hi + (((hi - 10) >> 7) & -7));
- ptr[1] = char(alphabase + lo + (((lo - 10) >> 7) & -7));
- ptr += 2;
- assert(ptr <= dest + dest_size);
- }
- return ptr;
-}
-
-::std::ostream &to_hex::output(::std::ostream &out) const {
- if (MDBX_LIKELY(!is_empty()))
- MDBX_CXX20_LIKELY {
- ::std::ostream::sentry sentry(out);
- auto src = source.byte_ptr();
- const char alphabase = (uppercase ? 'A' : 'a') - 10;
- unsigned width = 0;
- for (const auto end = source.end_byte_ptr(); src != end; ++src) {
- if (wrap_width && width >= wrap_width) {
- out << ::std::endl;
- width = 0;
- }
- const int8_t hi = *src >> 4;
- const int8_t lo = *src & 15;
- out.put(char(alphabase + hi + (((hi - 10) >> 7) & -7)));
- out.put(char(alphabase + lo + (((lo - 10) >> 7) & -7)));
- width += 2;
- }
- }
- return out;
-}
-
-char *from_hex::write_bytes(char *__restrict const dest,
- size_t dest_size) const {
- if (MDBX_UNLIKELY(source.length() % 2 && !ignore_spaces))
- MDBX_CXX20_UNLIKELY throw std::domain_error(
- "mdbx::from_hex:: odd length of hexadecimal string");
- if (MDBX_UNLIKELY(envisage_result_length() > dest_size))
- MDBX_CXX20_UNLIKELY throw_too_small_target_buffer();
-
- auto ptr = dest;
- auto src = source.byte_ptr();
- for (auto left = source.length(); left > 0;) {
- if (MDBX_UNLIKELY(*src <= ' ') &&
- MDBX_LIKELY(ignore_spaces && isspace(*src))) {
- ++src;
- --left;
- continue;
- }
-
- if (MDBX_UNLIKELY(left < 1 || !isxdigit(src[0]) || !isxdigit(src[1])))
- MDBX_CXX20_UNLIKELY throw std::domain_error(
- "mdbx::from_hex:: invalid hexadecimal string");
-
- int8_t hi = src[0];
- hi = (hi | 0x20) - 'a';
- hi += 10 + ((hi >> 7) & 7);
-
- int8_t lo = src[1];
- lo = (lo | 0x20) - 'a';
- lo += 10 + ((lo >> 7) & 7);
-
- *ptr++ = hi << 4 | lo;
- src += 2;
- left -= 2;
- assert(ptr <= dest + dest_size);
- }
- return ptr;
-}
-
-bool from_hex::is_erroneous() const noexcept {
- if (MDBX_UNLIKELY(source.length() % 2 && !ignore_spaces))
- MDBX_CXX20_UNLIKELY return true;
-
- bool got = false;
- auto src = source.byte_ptr();
- for (auto left = source.length(); left > 0;) {
- if (MDBX_UNLIKELY(*src <= ' ') &&
- MDBX_LIKELY(ignore_spaces && isspace(*src))) {
- ++src;
- --left;
- continue;
- }
-
- if (MDBX_UNLIKELY(left < 1 || !isxdigit(src[0]) || !isxdigit(src[1])))
- MDBX_CXX20_UNLIKELY return true;
-
- got = true;
- src += 2;
- left -= 2;
- }
- return !got;
-}
-
-//------------------------------------------------------------------------------
-
-enum : signed char {
- OO /* ASCII NUL */ = -8,
- EQ /* BASE64 '=' pad */ = -4,
- SP /* SPACE */ = -2,
- IL /* invalid */ = -1
-};
-
-static const byte b58_alphabet[58] = {
- '1', '2', '3', '4', '5', '6', '7', '8', '9', 'A', 'B', 'C', 'D', 'E', 'F',
- 'G', 'H', 'J', 'K', 'L', 'M', 'N', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W',
- 'X', 'Y', 'Z', 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'm',
- 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z'};
-
-#ifndef bswap64
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
-static inline uint64_t bswap64(uint64_t v) noexcept {
-#if __GNUC_PREREQ(4, 4) || __CLANG_PREREQ(4, 0) || \
- __has_builtin(__builtin_bswap64)
- return __builtin_bswap64(v);
-#elif defined(_MSC_VER) && !defined(__clang__)
- return _byteswap_uint64(v);
-#elif defined(__bswap_64)
- return __bswap_64(v);
-#elif defined(bswap_64)
- return bswap_64(v);
-#else
- return v << 56 | v >> 56 | ((v << 40) & UINT64_C(0x00ff000000000000)) |
- ((v << 24) & UINT64_C(0x0000ff0000000000)) |
- ((v << 8) & UINT64_C(0x000000ff00000000)) |
- ((v >> 8) & UINT64_C(0x00000000ff000000)) |
- ((v >> 24) & UINT64_C(0x0000000000ff0000)) |
- ((v >> 40) & UINT64_C(0x000000000000ff00));
-#endif
-}
-#endif /* __BYTE_ORDER__ */
-#endif /* ifndef bswap64 */
-
-static inline char b58_8to11(uint64_t &v) noexcept {
- const unsigned i = unsigned(v % 58);
- v /= 58;
- return b58_alphabet[i];
-}
-
-char *to_base58::write_bytes(char *__restrict const dest,
- size_t dest_size) const {
- if (MDBX_UNLIKELY(envisage_result_length() > dest_size))
- MDBX_CXX20_UNLIKELY throw_too_small_target_buffer();
-
- auto ptr = dest;
- auto src = source.byte_ptr();
- size_t left = source.length();
- auto line = ptr;
- while (MDBX_LIKELY(left > 7)) {
- uint64_t v;
- std::memcpy(&v, src, 8);
- src += 8;
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- v = bswap64(v);
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- ptr[10] = b58_8to11(v);
- ptr[9] = b58_8to11(v);
- ptr[8] = b58_8to11(v);
- ptr[7] = b58_8to11(v);
- ptr[6] = b58_8to11(v);
- ptr[5] = b58_8to11(v);
- ptr[4] = b58_8to11(v);
- ptr[3] = b58_8to11(v);
- ptr[2] = b58_8to11(v);
- ptr[1] = b58_8to11(v);
- ptr[0] = b58_8to11(v);
- assert(v == 0);
- ptr += 11;
- left -= 8;
- if (wrap_width && size_t(ptr - line) >= wrap_width && left) {
- *ptr = '\n';
- line = ++ptr;
- }
- assert(ptr <= dest + dest_size);
- }
-
- if (left) {
- uint64_t v = 0;
- unsigned parrots = 31;
- do {
- v = (v << 8) + *src++;
- parrots += 43;
- } while (--left);
-
- auto tail = ptr += parrots >> 5;
- assert(ptr <= dest + dest_size);
- do {
- *--tail = b58_8to11(v);
- parrots -= 32;
- } while (parrots > 31);
- assert(v == 0);
- }
-
- return ptr;
-}
-
-::std::ostream &to_base58::output(::std::ostream &out) const {
- if (MDBX_LIKELY(!is_empty()))
- MDBX_CXX20_LIKELY {
- ::std::ostream::sentry sentry(out);
- auto src = source.byte_ptr();
- size_t left = source.length();
- unsigned width = 0;
- std::array<char, 11> buf;
-
- while (MDBX_LIKELY(left > 7)) {
- uint64_t v;
- std::memcpy(&v, src, 8);
- src += 8;
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- v = bswap64(v);
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- buf[10] = b58_8to11(v);
- buf[9] = b58_8to11(v);
- buf[8] = b58_8to11(v);
- buf[7] = b58_8to11(v);
- buf[6] = b58_8to11(v);
- buf[5] = b58_8to11(v);
- buf[4] = b58_8to11(v);
- buf[3] = b58_8to11(v);
- buf[2] = b58_8to11(v);
- buf[1] = b58_8to11(v);
- buf[0] = b58_8to11(v);
- assert(v == 0);
- out.write(&buf.front(), 11);
- left -= 8;
- if (wrap_width && (width += 11) >= wrap_width && left) {
- out << ::std::endl;
- width = 0;
- }
- }
-
- if (left) {
- uint64_t v = 0;
- unsigned parrots = 31;
- do {
- v = (v << 8) + *src++;
- parrots += 43;
- } while (--left);
-
- auto ptr = buf.end();
- do {
- *--ptr = b58_8to11(v);
- parrots -= 32;
- } while (parrots > 31);
- assert(v == 0);
- out.write(&*ptr, buf.end() - ptr);
- }
- }
- return out;
-}
-
-const signed char b58_map[256] = {
- // 1 2 3 4 5 6 7 8 9 a b c d e f
- OO, IL, IL, IL, IL, IL, IL, IL, IL, SP, SP, SP, SP, SP, IL, IL, // 00
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // 10
- SP, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // 20
- IL, 0, 1, 2, 3, 4, 5, 6, 7, 8, IL, IL, IL, IL, IL, IL, // 30
- IL, 9, 10, 11, 12, 13, 14, 15, 16, IL, 17, 18, 19, 20, 21, IL, // 40
- 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, IL, IL, IL, IL, IL, // 50
- IL, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, IL, 44, 45, 46, // 60
- 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, IL, IL, IL, IL, IL, // 70
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // 80
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // 90
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // a0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // b0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // c0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // d0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // e0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL // f0
-};
-
-static inline signed char b58_11to8(uint64_t &v, const byte c) noexcept {
- const signed char m = b58_map[c];
- v = v * 58 + m;
- return m;
-}
-
-char *from_base58::write_bytes(char *__restrict const dest,
- size_t dest_size) const {
- if (MDBX_UNLIKELY(envisage_result_length() > dest_size))
- MDBX_CXX20_UNLIKELY throw_too_small_target_buffer();
-
- auto ptr = dest;
- auto src = source.byte_ptr();
- for (auto left = source.length(); left > 0;) {
- if (MDBX_UNLIKELY(isspace(*src)) && ignore_spaces) {
- ++src;
- --left;
- continue;
- }
-
- if (MDBX_LIKELY(left > 10)) {
- uint64_t v = 0;
- if (MDBX_UNLIKELY((b58_11to8(v, src[0]) | b58_11to8(v, src[1]) |
- b58_11to8(v, src[2]) | b58_11to8(v, src[3]) |
- b58_11to8(v, src[4]) | b58_11to8(v, src[5]) |
- b58_11to8(v, src[6]) | b58_11to8(v, src[7]) |
- b58_11to8(v, src[8]) | b58_11to8(v, src[9]) |
- b58_11to8(v, src[10])) < 0))
- MDBX_CXX20_UNLIKELY goto bailout;
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- v = bswap64(v);
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- std::memcpy(ptr, &v, 8);
- ptr += 8;
- src += 11;
- left -= 11;
- assert(ptr <= dest + dest_size);
- continue;
- }
-
- constexpr unsigned invalid_length_mask = 1 << 1 | 1 << 4 | 1 << 8;
- if (MDBX_UNLIKELY(invalid_length_mask & (1 << left)))
- MDBX_CXX20_UNLIKELY goto bailout;
-
- uint64_t v = 1;
- unsigned parrots = 0;
- do {
- if (MDBX_UNLIKELY(b58_11to8(v, *src++) < 0))
- MDBX_CXX20_UNLIKELY goto bailout;
- parrots += 32;
- } while (--left);
-
- auto tail = ptr += parrots / 43;
- assert(ptr <= dest + dest_size);
- do {
- *--tail = byte(v);
- v >>= 8;
- } while (v > 255);
- break;
- }
- return ptr;
-
-bailout:
- throw std::domain_error("mdbx::from_base58:: invalid base58 string");
-}
-
-bool from_base58::is_erroneous() const noexcept {
- bool got = false;
- auto src = source.byte_ptr();
- for (auto left = source.length(); left > 0;) {
- if (MDBX_UNLIKELY(*src <= ' ') &&
- MDBX_LIKELY(ignore_spaces && isspace(*src))) {
- ++src;
- --left;
- continue;
- }
-
- if (MDBX_LIKELY(left > 10)) {
- if (MDBX_UNLIKELY((b58_map[src[0]] | b58_map[src[1]] | b58_map[src[2]] |
- b58_map[src[3]] | b58_map[src[4]] | b58_map[src[5]] |
- b58_map[src[6]] | b58_map[src[7]] | b58_map[src[8]] |
- b58_map[src[9]] | b58_map[src[10]]) < 0))
- MDBX_CXX20_UNLIKELY return true;
- src += 11;
- left -= 11;
- got = true;
- continue;
- }
-
- constexpr unsigned invalid_length_mask = 1 << 1 | 1 << 4 | 1 << 8;
- if (invalid_length_mask & (1 << left))
- return false;
-
- do
- if (MDBX_UNLIKELY(b58_map[*src++] < 0))
- MDBX_CXX20_UNLIKELY return true;
- while (--left);
- got = true;
- break;
- }
- return !got;
-}
-
-//------------------------------------------------------------------------------
-
-static inline void b64_3to4(const byte x, const byte y, const byte z,
- char *__restrict dest) noexcept {
- static const byte alphabet[64] = {
- 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J', 'K', 'L', 'M',
- 'N', 'O', 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', 'X', 'Y', 'Z',
- 'a', 'b', 'c', 'd', 'e', 'f', 'g', 'h', 'i', 'j', 'k', 'l', 'm',
- 'n', 'o', 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', 'x', 'y', 'z',
- '0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '+', '/'};
- dest[0] = alphabet[(x & 0xfc) >> 2];
- dest[1] = alphabet[((x & 0x03) << 4) + ((y & 0xf0) >> 4)];
- dest[2] = alphabet[((y & 0x0f) << 2) + ((z & 0xc0) >> 6)];
- dest[3] = alphabet[z & 0x3f];
-}
-
-char *to_base64::write_bytes(char *__restrict const dest,
- size_t dest_size) const {
- if (MDBX_UNLIKELY(envisage_result_length() > dest_size))
- MDBX_CXX20_UNLIKELY throw_too_small_target_buffer();
-
- auto ptr = dest;
- auto src = source.byte_ptr();
- size_t left = source.length();
- auto line = ptr;
- while (true) {
- switch (left) {
- default:
- MDBX_CXX20_LIKELY left -= 3;
- b64_3to4(src[0], src[1], src[2], ptr);
- ptr += 4;
- src += 3;
- if (wrap_width && size_t(ptr - line) >= wrap_width && left) {
- *ptr = '\n';
- line = ++ptr;
- }
- assert(ptr <= dest + dest_size);
- continue;
- case 2:
- b64_3to4(src[0], src[1], 0, ptr);
- ptr[3] = '=';
- assert(ptr + 4 <= dest + dest_size);
- return ptr + 4;
- case 1:
- b64_3to4(src[0], 0, 0, ptr);
- ptr[2] = ptr[3] = '=';
- assert(ptr + 4 <= dest + dest_size);
- return ptr + 4;
- case 0:
- return ptr;
- }
- }
-}
-
-::std::ostream &to_base64::output(::std::ostream &out) const {
- if (MDBX_LIKELY(!is_empty()))
- MDBX_CXX20_LIKELY {
- ::std::ostream::sentry sentry(out);
- auto src = source.byte_ptr();
- size_t left = source.length();
- unsigned width = 0;
- std::array<char, 4> buf;
-
- while (true) {
- switch (left) {
- default:
- MDBX_CXX20_LIKELY left -= 3;
- b64_3to4(src[0], src[1], src[2], &buf.front());
- src += 3;
- out.write(&buf.front(), 4);
- if (wrap_width && (width += 4) >= wrap_width && left) {
- out << ::std::endl;
- width = 0;
- }
- continue;
- case 2:
- b64_3to4(src[0], src[1], 0, &buf.front());
- buf[3] = '=';
- return out.write(&buf.front(), 4);
- case 1:
- b64_3to4(src[0], 0, 0, &buf.front());
- buf[2] = buf[3] = '=';
- return out.write(&buf.front(), 4);
- case 0:
- return out;
- }
- }
- }
- return out;
-}
-
-static const signed char b64_map[256] = {
- // 1 2 3 4 5 6 7 8 9 a b c d e f
- OO, IL, IL, IL, IL, IL, IL, IL, IL, SP, SP, SP, SP, SP, IL, IL, // 00
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // 10
- SP, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, 62, IL, IL, IL, 63, // 20
- 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, IL, IL, IL, EQ, IL, IL, // 30
- IL, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, // 40
- 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, IL, IL, IL, IL, IL, // 50
- IL, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, // 60
- 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, IL, IL, IL, IL, IL, // 70
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // 80
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // 90
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // a0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // b0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // c0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // d0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, // e0
- IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL, IL // f0
-};
-
-static inline signed char b64_4to3(signed char a, signed char b, signed char c,
- signed char d,
- char *__restrict dest) noexcept {
- dest[0] = byte((a << 2) + ((b & 0x30) >> 4));
- dest[1] = byte(((b & 0xf) << 4) + ((c & 0x3c) >> 2));
- dest[2] = byte(((c & 0x3) << 6) + d);
- return a | b | c | d;
-}
-
-char *from_base64::write_bytes(char *__restrict const dest,
- size_t dest_size) const {
- if (MDBX_UNLIKELY(source.length() % 4 && !ignore_spaces))
- MDBX_CXX20_UNLIKELY throw std::domain_error(
- "mdbx::from_base64:: odd length of base64 string");
- if (MDBX_UNLIKELY(envisage_result_length() > dest_size))
- MDBX_CXX20_UNLIKELY throw_too_small_target_buffer();
-
- auto ptr = dest;
- auto src = source.byte_ptr();
- for (auto left = source.length(); left > 0;) {
- if (MDBX_UNLIKELY(*src <= ' ') &&
- MDBX_LIKELY(ignore_spaces && isspace(*src))) {
- ++src;
- --left;
- continue;
- }
-
- if (MDBX_UNLIKELY(left < 3))
- MDBX_CXX20_UNLIKELY {
- bailout:
- throw std::domain_error("mdbx::from_base64:: invalid base64 string");
- }
- const signed char a = b64_map[src[0]], b = b64_map[src[1]],
- c = b64_map[src[2]], d = b64_map[src[3]];
- if (MDBX_UNLIKELY(b64_4to3(a, b, c, d, ptr) < 0)) {
- if (left == 4 && (a | b) >= 0 && d == EQ) {
- if (c >= 0) {
- assert(ptr + 2 <= dest + dest_size);
- return ptr + 2;
- }
- if (c == d) {
- assert(ptr + 1 <= dest + dest_size);
- return ptr + 1;
- }
- }
- MDBX_CXX20_UNLIKELY goto bailout;
- }
- src += 4;
- left -= 4;
- ptr += 3;
- assert(ptr <= dest + dest_size);
- }
- return ptr;
-}
-
-bool from_base64::is_erroneous() const noexcept {
- if (MDBX_UNLIKELY(source.length() % 4 && !ignore_spaces))
- MDBX_CXX20_UNLIKELY return true;
-
- bool got = false;
- auto src = source.byte_ptr();
- for (auto left = source.length(); left > 0;) {
- if (MDBX_UNLIKELY(*src <= ' ') &&
- MDBX_LIKELY(ignore_spaces && isspace(*src))) {
- ++src;
- --left;
- continue;
- }
-
- if (MDBX_UNLIKELY(left < 3))
- MDBX_CXX20_UNLIKELY return false;
- const signed char a = b64_map[src[0]], b = b64_map[src[1]],
- c = b64_map[src[2]], d = b64_map[src[3]];
- if (MDBX_UNLIKELY((a | b | c | d) < 0))
- MDBX_CXX20_UNLIKELY {
- if (left == 4 && (a | b) >= 0 && d == EQ && (c >= 0 || c == d))
- return false;
- return true;
- }
- got = true;
- src += 4;
- left -= 4;
- }
- return !got;
-}
-
-//------------------------------------------------------------------------------
-
-template class LIBMDBX_API_TYPE buffer<legacy_allocator>;
-
-#if defined(__cpp_lib_memory_resource) && \
- __cpp_lib_memory_resource >= 201603L && _GLIBCXX_USE_CXX11_ABI
-template class LIBMDBX_API_TYPE buffer<polymorphic_allocator>;
-#endif /* __cpp_lib_memory_resource >= 201603L */
-
-//------------------------------------------------------------------------------
-
-static inline MDBX_env_flags_t mode2flags(env::mode mode) {
- switch (mode) {
- default:
- MDBX_CXX20_UNLIKELY throw std::invalid_argument("db::mode is invalid");
- case env::mode::readonly:
- return MDBX_RDONLY;
- case env::mode::write_file_io:
- return MDBX_ENV_DEFAULTS;
- case env::mode::write_mapped_io:
- return MDBX_WRITEMAP;
- }
-}
-
-__cold MDBX_env_flags_t
-env::operate_parameters::make_flags(bool accede, bool use_subdirectory) const {
- MDBX_env_flags_t flags = mode2flags(mode);
- if (accede)
- flags |= MDBX_ACCEDE;
- if (!use_subdirectory)
- flags |= MDBX_NOSUBDIR;
- if (options.exclusive)
- flags |= MDBX_EXCLUSIVE;
- if (options.orphan_read_transactions)
- flags |= MDBX_NOTLS;
- if (options.disable_readahead)
- flags |= MDBX_NORDAHEAD;
- if (options.disable_clear_memory)
- flags |= MDBX_NOMEMINIT;
-
- if (mode != readonly) {
- if (options.nested_write_transactions)
- flags &= ~MDBX_WRITEMAP;
- if (reclaiming.coalesce)
- flags |= MDBX_COALESCE;
- if (reclaiming.lifo)
- flags |= MDBX_LIFORECLAIM;
- switch (durability) {
- default:
- MDBX_CXX20_UNLIKELY throw std::invalid_argument(
- "db::durability is invalid");
- case env::durability::robust_synchronous:
- break;
- case env::durability::half_synchronous_weak_last:
- flags |= MDBX_NOMETASYNC;
- break;
- case env::durability::lazy_weak_tail:
- static_assert(MDBX_MAPASYNC == MDBX_SAFE_NOSYNC, "WTF? Obsolete C API?");
- flags |= MDBX_SAFE_NOSYNC;
- break;
- case env::durability::whole_fragile:
- flags |= MDBX_UTTERLY_NOSYNC;
- break;
- }
- }
- return flags;
-}
-
-env::mode
-env::operate_parameters::mode_from_flags(MDBX_env_flags_t flags) noexcept {
- if (flags & MDBX_RDONLY)
- return env::mode::readonly;
- return (flags & MDBX_WRITEMAP) ? env::mode::write_mapped_io
- : env::mode::write_file_io;
-}
-
-env::durability env::operate_parameters::durability_from_flags(
- MDBX_env_flags_t flags) noexcept {
- if ((flags & MDBX_UTTERLY_NOSYNC) == MDBX_UTTERLY_NOSYNC)
- return env::durability::whole_fragile;
- if (flags & MDBX_SAFE_NOSYNC)
- return env::durability::lazy_weak_tail;
- if (flags & MDBX_NOMETASYNC)
- return env::durability::half_synchronous_weak_last;
- return env::durability::robust_synchronous;
-}
-
-env::reclaiming_options::reclaiming_options(MDBX_env_flags_t flags) noexcept
- : lifo((flags & MDBX_LIFORECLAIM) ? true : false),
- coalesce((flags & MDBX_COALESCE) ? true : false) {}
-
-env::operate_options::operate_options(MDBX_env_flags_t flags) noexcept
- : orphan_read_transactions(
- ((flags & (MDBX_NOTLS | MDBX_EXCLUSIVE)) == MDBX_NOTLS) ? true
- : false),
- nested_write_transactions((flags & (MDBX_WRITEMAP | MDBX_RDONLY)) ? false
- : true),
- exclusive((flags & MDBX_EXCLUSIVE) ? true : false),
- disable_readahead((flags & MDBX_NORDAHEAD) ? true : false),
- disable_clear_memory((flags & MDBX_NOMEMINIT) ? true : false) {}
-
-bool env::is_pristine() const {
- return get_stat().ms_mod_txnid == 0 &&
- get_info().mi_recent_txnid == INITIAL_TXNID;
-}
-
-bool env::is_empty() const { return get_stat().ms_leaf_pages == 0; }
-
-#ifdef MDBX_STD_FILESYSTEM_PATH
-env &env::copy(const MDBX_STD_FILESYSTEM_PATH &destination, bool compactify,
- bool force_dynamic_size) {
- const path_to_pchar<MDBX_STD_FILESYSTEM_PATH> utf8(destination);
- error::success_or_throw(
- ::mdbx_env_copy(handle_, utf8,
- (compactify ? MDBX_CP_COMPACT : MDBX_CP_DEFAULTS) |
- (force_dynamic_size ? MDBX_CP_FORCE_DYNAMIC_SIZE
- : MDBX_CP_DEFAULTS)));
- return *this;
-}
-#endif /* MDBX_STD_FILESYSTEM_PATH */
-
-#if defined(_WIN32) || defined(_WIN64)
-env &env::copy(const ::std::wstring &destination, bool compactify,
- bool force_dynamic_size) {
- const path_to_pchar<::std::wstring> utf8(destination);
- error::success_or_throw(
- ::mdbx_env_copy(handle_, utf8,
- (compactify ? MDBX_CP_COMPACT : MDBX_CP_DEFAULTS) |
- (force_dynamic_size ? MDBX_CP_FORCE_DYNAMIC_SIZE
- : MDBX_CP_DEFAULTS)));
- return *this;
-}
-#endif /* Windows */
-
-env &env::copy(const ::std::string &destination, bool compactify,
- bool force_dynamic_size) {
- const path_to_pchar<::std::string> utf8(destination);
- error::success_or_throw(
- ::mdbx_env_copy(handle_, utf8,
- (compactify ? MDBX_CP_COMPACT : MDBX_CP_DEFAULTS) |
- (force_dynamic_size ? MDBX_CP_FORCE_DYNAMIC_SIZE
- : MDBX_CP_DEFAULTS)));
- return *this;
-}
-
-env &env::copy(filehandle fd, bool compactify, bool force_dynamic_size) {
- error::success_or_throw(
- ::mdbx_env_copy2fd(handle_, fd,
- (compactify ? MDBX_CP_COMPACT : MDBX_CP_DEFAULTS) |
- (force_dynamic_size ? MDBX_CP_FORCE_DYNAMIC_SIZE
- : MDBX_CP_DEFAULTS)));
- return *this;
-}
-
-path env::get_path() const {
- const char *c_str;
- error::success_or_throw(::mdbx_env_get_path(handle_, &c_str));
- return pchar_to_path<path>(c_str);
-}
-
-#ifdef MDBX_STD_FILESYSTEM_PATH
-bool env::remove(const MDBX_STD_FILESYSTEM_PATH &pathname,
- const remove_mode mode) {
- const path_to_pchar<MDBX_STD_FILESYSTEM_PATH> utf8(pathname);
- return error::boolean_or_throw(
- ::mdbx_env_delete(utf8, MDBX_env_delete_mode_t(mode)));
-}
-#endif /* MDBX_STD_FILESYSTEM_PATH */
-
-#if defined(_WIN32) || defined(_WIN64)
-bool env::remove(const ::std::wstring &pathname, const remove_mode mode) {
- const path_to_pchar<::std::wstring> utf8(pathname);
- return error::boolean_or_throw(
- ::mdbx_env_delete(utf8, MDBX_env_delete_mode_t(mode)));
-}
-#endif /* Windows */
-
-bool env::remove(const ::std::string &pathname, const remove_mode mode) {
- const path_to_pchar<::std::string> utf8(pathname);
- return error::boolean_or_throw(
- ::mdbx_env_delete(utf8, MDBX_env_delete_mode_t(mode)));
-}
-
-//------------------------------------------------------------------------------
-
-static inline MDBX_env *create_env() {
- MDBX_env *ptr;
- error::success_or_throw(::mdbx_env_create(&ptr));
- assert(ptr != nullptr);
- return ptr;
-}
-
-env_managed::~env_managed() noexcept {
- if (MDBX_UNLIKELY(handle_))
- MDBX_CXX20_UNLIKELY error::success_or_panic(
- ::mdbx_env_close(handle_), "mdbx::~env()", "mdbx_env_close");
-}
-
-void env_managed::close(bool dont_sync) {
- const error rc =
- static_cast<MDBX_error_t>(::mdbx_env_close_ex(handle_, dont_sync));
- switch (rc.code()) {
- case MDBX_EBADSIGN:
- MDBX_CXX20_UNLIKELY handle_ = nullptr;
- __fallthrough /* fall through */;
- default:
- MDBX_CXX20_UNLIKELY rc.throw_exception();
- case MDBX_SUCCESS:
- MDBX_CXX20_LIKELY handle_ = nullptr;
- }
-}
-
-__cold void env_managed::setup(unsigned max_maps, unsigned max_readers) {
- if (max_readers > 0)
- error::success_or_throw(::mdbx_env_set_maxreaders(handle_, max_readers));
- if (max_maps > 0)
- error::success_or_throw(::mdbx_env_set_maxdbs(handle_, max_maps));
-}
-
-#ifdef MDBX_STD_FILESYSTEM_PATH
-__cold env_managed::env_managed(const MDBX_STD_FILESYSTEM_PATH &pathname,
- const operate_parameters &op, bool accede)
- : env_managed(create_env()) {
- setup(op.max_maps, op.max_readers);
- const path_to_pchar<MDBX_STD_FILESYSTEM_PATH> utf8(pathname);
- error::success_or_throw(
- ::mdbx_env_open(handle_, utf8, op.make_flags(accede), 0));
-
- if (op.options.nested_write_transactions &&
- !get_options().nested_write_transactions)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_INCOMPATIBLE);
-}
-
-__cold env_managed::env_managed(const MDBX_STD_FILESYSTEM_PATH &pathname,
- const env_managed::create_parameters &cp,
- const env::operate_parameters &op, bool accede)
- : env_managed(create_env()) {
- setup(op.max_maps, op.max_readers);
- const path_to_pchar<MDBX_STD_FILESYSTEM_PATH> utf8(pathname);
- set_geometry(cp.geometry);
- error::success_or_throw(
- ::mdbx_env_open(handle_, utf8, op.make_flags(accede, cp.use_subdirectory),
- cp.file_mode_bits));
-
- if (op.options.nested_write_transactions &&
- !get_options().nested_write_transactions)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_INCOMPATIBLE);
-}
-#endif /* MDBX_STD_FILESYSTEM_PATH */
-
-#if defined(_WIN32) || defined(_WIN64)
-__cold env_managed::env_managed(const ::std::wstring &pathname,
- const operate_parameters &op, bool accede)
- : env_managed(create_env()) {
- setup(op.max_maps, op.max_readers);
- const path_to_pchar<::std::wstring> utf8(pathname);
- error::success_or_throw(
- ::mdbx_env_open(handle_, utf8, op.make_flags(accede), 0));
-
- if (op.options.nested_write_transactions &&
- !get_options().nested_write_transactions)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_INCOMPATIBLE);
-}
-
-__cold env_managed::env_managed(const ::std::wstring &pathname,
- const env_managed::create_parameters &cp,
- const env::operate_parameters &op, bool accede)
- : env_managed(create_env()) {
- setup(op.max_maps, op.max_readers);
- const path_to_pchar<::std::wstring> utf8(pathname);
- set_geometry(cp.geometry);
- error::success_or_throw(
- ::mdbx_env_open(handle_, utf8, op.make_flags(accede, cp.use_subdirectory),
- cp.file_mode_bits));
-
- if (op.options.nested_write_transactions &&
- !get_options().nested_write_transactions)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_INCOMPATIBLE);
-}
-#endif /* Windows */
-
-__cold env_managed::env_managed(const ::std::string &pathname,
- const operate_parameters &op, bool accede)
- : env_managed(create_env()) {
- setup(op.max_maps, op.max_readers);
- const path_to_pchar<::std::string> utf8(pathname);
- error::success_or_throw(
- ::mdbx_env_open(handle_, utf8, op.make_flags(accede), 0));
-
- if (op.options.nested_write_transactions &&
- !get_options().nested_write_transactions)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_INCOMPATIBLE);
-}
-
-__cold env_managed::env_managed(const ::std::string &pathname,
- const env_managed::create_parameters &cp,
- const env::operate_parameters &op, bool accede)
- : env_managed(create_env()) {
- setup(op.max_maps, op.max_readers);
- const path_to_pchar<::std::string> utf8(pathname);
- set_geometry(cp.geometry);
- error::success_or_throw(
- ::mdbx_env_open(handle_, utf8, op.make_flags(accede, cp.use_subdirectory),
- cp.file_mode_bits));
-
- if (op.options.nested_write_transactions &&
- !get_options().nested_write_transactions)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_INCOMPATIBLE);
-}
-
-//------------------------------------------------------------------------------
-
-txn_managed txn::start_nested() {
- MDBX_txn *nested;
- error::throw_on_nullptr(handle_, MDBX_BAD_TXN);
- error::success_or_throw(::mdbx_txn_begin(mdbx_txn_env(handle_), handle_,
- MDBX_TXN_READWRITE, &nested));
- assert(nested != nullptr);
- return txn_managed(nested);
-}
-
-txn_managed::~txn_managed() noexcept {
- if (MDBX_UNLIKELY(handle_))
- MDBX_CXX20_UNLIKELY error::success_or_panic(::mdbx_txn_abort(handle_),
- "mdbx::~txn", "mdbx_txn_abort");
-}
-
-void txn_managed::abort() {
- const error err = static_cast<MDBX_error_t>(::mdbx_txn_abort(handle_));
- if (MDBX_LIKELY(err.code() != MDBX_THREAD_MISMATCH))
- MDBX_CXX20_LIKELY handle_ = nullptr;
- if (MDBX_UNLIKELY(err.code() != MDBX_SUCCESS))
- MDBX_CXX20_UNLIKELY err.throw_exception();
-}
-
-void txn_managed::commit() {
- const error err = static_cast<MDBX_error_t>(::mdbx_txn_commit(handle_));
- if (MDBX_LIKELY(err.code() != MDBX_THREAD_MISMATCH))
- MDBX_CXX20_LIKELY handle_ = nullptr;
- if (MDBX_UNLIKELY(err.code() != MDBX_SUCCESS))
- MDBX_CXX20_UNLIKELY err.throw_exception();
-}
-
-//------------------------------------------------------------------------------
-
-bool txn::drop_map(const char *name, bool throw_if_absent) {
- map_handle map;
- const int err = ::mdbx_dbi_open(handle_, name, MDBX_DB_ACCEDE, &map.dbi);
- switch (err) {
- case MDBX_SUCCESS:
- drop_map(map);
- return true;
- case MDBX_NOTFOUND:
- case MDBX_BAD_DBI:
- if (!throw_if_absent)
- return false;
- MDBX_CXX17_FALLTHROUGH /* fallthrough */;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-bool txn::clear_map(const char *name, bool throw_if_absent) {
- map_handle map;
- const int err = ::mdbx_dbi_open(handle_, name, MDBX_DB_ACCEDE, &map.dbi);
- switch (err) {
- case MDBX_SUCCESS:
- clear_map(map);
- return true;
- case MDBX_NOTFOUND:
- case MDBX_BAD_DBI:
- if (!throw_if_absent)
- return false;
- MDBX_CXX17_FALLTHROUGH /* fallthrough */;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-//------------------------------------------------------------------------------
-
-void cursor_managed::close() {
- if (MDBX_UNLIKELY(!handle_))
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_EINVAL);
- ::mdbx_cursor_close(handle_);
- handle_ = nullptr;
-}
-
-//------------------------------------------------------------------------------
-
-__cold ::std::ostream &operator<<(::std::ostream &out, const slice &it) {
- out << "{";
- if (!it.is_valid())
- out << "INVALID." << it.length();
- else if (it.is_null())
- out << "NULL";
- else if (it.empty())
- out << "EMPTY->" << it.data();
- else {
- const slice root(it.head(std::min(it.length(), size_t(64))));
- out << it.length() << ".";
- if (root.is_printable())
- (out << "\"").write(root.char_ptr(), root.length()) << "\"";
- else
- out << root.encode_base58();
- if (root.length() < it.length())
- out << "...";
- }
- return out << "}";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out, const pair &it) {
- return out << "{" << it.key << " => " << it.value << "}";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out, const pair_result &it) {
- return out << "{" << (it.done ? "done: " : "non-done: ") << it.key << " => "
- << it.value << "}";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const ::mdbx::env::geometry::size &it) {
- switch (it.bytes) {
- case ::mdbx::env::geometry::default_value:
- return out << "default";
- case ::mdbx::env::geometry::minimal_value:
- return out << "minimal";
- case ::mdbx::env::geometry::maximal_value:
- return out << "maximal";
- }
-
- const auto bytes = (it.bytes < 0) ? out << "-",
- size_t(-it.bytes) : size_t(it.bytes);
- struct {
- size_t one;
- const char *suffix;
- } static const scales[] = {
-#if MDBX_WORDBITS > 32
- {env_managed::geometry::EiB, "EiB"},
- {env_managed::geometry::EB, "EB"},
- {env_managed::geometry::PiB, "PiB"},
- {env_managed::geometry::PB, "PB"},
- {env_managed::geometry::TiB, "TiB"},
- {env_managed::geometry::TB, "TB"},
-#endif
- {env_managed::geometry::GiB, "GiB"},
- {env_managed::geometry::GB, "GB"},
- {env_managed::geometry::MiB, "MiB"},
- {env_managed::geometry::MB, "MB"},
- {env_managed::geometry::KiB, "KiB"},
- {env_managed::geometry::kB, "kB"},
- {1, " bytes"}
- };
-
- for (const auto i : scales)
- if (bytes % i.one == 0)
- return out << bytes / i.one << i.suffix;
-
- assert(false);
- __unreachable();
- return out;
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const env::geometry &it) {
- return //
- out << "\tlower " << env::geometry::size(it.size_lower) //
- << ",\n\tnow " << env::geometry::size(it.size_now) //
- << ",\n\tupper " << env::geometry::size(it.size_upper) //
- << ",\n\tgrowth " << env::geometry::size(it.growth_step) //
- << ",\n\tshrink " << env::geometry::size(it.shrink_threshold) //
- << ",\n\tpagesize " << env::geometry::size(it.pagesize) << "\n";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const env::operate_parameters &it) {
- return out << "{\n" //
- << "\tmax_maps " << it.max_maps //
- << ",\n\tmax_readers " << it.max_readers //
- << ",\n\tmode " << it.mode //
- << ",\n\tdurability " << it.durability //
- << ",\n\treclaiming " << it.reclaiming //
- << ",\n\toptions " << it.options //
- << "\n}";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out, const env::mode &it) {
- switch (it) {
- case env::mode::readonly:
- return out << "readonly";
- case env::mode::write_file_io:
- return out << "write_file_io";
- case env::mode::write_mapped_io:
- return out << "write_mapped_io";
- default:
- return out << "mdbx::env::mode::invalid";
- }
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const env::durability &it) {
- switch (it) {
- case env::durability::robust_synchronous:
- return out << "robust_synchronous";
- case env::durability::half_synchronous_weak_last:
- return out << "half_synchronous_weak_last";
- case env::durability::lazy_weak_tail:
- return out << "lazy_weak_tail";
- case env::durability::whole_fragile:
- return out << "whole_fragile";
- default:
- return out << "mdbx::env::durability::invalid";
- }
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const env::reclaiming_options &it) {
- return out << "{" //
- << "lifo: " << (it.lifo ? "yes" : "no") //
- << ", coalesce: " << (it.coalesce ? "yes" : "no") //
- << "}";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const env::operate_options &it) {
- static const char comma[] = ", ";
- const char *delimiter = "";
- out << "{";
- if (it.orphan_read_transactions) {
- out << delimiter << "orphan_read_transactions";
- delimiter = comma;
- }
- if (it.nested_write_transactions) {
- out << delimiter << "nested_write_transactions";
- delimiter = comma;
- }
- if (it.exclusive) {
- out << delimiter << "exclusive";
- delimiter = comma;
- }
- if (it.disable_readahead) {
- out << delimiter << "disable_readahead";
- delimiter = comma;
- }
- if (it.disable_clear_memory) {
- out << delimiter << "disable_clear_memory";
- delimiter = comma;
- }
- if (delimiter != comma)
- out << "default";
- return out << "}";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const env_managed::create_parameters &it) {
- return out << "{\n" //
- << "\tfile_mode " << std::oct << it.file_mode_bits << std::dec //
- << ",\n\tsubdirectory " << (it.use_subdirectory ? "yes" : "no") //
- << ",\n"
- << it.geometry << "}";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const MDBX_log_level_t &it) {
- switch (it) {
- case MDBX_LOG_FATAL:
- return out << "LOG_FATAL";
- case MDBX_LOG_ERROR:
- return out << "LOG_ERROR";
- case MDBX_LOG_WARN:
- return out << "LOG_WARN";
- case MDBX_LOG_NOTICE:
- return out << "LOG_NOTICE";
- case MDBX_LOG_VERBOSE:
- return out << "LOG_VERBOSE";
- case MDBX_LOG_DEBUG:
- return out << "LOG_DEBUG";
- case MDBX_LOG_TRACE:
- return out << "LOG_TRACE";
- case MDBX_LOG_EXTRA:
- return out << "LOG_EXTRA";
- case MDBX_LOG_DONTCHANGE:
- return out << "LOG_DONTCHANGE";
- default:
- return out << "mdbx::log_level::invalid";
- }
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const MDBX_debug_flags_t &it) {
- if (it == MDBX_DBG_DONTCHANGE)
- return out << "DBG_DONTCHANGE";
-
- static const char comma[] = "|";
- const char *delimiter = "";
- out << "{";
- if (it & MDBX_DBG_ASSERT) {
- out << delimiter << "DBG_ASSERT";
- delimiter = comma;
- }
- if (it & MDBX_DBG_AUDIT) {
- out << delimiter << "DBG_AUDIT";
- delimiter = comma;
- }
- if (it & MDBX_DBG_JITTER) {
- out << delimiter << "DBG_JITTER";
- delimiter = comma;
- }
- if (it & MDBX_DBG_DUMP) {
- out << delimiter << "DBG_DUMP";
- delimiter = comma;
- }
- if (it & MDBX_DBG_LEGACY_MULTIOPEN) {
- out << delimiter << "DBG_LEGACY_MULTIOPEN";
- delimiter = comma;
- }
- if (it & MDBX_DBG_LEGACY_OVERLAP) {
- out << delimiter << "DBG_LEGACY_OVERLAP";
- delimiter = comma;
- }
- if (delimiter != comma)
- out << "DBG_NONE";
- return out << "}";
-}
-
-__cold ::std::ostream &operator<<(::std::ostream &out,
- const ::mdbx::error &err) {
- return out << err.what() << " (" << long(err.code()) << ")";
-}
-
-} // namespace mdbx
diff --git a/3rdparty/libmdbx/mdbx.h b/3rdparty/libmdbx/mdbx.h
deleted file mode 100644
index 2fc7e2de0ad17f646eece08e7189951da0a20a9f..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx.h
+++ /dev/null
@@ -1,5333 +0,0 @@
-/**
-
-_libmdbx_ is an extremely fast, compact, powerful, embedded,
-transactional [key-value
-store](https://en.wikipedia.org/wiki/Key-value_database) database, with
-[permissive license](./LICENSE). _MDBX_ has a specific set of properties and
-capabilities, focused on creating unique lightweight solutions with
-extraordinary performance.
-
-_libmdbx_ is superior to [LMDB](https://bit.ly/26ts7tL) in terms of features
-and reliability, not inferior in performance. In comparison to LMDB, _libmdbx_
-makes many things just work perfectly, not silently and catastrophically
-break down. _libmdbx_ supports Linux, Windows, MacOS, OSX, iOS, Android,
-FreeBSD, DragonFly, Solaris, OpenSolaris, OpenIndiana, NetBSD, OpenBSD and other
-systems compliant with POSIX.1-2008.
-
-The origin has been migrated to
-[GitFlic](https://gitflic.ru/project/erthink/libmdbx) since on 2022-04-15
-the Github administration, without any warning nor explanation, deleted libmdbx
-along with a lot of other projects, simultaneously blocking access for many
-developers. For the same reason ~~Github~~ is blacklisted forever.
-
-_The Future will (be) [Positive](https://www.ptsecurity.com). Ð’ÑÑ‘ будет хорошо._
-
-
-\section copyright LICENSE & COPYRIGHT
-
-\authors Copyright (c) 2015-2022, Leonid Yuriev <leo@yuriev.ru>
-and other _libmdbx_ authors: please see [AUTHORS](./AUTHORS) file.
-
-\copyright Redistribution and use in source and binary forms, with or without
-modification, are permitted only as authorized by the OpenLDAP Public License.
-
-A copy of this license is available in the file LICENSE in the
-top-level directory of the distribution or, alternatively, at
-<http://www.OpenLDAP.org/license.html>.
-
- ---
-
-This code is derived from "LMDB engine" written by
-Howard Chu (Symas Corporation), which itself derived from btree.c
-written by Martin Hedenfalk.
-
- ---
-
-Portions Copyright 2011-2015 Howard Chu, Symas Corp. All rights reserved.
-
-Redistribution and use in source and binary forms, with or without
-modification, are permitted only as authorized by the OpenLDAP
-Public License.
-
-A copy of this license is available in the file LICENSE in the
-top-level directory of the distribution or, alternatively, at
-<http://www.OpenLDAP.org/license.html>.
-
- ---
-
-Portions Copyright (c) 2009, 2010 Martin Hedenfalk <martin@bzero.se>
-
-Permission to use, copy, modify, and distribute this software for any
-purpose with or without fee is hereby granted, provided that the above
-copyright notice and this permission notice appear in all copies.
-
-THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
-WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
-MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
-ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
-WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
-ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
-OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
-
-*******************************************************************************/
-
-#pragma once
-#ifndef LIBMDBX_H
-#define LIBMDBX_H
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-/**
- \file mdbx.h
- \brief The libmdbx C API header file
-
- \defgroup c_api C API
- @{
- \defgroup c_err Error handling
- \defgroup c_opening Opening & Closing
- \defgroup c_transactions Transactions
- \defgroup c_dbi Databases
- \defgroup c_crud Create/Read/Update/Delete (see Quick Reference in details)
-
- \details
- \anchor c_crud_hints
-# Quick Reference for Insert/Update/Delete operations
-
-Historically, libmdbx inherits the API basis from LMDB, where it is often
-difficult to select flags/options and functions for the desired operation.
-So it is recommend using this hints.
-
-## Databases with UNIQUE keys
-
-In databases created without the \ref MDBX_DUPSORT option, keys are always
-unique. Thus always a single value corresponds to the each key, and so there
-are only a few cases of changing data.
-
-| Case | Flags to use | Result |
-|---------------------------------------------|---------------------|------------------------|
-| _INSERTING_|||
-|Key is absent → Insertion |\ref MDBX_NOOVERWRITE|Insertion |
-|Key exist → Error since key present |\ref MDBX_NOOVERWRITE|Error \ref MDBX_KEYEXIST and return Present value|
-| _UPSERTING_|||
-|Key is absent → Insertion |\ref MDBX_UPSERT |Insertion |
-|Key exist → Update |\ref MDBX_UPSERT |Update |
-| _UPDATING_|||
-|Key is absent → Error since no such key |\ref MDBX_CURRENT |Error \ref MDBX_NOTFOUND|
-|Key exist → Update |\ref MDBX_CURRENT |Update value |
-| _DELETING_|||
-|Key is absent → Error since no such key |\ref mdbx_del() or \ref mdbx_replace()|Error \ref MDBX_NOTFOUND|
-|Key exist → Delete by key |\ref mdbx_del() with the parameter `data = NULL`|Deletion|
-|Key exist → Delete by key with with data matching check|\ref mdbx_del() with the parameter `data` filled with the value which should be match for deletion|Deletion or \ref MDBX_NOTFOUND if the value does not match|
-|Delete at the current cursor position |\ref mdbx_cursor_del() with \ref MDBX_CURRENT flag|Deletion|
-|Extract (read & delete) value by the key |\ref mdbx_replace() with zero flag and parameter `new_data = NULL`|Returning a deleted value|
-
-
-## Databases with NON-UNIQUE keys
-
-In databases created with the \ref MDBX_DUPSORT (Sorted Duplicates) option, keys
-may be non unique. Such non-unique keys in a key-value database may be treated
-as a duplicates or as like a multiple values corresponds to keys.
-
-
-| Case | Flags to use | Result |
-|---------------------------------------------|---------------------|------------------------|
-| _INSERTING_|||
-|Key is absent → Insertion |\ref MDBX_NOOVERWRITE|Insertion|
-|Key exist → Needn't to add new values |\ref MDBX_NOOVERWRITE|Error \ref MDBX_KEYEXIST with returning the first value from those already present|
-| _UPSERTING_|||
-|Key is absent → Insertion |\ref MDBX_UPSERT |Insertion|
-|Key exist → Wanna to add new values |\ref MDBX_UPSERT |Add one more value to the key|
-|Key exist → Replace all values with a new one|\ref MDBX_UPSERT + \ref MDBX_ALLDUPS|Overwrite by single new value|
-| _UPDATING_|||
-|Key is absent → Error since no such key |\ref MDBX_CURRENT |Error \ref MDBX_NOTFOUND|
-|Key exist, Single value → Update |\ref MDBX_CURRENT |Update single value |
-|Key exist, Multiple values → Replace all values with a new one|\ref MDBX_CURRENT + \ref MDBX_ALLDUPS|Overwrite by single new value|
-|Key exist, Multiple values → Error since it is unclear which of the values should be updated|\ref mdbx_put() with \ref MDBX_CURRENT|Error \ref MDBX_EMULTIVAL|
-|Key exist, Multiple values → Update particular entry of multi-value|\ref mdbx_replace() with \ref MDBX_CURRENT + \ref MDBX_NOOVERWRITE and the parameter `old_value` filled with the value that wanna to update|Update one multi-value entry|
-|Key exist, Multiple values → Update the current entry of multi-value|\ref mdbx_cursor_put() with \ref MDBX_CURRENT|Update one multi-value entry|
-| _DELETING_|||
-|Key is absent → Error since no such key |\ref mdbx_del() or \ref mdbx_replace()|Error \ref MDBX_NOTFOUND|
-|Key exist → Delete all values corresponds given key|\ref mdbx_del() with the parameter `data = NULL`|Deletion|
-|Key exist → Delete particular value corresponds given key|\ref mdbx_del() with the parameter `data` filled with the value that wanna to delete, or \ref mdbx_replace() with \ref MDBX_CURRENT + \ref MDBX_NOOVERWRITE and the `old_value` parameter filled with the value that wanna to delete and `new_data = NULL`| Deletion or \ref MDBX_NOTFOUND if no such key-value pair|
-|Delete one value at the current cursor position|\ref mdbx_cursor_del() with \ref MDBX_CURRENT flag|Deletion only the current entry|
-|Delete all values of key at the current cursor position|\ref mdbx_cursor_del() with with \ref MDBX_ALLDUPS flag|Deletion all duplicates of key (all multi-values) at the current cursor position|
-
- \defgroup c_cursors Cursors
- \defgroup c_statinfo Statistics & Information
- \defgroup c_settings Settings
- \defgroup c_debug Logging and runtime debug
- \defgroup c_rqest Range query estimation
- \defgroup c_extra Extra operations
-*/
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#include <stdarg.h>
-#include <stddef.h>
-#include <stdint.h>
-#if !defined(NDEBUG) && !defined(assert)
-#include <assert.h>
-#endif /* NDEBUG */
-
-#if defined(_WIN32) || defined(_WIN64)
-#include <windows.h>
-#include <winnt.h>
-#ifndef __mode_t_defined
-typedef unsigned short mdbx_mode_t;
-#else
-typedef mode_t mdbx_mode_t;
-#endif /* __mode_t_defined */
-typedef HANDLE mdbx_filehandle_t;
-typedef DWORD mdbx_pid_t;
-typedef DWORD mdbx_tid_t;
-#else /* Windows */
-#include <errno.h> /* for error codes */
-#include <pthread.h> /* for pthread_t */
-#include <sys/types.h> /* for pid_t */
-#include <sys/uio.h> /* for struct iovec */
-#define HAVE_STRUCT_IOVEC 1
-typedef int mdbx_filehandle_t;
-typedef pid_t mdbx_pid_t;
-typedef pthread_t mdbx_tid_t;
-typedef mode_t mdbx_mode_t;
-#endif /* !Windows */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-/** @} close c_api
- * \defgroup api_macros Common Macros
- * @{ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __has_attribute
-#define __has_attribute(x) (0)
-#endif /* __has_attribute */
-
-#ifndef __has_cpp_attribute
-#define __has_cpp_attribute(x) 0
-#endif /* __has_cpp_attribute */
-
-#ifndef __has_feature
-#define __has_feature(x) (0)
-#endif /* __has_feature */
-
-#ifndef __has_extension
-#define __has_extension(x) (0)
-#endif /* __has_extension */
-
-#ifndef __has_builtin
-#define __has_builtin(x) (0)
-#endif /* __has_builtin */
-
-/** Many functions have no effects except the return value and their
- * return value depends only on the parameters and/or global variables.
- * Such a function can be subject to common subexpression elimination
- * and loop optimization just as an arithmetic operator would be.
- * These functions should be declared with the attribute pure. */
-#if (defined(__GNUC__) || __has_attribute(__pure__)) && \
- (!defined(__clang__) /* https://bugs.llvm.org/show_bug.cgi?id=43275 */ \
- || !defined(__cplusplus) || !__has_feature(cxx_exceptions))
-#define MDBX_PURE_FUNCTION __attribute__((__pure__))
-#elif defined(_MSC_VER) && !defined(__clang__) && _MSC_VER >= 1920
-#define MDBX_PURE_FUNCTION
-#elif defined(__cplusplus) && __has_cpp_attribute(gnu::pure) && \
- (!defined(__clang__) || !__has_feature(cxx_exceptions))
-#define MDBX_PURE_FUNCTION [[gnu::pure]]
-#else
-#define MDBX_PURE_FUNCTION
-#endif /* MDBX_PURE_FUNCTION */
-
-/** Like \ref MDBX_PURE_FUNCTION with addition `noexcept` restriction
- * that is compatible to CLANG and proposed [[pure]]. */
-#if defined(__GNUC__) || \
- (__has_attribute(__pure__) && __has_attribute(__nothrow__))
-#define MDBX_NOTHROW_PURE_FUNCTION __attribute__((__pure__, __nothrow__))
-#elif defined(_MSC_VER) && !defined(__clang__) && _MSC_VER >= 1920
-#if __has_cpp_attribute(pure)
-#define MDBX_NOTHROW_PURE_FUNCTION [[pure]]
-#else
-#define MDBX_NOTHROW_PURE_FUNCTION
-#endif
-#elif defined(__cplusplus) && __has_cpp_attribute(gnu::pure)
-#if __has_cpp_attribute(gnu::nothrow)
-#define MDBX_NOTHROW_PURE_FUNCTION [[gnu::pure, gnu::nothrow]]
-#else
-#define MDBX_NOTHROW_PURE_FUNCTION [[gnu::pure]]
-#endif
-#elif defined(__cplusplus) && __has_cpp_attribute(pure)
-#define MDBX_NOTHROW_PURE_FUNCTION [[pure]]
-#else
-#define MDBX_NOTHROW_PURE_FUNCTION
-#endif /* MDBX_NOTHROW_PURE_FUNCTION */
-
-/** Many functions do not examine any values except their arguments,
- * and have no effects except the return value. Basically this is just
- * slightly more strict class than the PURE attribute, since function
- * is not allowed to read global memory.
- *
- * Note that a function that has pointer arguments and examines the
- * data pointed to must not be declared const. Likewise, a function
- * that calls a non-const function usually must not be const.
- * It does not make sense for a const function to return void. */
-#if (defined(__GNUC__) || __has_attribute(__pure__)) && \
- (!defined(__clang__) /* https://bugs.llvm.org/show_bug.cgi?id=43275 */ \
- || !defined(__cplusplus) || !__has_feature(cxx_exceptions))
-#define MDBX_CONST_FUNCTION __attribute__((__const__))
-#elif defined(_MSC_VER) && !defined(__clang__) && _MSC_VER >= 1920
-#define MDBX_CONST_FUNCTION MDBX_PURE_FUNCTION
-#elif defined(__cplusplus) && __has_cpp_attribute(gnu::const) && \
- (!defined(__clang__) || !__has_feature(cxx_exceptions))
-#define MDBX_CONST_FUNCTION [[gnu::const]]
-#else
-#define MDBX_CONST_FUNCTION MDBX_PURE_FUNCTION
-#endif /* MDBX_CONST_FUNCTION */
-
-/** Like \ref MDBX_CONST_FUNCTION with addition `noexcept` restriction
- * that is compatible to CLANG and future [[const]]. */
-#if defined(__GNUC__) || \
- (__has_attribute(__const__) && __has_attribute(__nothrow__))
-#define MDBX_NOTHROW_CONST_FUNCTION __attribute__((__const__, __nothrow__))
-#elif defined(_MSC_VER) && !defined(__clang__) && _MSC_VER >= 1920
-#define MDBX_NOTHROW_CONST_FUNCTION MDBX_NOTHROW_PURE_FUNCTION
-#elif defined(__cplusplus) && __has_cpp_attribute(gnu::const)
-#if __has_cpp_attribute(gnu::nothrow)
-#define MDBX_NOTHROW_PURE_FUNCTION [[gnu::const, gnu::nothrow]]
-#else
-#define MDBX_NOTHROW_PURE_FUNCTION [[gnu::const]]
-#endif
-#elif defined(__cplusplus) && __has_cpp_attribute(const)
-#define MDBX_NOTHROW_CONST_FUNCTION [[const]]
-#else
-#define MDBX_NOTHROW_CONST_FUNCTION MDBX_NOTHROW_PURE_FUNCTION
-#endif /* MDBX_NOTHROW_CONST_FUNCTION */
-
-#ifndef MDBX_DEPRECATED /* may be predefined to avoid warnings "deprecated" */
-#ifdef __deprecated
-#define MDBX_DEPRECATED __deprecated
-#elif defined(__GNUC__) || __has_attribute(__deprecated__)
-#define MDBX_DEPRECATED __attribute__((__deprecated__))
-#elif defined(_MSC_VER)
-#define MDBX_DEPRECATED __declspec(deprecated)
-#else
-#define MDBX_DEPRECATED
-#endif
-#endif /* MDBX_DEPRECATED */
-
-#ifndef __dll_export
-#if defined(_WIN32) || defined(_WIN64) || defined(__CYGWIN__) || \
- defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)
-#if defined(__GNUC__) || __has_attribute(__dllexport__)
-#define __dll_export __attribute__((__dllexport__))
-#elif defined(_MSC_VER)
-#define __dll_export __declspec(dllexport)
-#else
-#define __dll_export
-#endif
-#elif defined(__GNUC__) || __has_attribute(__visibility__)
-#define __dll_export __attribute__((__visibility__("default")))
-#else
-#define __dll_export
-#endif
-#endif /* __dll_export */
-
-#ifndef __dll_import
-#if defined(_WIN32) || defined(_WIN64) || defined(__CYGWIN__) || \
- defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)
-#if defined(__GNUC__) || __has_attribute(__dllimport__)
-#define __dll_import __attribute__((__dllimport__))
-#elif defined(_MSC_VER)
-#define __dll_import __declspec(dllimport)
-#else
-#define __dll_import
-#endif
-#else
-#define __dll_import
-#endif
-#endif /* __dll_import */
-
-/** \brief Auxiliary macro for robustly define the both inline version of API
- * function and non-inline fallback dll-exported version for applications linked
- * with old version of libmdbx, with a strictly ODR-common implementation. */
-#if defined(LIBMDBX_INTERNALS) && !defined(LIBMDBX_NO_EXPORTS_LEGACY_API)
-#define LIBMDBX_INLINE_API(TYPE, NAME, ARGS) \
- /* proto of exported which uses common impl */ LIBMDBX_API TYPE NAME ARGS; \
- /* definition of common impl */ static __inline TYPE __inline_##NAME ARGS
-#else
-#define LIBMDBX_INLINE_API(TYPE, NAME, ARGS) static __inline TYPE NAME ARGS
-#endif /* LIBMDBX_INLINE_API */
-
-/** \brief Converts a macro argument into a string constant. */
-#ifndef MDBX_STRINGIFY
-#define MDBX_STRINGIFY_HELPER(x) #x
-#define MDBX_STRINGIFY(x) MDBX_STRINGIFY_HELPER(x)
-#endif /* MDBX_STRINGIFY */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __cplusplus
-#ifndef bool
-#define bool _Bool
-#endif
-#ifndef true
-#define true (1)
-#endif
-#ifndef false
-#define false (0)
-#endif
-#endif /* bool without __cplusplus */
-
-#if !defined(DOXYGEN) && (!defined(__cpp_noexcept_function_type) || \
- __cpp_noexcept_function_type < 201510L)
-#define MDBX_CXX17_NOEXCEPT
-#else
-#define MDBX_CXX17_NOEXCEPT noexcept
-#endif /* MDBX_CXX17_NOEXCEPT */
-
-/* Workaround for old compilers without properly support for constexpr. */
-#if !defined(__cplusplus)
-#define MDBX_CXX01_CONSTEXPR __inline
-#define MDBX_CXX01_CONSTEXPR_VAR const
-#elif !defined(DOXYGEN) && \
- ((__cplusplus < 201103L && defined(__cpp_constexpr) && \
- __cpp_constexpr < 200704L) || \
- (defined(__LCC__) && __LCC__ < 124) || \
- (defined(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ < 407) && \
- !defined(__clang__) && !defined(__LCC__)) || \
- (defined(_MSC_VER) && _MSC_VER < 1910) || \
- (defined(__clang__) && __clang_major__ < 4))
-#define MDBX_CXX01_CONSTEXPR inline
-#define MDBX_CXX01_CONSTEXPR_VAR const
-#else
-#define MDBX_CXX01_CONSTEXPR constexpr
-#define MDBX_CXX01_CONSTEXPR_VAR constexpr
-#endif /* MDBX_CXX01_CONSTEXPR */
-
-#if !defined(__cplusplus)
-#define MDBX_CXX11_CONSTEXPR __inline
-#define MDBX_CXX11_CONSTEXPR_VAR const
-#elif !defined(DOXYGEN) && \
- (!defined(__cpp_constexpr) || __cpp_constexpr < 201304L || \
- (defined(__LCC__) && __LCC__ < 124) || \
- (defined(__GNUC__) && __GNUC__ < 6 && !defined(__clang__) && \
- !defined(__LCC__)) || \
- (defined(_MSC_VER) && _MSC_VER < 1910) || \
- (defined(__clang__) && __clang_major__ < 5))
-#define MDBX_CXX11_CONSTEXPR inline
-#define MDBX_CXX11_CONSTEXPR_VAR const
-#else
-#define MDBX_CXX11_CONSTEXPR constexpr
-#define MDBX_CXX11_CONSTEXPR_VAR constexpr
-#endif /* MDBX_CXX11_CONSTEXPR */
-
-#if !defined(__cplusplus)
-#define MDBX_CXX14_CONSTEXPR __inline
-#define MDBX_CXX14_CONSTEXPR_VAR const
-#elif defined(DOXYGEN) || \
- defined(__cpp_constexpr) && __cpp_constexpr >= 201304L && \
- ((defined(_MSC_VER) && _MSC_VER >= 1910) || \
- (defined(__clang__) && __clang_major__ > 4) || \
- (defined(__GNUC__) && __GNUC__ > 6) || \
- (!defined(__GNUC__) && !defined(__clang__) && !defined(_MSC_VER)))
-#define MDBX_CXX14_CONSTEXPR constexpr
-#define MDBX_CXX14_CONSTEXPR_VAR constexpr
-#else
-#define MDBX_CXX14_CONSTEXPR inline
-#define MDBX_CXX14_CONSTEXPR_VAR const
-#endif /* MDBX_CXX14_CONSTEXPR */
-
-#if defined(__noreturn)
-#define MDBX_NORETURN __noreturn
-#elif defined(_Noreturn)
-#define MDBX_NORETURN _Noreturn
-#elif defined(__GNUC__) || __has_attribute(__noreturn__)
-#define MDBX_NORETURN __attribute__((__noreturn__))
-#elif defined(_MSC_VER) && !defined(__clang__)
-#define MDBX_NORETURN __declspec(noreturn)
-#else
-#define MDBX_NORETURN
-#endif /* MDBX_NORETURN */
-
-#ifndef MDBX_PRINTF_ARGS
-#if defined(__GNUC__) || __has_attribute(__format__)
-#if defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)
-#define MDBX_PRINTF_ARGS(format_index, first_arg) \
- __attribute__((__format__(__gnu_printf__, format_index, first_arg)))
-#else
-#define MDBX_PRINTF_ARGS(format_index, first_arg) \
- __attribute__((__format__(__printf__, format_index, first_arg)))
-#endif /* MinGW */
-#else
-#define MDBX_PRINTF_ARGS(format_index, first_arg)
-#endif
-#endif /* MDBX_PRINTF_ARGS */
-
-#if defined(DOXYGEN) || \
- (defined(__cplusplus) && __cplusplus >= 201603 && \
- __has_cpp_attribute(maybe_unused) && \
- __has_cpp_attribute(maybe_unused) >= 201603) || \
- (!defined(__cplusplus) && defined(__STDC_VERSION__) && \
- __STDC_VERSION__ > 202005L)
-#define MDBX_MAYBE_UNUSED [[maybe_unused]]
-#elif defined(__GNUC__) || __has_attribute(__unused__)
-#define MDBX_MAYBE_UNUSED __attribute__((__unused__))
-#else
-#define MDBX_MAYBE_UNUSED
-#endif /* MDBX_MAYBE_UNUSED */
-
-#if __has_attribute(no_sanitize)
-#define MDBX_NOSANITIZE_ENUM __attribute((__no_sanitize__("enum")))
-#else
-#define MDBX_NOSANITIZE_ENUM
-#endif /* MDBX_NOSANITIZE_ENUM */
-
-/* Oh, below are some songs and dances since:
- * - C++ requires explicit definition of the necessary operators.
- * - the proper implementation of DEFINE_ENUM_FLAG_OPERATORS for C++ required
- * the constexpr feature which is broken in most old compilers;
- * - DEFINE_ENUM_FLAG_OPERATORS may be defined broken as in the Windows SDK. */
-#ifndef DEFINE_ENUM_FLAG_OPERATORS
-
-#ifdef __cplusplus
-#if !defined(__cpp_constexpr) || __cpp_constexpr < 200704L || \
- (defined(__LCC__) && __LCC__ < 124) || \
- (defined(__GNUC__) && (__GNUC__ * 100 + __GNUC_MINOR__ < 407) && \
- !defined(__clang__) && !defined(__LCC__)) || \
- (defined(_MSC_VER) && _MSC_VER < 1910) || \
- (defined(__clang__) && __clang_major__ < 4)
-/* The constexpr feature is not available or (may be) broken */
-#define CONSTEXPR_ENUM_FLAGS_OPERATIONS 0
-#else
-/* C always allows these operators for enums */
-#define CONSTEXPR_ENUM_FLAGS_OPERATIONS 1
-#endif /* __cpp_constexpr */
-
-/// Define operator overloads to enable bit operations on enum values that are
-/// used to define flags (based on Microsoft's DEFINE_ENUM_FLAG_OPERATORS).
-#define DEFINE_ENUM_FLAG_OPERATORS(ENUM) \
- extern "C++" { \
- MDBX_NOSANITIZE_ENUM MDBX_CXX01_CONSTEXPR ENUM operator|(ENUM a, ENUM b) { \
- return ENUM(unsigned(a) | unsigned(b)); \
- } \
- MDBX_NOSANITIZE_ENUM MDBX_CXX14_CONSTEXPR ENUM &operator|=(ENUM &a, \
- ENUM b) { \
- return a = a | b; \
- } \
- MDBX_NOSANITIZE_ENUM MDBX_CXX01_CONSTEXPR ENUM operator&(ENUM a, ENUM b) { \
- return ENUM(unsigned(a) & unsigned(b)); \
- } \
- MDBX_NOSANITIZE_ENUM MDBX_CXX01_CONSTEXPR ENUM operator&(ENUM a, \
- unsigned b) { \
- return ENUM(unsigned(a) & b); \
- } \
- MDBX_NOSANITIZE_ENUM MDBX_CXX01_CONSTEXPR ENUM operator&(unsigned a, \
- ENUM b) { \
- return ENUM(a & unsigned(b)); \
- } \
- MDBX_NOSANITIZE_ENUM MDBX_CXX14_CONSTEXPR ENUM &operator&=(ENUM &a, \
- ENUM b) { \
- return a = a & b; \
- } \
- MDBX_NOSANITIZE_ENUM MDBX_CXX14_CONSTEXPR ENUM &operator&=(ENUM &a, \
- unsigned b) { \
- return a = a & b; \
- } \
- MDBX_CXX01_CONSTEXPR unsigned operator~(ENUM a) { return ~unsigned(a); } \
- MDBX_NOSANITIZE_ENUM MDBX_CXX01_CONSTEXPR ENUM operator^(ENUM a, ENUM b) { \
- return ENUM(unsigned(a) ^ unsigned(b)); \
- } \
- MDBX_NOSANITIZE_ENUM MDBX_CXX14_CONSTEXPR ENUM &operator^=(ENUM &a, \
- ENUM b) { \
- return a = a ^ b; \
- } \
- }
-#else /* __cplusplus */
-/* nope for C since it always allows these operators for enums */
-#define DEFINE_ENUM_FLAG_OPERATORS(ENUM)
-#define CONSTEXPR_ENUM_FLAGS_OPERATIONS 1
-#endif /* !__cplusplus */
-
-#elif !defined(CONSTEXPR_ENUM_FLAGS_OPERATIONS)
-
-#ifdef __cplusplus
-/* DEFINE_ENUM_FLAG_OPERATORS may be defined broken as in the Windows SDK */
-#define CONSTEXPR_ENUM_FLAGS_OPERATIONS 0
-#else
-/* C always allows these operators for enums */
-#define CONSTEXPR_ENUM_FLAGS_OPERATIONS 1
-#endif
-
-#endif /* DEFINE_ENUM_FLAG_OPERATORS */
-
-/** @} end of Common Macros */
-
-/*----------------------------------------------------------------------------*/
-
-/** \addtogroup c_api
- * @{ */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* MDBX version 0.11.x */
-#define MDBX_VERSION_MAJOR 0
-#define MDBX_VERSION_MINOR 11
-
-#ifndef LIBMDBX_API
-#if defined(LIBMDBX_EXPORTS)
-#define LIBMDBX_API __dll_export
-#elif defined(LIBMDBX_IMPORTS)
-#define LIBMDBX_API __dll_import
-#else
-#define LIBMDBX_API
-#endif
-#endif /* LIBMDBX_API */
-
-#ifdef __cplusplus
-#if defined(__clang__) || __has_attribute(type_visibility)
-#define LIBMDBX_API_TYPE LIBMDBX_API __attribute__((type_visibility("default")))
-#else
-#define LIBMDBX_API_TYPE LIBMDBX_API
-#endif
-#else
-#define LIBMDBX_API_TYPE
-#endif /* LIBMDBX_API_TYPE */
-
-#if defined(LIBMDBX_IMPORTS)
-#define LIBMDBX_VERINFO_API __dll_import
-#else
-#define LIBMDBX_VERINFO_API __dll_export
-#endif /* LIBMDBX_VERINFO_API */
-
-/** \brief libmdbx version information */
-extern LIBMDBX_VERINFO_API const struct MDBX_version_info {
- uint8_t major; /**< Major version number */
- uint8_t minor; /**< Minor version number */
- uint16_t release; /**< Release number of Major.Minor */
- uint32_t revision; /**< Revision number of Release */
- struct {
- const char *datetime; /**< committer date, strict ISO-8601 format */
- const char *tree; /**< commit hash (hexadecimal digits) */
- const char *commit; /**< tree hash, i.e. digest of the source code */
- const char *describe; /**< git-describe string */
- } git; /**< source information from git */
- const char *sourcery; /**< sourcery anchor for pinning */
-} /** \brief libmdbx version information */ mdbx_version;
-
-/** \brief libmdbx build information
- * \attention Some strings could be NULL in case no corresponding information
- * was provided at build time (i.e. flags). */
-extern LIBMDBX_VERINFO_API const struct MDBX_build_info {
- const char *datetime; /**< build timestamp (ISO-8601 or __DATE__ __TIME__) */
- const char *target; /**< cpu/arch-system-config triplet */
- const char *options; /**< mdbx-related options */
- const char *compiler; /**< compiler */
- const char *flags; /**< CFLAGS and CXXFLAGS */
-} /** \brief libmdbx build information */ mdbx_build;
-
-#if (defined(_WIN32) || defined(_WIN64)) && !MDBX_BUILD_SHARED_LIBRARY
-/* MDBX internally uses global and thread local storage destructors to
- * automatically (de)initialization, releasing reader lock table slots
- * and so on.
- *
- * If MDBX builded as a DLL this is done out-of-the-box by DllEntry() function,
- * which called automatically by Windows core with passing corresponding reason
- * argument.
- *
- * Otherwise, if MDBX was builded not as a DLL, some black magic
- * may be required depending of Windows version:
- *
- * - Modern Windows versions, including Windows Vista and later, provides
- * support for "TLS Directory" (e.g .CRT$XL[A-Z] sections in executable
- * or dll file). In this case, MDBX capable of doing all automatically,
- * therefore you DON'T NEED to call mdbx_module_handler()
- * so the MDBX_MANUAL_MODULE_HANDLER defined as 0.
- *
- * - Obsolete versions of Windows, prior to Windows Vista, REQUIRES calling
- * mdbx_module_handler() manually from corresponding DllMain() or WinMain()
- * of your DLL or application,
- * so the MDBX_MANUAL_MODULE_HANDLER defined as 1.
- *
- * Therefore, building MDBX as a DLL is recommended for all version of Windows.
- * So, if you doubt, just build MDBX as the separate DLL and don't care about
- * the MDBX_MANUAL_MODULE_HANDLER. */
-
-#ifndef _WIN32_WINNT
-#error Non-dll build libmdbx requires target Windows version \
- to be explicitly defined via _WIN32_WINNT for properly \
- handling thread local storage destructors.
-#endif /* _WIN32_WINNT */
-
-#if _WIN32_WINNT >= 0x0600 /* Windows Vista */
-/* As described above mdbx_module_handler() is NOT needed for Windows Vista
- * and later. */
-#define MDBX_MANUAL_MODULE_HANDLER 0
-#else
-/* As described above mdbx_module_handler() IS REQUIRED for Windows versions
- * prior to Windows Vista. */
-#define MDBX_MANUAL_MODULE_HANDLER 1
-void LIBMDBX_API NTAPI mdbx_module_handler(PVOID module, DWORD reason,
- PVOID reserved);
-#endif
-
-#endif /* Windows && !DLL && MDBX_MANUAL_MODULE_HANDLER */
-
-/* OPACITY STRUCTURES *********************************************************/
-
-/** \brief Opaque structure for a database environment.
- * \details An environment supports multiple key-value sub-databases (aka
- * key-value spaces or tables), all residing in the same shared-memory map.
- * \see mdbx_env_create() \see mdbx_env_close() */
-#ifndef __cplusplus
-typedef struct MDBX_env MDBX_env;
-#else
-struct MDBX_env;
-#endif
-
-/** \brief Opaque structure for a transaction handle.
- * \ingroup c_transactions
- * \details All database operations require a transaction handle. Transactions
- * may be read-only or read-write.
- * \see mdbx_txn_begin() \see mdbx_txn_commit() \see mdbx_txn_abort() */
-#ifndef __cplusplus
-typedef struct MDBX_txn MDBX_txn;
-#else
-struct MDBX_txn;
-#endif
-
-/** \brief A handle for an individual database (key-value spaces) in the
- * environment.
- * \ingroup c_dbi
- * \details Zero handle is used internally (hidden Garbage Collection subDB).
- * So, any valid DBI-handle great than 0 and less than or equal
- * \ref MDBX_MAX_DBI.
- * \see mdbx_dbi_open() \see mdbx_dbi_close() */
-typedef uint32_t MDBX_dbi;
-
-/** \brief Opaque structure for navigating through a database
- * \ingroup c_cursors
- * \see mdbx_cursor_create() \see mdbx_cursor_bind() \see mdbx_cursor_close()
- */
-#ifndef __cplusplus
-typedef struct MDBX_cursor MDBX_cursor;
-#else
-struct MDBX_cursor;
-#endif
-
-/** \brief Generic structure used for passing keys and data in and out of the
- * database.
- * \anchor MDBX_val \see mdbx::slice \see mdbx::buffer
- *
- * \details Values returned from the database are valid only until a subsequent
- * update operation, or the end of the transaction. Do not modify or
- * free them, they commonly point into the database itself.
- *
- * Key sizes must be between 0 and \ref mdbx_env_get_maxkeysize() inclusive.
- * The same applies to data sizes in databases with the \ref MDBX_DUPSORT flag.
- * Other data items can in theory be from 0 to \ref MDBX_MAXDATASIZE bytes long.
- *
- * \note The notable difference between MDBX and LMDB is that MDBX support zero
- * length keys. */
-#ifndef HAVE_STRUCT_IOVEC
-struct iovec {
- void *iov_base; /**< pointer to some data */
- size_t iov_len; /**< the length of data in bytes */
-};
-#define HAVE_STRUCT_IOVEC
-#endif /* HAVE_STRUCT_IOVEC */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-/* The `iov_len` is signed on Sun/Solaris.
- * So define custom MDBX_val to avoid a lot of warnings. */
-struct MDBX_val {
- void *iov_base; /**< pointer to some data */
- size_t iov_len; /**< the length of data in bytes */
-};
-#ifndef __cplusplus
-typedef struct MDBX_val MDBX_val;
-#endif
-#else /* SunOS */
-typedef struct iovec MDBX_val;
-#endif /* ! SunOS */
-
-enum MDBX_constants {
- /** The hard limit for DBI handles */
- MDBX_MAX_DBI = UINT32_C(32765),
-
- /** The maximum size of a data item. */
- MDBX_MAXDATASIZE = UINT32_C(0x7fff0000),
-
- /** The minimal database page size in bytes. */
- MDBX_MIN_PAGESIZE = 256,
-
- /** The maximal database page size in bytes. */
- MDBX_MAX_PAGESIZE = 65536,
-};
-
-/* THE FILES *******************************************************************
- * At the file system level, the environment corresponds to a pair of files. */
-
-/** \brief The name of the lock file in the environment */
-#define MDBX_LOCKNAME "/mdbx.lck"
-/** \brief The name of the data file in the environment */
-#define MDBX_DATANAME "/mdbx.dat"
-
-/** \brief The suffix of the lock file when \ref MDBX_NOSUBDIR is used */
-#define MDBX_LOCK_SUFFIX "-lck"
-
-/* DEBUG & LOGGING ************************************************************/
-
-/** \addtogroup c_debug
- * \note Most of debug feature enabled only when libmdbx builded with
- * \ref MDBX_DEBUG build option. @{ */
-
-/** Log level
- * \note Levels detailed than (great than) \ref MDBX_LOG_NOTICE
- * requires build libmdbx with \ref MDBX_DEBUG option. */
-enum MDBX_log_level_t {
- /** Critical conditions, i.e. assertion failures.
- * \note libmdbx always produces such messages regardless
- * of \ref MDBX_DEBUG build option. */
- MDBX_LOG_FATAL = 0,
-
- /** Enables logging for error conditions
- * and \ref MDBX_LOG_FATAL.
- * \note libmdbx always produces such messages regardless
- * of \ref MDBX_DEBUG build option. */
- MDBX_LOG_ERROR = 1,
-
- /** Enables logging for warning conditions
- * and \ref MDBX_LOG_ERROR ... \ref MDBX_LOG_FATAL.
- * \note libmdbx always produces such messages regardless
- * of \ref MDBX_DEBUG build option. */
- MDBX_LOG_WARN = 2,
-
- /** Enables logging for normal but significant condition
- * and \ref MDBX_LOG_WARN ... \ref MDBX_LOG_FATAL.
- * \note libmdbx always produces such messages regardless
- * of \ref MDBX_DEBUG build option. */
- MDBX_LOG_NOTICE = 3,
-
- /** Enables logging for verbose informational
- * and \ref MDBX_LOG_NOTICE ... \ref MDBX_LOG_FATAL.
- * \note Requires build libmdbx with \ref MDBX_DEBUG option. */
- MDBX_LOG_VERBOSE = 4,
-
- /** Enables logging for debug-level messages
- * and \ref MDBX_LOG_VERBOSE ... \ref MDBX_LOG_FATAL.
- * \note Requires build libmdbx with \ref MDBX_DEBUG option. */
- MDBX_LOG_DEBUG = 5,
-
- /** Enables logging for trace debug-level messages
- * and \ref MDBX_LOG_DEBUG ... \ref MDBX_LOG_FATAL.
- * \note Requires build libmdbx with \ref MDBX_DEBUG option. */
- MDBX_LOG_TRACE = 6,
-
- /** Enables extra debug-level messages (dump pgno lists)
- * and all other log-messages.
- * \note Requires build libmdbx with \ref MDBX_DEBUG option. */
- MDBX_LOG_EXTRA = 7,
-
-#ifdef ENABLE_UBSAN
- MDBX_LOG_MAX = 7 /* avoid UBSAN false-positive trap by a tests */,
-#endif /* ENABLE_UBSAN */
-
- /** for \ref mdbx_setup_debug() only: Don't change current settings */
- MDBX_LOG_DONTCHANGE = -1
-};
-#ifndef __cplusplus
-typedef enum MDBX_log_level_t MDBX_log_level_t;
-#endif
-
-/** \brief Runtime debug flags
- *
- * \details `MDBX_DBG_DUMP` and `MDBX_DBG_LEGACY_MULTIOPEN` always have an
- * effect, but `MDBX_DBG_ASSERT`, `MDBX_DBG_AUDIT` and `MDBX_DBG_JITTER` only if
- * libmdbx builded with \ref MDBX_DEBUG. */
-enum MDBX_debug_flags_t {
- MDBX_DBG_NONE = 0,
-
- /** Enable assertion checks.
- * \note Always enabled for builds with `MDBX_FORCE_ASSERTIONS` option,
- * otherwise requires build with \ref MDBX_DEBUG > 0 */
- MDBX_DBG_ASSERT = 1,
-
- /** Enable pages usage audit at commit transactions.
- * \note Requires build with \ref MDBX_DEBUG > 0 */
- MDBX_DBG_AUDIT = 2,
-
- /** Enable small random delays in critical points.
- * \note Requires build with \ref MDBX_DEBUG > 0 */
- MDBX_DBG_JITTER = 4,
-
- /** Include or not meta-pages in coredump files.
- * \note May affect performance in \ref MDBX_WRITEMAP mode */
- MDBX_DBG_DUMP = 8,
-
- /** Allow multi-opening environment(s) */
- MDBX_DBG_LEGACY_MULTIOPEN = 16,
-
- /** Allow read and write transactions overlapping for the same thread. */
- MDBX_DBG_LEGACY_OVERLAP = 32,
-
- /** Don't auto-upgrade format signature.
- * \note However a new write transactions will use and store
- * the last signature regardless this flag */
- MDBX_DBG_DONT_UPGRADE = 64,
-
-#ifdef ENABLE_UBSAN
- MDBX_DBG_MAX = ((unsigned)MDBX_LOG_MAX) << 16 |
- 127 /* avoid UBSAN false-positive trap by a tests */,
-#endif /* ENABLE_UBSAN */
-
- /** for mdbx_setup_debug() only: Don't change current settings */
- MDBX_DBG_DONTCHANGE = -1
-};
-#ifndef __cplusplus
-typedef enum MDBX_debug_flags_t MDBX_debug_flags_t;
-#else
-DEFINE_ENUM_FLAG_OPERATORS(MDBX_debug_flags_t)
-#endif
-
-/** \brief A debug-logger callback function,
- * called before printing the message and aborting.
- * \see mdbx_setup_debug()
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] msg The assertion message, not including newline. */
-typedef void MDBX_debug_func(MDBX_log_level_t loglevel, const char *function,
- int line, const char *fmt,
- va_list args) MDBX_CXX17_NOEXCEPT;
-
-/** \brief The "don't change `logger`" value for mdbx_setup_debug() */
-#define MDBX_LOGGER_DONTCHANGE ((MDBX_debug_func *)(intptr_t)-1)
-
-/** \brief Setup global log-level, debug options and debug logger.
- * \returns The previously `debug_flags` in the 0-15 bits
- * and `log_level` in the 16-31 bits. */
-LIBMDBX_API int mdbx_setup_debug(MDBX_log_level_t log_level,
- MDBX_debug_flags_t debug_flags,
- MDBX_debug_func *logger);
-
-/** \brief A callback function for most MDBX assert() failures,
- * called before printing the message and aborting.
- * \see mdbx_env_set_assert()
- *
- * \param [in] env An environment handle returned by mdbx_env_create().
- * \param [in] msg The assertion message, not including newline. */
-typedef void MDBX_assert_func(const MDBX_env *env, const char *msg,
- const char *function,
- unsigned line) MDBX_CXX17_NOEXCEPT;
-
-/** \brief Set or reset the assert() callback of the environment.
- *
- * Does nothing if libmdbx was built with MDBX_DEBUG=0 or with NDEBUG,
- * and will return `MDBX_ENOSYS` in such case.
- *
- * \param [in] env An environment handle returned by mdbx_env_create().
- * \param [in] func An MDBX_assert_func function, or 0.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_set_assert(MDBX_env *env, MDBX_assert_func *func);
-
-/** \brief Dump given MDBX_val to the buffer
- *
- * Dumps it as string if value is printable (all bytes in the range 0x20..0x7E),
- * otherwise made hexadecimal dump. Requires at least 4 byte length buffer.
- *
- * \returns One of:
- * - NULL if given buffer size less than 4 bytes;
- * - pointer to constant string if given value NULL or empty;
- * - otherwise pointer to given buffer. */
-LIBMDBX_API const char *mdbx_dump_val(const MDBX_val *key, char *const buf,
- const size_t bufsize);
-
-/** \brief Panics with message and causes abnormal process termination. */
-LIBMDBX_API void mdbx_panic(const char *fmt, ...) MDBX_PRINTF_ARGS(1, 2);
-
-/** @} end of logging & debug */
-
-/** \brief Environment flags
- * \ingroup c_opening
- * \anchor env_flags
- * \see mdbx_env_open() \see mdbx_env_set_flags() */
-enum MDBX_env_flags_t {
- MDBX_ENV_DEFAULTS = 0,
-
- /** No environment directory.
- *
- * By default, MDBX creates its environment in a directory whose pathname is
- * given in path, and creates its data and lock files under that directory.
- * With this option, path is used as-is for the database main data file.
- * The database lock file is the path with "-lck" appended.
- *
- * - with `MDBX_NOSUBDIR` = in a filesystem we have the pair of MDBX-files
- * which names derived from given pathname by appending predefined suffixes.
- *
- * - without `MDBX_NOSUBDIR` = in a filesystem we have the MDBX-directory with
- * given pathname, within that a pair of MDBX-files with predefined names.
- *
- * This flag affects only at new environment creating by \ref mdbx_env_open(),
- * otherwise at opening an existing environment libmdbx will choice this
- * automatically. */
- MDBX_NOSUBDIR = UINT32_C(0x4000),
-
- /** Read only mode.
- *
- * Open the environment in read-only mode. No write operations will be
- * allowed. MDBX will still modify the lock file - except on read-only
- * filesystems, where MDBX does not use locks.
- *
- * - with `MDBX_RDONLY` = open environment in read-only mode.
- * MDBX supports pure read-only mode (i.e. without opening LCK-file) only
- * when environment directory and/or both files are not writable (and the
- * LCK-file may be missing). In such case allowing file(s) to be placed
- * on a network read-only share.
- *
- * - without `MDBX_RDONLY` = open environment in read-write mode.
- *
- * This flag affects only at environment opening but can't be changed after.
- */
- MDBX_RDONLY = UINT32_C(0x20000),
-
- /** Open environment in exclusive/monopolistic mode.
- *
- * `MDBX_EXCLUSIVE` flag can be used as a replacement for `MDB_NOLOCK`,
- * which don't supported by MDBX.
- * In this way, you can get the minimal overhead, but with the correct
- * multi-process and multi-thread locking.
- *
- * - with `MDBX_EXCLUSIVE` = open environment in exclusive/monopolistic mode
- * or return \ref MDBX_BUSY if environment already used by other process.
- * The main feature of the exclusive mode is the ability to open the
- * environment placed on a network share.
- *
- * - without `MDBX_EXCLUSIVE` = open environment in cooperative mode,
- * i.e. for multi-process access/interaction/cooperation.
- * The main requirements of the cooperative mode are:
- *
- * 1. data files MUST be placed in the LOCAL file system,
- * but NOT on a network share.
- * 2. environment MUST be opened only by LOCAL processes,
- * but NOT over a network.
- * 3. OS kernel (i.e. file system and memory mapping implementation) and
- * all processes that open the given environment MUST be running
- * in the physically single RAM with cache-coherency. The only
- * exception for cache-consistency requirement is Linux on MIPS
- * architecture, but this case has not been tested for a long time).
- *
- * This flag affects only at environment opening but can't be changed after.
- */
- MDBX_EXCLUSIVE = UINT32_C(0x400000),
-
- /** Using database/environment which already opened by another process(es).
- *
- * The `MDBX_ACCEDE` flag is useful to avoid \ref MDBX_INCOMPATIBLE error
- * while opening the database/environment which is already used by another
- * process(es) with unknown mode/flags. In such cases, if there is a
- * difference in the specified flags (\ref MDBX_NOMETASYNC,
- * \ref MDBX_SAFE_NOSYNC, \ref MDBX_UTTERLY_NOSYNC, \ref MDBX_LIFORECLAIM,
- * \ref MDBX_COALESCE and \ref MDBX_NORDAHEAD), instead of returning an error,
- * the database will be opened in a compatibility with the already used mode.
- *
- * `MDBX_ACCEDE` has no effect if the current process is the only one either
- * opening the DB in read-only mode or other process(es) uses the DB in
- * read-only mode. */
- MDBX_ACCEDE = UINT32_C(0x40000000),
-
- /** Map data into memory with write permission.
- *
- * Use a writeable memory map unless \ref MDBX_RDONLY is set. This uses fewer
- * mallocs and requires much less work for tracking database pages, but
- * loses protection from application bugs like wild pointer writes and other
- * bad updates into the database. This may be slightly faster for DBs that
- * fit entirely in RAM, but is slower for DBs larger than RAM. Also adds the
- * possibility for stray application writes thru pointers to silently
- * corrupt the database.
- *
- * - with `MDBX_WRITEMAP` = all data will be mapped into memory in the
- * read-write mode. This offers a significant performance benefit, since the
- * data will be modified directly in mapped memory and then flushed to disk
- * by single system call, without any memory management nor copying.
- *
- * - without `MDBX_WRITEMAP` = data will be mapped into memory in the
- * read-only mode. This requires stocking all modified database pages in
- * memory and then writing them to disk through file operations.
- *
- * \warning On the other hand, `MDBX_WRITEMAP` adds the possibility for stray
- * application writes thru pointers to silently corrupt the database.
- *
- * \note The `MDBX_WRITEMAP` mode is incompatible with nested transactions,
- * since this is unreasonable. I.e. nested transactions requires mallocation
- * of database pages and more work for tracking ones, which neuters a
- * performance boost caused by the `MDBX_WRITEMAP` mode.
- *
- * This flag affects only at environment opening but can't be changed after.
- */
- MDBX_WRITEMAP = UINT32_C(0x80000),
-
- /** Tie reader locktable slots to read-only transactions
- * instead of to threads.
- *
- * Don't use Thread-Local Storage, instead tie reader locktable slots to
- * \ref MDBX_txn objects instead of to threads. So, \ref mdbx_txn_reset()
- * keeps the slot reserved for the \ref MDBX_txn object. A thread may use
- * parallel read-only transactions. And a read-only transaction may span
- * threads if you synchronizes its use.
- *
- * Applications that multiplex many user threads over individual OS threads
- * need this option. Such an application must also serialize the write
- * transactions in an OS thread, since MDBX's write locking is unaware of
- * the user threads.
- *
- * \note Regardless to `MDBX_NOTLS` flag a write transaction entirely should
- * always be used in one thread from start to finish. MDBX checks this in a
- * reasonable manner and return the \ref MDBX_THREAD_MISMATCH error in rules
- * violation.
- *
- * This flag affects only at environment opening but can't be changed after.
- */
- MDBX_NOTLS = UINT32_C(0x200000),
-
- /** Don't do readahead.
- *
- * Turn off readahead. Most operating systems perform readahead on read
- * requests by default. This option turns it off if the OS supports it.
- * Turning it off may help random read performance when the DB is larger
- * than RAM and system RAM is full.
- *
- * By default libmdbx dynamically enables/disables readahead depending on
- * the actual database size and currently available memory. On the other
- * hand, such automation has some limitation, i.e. could be performed only
- * when DB size changing but can't tracks and reacts changing a free RAM
- * availability, since it changes independently and asynchronously.
- *
- * \note The mdbx_is_readahead_reasonable() function allows to quickly find
- * out whether to use readahead or not based on the size of the data and the
- * amount of available memory.
- *
- * This flag affects only at environment opening and can't be changed after.
- */
- MDBX_NORDAHEAD = UINT32_C(0x800000),
-
- /** Don't initialize malloc'ed memory before writing to datafile.
- *
- * Don't initialize malloc'ed memory before writing to unused spaces in the
- * data file. By default, memory for pages written to the data file is
- * obtained using malloc. While these pages may be reused in subsequent
- * transactions, freshly malloc'ed pages will be initialized to zeroes before
- * use. This avoids persisting leftover data from other code (that used the
- * heap and subsequently freed the memory) into the data file.
- *
- * Note that many other system libraries may allocate and free memory from
- * the heap for arbitrary uses. E.g., stdio may use the heap for file I/O
- * buffers. This initialization step has a modest performance cost so some
- * applications may want to disable it using this flag. This option can be a
- * problem for applications which handle sensitive data like passwords, and
- * it makes memory checkers like Valgrind noisy. This flag is not needed
- * with \ref MDBX_WRITEMAP, which writes directly to the mmap instead of using
- * malloc for pages. The initialization is also skipped if \ref MDBX_RESERVE
- * is used; the caller is expected to overwrite all of the memory that was
- * reserved in that case.
- *
- * This flag may be changed at any time using `mdbx_env_set_flags()`. */
- MDBX_NOMEMINIT = UINT32_C(0x1000000),
-
- /** Aims to coalesce a Garbage Collection items.
- *
- * With `MDBX_COALESCE` flag MDBX will aims to coalesce items while recycling
- * a Garbage Collection. Technically, when possible short lists of pages
- * will be combined into longer ones, but to fit on one database page. As a
- * result, there will be fewer items in Garbage Collection and a page lists
- * are longer, which slightly increases the likelihood of returning pages to
- * Unallocated space and reducing the database file.
- *
- * This flag may be changed at any time using mdbx_env_set_flags(). */
- MDBX_COALESCE = UINT32_C(0x2000000),
-
- /** LIFO policy for recycling a Garbage Collection items.
- *
- * `MDBX_LIFORECLAIM` flag turns on LIFO policy for recycling a Garbage
- * Collection items, instead of FIFO by default. On systems with a disk
- * write-back cache, this can significantly increase write performance, up
- * to several times in a best case scenario.
- *
- * LIFO recycling policy means that for reuse pages will be taken which became
- * unused the lastest (i.e. just now or most recently). Therefore the loop of
- * database pages circulation becomes as short as possible. In other words,
- * the number of pages, that are overwritten in memory and on disk during a
- * series of write transactions, will be as small as possible. Thus creates
- * ideal conditions for the efficient operation of the disk write-back cache.
- *
- * \ref MDBX_LIFORECLAIM is compatible with all no-sync flags, but gives NO
- * noticeable impact in combination with \ref MDBX_SAFE_NOSYNC or
- * \ref MDBX_UTTERLY_NOSYNC. Because MDBX will reused pages only before the
- * last "steady" MVCC-snapshot, i.e. the loop length of database pages
- * circulation will be mostly defined by frequency of calling
- * \ref mdbx_env_sync() rather than LIFO and FIFO difference.
- *
- * This flag may be changed at any time using mdbx_env_set_flags(). */
- MDBX_LIFORECLAIM = UINT32_C(0x4000000),
-
- /** Debugging option, fill/perturb released pages. */
- MDBX_PAGEPERTURB = UINT32_C(0x8000000),
-
- /* SYNC MODES****************************************************************/
- /** \defgroup sync_modes SYNC MODES
- *
- * \attention Using any combination of \ref MDBX_SAFE_NOSYNC, \ref
- * MDBX_NOMETASYNC and especially \ref MDBX_UTTERLY_NOSYNC is always a deal to
- * reduce durability for gain write performance. You must know exactly what
- * you are doing and what risks you are taking!
- *
- * \note for LMDB users: \ref MDBX_SAFE_NOSYNC is NOT similar to LMDB_NOSYNC,
- * but \ref MDBX_UTTERLY_NOSYNC is exactly match LMDB_NOSYNC. See details
- * below.
- *
- * THE SCENE:
- * - The DAT-file contains several MVCC-snapshots of B-tree at same time,
- * each of those B-tree has its own root page.
- * - Each of meta pages at the beginning of the DAT file contains a
- * pointer to the root page of B-tree which is the result of the particular
- * transaction, and a number of this transaction.
- * - For data durability, MDBX must first write all MVCC-snapshot data
- * pages and ensure that are written to the disk, then update a meta page
- * with the new transaction number and a pointer to the corresponding new
- * root page, and flush any buffers yet again.
- * - Thus during commit a I/O buffers should be flushed to the disk twice;
- * i.e. fdatasync(), FlushFileBuffers() or similar syscall should be
- * called twice for each commit. This is very expensive for performance,
- * but guaranteed durability even on unexpected system failure or power
- * outage. Of course, provided that the operating system and the
- * underlying hardware (e.g. disk) work correctly.
- *
- * TRADE-OFF:
- * By skipping some stages described above, you can significantly benefit in
- * speed, while partially or completely losing in the guarantee of data
- * durability and/or consistency in the event of system or power failure.
- * Moreover, if for any reason disk write order is not preserved, then at
- * moment of a system crash, a meta-page with a pointer to the new B-tree may
- * be written to disk, while the itself B-tree not yet. In that case, the
- * database will be corrupted!
- *
- * \see MDBX_SYNC_DURABLE \see MDBX_NOMETASYNC \see MDBX_SAFE_NOSYNC
- * \see MDBX_UTTERLY_NOSYNC
- *
- * @{ */
-
- /** Default robust and durable sync mode.
- *
- * Metadata is written and flushed to disk after a data is written and
- * flushed, which guarantees the integrity of the database in the event
- * of a crash at any time.
- *
- * \attention Please do not use other modes until you have studied all the
- * details and are sure. Otherwise, you may lose your users' data, as happens
- * in [Miranda NG](https://www.miranda-ng.org/) messenger. */
- MDBX_SYNC_DURABLE = 0,
-
- /** Don't sync the meta-page after commit.
- *
- * Flush system buffers to disk only once per transaction commit, omit the
- * metadata flush. Defer that until the system flushes files to disk,
- * or next non-\ref MDBX_RDONLY commit or \ref mdbx_env_sync(). Depending on
- * the platform and hardware, with \ref MDBX_NOMETASYNC you may get a doubling
- * of write performance.
- *
- * This trade-off maintains database integrity, but a system crash may
- * undo the last committed transaction. I.e. it preserves the ACI
- * (atomicity, consistency, isolation) but not D (durability) database
- * property.
- *
- * `MDBX_NOMETASYNC` flag may be changed at any time using
- * \ref mdbx_env_set_flags() or by passing to \ref mdbx_txn_begin() for
- * particular write transaction. \see sync_modes */
- MDBX_NOMETASYNC = UINT32_C(0x40000),
-
- /** Don't sync anything but keep previous steady commits.
- *
- * Like \ref MDBX_UTTERLY_NOSYNC the `MDBX_SAFE_NOSYNC` flag disable similarly
- * flush system buffers to disk when committing a transaction. But there is a
- * huge difference in how are recycled the MVCC snapshots corresponding to
- * previous "steady" transactions (see below).
- *
- * With \ref MDBX_WRITEMAP the `MDBX_SAFE_NOSYNC` instructs MDBX to use
- * asynchronous mmap-flushes to disk. Asynchronous mmap-flushes means that
- * actually all writes will scheduled and performed by operation system on it
- * own manner, i.e. unordered. MDBX itself just notify operating system that
- * it would be nice to write data to disk, but no more.
- *
- * Depending on the platform and hardware, with `MDBX_SAFE_NOSYNC` you may get
- * a multiple increase of write performance, even 10 times or more.
- *
- * In contrast to \ref MDBX_UTTERLY_NOSYNC mode, with `MDBX_SAFE_NOSYNC` flag
- * MDBX will keeps untouched pages within B-tree of the last transaction
- * "steady" which was synced to disk completely. This has big implications for
- * both data durability and (unfortunately) performance:
- * - a system crash can't corrupt the database, but you will lose the last
- * transactions; because MDBX will rollback to last steady commit since it
- * kept explicitly.
- * - the last steady transaction makes an effect similar to "long-lived" read
- * transaction (see above in the \ref restrictions section) since prevents
- * reuse of pages freed by newer write transactions, thus the any data
- * changes will be placed in newly allocated pages.
- * - to avoid rapid database growth, the system will sync data and issue
- * a steady commit-point to resume reuse pages, each time there is
- * insufficient space and before increasing the size of the file on disk.
- *
- * In other words, with `MDBX_SAFE_NOSYNC` flag MDBX insures you from the
- * whole database corruption, at the cost increasing database size and/or
- * number of disk IOPs. So, `MDBX_SAFE_NOSYNC` flag could be used with
- * \ref mdbx_env_sync() as alternatively for batch committing or nested
- * transaction (in some cases). As well, auto-sync feature exposed by
- * \ref mdbx_env_set_syncbytes() and \ref mdbx_env_set_syncperiod() functions
- * could be very useful with `MDBX_SAFE_NOSYNC` flag.
- *
- * The number and volume of of disk IOPs with MDBX_SAFE_NOSYNC flag will
- * exactly the as without any no-sync flags. However, you should expect a
- * larger process's [work set](https://bit.ly/2kA2tFX) and significantly worse
- * a [locality of reference](https://bit.ly/2mbYq2J), due to the more
- * intensive allocation of previously unused pages and increase the size of
- * the database.
- *
- * `MDBX_SAFE_NOSYNC` flag may be changed at any time using
- * \ref mdbx_env_set_flags() or by passing to \ref mdbx_txn_begin() for
- * particular write transaction. */
- MDBX_SAFE_NOSYNC = UINT32_C(0x10000),
-
- /** \deprecated Please use \ref MDBX_SAFE_NOSYNC instead of `MDBX_MAPASYNC`.
- *
- * Since version 0.9.x the `MDBX_MAPASYNC` is deprecated and has the same
- * effect as \ref MDBX_SAFE_NOSYNC with \ref MDBX_WRITEMAP. This just API
- * simplification is for convenience and clarity. */
- MDBX_MAPASYNC = MDBX_SAFE_NOSYNC,
-
- /** Don't sync anything and wipe previous steady commits.
- *
- * Don't flush system buffers to disk when committing a transaction. This
- * optimization means a system crash can corrupt the database, if buffers are
- * not yet flushed to disk. Depending on the platform and hardware, with
- * `MDBX_UTTERLY_NOSYNC` you may get a multiple increase of write performance,
- * even 100 times or more.
- *
- * If the filesystem preserves write order (which is rare and never provided
- * unless explicitly noted) and the \ref MDBX_WRITEMAP and \ref
- * MDBX_LIFORECLAIM flags are not used, then a system crash can't corrupt the
- * database, but you can lose the last transactions, if at least one buffer is
- * not yet flushed to disk. The risk is governed by how often the system
- * flushes dirty buffers to disk and how often \ref mdbx_env_sync() is called.
- * So, transactions exhibit ACI (atomicity, consistency, isolation) properties
- * and only lose `D` (durability). I.e. database integrity is maintained, but
- * a system crash may undo the final transactions.
- *
- * Otherwise, if the filesystem not preserves write order (which is
- * typically) or \ref MDBX_WRITEMAP or \ref MDBX_LIFORECLAIM flags are used,
- * you should expect the corrupted database after a system crash.
- *
- * So, most important thing about `MDBX_UTTERLY_NOSYNC`:
- * - a system crash immediately after commit the write transaction
- * high likely lead to database corruption.
- * - successful completion of mdbx_env_sync(force = true) after one or
- * more committed transactions guarantees consistency and durability.
- * - BUT by committing two or more transactions you back database into
- * a weak state, in which a system crash may lead to database corruption!
- * In case single transaction after mdbx_env_sync, you may lose transaction
- * itself, but not a whole database.
- *
- * Nevertheless, `MDBX_UTTERLY_NOSYNC` provides "weak" durability in case
- * of an application crash (but no durability on system failure), and
- * therefore may be very useful in scenarios where data durability is
- * not required over a system failure (e.g for short-lived data), or if you
- * can take such risk.
- *
- * `MDBX_UTTERLY_NOSYNC` flag may be changed at any time using
- * \ref mdbx_env_set_flags(), but don't has effect if passed to
- * \ref mdbx_txn_begin() for particular write transaction. \see sync_modes */
- MDBX_UTTERLY_NOSYNC = MDBX_SAFE_NOSYNC | UINT32_C(0x100000),
-
- /** @} end of SYNC MODES */
-};
-#ifndef __cplusplus
-/** \ingroup c_opening */
-typedef enum MDBX_env_flags_t MDBX_env_flags_t;
-#else
-DEFINE_ENUM_FLAG_OPERATORS(MDBX_env_flags_t)
-#endif
-
-/** Transaction flags
- * \ingroup c_transactions
- * \anchor txn_flags
- * \see mdbx_txn_begin() \see mdbx_txn_flags() */
-enum MDBX_txn_flags_t {
- /** Start read-write transaction.
- *
- * Only one write transaction may be active at a time. Writes are fully
- * serialized, which guarantees that writers can never deadlock. */
- MDBX_TXN_READWRITE = 0,
-
- /** Start read-only transaction.
- *
- * There can be multiple read-only transactions simultaneously that do not
- * block each other and a write transactions. */
- MDBX_TXN_RDONLY = MDBX_RDONLY,
-
-/** Prepare but not start read-only transaction.
- *
- * Transaction will not be started immediately, but created transaction handle
- * will be ready for use with \ref mdbx_txn_renew(). This flag allows to
- * preallocate memory and assign a reader slot, thus avoiding these operations
- * at the next start of the transaction. */
-#if CONSTEXPR_ENUM_FLAGS_OPERATIONS || defined(DOXYGEN)
- MDBX_TXN_RDONLY_PREPARE = MDBX_RDONLY | MDBX_NOMEMINIT,
-#else
- MDBX_TXN_RDONLY_PREPARE = uint32_t(MDBX_RDONLY) | uint32_t(MDBX_NOMEMINIT),
-#endif
-
- /** Do not block when starting a write transaction. */
- MDBX_TXN_TRY = UINT32_C(0x10000000),
-
- /** Exactly the same as \ref MDBX_NOMETASYNC,
- * but for this transaction only */
- MDBX_TXN_NOMETASYNC = MDBX_NOMETASYNC,
-
- /** Exactly the same as \ref MDBX_SAFE_NOSYNC,
- * but for this transaction only */
- MDBX_TXN_NOSYNC = MDBX_SAFE_NOSYNC
-};
-#ifndef __cplusplus
-typedef enum MDBX_txn_flags_t MDBX_txn_flags_t;
-#else
-DEFINE_ENUM_FLAG_OPERATORS(MDBX_txn_flags_t)
-#endif
-
-/** \brief Database flags
- * \ingroup c_dbi
- * \anchor db_flags
- * \see mdbx_dbi_open() */
-enum MDBX_db_flags_t {
- /** Variable length unique keys with usual byte-by-byte string comparison. */
- MDBX_DB_DEFAULTS = 0,
-
- /** Use reverse string comparison for keys. */
- MDBX_REVERSEKEY = UINT32_C(0x02),
-
- /** Use sorted duplicates, i.e. allow multi-values for a keys. */
- MDBX_DUPSORT = UINT32_C(0x04),
-
- /** Numeric keys in native byte order either uint32_t or uint64_t
- * (must be one of uint32_t or uint64_t, other integer types, for example,
- * signed integer or uint16_t will not work).
- * The keys must all be of the same size and must be aligned while passing as
- * arguments. */
- MDBX_INTEGERKEY = UINT32_C(0x08),
-
- /** With \ref MDBX_DUPSORT; sorted dup items have fixed size. The data values
- * must all be of the same size. */
- MDBX_DUPFIXED = UINT32_C(0x10),
-
- /** With \ref MDBX_DUPSORT and with \ref MDBX_DUPFIXED; dups are fixed size
- * like \ref MDBX_INTEGERKEY -style integers. The data values must all be of
- * the same size and must be aligned while passing as arguments. */
- MDBX_INTEGERDUP = UINT32_C(0x20),
-
- /** With \ref MDBX_DUPSORT; use reverse string comparison for data values. */
- MDBX_REVERSEDUP = UINT32_C(0x40),
-
- /** Create DB if not already existing. */
- MDBX_CREATE = UINT32_C(0x40000),
-
- /** Opens an existing sub-database created with unknown flags.
- *
- * The `MDBX_DB_ACCEDE` flag is intend to open a existing sub-database which
- * was created with unknown flags (\ref MDBX_REVERSEKEY, \ref MDBX_DUPSORT,
- * \ref MDBX_INTEGERKEY, \ref MDBX_DUPFIXED, \ref MDBX_INTEGERDUP and
- * \ref MDBX_REVERSEDUP).
- *
- * In such cases, instead of returning the \ref MDBX_INCOMPATIBLE error, the
- * sub-database will be opened with flags which it was created, and then an
- * application could determine the actual flags by \ref mdbx_dbi_flags(). */
- MDBX_DB_ACCEDE = MDBX_ACCEDE
-};
-#ifndef __cplusplus
-/** \ingroup c_dbi */
-typedef enum MDBX_db_flags_t MDBX_db_flags_t;
-#else
-DEFINE_ENUM_FLAG_OPERATORS(MDBX_db_flags_t)
-#endif
-
-/** \brief Data changing flags
- * \ingroup c_crud
- * \see \ref c_crud_hints "Quick reference for Insert/Update/Delete operations"
- * \see mdbx_put() \see mdbx_cursor_put() \see mdbx_replace() */
-enum MDBX_put_flags_t {
- /** Upsertion by default (without any other flags) */
- MDBX_UPSERT = 0,
-
- /** For insertion: Don't write if the key already exists. */
- MDBX_NOOVERWRITE = UINT32_C(0x10),
-
- /** Has effect only for \ref MDBX_DUPSORT databases.
- * For upsertion: don't write if the key-value pair already exist.
- * For deletion: remove all values for key. */
- MDBX_NODUPDATA = UINT32_C(0x20),
-
- /** For upsertion: overwrite the current key/data pair.
- * MDBX allows this flag for \ref mdbx_put() for explicit overwrite/update
- * without insertion.
- * For deletion: remove only single entry at the current cursor position. */
- MDBX_CURRENT = UINT32_C(0x40),
-
- /** Has effect only for \ref MDBX_DUPSORT databases.
- * For deletion: remove all multi-values (aka duplicates) for given key.
- * For upsertion: replace all multi-values for given key with a new one. */
- MDBX_ALLDUPS = UINT32_C(0x80),
-
- /** For upsertion: Just reserve space for data, don't copy it.
- * Return a pointer to the reserved space. */
- MDBX_RESERVE = UINT32_C(0x10000),
-
- /** Data is being appended.
- * Don't split full pages, continue on a new instead. */
- MDBX_APPEND = UINT32_C(0x20000),
-
- /** Has effect only for \ref MDBX_DUPSORT databases.
- * Duplicate data is being appended.
- * Don't split full pages, continue on a new instead. */
- MDBX_APPENDDUP = UINT32_C(0x40000),
-
- /** Only for \ref MDBX_DUPFIXED.
- * Store multiple data items in one call. */
- MDBX_MULTIPLE = UINT32_C(0x80000)
-};
-#ifndef __cplusplus
-/** \ingroup c_crud */
-typedef enum MDBX_put_flags_t MDBX_put_flags_t;
-#else
-DEFINE_ENUM_FLAG_OPERATORS(MDBX_put_flags_t)
-#endif
-
-/** \brief Environment copy flags
- * \ingroup c_extra
- * \see mdbx_env_copy() \see mdbx_env_copy2fd() */
-enum MDBX_copy_flags_t {
- MDBX_CP_DEFAULTS = 0,
-
- /** Copy with compactification: Omit free space from copy and renumber all
- * pages sequentially */
- MDBX_CP_COMPACT = 1u,
-
- /** Force to make resizeable copy, i.e. dynamic size instead of fixed */
- MDBX_CP_FORCE_DYNAMIC_SIZE = 2u
-};
-#ifndef __cplusplus
-/** \ingroup c_extra */
-typedef enum MDBX_copy_flags_t MDBX_copy_flags_t;
-#else
-DEFINE_ENUM_FLAG_OPERATORS(MDBX_copy_flags_t)
-#endif
-
-/** \brief Cursor operations
- * \ingroup c_cursors
- * This is the set of all operations for retrieving data using a cursor.
- * \see mdbx_cursor_get() */
-enum MDBX_cursor_op {
- /** Position at first key/data item */
- MDBX_FIRST,
-
- /** \ref MDBX_DUPSORT -only: Position at first data item of current key. */
- MDBX_FIRST_DUP,
-
- /** \ref MDBX_DUPSORT -only: Position at key/data pair. */
- MDBX_GET_BOTH,
-
- /** \ref MDBX_DUPSORT -only: Position at given key and at first data greater
- * than or equal to specified data. */
- MDBX_GET_BOTH_RANGE,
-
- /** Return key/data at current cursor position */
- MDBX_GET_CURRENT,
-
- /** \ref MDBX_DUPFIXED -only: Return up to a page of duplicate data items
- * from current cursor position. Move cursor to prepare
- * for \ref MDBX_NEXT_MULTIPLE. */
- MDBX_GET_MULTIPLE,
-
- /** Position at last key/data item */
- MDBX_LAST,
-
- /** \ref MDBX_DUPSORT -only: Position at last data item of current key. */
- MDBX_LAST_DUP,
-
- /** Position at next data item */
- MDBX_NEXT,
-
- /** \ref MDBX_DUPSORT -only: Position at next data item of current key. */
- MDBX_NEXT_DUP,
-
- /** \ref MDBX_DUPFIXED -only: Return up to a page of duplicate data items
- * from next cursor position. Move cursor to prepare
- * for `MDBX_NEXT_MULTIPLE`. */
- MDBX_NEXT_MULTIPLE,
-
- /** Position at first data item of next key */
- MDBX_NEXT_NODUP,
-
- /** Position at previous data item */
- MDBX_PREV,
-
- /** \ref MDBX_DUPSORT -only: Position at previous data item of current key. */
- MDBX_PREV_DUP,
-
- /** Position at last data item of previous key */
- MDBX_PREV_NODUP,
-
- /** Position at specified key */
- MDBX_SET,
-
- /** Position at specified key, return both key and data */
- MDBX_SET_KEY,
-
- /** Position at first key greater than or equal to specified key. */
- MDBX_SET_RANGE,
-
- /** \ref MDBX_DUPFIXED -only: Position at previous page and return up to
- * a page of duplicate data items. */
- MDBX_PREV_MULTIPLE,
-
- /** Positions cursor at first key-value pair greater than or equal to
- * specified, return both key and data, and the return code depends on whether
- * a exact match.
- *
- * For non DUPSORT-ed collections this work the same to \ref MDBX_SET_RANGE,
- * but returns \ref MDBX_SUCCESS if key found exactly or
- * \ref MDBX_RESULT_TRUE if greater key was found.
- *
- * For DUPSORT-ed a data value is taken into account for duplicates,
- * i.e. for a pairs/tuples of a key and an each data value of duplicates.
- * Returns \ref MDBX_SUCCESS if key-value pair found exactly or
- * \ref MDBX_RESULT_TRUE if the next pair was returned. */
- MDBX_SET_LOWERBOUND,
-
- /** Positions cursor at first key-value pair greater than specified,
- * return both key and data, and the return code depends on whether a
- * upper-bound was found.
- *
- * For non DUPSORT-ed collections this work the same to \ref MDBX_SET_RANGE,
- * but returns \ref MDBX_SUCCESS if the greater key was found or
- * \ref MDBX_NOTFOUND otherwise.
- *
- * For DUPSORT-ed a data value is taken into account for duplicates,
- * i.e. for a pairs/tuples of a key and an each data value of duplicates.
- * Returns \ref MDBX_SUCCESS if the greater pair was returned or
- * \ref MDBX_NOTFOUND otherwise. */
- MDBX_SET_UPPERBOUND
-};
-#ifndef __cplusplus
-/** \ingroup c_cursors */
-typedef enum MDBX_cursor_op MDBX_cursor_op;
-#endif
-
-/** \brief Errors and return codes
- * \ingroup c_err
- *
- * BerkeleyDB uses -30800 to -30999, we'll go under them
- * \see mdbx_strerror() \see mdbx_strerror_r() \see mdbx_liberr2str() */
-enum MDBX_error_t {
- /** Successful result */
- MDBX_SUCCESS = 0,
-
- /** Alias for \ref MDBX_SUCCESS */
- MDBX_RESULT_FALSE = MDBX_SUCCESS,
-
- /** Successful result with special meaning or a flag */
- MDBX_RESULT_TRUE = -1,
-
- /** key/data pair already exists */
- MDBX_KEYEXIST = -30799,
-
- /** The first LMDB-compatible defined error code */
- MDBX_FIRST_LMDB_ERRCODE = MDBX_KEYEXIST,
-
- /** key/data pair not found (EOF) */
- MDBX_NOTFOUND = -30798,
-
- /** Requested page not found - this usually indicates corruption */
- MDBX_PAGE_NOTFOUND = -30797,
-
- /** Database is corrupted (page was wrong type and so on) */
- MDBX_CORRUPTED = -30796,
-
- /** Environment had fatal error,
- * i.e. update of meta page failed and so on. */
- MDBX_PANIC = -30795,
-
- /** DB file version mismatch with libmdbx */
- MDBX_VERSION_MISMATCH = -30794,
-
- /** File is not a valid MDBX file */
- MDBX_INVALID = -30793,
-
- /** Environment mapsize reached */
- MDBX_MAP_FULL = -30792,
-
- /** Environment maxdbs reached */
- MDBX_DBS_FULL = -30791,
-
- /** Environment maxreaders reached */
- MDBX_READERS_FULL = -30790,
-
- /** Transaction has too many dirty pages, i.e transaction too big */
- MDBX_TXN_FULL = -30788,
-
- /** Cursor stack too deep - this usually indicates corruption,
- * i.e branch-pages loop */
- MDBX_CURSOR_FULL = -30787,
-
- /** Page has not enough space - internal error */
- MDBX_PAGE_FULL = -30786,
-
- /** Database engine was unable to extend mapping, e.g. since address space
- * is unavailable or busy. This can mean:
- * - Database size extended by other process beyond to environment mapsize
- * and engine was unable to extend mapping while starting read
- * transaction. Environment should be reopened to continue.
- * - Engine was unable to extend mapping during write transaction
- * or explicit call of \ref mdbx_env_set_geometry(). */
- MDBX_UNABLE_EXTEND_MAPSIZE = -30785,
-
- /** Environment or database is not compatible with the requested operation
- * or the specified flags. This can mean:
- * - The operation expects an \ref MDBX_DUPSORT / \ref MDBX_DUPFIXED
- * database.
- * - Opening a named DB when the unnamed DB has \ref MDBX_DUPSORT /
- * \ref MDBX_INTEGERKEY.
- * - Accessing a data record as a database, or vice versa.
- * - The database was dropped and recreated with different flags. */
- MDBX_INCOMPATIBLE = -30784,
-
- /** Invalid reuse of reader locktable slot,
- * e.g. read-transaction already run for current thread */
- MDBX_BAD_RSLOT = -30783,
-
- /** Transaction is not valid for requested operation,
- * e.g. had errored and be must aborted, has a child, or is invalid */
- MDBX_BAD_TXN = -30782,
-
- /** Invalid size or alignment of key or data for target database,
- * either invalid subDB name */
- MDBX_BAD_VALSIZE = -30781,
-
- /** The specified DBI-handle is invalid
- * or changed by another thread/transaction */
- MDBX_BAD_DBI = -30780,
-
- /** Unexpected internal error, transaction should be aborted */
- MDBX_PROBLEM = -30779,
-
- /** The last LMDB-compatible defined error code */
- MDBX_LAST_LMDB_ERRCODE = MDBX_PROBLEM,
-
- /** Another write transaction is running or environment is already used while
- * opening with \ref MDBX_EXCLUSIVE flag */
- MDBX_BUSY = -30778,
-
- /** The first of MDBX-added error codes */
- MDBX_FIRST_ADDED_ERRCODE = MDBX_BUSY,
-
- /** The specified key has more than one associated value */
- MDBX_EMULTIVAL = -30421,
-
- /** Bad signature of a runtime object(s), this can mean:
- * - memory corruption or double-free;
- * - ABI version mismatch (rare case); */
- MDBX_EBADSIGN = -30420,
-
- /** Database should be recovered, but this could NOT be done for now
- * since it opened in read-only mode */
- MDBX_WANNA_RECOVERY = -30419,
-
- /** The given key value is mismatched to the current cursor position */
- MDBX_EKEYMISMATCH = -30418,
-
- /** Database is too large for current system,
- * e.g. could NOT be mapped into RAM. */
- MDBX_TOO_LARGE = -30417,
-
- /** A thread has attempted to use a not owned object,
- * e.g. a transaction that started by another thread. */
- MDBX_THREAD_MISMATCH = -30416,
-
- /** Overlapping read and write transactions for the current thread */
- MDBX_TXN_OVERLAPPING = -30415,
-
- /* The last of MDBX-added error codes */
- MDBX_LAST_ADDED_ERRCODE = MDBX_TXN_OVERLAPPING,
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_ENODATA = ERROR_HANDLE_EOF,
- MDBX_EINVAL = ERROR_INVALID_PARAMETER,
- MDBX_EACCESS = ERROR_ACCESS_DENIED,
- MDBX_ENOMEM = ERROR_OUTOFMEMORY,
- MDBX_EROFS = ERROR_FILE_READ_ONLY,
- MDBX_ENOSYS = ERROR_NOT_SUPPORTED,
- MDBX_EIO = ERROR_WRITE_FAULT,
- MDBX_EPERM = ERROR_INVALID_FUNCTION,
- MDBX_EINTR = ERROR_CANCELLED,
- MDBX_ENOFILE = ERROR_FILE_NOT_FOUND,
- MDBX_EREMOTE = ERROR_REMOTE_STORAGE_MEDIA_ERROR
-#else /* Windows */
-#ifdef ENODATA
- MDBX_ENODATA = ENODATA,
-#else
- MDBX_ENODATA = 9919 /* for compatibility with LLVM's C++ libraries/headers */,
-#endif /* ENODATA */
- MDBX_EINVAL = EINVAL,
- MDBX_EACCESS = EACCES,
- MDBX_ENOMEM = ENOMEM,
- MDBX_EROFS = EROFS,
- MDBX_ENOSYS = ENOSYS,
- MDBX_EIO = EIO,
- MDBX_EPERM = EPERM,
- MDBX_EINTR = EINTR,
- MDBX_ENOFILE = ENOENT,
- MDBX_EREMOTE = ENOTBLK
-#endif /* !Windows */
-};
-#ifndef __cplusplus
-/** \ingroup c_err */
-typedef enum MDBX_error_t MDBX_error_t;
-#endif
-
-/** MDBX_MAP_RESIZED
- * \ingroup c_err
- * \deprecated Please review your code to use MDBX_UNABLE_EXTEND_MAPSIZE
- * instead. */
-MDBX_DEPRECATED static __inline int MDBX_MAP_RESIZED_is_deprecated() {
- return MDBX_UNABLE_EXTEND_MAPSIZE;
-}
-#define MDBX_MAP_RESIZED MDBX_MAP_RESIZED_is_deprecated()
-
-/** \brief Return a string describing a given error code.
- * \ingroup c_err
- *
- * This function is a superset of the ANSI C X3.159-1989 (ANSI C) `strerror()`
- * function. If the error code is greater than or equal to 0, then the string
- * returned by the system function `strerror()` is returned. If the error code
- * is less than 0, an error string corresponding to the MDBX library error is
- * returned. See errors for a list of MDBX-specific error codes.
- *
- * `mdbx_strerror()` is NOT thread-safe because may share common internal buffer
- * for system messages. The returned string must NOT be modified by the
- * application, but MAY be modified by a subsequent call to
- * \ref mdbx_strerror(), `strerror()` and other related functions.
- * \see mdbx_strerror_r()
- *
- * \param [in] errnum The error code.
- *
- * \returns "error message" The description of the error. */
-LIBMDBX_API const char *mdbx_strerror(int errnum);
-
-/** \brief Return a string describing a given error code.
- * \ingroup c_err
- *
- * This function is a superset of the ANSI C X3.159-1989 (ANSI C) `strerror()`
- * function. If the error code is greater than or equal to 0, then the string
- * returned by the system function `strerror()` is returned. If the error code
- * is less than 0, an error string corresponding to the MDBX library error is
- * returned. See errors for a list of MDBX-specific error codes.
- *
- * `mdbx_strerror_r()` is thread-safe since uses user-supplied buffer where
- * appropriate. The returned string must NOT be modified by the application,
- * since it may be pointer to internal constant string. However, there is no
- * restriction if the returned string points to the supplied buffer.
- * \see mdbx_strerror()
- *
- * mdbx_liberr2str() returns string describing only MDBX error numbers but NULL
- * for non-MDBX error codes. This function is thread-safe since return pointer
- * to constant non-localized strings.
- *
- * \param [in] errnum The error code.
- * \param [in,out] buf Buffer to store the error message.
- * \param [in] buflen The size of buffer to store the message.
- *
- * \returns "error message" The description of the error. */
-LIBMDBX_API const char *mdbx_strerror_r(int errnum, char *buf, size_t buflen);
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API const char *mdbx_liberr2str(int errnum);
-
-#if defined(_WIN32) || defined(_WIN64) || defined(DOXYGEN)
-/** Bit of Windows' madness. The similar to \ref mdbx_strerror() but returns
- * Windows error-messages in the OEM-encoding for console utilities.
- * \ingroup c_err
- * \see mdbx_strerror_r_ANSI2OEM() */
-LIBMDBX_API const char *mdbx_strerror_ANSI2OEM(int errnum);
-
-/** Bit of Windows' madness. The similar to \ref mdbx_strerror_r() but returns
- * Windows error-messages in the OEM-encoding for console utilities.
- * \ingroup c_err
- * \see mdbx_strerror_ANSI2OEM() */
-LIBMDBX_API const char *mdbx_strerror_r_ANSI2OEM(int errnum, char *buf,
- size_t buflen);
-#endif /* Bit of Windows' madness */
-
-/** \brief Create an MDBX environment instance.
- * \ingroup c_opening
- *
- * This function allocates memory for a \ref MDBX_env structure. To release
- * the allocated memory and discard the handle, call \ref mdbx_env_close().
- * Before the handle may be used, it must be opened using \ref mdbx_env_open().
- *
- * Various other options may also need to be set before opening the handle,
- * e.g. \ref mdbx_env_set_geometry(), \ref mdbx_env_set_maxreaders(),
- * \ref mdbx_env_set_maxdbs(), depending on usage requirements.
- *
- * \param [out] penv The address where the new handle will be stored.
- *
- * \returns a non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_create(MDBX_env **penv);
-
-/** \brief MDBX environment options. */
-enum MDBX_option_t {
- /** \brief Controls the maximum number of named databases for the environment.
- *
- * \details By default only unnamed key-value database could used and
- * appropriate value should set by `MDBX_opt_max_db` to using any more named
- * subDB(s). To reduce overhead, use the minimum sufficient value. This option
- * may only set after \ref mdbx_env_create() and before \ref mdbx_env_open().
- *
- * \see mdbx_env_set_maxdbs() \see mdbx_env_get_maxdbs() */
- MDBX_opt_max_db,
-
- /** \brief Defines the maximum number of threads/reader slots
- * for all processes interacting with the database.
- *
- * \details This defines the number of slots in the lock table that is used to
- * track readers in the the environment. The default is about 100 for 4K
- * system page size. Starting a read-only transaction normally ties a lock
- * table slot to the current thread until the environment closes or the thread
- * exits. If \ref MDBX_NOTLS is in use, \ref mdbx_txn_begin() instead ties the
- * slot to the \ref MDBX_txn object until it or the \ref MDBX_env object is
- * destroyed. This option may only set after \ref mdbx_env_create() and before
- * \ref mdbx_env_open(), and has an effect only when the database is opened by
- * the first process interacts with the database.
- *
- * \see mdbx_env_set_maxreaders() \see mdbx_env_get_maxreaders() */
- MDBX_opt_max_readers,
-
- /** \brief Controls interprocess/shared threshold to force flush the data
- * buffers to disk, if \ref MDBX_SAFE_NOSYNC is used.
- *
- * \see mdbx_env_set_syncbytes() \see mdbx_env_get_syncbytes() */
- MDBX_opt_sync_bytes,
-
- /** \brief Controls interprocess/shared relative period since the last
- * unsteady commit to force flush the data buffers to disk,
- * if \ref MDBX_SAFE_NOSYNC is used.
- * \see mdbx_env_set_syncperiod() \see mdbx_env_get_syncperiod() */
- MDBX_opt_sync_period,
-
- /** \brief Controls the in-process limit to grow a list of reclaimed/recycled
- * page's numbers for finding a sequence of contiguous pages for large data
- * items.
- *
- * \details A long values requires allocation of contiguous database pages.
- * To find such sequences, it may be necessary to accumulate very large lists,
- * especially when placing very long values (more than a megabyte) in a large
- * databases (several tens of gigabytes), which is much expensive in extreme
- * cases. This threshold allows you to avoid such costs by allocating new
- * pages at the end of the database (with its possible growth on disk),
- * instead of further accumulating/reclaiming Garbage Collection records.
- *
- * On the other hand, too small threshold will lead to unreasonable database
- * growth, or/and to the inability of put long values.
- *
- * The `MDBX_opt_rp_augment_limit` controls described limit for the current
- * process. Default is 262144, it is usually enough for most cases. */
- MDBX_opt_rp_augment_limit,
-
- /** \brief Controls the in-process limit to grow a cache of dirty
- * pages for reuse in the current transaction.
- *
- * \details A 'dirty page' refers to a page that has been updated in memory
- * only, the changes to a dirty page are not yet stored on disk.
- * To reduce overhead, it is reasonable to release not all such pages
- * immediately, but to leave some ones in cache for reuse in the current
- * transaction.
- *
- * The `MDBX_opt_loose_limit` allows you to set a limit for such cache inside
- * the current process. Should be in the range 0..255, default is 64. */
- MDBX_opt_loose_limit,
-
- /** \brief Controls the in-process limit of a pre-allocated memory items
- * for dirty pages.
- *
- * \details A 'dirty page' refers to a page that has been updated in memory
- * only, the changes to a dirty page are not yet stored on disk.
- * Without \ref MDBX_WRITEMAP dirty pages are allocated from memory and
- * released when a transaction is committed. To reduce overhead, it is
- * reasonable to release not all ones, but to leave some allocations in
- * reserve for reuse in the next transaction(s).
- *
- * The `MDBX_opt_dp_reserve_limit` allows you to set a limit for such reserve
- * inside the current process. Default is 1024. */
- MDBX_opt_dp_reserve_limit,
-
- /** \brief Controls the in-process limit of dirty pages
- * for a write transaction.
- *
- * \details A 'dirty page' refers to a page that has been updated in memory
- * only, the changes to a dirty page are not yet stored on disk.
- * Without \ref MDBX_WRITEMAP dirty pages are allocated from memory and will
- * be busy until are written to disk. Therefore for a large transactions is
- * reasonable to limit dirty pages collecting above an some threshold but
- * spill to disk instead.
- *
- * The `MDBX_opt_txn_dp_limit` controls described threshold for the current
- * process. Default is 65536, it is usually enough for most cases. */
- MDBX_opt_txn_dp_limit,
-
- /** \brief Controls the in-process initial allocation size for dirty pages
- * list of a write transaction. Default is 1024. */
- MDBX_opt_txn_dp_initial,
-
- /** \brief Controls the in-process how maximal part of the dirty pages may be
- * spilled when necessary.
- *
- * \details The `MDBX_opt_spill_max_denominator` defines the denominator for
- * limiting from the top for part of the current dirty pages may be spilled
- * when the free room for a new dirty pages (i.e. distance to the
- * `MDBX_opt_txn_dp_limit` threshold) is not enough to perform requested
- * operation.
- * Exactly `max_pages_to_spill = dirty_pages - dirty_pages / N`,
- * where `N` is the value set by `MDBX_opt_spill_max_denominator`.
- *
- * Should be in the range 0..255, where zero means no limit, i.e. all dirty
- * pages could be spilled. Default is 8, i.e. no more than 7/8 of the current
- * dirty pages may be spilled when reached the condition described above. */
- MDBX_opt_spill_max_denominator,
-
- /** \brief Controls the in-process how minimal part of the dirty pages should
- * be spilled when necessary.
- *
- * \details The `MDBX_opt_spill_min_denominator` defines the denominator for
- * limiting from the bottom for part of the current dirty pages should be
- * spilled when the free room for a new dirty pages (i.e. distance to the
- * `MDBX_opt_txn_dp_limit` threshold) is not enough to perform requested
- * operation.
- * Exactly `min_pages_to_spill = dirty_pages / N`,
- * where `N` is the value set by `MDBX_opt_spill_min_denominator`.
- *
- * Should be in the range 0..255, where zero means no restriction at the
- * bottom. Default is 8, i.e. at least the 1/8 of the current dirty pages
- * should be spilled when reached the condition described above. */
- MDBX_opt_spill_min_denominator,
-
- /** \brief Controls the in-process how much of the parent transaction dirty
- * pages will be spilled while start each child transaction.
- *
- * \details The `MDBX_opt_spill_parent4child_denominator` defines the
- * denominator to determine how much of parent transaction dirty pages will be
- * spilled explicitly while start each child transaction.
- * Exactly `pages_to_spill = dirty_pages / N`,
- * where `N` is the value set by `MDBX_opt_spill_parent4child_denominator`.
- *
- * For a stack of nested transactions each dirty page could be spilled only
- * once, and parent's dirty pages couldn't be spilled while child
- * transaction(s) are running. Therefore a child transaction could reach
- * \ref MDBX_TXN_FULL when parent(s) transaction has spilled too less (and
- * child reach the limit of dirty pages), either when parent(s) has spilled
- * too more (since child can't spill already spilled pages). So there is no
- * universal golden ratio.
- *
- * Should be in the range 0..255, where zero means no explicit spilling will
- * be performed during starting nested transactions.
- * Default is 0, i.e. by default no spilling performed during starting nested
- * transactions, that correspond historically behaviour. */
- MDBX_opt_spill_parent4child_denominator,
-
- /** \brief Controls the in-process threshold of semi-empty pages merge.
- * \warning This is experimental option and subject for change or removal.
- * \details This option controls the in-process threshold of minimum page
- * fill, as used space of percentage of a page. Neighbour pages emptier than
- * this value are candidates for merging. The threshold value is specified
- * in 1/65536 of percent, which is equivalent to the 16-dot-16 fixed point
- * format. The specified value must be in the range from 12.5% (almost empty)
- * to 50% (half empty) which corresponds to the range from 8192 and to 32768
- * in units respectively. */
- MDBX_opt_merge_threshold_16dot16_percent,
-};
-#ifndef __cplusplus
-/** \ingroup c_settings */
-typedef enum MDBX_option_t MDBX_option_t;
-#endif
-
-/** \brief Sets the value of a runtime options for an environment.
- * \ingroup c_settings
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] option The option from \ref MDBX_option_t to set value of it.
- * \param [in] value The value of option to be set.
- *
- * \see MDBX_option_t
- * \see mdbx_env_get_option()
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_set_option(MDBX_env *env, const MDBX_option_t option,
- uint64_t value);
-
-/** \brief Gets the value of runtime options from an environment.
- * \ingroup c_settings
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] option The option from \ref MDBX_option_t to get value of it.
- * \param [out] pvalue The address where the option's value will be stored.
- *
- * \see MDBX_option_t
- * \see mdbx_env_get_option()
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_get_option(const MDBX_env *env,
- const MDBX_option_t option,
- uint64_t *pvalue);
-
-/** \brief Open an environment instance.
- * \ingroup c_opening
- *
- * Indifferently this function will fails or not, the \ref mdbx_env_close() must
- * be called later to discard the \ref MDBX_env handle and release associated
- * resources.
- *
- * \param [in] env An environment handle returned
- * by \ref mdbx_env_create()
- *
- * \param [in] pathname The pathname for the database or the directory in which
- * the database files reside. In the case of directory it
- * must already exist and be writable.
- *
- * \param [in] flags Specifies options for this environment.
- * This parameter must be bitwise OR'ing together
- * any constants described above in the \ref env_flags
- * and \ref sync_modes sections.
- *
- * Flags set by mdbx_env_set_flags() are also used:
- * - \ref MDBX_ENV_DEFAULTS, \ref MDBX_NOSUBDIR, \ref MDBX_RDONLY,
- * \ref MDBX_EXCLUSIVE, \ref MDBX_WRITEMAP, \ref MDBX_NOTLS,
- * \ref MDBX_NORDAHEAD, \ref MDBX_NOMEMINIT, \ref MDBX_COALESCE,
- * \ref MDBX_LIFORECLAIM. See \ref env_flags section.
- *
- * - \ref MDBX_SYNC_DURABLE, \ref MDBX_NOMETASYNC, \ref MDBX_SAFE_NOSYNC,
- * \ref MDBX_UTTERLY_NOSYNC. See \ref sync_modes section.
- *
- * \note `MDB_NOLOCK` flag don't supported by MDBX,
- * try use \ref MDBX_EXCLUSIVE as a replacement.
- *
- * \note MDBX don't allow to mix processes with different \ref MDBX_SAFE_NOSYNC
- * flags on the same environment.
- * In such case \ref MDBX_INCOMPATIBLE will be returned.
- *
- * If the database is already exist and parameters specified early by
- * \ref mdbx_env_set_geometry() are incompatible (i.e. for instance, different
- * page size) then \ref mdbx_env_open() will return \ref MDBX_INCOMPATIBLE
- * error.
- *
- * \param [in] mode The UNIX permissions to set on created files.
- * Zero value means to open existing, but do not create.
- *
- * \return A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_VERSION_MISMATCH The version of the MDBX library doesn't match
- * the version that created the database environment.
- * \retval MDBX_INVALID The environment file headers are corrupted.
- * \retval MDBX_ENOENT The directory specified by the path parameter
- * doesn't exist.
- * \retval MDBX_EACCES The user didn't have permission to access
- * the environment files.
- * \retval MDBX_EAGAIN The environment was locked by another process.
- * \retval MDBX_BUSY The \ref MDBX_EXCLUSIVE flag was specified and the
- * environment is in use by another process,
- * or the current process tries to open environment
- * more than once.
- * \retval MDBX_INCOMPATIBLE Environment is already opened by another process,
- * but with different set of \ref MDBX_SAFE_NOSYNC,
- * \ref MDBX_UTTERLY_NOSYNC flags.
- * Or if the database is already exist and parameters
- * specified early by \ref mdbx_env_set_geometry()
- * are incompatible (i.e. different pagesize, etc).
- *
- * \retval MDBX_WANNA_RECOVERY The \ref MDBX_RDONLY flag was specified but
- * read-write access is required to rollback
- * inconsistent state after a system crash.
- *
- * \retval MDBX_TOO_LARGE Database is too large for this process,
- * i.e. 32-bit process tries to open >4Gb database.
- */
-LIBMDBX_API int mdbx_env_open(MDBX_env *env, const char *pathname,
- MDBX_env_flags_t flags, mdbx_mode_t mode);
-
-/** \brief Deletion modes for \ref mdbx_env_delete().
- * \ingroup c_extra
- * \see mdbx_env_delete() */
-enum MDBX_env_delete_mode_t {
- /** \brief Just delete the environment's files and directory if any.
- * \note On POSIX systems, processes already working with the database will
- * continue to work without interference until it close the environment.
- * \note On Windows, the behavior of `MDB_ENV_JUST_DELETE` is different
- * because the system does not support deleting files that are currently
- * memory mapped. */
- MDBX_ENV_JUST_DELETE = 0,
- /** \brief Make sure that the environment is not being used by other
- * processes, or return an error otherwise. */
- MDBX_ENV_ENSURE_UNUSED = 1,
- /** \brief Wait until other processes closes the environment before deletion.
- */
- MDBX_ENV_WAIT_FOR_UNUSED = 2,
-};
-#ifndef __cplusplus
-/** \ingroup c_extra */
-typedef enum MDBX_env_delete_mode_t MDBX_env_delete_mode_t;
-#endif
-
-/** \brief Delete the environment's files in a proper and multiprocess-safe way.
- * \ingroup c_extra
- *
- * \param [in] pathname The pathname for the database or the directory in which
- * the database files reside.
- *
- * \param [in] mode Specifies deletion mode for the environment. This
- * parameter must be set to one of the constants described
- * above in the \ref MDBX_env_delete_mode_t section.
- *
- * \note The \ref MDBX_ENV_JUST_DELETE don't supported on Windows since system
- * unable to delete a memory-mapped files.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_RESULT_TRUE No corresponding files or directories were found,
- * so no deletion was performed. */
-LIBMDBX_API int mdbx_env_delete(const char *pathname,
- MDBX_env_delete_mode_t mode);
-
-/** \brief Copy an MDBX environment to the specified path, with options.
- * \ingroup c_extra
- *
- * This function may be used to make a backup of an existing environment.
- * No lockfile is created, since it gets recreated at need.
- * \note This call can trigger significant file size growth if run in
- * parallel with write transactions, because it employs a read-only
- * transaction. See long-lived transactions under \ref restrictions section.
- *
- * \param [in] env An environment handle returned by mdbx_env_create().
- * It must have already been opened successfully.
- * \param [in] dest The pathname of a file in which the copy will reside.
- * This file must not be already exist, but parent directory
- * must be writable.
- * \param [in] flags Specifies options for this operation. This parameter
- * must be bitwise OR'ing together any of the constants
- * described here:
- *
- * - \ref MDBX_CP_DEFAULTS
- * Perform copy as-is without compaction, etc.
- *
- * - \ref MDBX_CP_COMPACT
- * Perform compaction while copying: omit free pages and sequentially
- * renumber all pages in output. This option consumes little bit more
- * CPU for processing, but may running quickly than the default, on
- * account skipping free pages.
- *
- * - \ref MDBX_CP_FORCE_DYNAMIC_SIZE
- * Force to make resizeable copy, i.e. dynamic size instead of fixed.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_copy(MDBX_env *env, const char *dest,
- MDBX_copy_flags_t flags);
-
-/** \brief Copy an environment to the specified file descriptor, with
- * options.
- * \ingroup c_extra
- *
- * This function may be used to make a backup of an existing environment.
- * No lockfile is created, since it gets recreated at need.
- * \see mdbx_env_copy()
- *
- * \note This call can trigger significant file size growth if run in
- * parallel with write transactions, because it employs a read-only
- * transaction. See long-lived transactions under \ref restrictions
- * section.
- *
- * \note Fails if the environment has suffered a page leak and the destination
- * file descriptor is associated with a pipe, socket, or FIFO.
- *
- * \param [in] env An environment handle returned by mdbx_env_create().
- * It must have already been opened successfully.
- * \param [in] fd The file descriptor to write the copy to. It must have
- * already been opened for Write access.
- * \param [in] flags Special options for this operation. \see mdbx_env_copy()
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_copy2fd(MDBX_env *env, mdbx_filehandle_t fd,
- MDBX_copy_flags_t flags);
-
-/** \brief Statistics for a database in the environment
- * \ingroup c_statinfo
- * \see mdbx_env_stat_ex() \see mdbx_dbi_stat() */
-struct MDBX_stat {
- uint32_t ms_psize; /**< Size of a database page. This is the same for all
- databases. */
- uint32_t ms_depth; /**< Depth (height) of the B-tree */
- uint64_t ms_branch_pages; /**< Number of internal (non-leaf) pages */
- uint64_t ms_leaf_pages; /**< Number of leaf pages */
- uint64_t ms_overflow_pages; /**< Number of overflow pages */
- uint64_t ms_entries; /**< Number of data items */
- uint64_t ms_mod_txnid; /**< Transaction ID of committed last modification */
-};
-#ifndef __cplusplus
-/** \ingroup c_statinfo */
-typedef struct MDBX_stat MDBX_stat;
-#endif
-
-/** \brief Return statistics about the MDBX environment.
- * \ingroup c_statinfo
- *
- * At least one of env or txn argument must be non-null. If txn is passed
- * non-null then stat will be filled accordingly to the given transaction.
- * Otherwise, if txn is null, then stat will be populated by a snapshot from
- * the last committed write transaction, and at next time, other information
- * can be returned.
- *
- * Legacy mdbx_env_stat() correspond to calling \ref mdbx_env_stat_ex() with the
- * null `txn` argument.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create()
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin()
- * \param [out] stat The address of an \ref MDBX_stat structure where
- * the statistics will be copied
- * \param [in] bytes The size of \ref MDBX_stat.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_stat_ex(const MDBX_env *env, const MDBX_txn *txn,
- MDBX_stat *stat, size_t bytes);
-
-/** \brief Return statistics about the MDBX environment.
- * \ingroup c_statinfo
- * \deprecated Please use mdbx_env_stat_ex() instead. */
-MDBX_DEPRECATED LIBMDBX_INLINE_API(int, mdbx_env_stat,
- (const MDBX_env *env, MDBX_stat *stat,
- size_t bytes)) {
- return mdbx_env_stat_ex(env, NULL, stat, bytes);
-}
-
-/** \brief Information about the environment
- * \ingroup c_statinfo
- * \see mdbx_env_info_ex() */
-struct MDBX_envinfo {
- struct {
- uint64_t lower; /**< Lower limit for datafile size */
- uint64_t upper; /**< Upper limit for datafile size */
- uint64_t current; /**< Current datafile size */
- uint64_t shrink; /**< Shrink threshold for datafile */
- uint64_t grow; /**< Growth step for datafile */
- } mi_geo;
- uint64_t mi_mapsize; /**< Size of the data memory map */
- uint64_t mi_last_pgno; /**< Number of the last used page */
- uint64_t mi_recent_txnid; /**< ID of the last committed transaction */
- uint64_t mi_latter_reader_txnid; /**< ID of the last reader transaction */
- uint64_t mi_self_latter_reader_txnid; /**< ID of the last reader transaction
- of caller process */
- uint64_t mi_meta0_txnid, mi_meta0_sign;
- uint64_t mi_meta1_txnid, mi_meta1_sign;
- uint64_t mi_meta2_txnid, mi_meta2_sign;
- uint32_t mi_maxreaders; /**< Total reader slots in the environment */
- uint32_t mi_numreaders; /**< Max reader slots used in the environment */
- uint32_t mi_dxb_pagesize; /**< Database pagesize */
- uint32_t mi_sys_pagesize; /**< System pagesize */
-
- /** \brief A mostly unique ID that is regenerated on each boot.
-
- As such it can be used to identify the local machine's current boot. MDBX
- uses such when open the database to determine whether rollback required to
- the last steady sync point or not. I.e. if current bootid is differ from the
- value within a database then the system was rebooted and all changes since
- last steady sync must be reverted for data integrity. Zeros mean that no
- relevant information is available from the system. */
- struct {
- struct {
- uint64_t x, y;
- } current, meta0, meta1, meta2;
- } mi_bootid;
-
- /** Bytes not explicitly synchronized to disk */
- uint64_t mi_unsync_volume;
- /** Current auto-sync threshold, see \ref mdbx_env_set_syncbytes(). */
- uint64_t mi_autosync_threshold;
- /** Time since the last steady sync in 1/65536 of second */
- uint32_t mi_since_sync_seconds16dot16;
- /** Current auto-sync period in 1/65536 of second,
- * see \ref mdbx_env_set_syncperiod(). */
- uint32_t mi_autosync_period_seconds16dot16;
- /** Time since the last readers check in 1/65536 of second,
- * see \ref mdbx_reader_check(). */
- uint32_t mi_since_reader_check_seconds16dot16;
- /** Current environment mode.
- * The same as \ref mdbx_env_get_flags() returns. */
- uint32_t mi_mode;
-
- /** Statistics of page operations.
- * \details Overall statistics of page operations of all (running, completed
- * and aborted) transactions in the current multi-process session (since the
- * first process opened the database after everyone had previously closed it).
- */
- struct {
- uint64_t newly; /**< Quantity of a new pages added */
- uint64_t cow; /**< Quantity of pages copied for update */
- uint64_t clone; /**< Quantity of parent's dirty pages clones
- for nested transactions */
- uint64_t split; /**< Page splits */
- uint64_t merge; /**< Page merges */
- uint64_t spill; /**< Quantity of spilled dirty pages */
- uint64_t unspill; /**< Quantity of unspilled/reloaded pages */
- uint64_t wops; /**< Number of explicit write operations (not a pages)
- to a disk */
- } mi_pgop_stat;
-};
-#ifndef __cplusplus
-/** \ingroup c_statinfo */
-typedef struct MDBX_envinfo MDBX_envinfo;
-#endif
-
-/** \brief Return information about the MDBX environment.
- * \ingroup c_statinfo
- *
- * At least one of env or txn argument must be non-null. If txn is passed
- * non-null then stat will be filled accordingly to the given transaction.
- * Otherwise, if txn is null, then stat will be populated by a snapshot from
- * the last committed write transaction, and at next time, other information
- * can be returned.
- *
- * Legacy \ref mdbx_env_info() correspond to calling \ref mdbx_env_info_ex()
- * with the null `txn` argument.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create()
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin()
- * \param [out] info The address of an \ref MDBX_envinfo structure
- * where the information will be copied
- * \param [in] bytes The size of \ref MDBX_envinfo.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_info_ex(const MDBX_env *env, const MDBX_txn *txn,
- MDBX_envinfo *info, size_t bytes);
-/** \brief Return information about the MDBX environment.
- * \ingroup c_statinfo
- * \deprecated Please use mdbx_env_info_ex() instead. */
-MDBX_DEPRECATED LIBMDBX_INLINE_API(int, mdbx_env_info,
- (const MDBX_env *env, MDBX_envinfo *info,
- size_t bytes)) {
- return mdbx_env_info_ex(env, NULL, info, bytes);
-}
-
-/** \brief Flush the environment data buffers to disk.
- * \ingroup c_extra
- *
- * Unless the environment was opened with no-sync flags (\ref MDBX_NOMETASYNC,
- * \ref MDBX_SAFE_NOSYNC and \ref MDBX_UTTERLY_NOSYNC), then
- * data is always written an flushed to disk when \ref mdbx_txn_commit() is
- * called. Otherwise \ref mdbx_env_sync() may be called to manually write and
- * flush unsynced data to disk.
- *
- * Besides, \ref mdbx_env_sync_ex() with argument `force=false` may be used to
- * provide polling mode for lazy/asynchronous sync in conjunction with
- * \ref mdbx_env_set_syncbytes() and/or \ref mdbx_env_set_syncperiod().
- *
- * \note This call is not valid if the environment was opened with MDBX_RDONLY.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create()
- * \param [in] force If non-zero, force a flush. Otherwise, If force is
- * zero, then will run in polling mode,
- * i.e. it will check the thresholds that were
- * set \ref mdbx_env_set_syncbytes()
- * and/or \ref mdbx_env_set_syncperiod() and perform flush
- * if at least one of the thresholds is reached.
- *
- * \param [in] nonblock Don't wait if write transaction
- * is running by other thread.
- *
- * \returns A non-zero error value on failure and \ref MDBX_RESULT_TRUE or 0 on
- * success. The \ref MDBX_RESULT_TRUE means no data pending for flush
- * to disk, and 0 otherwise. Some possible errors are:
- *
- * \retval MDBX_EACCES the environment is read-only.
- * \retval MDBX_BUSY the environment is used by other thread
- * and `nonblock=true`.
- * \retval MDBX_EINVAL an invalid parameter was specified.
- * \retval MDBX_EIO an error occurred during synchronization. */
-LIBMDBX_API int mdbx_env_sync_ex(MDBX_env *env, bool force, bool nonblock);
-
-/** \brief The shortcut to calling \ref mdbx_env_sync_ex() with
- * the `force=true` and `nonblock=false` arguments.
- * \ingroup c_extra */
-LIBMDBX_INLINE_API(int, mdbx_env_sync, (MDBX_env * env)) {
- return mdbx_env_sync_ex(env, true, false);
-}
-
-/** \brief The shortcut to calling \ref mdbx_env_sync_ex() with
- * the `force=false` and `nonblock=true` arguments.
- * \ingroup c_extra */
-LIBMDBX_INLINE_API(int, mdbx_env_sync_poll, (MDBX_env * env)) {
- return mdbx_env_sync_ex(env, false, true);
-}
-
-/** \brief Sets threshold to force flush the data buffers to disk, even any of
- * \ref MDBX_SAFE_NOSYNC flag in the environment.
- * \ingroup c_settings
- * \see mdbx_env_get_syncbytes \see MDBX_opt_sync_bytes
- *
- * The threshold value affects all processes which operates with given
- * environment until the last process close environment or a new value will be
- * settled.
- *
- * Data is always written to disk when \ref mdbx_txn_commit() is called, but
- * the operating system may keep it buffered. MDBX always flushes the OS buffers
- * upon commit as well, unless the environment was opened with
- * \ref MDBX_SAFE_NOSYNC, \ref MDBX_UTTERLY_NOSYNC
- * or in part \ref MDBX_NOMETASYNC.
- *
- * The default is 0, than mean no any threshold checked, and no additional
- * flush will be made.
- *
- * \param [in] env An environment handle returned by mdbx_env_create().
- * \param [in] threshold The size in bytes of summary changes when
- * a synchronous flush would be made.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_INLINE_API(int, mdbx_env_set_syncbytes,
- (MDBX_env * env, size_t threshold)) {
- return mdbx_env_set_option(env, MDBX_opt_sync_bytes, threshold);
-}
-
-/** \brief Get threshold to force flush the data buffers to disk, even any of
- * \ref MDBX_SAFE_NOSYNC flag in the environment.
- * \ingroup c_statinfo
- * \see mdbx_env_set_syncbytes() \see MDBX_opt_sync_bytes
- *
- * \param [in] env An environment handle returned
- * by \ref mdbx_env_create().
- * \param [out] threshold Address of an size_t to store
- * the number of bytes of summary changes when
- * a synchronous flush would be made.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_INLINE_API(int, mdbx_env_get_syncbytes,
- (const MDBX_env *env, size_t *threshold)) {
- int rc = MDBX_EINVAL;
- if (threshold) {
- uint64_t proxy = 0;
- rc = mdbx_env_get_option(env, MDBX_opt_sync_bytes, &proxy);
-#ifdef assert
- assert(proxy <= SIZE_MAX);
-#endif /* assert */
- *threshold = (size_t)proxy;
- }
- return rc;
-}
-
-/** \brief Sets relative period since the last unsteady commit to force flush
- * the data buffers to disk, even of \ref MDBX_SAFE_NOSYNC flag in the
- * environment.
- * \ingroup c_settings
- * \see mdbx_env_get_syncperiod \see MDBX_opt_sync_period
- *
- * The relative period value affects all processes which operates with given
- * environment until the last process close environment or a new value will be
- * settled.
- *
- * Data is always written to disk when \ref mdbx_txn_commit() is called, but the
- * operating system may keep it buffered. MDBX always flushes the OS buffers
- * upon commit as well, unless the environment was opened with
- * \ref MDBX_SAFE_NOSYNC or in part \ref MDBX_NOMETASYNC.
- *
- * Settled period don't checked asynchronously, but only by the
- * \ref mdbx_txn_commit() and \ref mdbx_env_sync() functions. Therefore, in
- * cases where transactions are committed infrequently and/or irregularly,
- * polling by \ref mdbx_env_sync() may be a reasonable solution to timeout
- * enforcement.
- *
- * The default is 0, than mean no any timeout checked, and no additional
- * flush will be made.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] seconds_16dot16 The period in 1/65536 of second when
- * a synchronous flush would be made since
- * the last unsteady commit.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_INLINE_API(int, mdbx_env_set_syncperiod,
- (MDBX_env * env, unsigned seconds_16dot16)) {
- return mdbx_env_set_option(env, MDBX_opt_sync_period, seconds_16dot16);
-}
-
-/** \brief Get relative period since the last unsteady commit to force flush
- * the data buffers to disk, even of \ref MDBX_SAFE_NOSYNC flag in the
- * environment.
- * \ingroup c_statinfo
- * \see mdbx_env_set_syncperiod() \see MDBX_opt_sync_period
- *
- * \param [in] env An environment handle returned
- * by \ref mdbx_env_create().
- * \param [out] period_seconds_16dot16 Address of an size_t to store
- * the period in 1/65536 of second when
- * a synchronous flush would be made since
- * the last unsteady commit.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_INLINE_API(int, mdbx_env_get_syncperiod,
- (const MDBX_env *env, unsigned *period_seconds_16dot16)) {
- int rc = MDBX_EINVAL;
- if (period_seconds_16dot16) {
- uint64_t proxy = 0;
- rc = mdbx_env_get_option(env, MDBX_opt_sync_period, &proxy);
-#ifdef assert
- assert(proxy <= UINT32_MAX);
-#endif /* assert */
- *period_seconds_16dot16 = (unsigned)proxy;
- }
- return rc;
-}
-
-/** \brief Close the environment and release the memory map.
- * \ingroup c_opening
- *
- * Only a single thread may call this function. All transactions, databases,
- * and cursors must already be closed before calling this function. Attempts
- * to use any such handles after calling this function will cause a `SIGSEGV`.
- * The environment handle will be freed and must not be used again after this
- * call.
- *
- * \param [in] env An environment handle returned by
- * \ref mdbx_env_create().
- *
- * \param [in] dont_sync A dont'sync flag, if non-zero the last checkpoint
- * will be kept "as is" and may be still "weak" in the
- * \ref MDBX_SAFE_NOSYNC or \ref MDBX_UTTERLY_NOSYNC
- * modes. Such "weak" checkpoint will be ignored on
- * opening next time, and transactions since the last
- * non-weak checkpoint (meta-page update) will rolledback
- * for consistency guarantee.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_BUSY The write transaction is running by other thread,
- * in such case \ref MDBX_env instance has NOT be destroyed
- * not released!
- * \note If any OTHER error code was returned then
- * given MDBX_env instance has been destroyed and released.
- *
- * \retval MDBX_EBADSIGN Environment handle already closed or not valid,
- * i.e. \ref mdbx_env_close() was already called for the
- * `env` or was not created by \ref mdbx_env_create().
- *
- * \retval MDBX_PANIC If \ref mdbx_env_close_ex() was called in the child
- * process after `fork()`. In this case \ref MDBX_PANIC
- * is expected, i.e. \ref MDBX_env instance was freed in
- * proper manner.
- *
- * \retval MDBX_EIO An error occurred during synchronization. */
-LIBMDBX_API int mdbx_env_close_ex(MDBX_env *env, bool dont_sync);
-
-/** \brief The shortcut to calling \ref mdbx_env_close_ex() with
- * the `dont_sync=false` argument.
- * \ingroup c_opening */
-LIBMDBX_INLINE_API(int, mdbx_env_close, (MDBX_env * env)) {
- return mdbx_env_close_ex(env, false);
-}
-
-/** \brief Set environment flags.
- * \ingroup c_settings
- *
- * This may be used to set some flags in addition to those from
- * mdbx_env_open(), or to unset these flags.
- * \see mdbx_env_get_flags()
- *
- * \note In contrast to LMDB, the MDBX serialize threads via mutex while
- * changing the flags. Therefore this function will be blocked while a write
- * transaction running by other thread, or \ref MDBX_BUSY will be returned if
- * function called within a write transaction.
- *
- * \param [in] env An environment handle returned
- * by \ref mdbx_env_create().
- * \param [in] flags The \ref env_flags to change, bitwise OR'ed together.
- * \param [in] onoff A non-zero value sets the flags, zero clears them.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_env_set_flags(MDBX_env *env, MDBX_env_flags_t flags,
- bool onoff);
-
-/** \brief Get environment flags.
- * \ingroup c_statinfo
- * \see mdbx_env_set_flags()
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [out] flags The address of an integer to store the flags.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_env_get_flags(const MDBX_env *env, unsigned *flags);
-
-/** \brief Return the path that was used in mdbx_env_open().
- * \ingroup c_statinfo
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create()
- * \param [out] dest Address of a string pointer to contain the path.
- * This is the actual string in the environment, not a
- * copy. It should not be altered in any way.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_env_get_path(const MDBX_env *env, const char **dest);
-
-/** \brief Return the file descriptor for the given environment.
- * \ingroup c_statinfo
- *
- * \note All MDBX file descriptors have `FD_CLOEXEC` and
- * couldn't be used after exec() and or `fork()`.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [out] fd Address of a int to contain the descriptor.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_env_get_fd(const MDBX_env *env, mdbx_filehandle_t *fd);
-
-/** \brief Set all size-related parameters of environment, including page size
- * and the min/max size of the memory map.
- * \ingroup c_settings
- *
- * In contrast to LMDB, the MDBX provide automatic size management of an
- * database according the given parameters, including shrinking and resizing
- * on the fly. From user point of view all of these just working. Nevertheless,
- * it is reasonable to know some details in order to make optimal decisions
- * when choosing parameters.
- *
- * Both \ref mdbx_env_set_geometry() and legacy \ref mdbx_env_set_mapsize() are
- * inapplicable to read-only opened environment.
- *
- * Both \ref mdbx_env_set_geometry() and legacy \ref mdbx_env_set_mapsize()
- * could be called either before or after \ref mdbx_env_open(), either within
- * the write transaction running by current thread or not:
- *
- * - In case \ref mdbx_env_set_geometry() or legacy \ref mdbx_env_set_mapsize()
- * was called BEFORE \ref mdbx_env_open(), i.e. for closed environment, then
- * the specified parameters will be used for new database creation,
- * or will be applied during opening if database exists and no other process
- * using it.
- *
- * If the database is already exist, opened with \ref MDBX_EXCLUSIVE or not
- * used by any other process, and parameters specified by
- * \ref mdbx_env_set_geometry() are incompatible (i.e. for instance,
- * different page size) then \ref mdbx_env_open() will return
- * \ref MDBX_INCOMPATIBLE error.
- *
- * In another way, if database will opened read-only or will used by other
- * process during calling \ref mdbx_env_open() that specified parameters will
- * silently discarded (open the database with \ref MDBX_EXCLUSIVE flag
- * to avoid this).
- *
- * - In case \ref mdbx_env_set_geometry() or legacy \ref mdbx_env_set_mapsize()
- * was called after \ref mdbx_env_open() WITHIN the write transaction running
- * by current thread, then specified parameters will be applied as a part of
- * write transaction, i.e. will not be completely visible to any others
- * processes until the current write transaction has been committed by the
- * current process. However, if transaction will be aborted, then the
- * database file will be reverted to the previous size not immediately, but
- * when a next transaction will be committed or when the database will be
- * opened next time.
- *
- * - In case \ref mdbx_env_set_geometry() or legacy \ref mdbx_env_set_mapsize()
- * was called after \ref mdbx_env_open() but OUTSIDE a write transaction,
- * then MDBX will execute internal pseudo-transaction to apply new parameters
- * (but only if anything has been changed), and changes be visible to any
- * others processes immediately after succesful completion of function.
- *
- * Essentially a concept of "automatic size management" is simple and useful:
- * - There are the lower and upper bounds of the database file size;
- * - There is the growth step by which the database file will be increased,
- * in case of lack of space;
- * - There is the threshold for unused space, beyond which the database file
- * will be shrunk;
- * - The size of the memory map is also the maximum size of the database;
- * - MDBX will automatically manage both the size of the database and the size
- * of memory map, according to the given parameters.
- *
- * So, there some considerations about choosing these parameters:
- * - The lower bound allows you to prevent database shrinking below certain
- * reasonable size to avoid unnecessary resizing costs.
- * - The upper bound allows you to prevent database growth above certain
- * reasonable size. Besides, the upper bound defines the linear address space
- * reservation in each process that opens the database. Therefore changing
- * the upper bound is costly and may be required reopening environment in
- * case of \ref MDBX_UNABLE_EXTEND_MAPSIZE errors, and so on. Therefore, this
- * value should be chosen reasonable large, to accommodate future growth of
- * the database.
- * - The growth step must be greater than zero to allow the database to grow,
- * but also reasonable not too small, since increasing the size by little
- * steps will result a large overhead.
- * - The shrink threshold must be greater than zero to allow the database
- * to shrink but also reasonable not too small (to avoid extra overhead) and
- * not less than growth step to avoid up-and-down flouncing.
- * - The current size (i.e. `size_now` argument) is an auxiliary parameter for
- * simulation legacy \ref mdbx_env_set_mapsize() and as workaround Windows
- * issues (see below).
- *
- * Unfortunately, Windows has is a several issue
- * with resizing of memory-mapped file:
- * - Windows unable shrinking a memory-mapped file (i.e memory-mapped section)
- * in any way except unmapping file entirely and then map again. Moreover,
- * it is impossible in any way when a memory-mapped file is used more than
- * one process.
- * - Windows does not provide the usual API to augment a memory-mapped file
- * (i.e. a memory-mapped partition), but only by using "Native API"
- * in an undocumented way.
- *
- * MDBX bypasses all Windows issues, but at a cost:
- * - Ability to resize database on the fly requires an additional lock
- * and release `SlimReadWriteLock` during each read-only transaction.
- * - During resize all in-process threads should be paused and then resumed.
- * - Shrinking of database file is performed only when it used by single
- * process, i.e. when a database closes by the last process or opened
- * by the first.
- * = Therefore, the size_now argument may be useful to set database size
- * by the first process which open a database, and thus avoid expensive
- * remapping further.
- *
- * For create a new database with particular parameters, including the page
- * size, \ref mdbx_env_set_geometry() should be called after
- * \ref mdbx_env_create() and before \ref mdbx_env_open(). Once the database is
- * created, the page size cannot be changed. If you do not specify all or some
- * of the parameters, the corresponding default values will be used. For
- * instance, the default for database size is 10485760 bytes.
- *
- * If the mapsize is increased by another process, MDBX silently and
- * transparently adopt these changes at next transaction start. However,
- * \ref mdbx_txn_begin() will return \ref MDBX_UNABLE_EXTEND_MAPSIZE if new
- * mapping size could not be applied for current process (for instance if
- * address space is busy). Therefore, in the case of
- * \ref MDBX_UNABLE_EXTEND_MAPSIZE error you need close and reopen the
- * environment to resolve error.
- *
- * \note Actual values may be different than your have specified because of
- * rounding to specified database page size, the system page size and/or the
- * size of the system virtual memory management unit. You can get actual values
- * by \ref mdbx_env_sync_ex() or see by using the tool `mdbx_chk` with the `-v`
- * option.
- *
- * Legacy \ref mdbx_env_set_mapsize() correspond to calling
- * \ref mdbx_env_set_geometry() with the arguments `size_lower`, `size_now`,
- * `size_upper` equal to the `size` and `-1` (i.e. default) for all other
- * parameters.
- *
- * \param [in] env An environment handle returned
- * by \ref mdbx_env_create()
- *
- * \param [in] size_lower The lower bound of database size in bytes.
- * Zero value means "minimal acceptable",
- * and negative means "keep current or use default".
- *
- * \param [in] size_now The size in bytes to setup the database size for
- * now. Zero value means "minimal acceptable", and
- * negative means "keep current or use default". So,
- * it is recommended always pass -1 in this argument
- * except some special cases.
- *
- * \param [in] size_upper The upper bound of database size in bytes.
- * Zero value means "minimal acceptable",
- * and negative means "keep current or use default".
- * It is recommended to avoid change upper bound while
- * database is used by other processes or threaded
- * (i.e. just pass -1 in this argument except absolutely
- * necessary). Otherwise you must be ready for
- * \ref MDBX_UNABLE_EXTEND_MAPSIZE error(s), unexpected
- * pauses during remapping and/or system errors like
- * "address busy", and so on. In other words, there
- * is no way to handle a growth of the upper bound
- * robustly because there may be a lack of appropriate
- * system resources (which are extremely volatile in
- * a multi-process multi-threaded environment).
- *
- * \param [in] growth_step The growth step in bytes, must be greater than
- * zero to allow the database to grow. Negative value
- * means "keep current or use default".
- *
- * \param [in] shrink_threshold The shrink threshold in bytes, must be greater
- * than zero to allow the database to shrink and
- * greater than growth_step to avoid shrinking
- * right after grow.
- * Negative value means "keep current
- * or use default". Default is 2*growth_step.
- *
- * \param [in] pagesize The database page size for new database
- * creation or -1 otherwise. Once the database
- * is created, the page size cannot be changed.
- * Must be power of 2 in the range between
- * \ref MDBX_MIN_PAGESIZE and
- * \ref MDBX_MAX_PAGESIZE. Zero value means
- * "minimal acceptable", and negative means
- * "keep current or use default".
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified,
- * or the environment has an active write transaction.
- * \retval MDBX_EPERM Two specific cases for Windows:
- * 1) Shrinking was disabled before via geometry settings
- * and now it enabled, but there are reading threads that
- * don't use the additional `SRWL` (which is required to
- * avoid Windows issues).
- * 2) Temporary close memory mapped is required to change
- * geometry, but there read transaction(s) is running
- * and no corresponding thread(s) could be suspended
- * since the \ref MDBX_NOTLS mode is used.
- * \retval MDBX_EACCESS The environment opened in read-only.
- * \retval MDBX_MAP_FULL Specified size smaller than the space already
- * consumed by the environment.
- * \retval MDBX_TOO_LARGE Specified size is too large, i.e. too many pages for
- * given size, or a 32-bit process requests too much
- * bytes for the 32-bit address space. */
-LIBMDBX_API int mdbx_env_set_geometry(MDBX_env *env, intptr_t size_lower,
- intptr_t size_now, intptr_t size_upper,
- intptr_t growth_step,
- intptr_t shrink_threshold,
- intptr_t pagesize);
-
-/** \deprecated Please use \ref mdbx_env_set_geometry() instead.
- * \ingroup c_settings */
-MDBX_DEPRECATED LIBMDBX_INLINE_API(int, mdbx_env_set_mapsize,
- (MDBX_env * env, size_t size)) {
- return mdbx_env_set_geometry(env, size, size, size, -1, -1, -1);
-}
-
-/** \brief Find out whether to use readahead or not, based on the given database
- * size and the amount of available memory.
- * \ingroup c_extra
- *
- * \param [in] volume The expected database size in bytes.
- * \param [in] redundancy Additional reserve or overload in case of negative
- * value.
- *
- * \returns A \ref MDBX_RESULT_TRUE or \ref MDBX_RESULT_FALSE value,
- * otherwise the error code:
- * \retval MDBX_RESULT_TRUE Readahead is reasonable.
- * \retval MDBX_RESULT_FALSE Readahead is NOT reasonable,
- * i.e. \ref MDBX_NORDAHEAD is useful to
- * open environment by \ref mdbx_env_open().
- * \retval Otherwise the error code. */
-LIBMDBX_API int mdbx_is_readahead_reasonable(size_t volume,
- intptr_t redundancy);
-
-/** \brief Returns the minimal database page size in bytes.
- * \ingroup c_statinfo */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_INLINE_API(intptr_t, mdbx_limits_pgsize_min,
- (void)) {
- return MDBX_MIN_PAGESIZE;
-}
-
-/** \brief Returns the maximal database page size in bytes.
- * \ingroup c_statinfo */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_INLINE_API(intptr_t, mdbx_limits_pgsize_max,
- (void)) {
- return MDBX_MAX_PAGESIZE;
-}
-
-/** \brief Returns minimal database size in bytes for given page size,
- * or -1 if pagesize is invalid.
- * \ingroup c_statinfo */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API intptr_t
-mdbx_limits_dbsize_min(intptr_t pagesize);
-
-/** \brief Returns maximal database size in bytes for given page size,
- * or -1 if pagesize is invalid.
- * \ingroup c_statinfo */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API intptr_t
-mdbx_limits_dbsize_max(intptr_t pagesize);
-
-/** \brief Returns maximal key size in bytes for given page size
- * and database flags, or -1 if pagesize is invalid.
- * \ingroup c_statinfo
- * \see db_flags */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API intptr_t
-mdbx_limits_keysize_max(intptr_t pagesize, MDBX_db_flags_t flags);
-
-/** \brief Returns maximal data size in bytes for given page size
- * and database flags, or -1 if pagesize is invalid.
- * \ingroup c_statinfo
- * \see db_flags */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API intptr_t
-mdbx_limits_valsize_max(intptr_t pagesize, MDBX_db_flags_t flags);
-
-/** \brief Returns maximal write transaction size (i.e. limit for summary volume
- * of dirty pages) in bytes for given page size, or -1 if pagesize is invalid.
- * \ingroup c_statinfo */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API intptr_t
-mdbx_limits_txnsize_max(intptr_t pagesize);
-
-/** \brief Set the maximum number of threads/reader slots for for all processes
- * interacts with the database.
- * \ingroup c_settings
- *
- * \details This defines the number of slots in the lock table that is used to
- * track readers in the the environment. The default is about 100 for 4K system
- * page size. Starting a read-only transaction normally ties a lock table slot
- * to the current thread until the environment closes or the thread exits. If
- * \ref MDBX_NOTLS is in use, \ref mdbx_txn_begin() instead ties the slot to the
- * \ref MDBX_txn object until it or the \ref MDBX_env object is destroyed.
- * This function may only be called after \ref mdbx_env_create() and before
- * \ref mdbx_env_open(), and has an effect only when the database is opened by
- * the first process interacts with the database.
- * \see mdbx_env_get_maxreaders()
- *
- * \param [in] env An environment handle returned
- * by \ref mdbx_env_create().
- * \param [in] readers The maximum number of reader lock table slots.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified.
- * \retval MDBX_EPERM The environment is already open. */
-LIBMDBX_INLINE_API(int, mdbx_env_set_maxreaders,
- (MDBX_env * env, unsigned readers)) {
- return mdbx_env_set_option(env, MDBX_opt_max_readers, readers);
-}
-
-/** \brief Get the maximum number of threads/reader slots for the environment.
- * \ingroup c_statinfo
- * \see mdbx_env_set_maxreaders()
- *
- * \param [in] env An environment handle returned
- * by \ref mdbx_env_create().
- * \param [out] readers Address of an integer to store the number of readers.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_INLINE_API(int, mdbx_env_get_maxreaders,
- (const MDBX_env *env, unsigned *readers)) {
- int rc = MDBX_EINVAL;
- if (readers) {
- uint64_t proxy = 0;
- rc = mdbx_env_get_option(env, MDBX_opt_max_readers, &proxy);
- *readers = (unsigned)proxy;
- }
- return rc;
-}
-
-/** \brief Set the maximum number of named databases for the environment.
- * \ingroup c_settings
- *
- * This function is only needed if multiple databases will be used in the
- * environment. Simpler applications that use the environment as a single
- * unnamed database can ignore this option.
- * This function may only be called after \ref mdbx_env_create() and before
- * \ref mdbx_env_open().
- *
- * Currently a moderate number of slots are cheap but a huge number gets
- * expensive: 7-120 words per transaction, and every \ref mdbx_dbi_open()
- * does a linear search of the opened slots.
- * \see mdbx_env_get_maxdbs()
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] dbs The maximum number of databases.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified.
- * \retval MDBX_EPERM The environment is already open. */
-LIBMDBX_INLINE_API(int, mdbx_env_set_maxdbs, (MDBX_env * env, MDBX_dbi dbs)) {
- return mdbx_env_set_option(env, MDBX_opt_max_db, dbs);
-}
-
-/** \brief Get the maximum number of named databases for the environment.
- * \ingroup c_statinfo
- * \see mdbx_env_set_maxdbs()
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [out] dbs Address to store the maximum number of databases.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_INLINE_API(int, mdbx_env_get_maxdbs,
- (const MDBX_env *env, MDBX_dbi *dbs)) {
- int rc = MDBX_EINVAL;
- if (dbs) {
- uint64_t proxy = 0;
- rc = mdbx_env_get_option(env, MDBX_opt_max_db, &proxy);
- *dbs = (MDBX_dbi)proxy;
- }
- return rc;
-}
-
-/** \brief Returns the default size of database page for the current system.
- * \ingroup c_statinfo
- * \details Default size of database page depends on the size of the system
- * page and usually exactly match it. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API size_t mdbx_default_pagesize(void);
-
-/** \brief Returns basic information about system RAM.
- * This function provides a portable way to get information about available RAM
- * and can be useful in that it returns the same information that libmdbx uses
- * internally to adjust various options and control readahead.
- * \ingroup c_statinfo
- *
- * \param [out] page_size Optional address where the system page size
- * will be stored.
- * \param [out] total_pages Optional address where the number of total RAM
- * pages will be stored.
- * \param [out] avail_pages Optional address where the number of
- * available/free RAM pages will be stored.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_get_sysraminfo(intptr_t *page_size, intptr_t *total_pages,
- intptr_t *avail_pages);
-
-/** \brief Returns the maximum size of keys can put.
- * \ingroup c_statinfo
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] flags Database options (\ref MDBX_DUPSORT, \ref MDBX_INTEGERKEY
- * and so on). \see db_flags
- *
- * \returns The maximum size of a key can write,
- * or -1 if something is wrong. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int
-mdbx_env_get_maxkeysize_ex(const MDBX_env *env, MDBX_db_flags_t flags);
-
-/** \brief Returns the maximum size of data we can put.
- * \ingroup c_statinfo
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] flags Database options (\ref MDBX_DUPSORT, \ref MDBX_INTEGERKEY
- * and so on). \see db_flags
- *
- * \returns The maximum size of a data can write,
- * or -1 if something is wrong. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int
-mdbx_env_get_maxvalsize_ex(const MDBX_env *env, MDBX_db_flags_t flags);
-
-/** \deprecated Please use \ref mdbx_env_get_maxkeysize_ex()
- * and/or \ref mdbx_env_get_maxvalsize_ex()
- * \ingroup c_statinfo */
-MDBX_NOTHROW_PURE_FUNCTION MDBX_DEPRECATED LIBMDBX_API int
-mdbx_env_get_maxkeysize(const MDBX_env *env);
-
-/** \brief Sets application information (a context pointer) associated with
- * the environment.
- * \see mdbx_env_get_userctx()
- * \ingroup c_settings
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] ctx An arbitrary pointer for whatever the application needs.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_set_userctx(MDBX_env *env, void *ctx);
-
-/** \brief Returns an application information (a context pointer) associated
- * with the environment.
- * \see mdbx_env_set_userctx()
- * \ingroup c_statinfo
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create()
- * \returns The pointer set by \ref mdbx_env_set_userctx()
- * or `NULL` if something wrong. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API void *
-mdbx_env_get_userctx(const MDBX_env *env);
-
-/** \brief Create a transaction with a user provided context pointer
- * for use with the environment.
- * \ingroup c_transactions
- *
- * The transaction handle may be discarded using \ref mdbx_txn_abort()
- * or \ref mdbx_txn_commit().
- * \see mdbx_txn_begin()
- *
- * \note A transaction and its cursors must only be used by a single thread,
- * and a thread may only have a single transaction at a time. If \ref MDBX_NOTLS
- * is in use, this does not apply to read-only transactions.
- *
- * \note Cursors may not span transactions.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- *
- * \param [in] parent If this parameter is non-NULL, the new transaction will
- * be a nested transaction, with the transaction indicated
- * by parent as its parent. Transactions may be nested
- * to any level. A parent transaction and its cursors may
- * not issue any other operations than mdbx_txn_commit and
- * \ref mdbx_txn_abort() while it has active child
- * transactions.
- *
- * \param [in] flags Special options for this transaction. This parameter
- * must be set to 0 or by bitwise OR'ing together one
- * or more of the values described here:
- * - \ref MDBX_RDONLY This transaction will not perform
- * any write operations.
- *
- * - \ref MDBX_TXN_TRY Do not block when starting
- * a write transaction.
- *
- * - \ref MDBX_SAFE_NOSYNC, \ref MDBX_NOMETASYNC.
- * Do not sync data to disk corresponding
- * to \ref MDBX_NOMETASYNC or \ref MDBX_SAFE_NOSYNC
- * description. \see sync_modes
- *
- * \param [out] txn Address where the new \ref MDBX_txn handle
- * will be stored.
- *
- * \param [in] context A pointer to application context to be associated with
- * created transaction and could be retrieved by
- * \ref mdbx_txn_get_userctx() until transaction finished.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_PANIC A fatal error occurred earlier and the
- * environment must be shut down.
- * \retval MDBX_UNABLE_EXTEND_MAPSIZE Another process wrote data beyond
- * this MDBX_env's mapsize and this
- * environment map must be resized as well.
- * See \ref mdbx_env_set_mapsize().
- * \retval MDBX_READERS_FULL A read-only transaction was requested and
- * the reader lock table is full.
- * See \ref mdbx_env_set_maxreaders().
- * \retval MDBX_ENOMEM Out of memory.
- * \retval MDBX_BUSY The write transaction is already started by the
- * current thread. */
-LIBMDBX_API int mdbx_txn_begin_ex(MDBX_env *env, MDBX_txn *parent,
- MDBX_txn_flags_t flags, MDBX_txn **txn,
- void *context);
-
-/** \brief Create a transaction for use with the environment.
- * \ingroup c_transactions
- *
- * The transaction handle may be discarded using \ref mdbx_txn_abort()
- * or \ref mdbx_txn_commit().
- * \see mdbx_txn_begin_ex()
- *
- * \note A transaction and its cursors must only be used by a single thread,
- * and a thread may only have a single transaction at a time. If \ref MDBX_NOTLS
- * is in use, this does not apply to read-only transactions.
- *
- * \note Cursors may not span transactions.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- *
- * \param [in] parent If this parameter is non-NULL, the new transaction will
- * be a nested transaction, with the transaction indicated
- * by parent as its parent. Transactions may be nested
- * to any level. A parent transaction and its cursors may
- * not issue any other operations than mdbx_txn_commit and
- * \ref mdbx_txn_abort() while it has active child
- * transactions.
- *
- * \param [in] flags Special options for this transaction. This parameter
- * must be set to 0 or by bitwise OR'ing together one
- * or more of the values described here:
- * - \ref MDBX_RDONLY This transaction will not perform
- * any write operations.
- *
- * - \ref MDBX_TXN_TRY Do not block when starting
- * a write transaction.
- *
- * - \ref MDBX_SAFE_NOSYNC, \ref MDBX_NOMETASYNC.
- * Do not sync data to disk corresponding
- * to \ref MDBX_NOMETASYNC or \ref MDBX_SAFE_NOSYNC
- * description. \see sync_modes
- *
- * \param [out] txn Address where the new \ref MDBX_txn handle
- * will be stored.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_PANIC A fatal error occurred earlier and the
- * environment must be shut down.
- * \retval MDBX_UNABLE_EXTEND_MAPSIZE Another process wrote data beyond
- * this MDBX_env's mapsize and this
- * environment map must be resized as well.
- * See \ref mdbx_env_set_mapsize().
- * \retval MDBX_READERS_FULL A read-only transaction was requested and
- * the reader lock table is full.
- * See \ref mdbx_env_set_maxreaders().
- * \retval MDBX_ENOMEM Out of memory.
- * \retval MDBX_BUSY The write transaction is already started by the
- * current thread. */
-LIBMDBX_INLINE_API(int, mdbx_txn_begin,
- (MDBX_env * env, MDBX_txn *parent, MDBX_txn_flags_t flags,
- MDBX_txn **txn)) {
- return mdbx_txn_begin_ex(env, parent, flags, txn, NULL);
-}
-
-/** \brief Sets application information associated (a context pointer) with the
- * transaction.
- * \ingroup c_transactions
- * \see mdbx_txn_get_userctx()
- *
- * \param [in] txn An transaction handle returned by \ref mdbx_txn_begin_ex()
- * or \ref mdbx_txn_begin().
- * \param [in] ctx An arbitrary pointer for whatever the application needs.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_txn_set_userctx(MDBX_txn *txn, void *ctx);
-
-/** \brief Returns an application information (a context pointer) associated
- * with the transaction.
- * \ingroup c_transactions
- * \see mdbx_txn_set_userctx()
- *
- * \param [in] txn An transaction handle returned by \ref mdbx_txn_begin_ex()
- * or \ref mdbx_txn_begin().
- * \returns The pointer which was passed via the `context` parameter
- * of `mdbx_txn_begin_ex()` or set by \ref mdbx_txn_set_userctx(),
- * or `NULL` if something wrong. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API void *
-mdbx_txn_get_userctx(const MDBX_txn *txn);
-
-/** \brief Information about the transaction
- * \ingroup c_statinfo
- * \see mdbx_txn_info */
-struct MDBX_txn_info {
- /** The ID of the transaction. For a READ-ONLY transaction, this corresponds
- to the snapshot being read. */
- uint64_t txn_id;
-
- /** For READ-ONLY transaction: the lag from a recent MVCC-snapshot, i.e. the
- number of committed transaction since read transaction started.
- For WRITE transaction (provided if `scan_rlt=true`): the lag of the oldest
- reader from current transaction (i.e. at least 1 if any reader running). */
- uint64_t txn_reader_lag;
-
- /** Used space by this transaction, i.e. corresponding to the last used
- * database page. */
- uint64_t txn_space_used;
-
- /** Current size of database file. */
- uint64_t txn_space_limit_soft;
-
- /** Upper bound for size the database file, i.e. the value `size_upper`
- argument of the appropriate call of \ref mdbx_env_set_geometry(). */
- uint64_t txn_space_limit_hard;
-
- /** For READ-ONLY transaction: The total size of the database pages that were
- retired by committed write transactions after the reader's MVCC-snapshot,
- i.e. the space which would be freed after the Reader releases the
- MVCC-snapshot for reuse by completion read transaction.
- For WRITE transaction: The summarized size of the database pages that were
- retired for now due Copy-On-Write during this transaction. */
- uint64_t txn_space_retired;
-
- /** For READ-ONLY transaction: the space available for writer(s) and that
- must be exhausted for reason to call the Handle-Slow-Readers callback for
- this read transaction.
- For WRITE transaction: the space inside transaction
- that left to `MDBX_TXN_FULL` error. */
- uint64_t txn_space_leftover;
-
- /** For READ-ONLY transaction (provided if `scan_rlt=true`): The space that
- actually become available for reuse when only this transaction will be
- finished.
- For WRITE transaction: The summarized size of the dirty database
- pages that generated during this transaction. */
- uint64_t txn_space_dirty;
-};
-#ifndef __cplusplus
-/** \ingroup c_statinfo */
-typedef struct MDBX_txn_info MDBX_txn_info;
-#endif
-
-/** \brief Return information about the MDBX transaction.
- * \ingroup c_statinfo
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin()
- * \param [out] info The address of an \ref MDBX_txn_info structure
- * where the information will be copied.
- * \param [in] scan_rlt The boolean flag controls the scan of the read lock
- * table to provide complete information. Such scan
- * is relatively expensive and you can avoid it
- * if corresponding fields are not needed.
- * See description of \ref MDBX_txn_info.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_txn_info(const MDBX_txn *txn, MDBX_txn_info *info,
- bool scan_rlt);
-
-/** \brief Returns the transaction's MDBX_env.
- * \ingroup c_transactions
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin() */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API MDBX_env *
-mdbx_txn_env(const MDBX_txn *txn);
-
-/** \brief Return the transaction's flags.
- * \ingroup c_transactions
- *
- * This returns the flags associated with this transaction.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- *
- * \returns A transaction flags, valid if input is an valid transaction,
- * otherwise -1. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int mdbx_txn_flags(const MDBX_txn *txn);
-
-/** \brief Return the transaction's ID.
- * \ingroup c_statinfo
- *
- * This returns the identifier associated with this transaction. For a
- * read-only transaction, this corresponds to the snapshot being read;
- * concurrent readers will frequently have the same transaction ID.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- *
- * \returns A transaction ID, valid if input is an active transaction,
- * otherwise 0. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API uint64_t
-mdbx_txn_id(const MDBX_txn *txn);
-
-/** \brief Latency of commit stages in 1/65536 of seconds units.
- * \warning This structure may be changed in future releases.
- * \see mdbx_txn_commit_ex() */
-struct MDBX_commit_latency {
- /** \brief Duration of preparation (commit child transactions, update
- * sub-databases records and cursors destroying). */
- uint32_t preparation;
- /** \brief Duration of GC/freeDB handling & updation. */
- uint32_t gc;
- /** \brief Duration of internal audit if enabled. */
- uint32_t audit;
- /** \brief Duration of writing dirty/modified data pages to a filesystem,
- * i.e. the summary duration of a `write()` syscalls during commit. */
- uint32_t write;
- /** \brief Duration of syncing written data to the disk/storage, i.e.
- * the duration of a `fdatasync()` or a `msync()` syscall during commit. */
- uint32_t sync;
- /** \brief Duration of transaction ending (releasing resources). */
- uint32_t ending;
- /** \brief The total duration of a commit. */
- uint32_t whole;
-};
-#ifndef __cplusplus
-/** \ingroup c_statinfo */
-typedef struct MDBX_commit_latency MDBX_commit_latency;
-#endif
-
-/** \brief Commit all the operations of a transaction into the database and
- * collect latency information.
- * \see mdbx_txn_commit()
- * \ingroup c_statinfo
- * \warning This function may be changed in future releases. */
-LIBMDBX_API int mdbx_txn_commit_ex(MDBX_txn *txn, MDBX_commit_latency *latency);
-
-/** \brief Commit all the operations of a transaction into the database.
- * \ingroup c_transactions
- *
- * If the current thread is not eligible to manage the transaction then
- * the \ref MDBX_THREAD_MISMATCH error will returned. Otherwise the transaction
- * will be committed and its handle is freed. If the transaction cannot
- * be committed, it will be aborted with the corresponding error returned.
- *
- * Thus, a result other than \ref MDBX_THREAD_MISMATCH means that the
- * transaction is terminated:
- * - Resources are released;
- * - Transaction handle is invalid;
- * - Cursor(s) associated with transaction must not be used, except with
- * mdbx_cursor_renew() and \ref mdbx_cursor_close().
- * Such cursor(s) must be closed explicitly by \ref mdbx_cursor_close()
- * before or after transaction commit, either can be reused with
- * \ref mdbx_cursor_renew() until it will be explicitly closed by
- * \ref mdbx_cursor_close().
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_RESULT_TRUE Transaction was aborted since it should
- * be aborted due to previous errors.
- * \retval MDBX_PANIC A fatal error occurred earlier
- * and the environment must be shut down.
- * \retval MDBX_BAD_TXN Transaction is already finished or never began.
- * \retval MDBX_EBADSIGN Transaction object has invalid signature,
- * e.g. transaction was already terminated
- * or memory was corrupted.
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL Transaction handle is NULL.
- * \retval MDBX_ENOSPC No more disk space.
- * \retval MDBX_EIO A system-level I/O error occurred.
- * \retval MDBX_ENOMEM Out of memory. */
-LIBMDBX_INLINE_API(int, mdbx_txn_commit, (MDBX_txn * txn)) {
- return mdbx_txn_commit_ex(txn, NULL);
-}
-
-/** \brief Abandon all the operations of the transaction instead of saving them.
- * \ingroup c_transactions
- *
- * The transaction handle is freed. It and its cursors must not be used again
- * after this call, except with \ref mdbx_cursor_renew() and
- * \ref mdbx_cursor_close().
- *
- * If the current thread is not eligible to manage the transaction then
- * the \ref MDBX_THREAD_MISMATCH error will returned. Otherwise the transaction
- * will be aborted and its handle is freed. Thus, a result other than
- * \ref MDBX_THREAD_MISMATCH means that the transaction is terminated:
- * - Resources are released;
- * - Transaction handle is invalid;
- * - Cursor(s) associated with transaction must not be used, except with
- * \ref mdbx_cursor_renew() and \ref mdbx_cursor_close().
- * Such cursor(s) must be closed explicitly by \ref mdbx_cursor_close()
- * before or after transaction abort, either can be reused with
- * \ref mdbx_cursor_renew() until it will be explicitly closed by
- * \ref mdbx_cursor_close().
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_PANIC A fatal error occurred earlier and
- * the environment must be shut down.
- * \retval MDBX_BAD_TXN Transaction is already finished or never began.
- * \retval MDBX_EBADSIGN Transaction object has invalid signature,
- * e.g. transaction was already terminated
- * or memory was corrupted.
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL Transaction handle is NULL. */
-LIBMDBX_API int mdbx_txn_abort(MDBX_txn *txn);
-
-/** \brief Marks transaction as broken.
- * \ingroup c_transactions
- *
- * Function keeps the transaction handle and corresponding locks, but makes
- * impossible to perform any operations within a broken transaction.
- * Broken transaction must then be aborted explicitly later.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- *
- * \see mdbx_txn_abort() \see mdbx_txn_reset() \see mdbx_txn_commit()
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_txn_break(MDBX_txn *txn);
-
-/** \brief Reset a read-only transaction.
- * \ingroup c_transactions
- *
- * Abort the read-only transaction like \ref mdbx_txn_abort(), but keep the
- * transaction handle. Therefore \ref mdbx_txn_renew() may reuse the handle.
- * This saves allocation overhead if the process will start a new read-only
- * transaction soon, and also locking overhead if \ref MDBX_NOTLS is in use. The
- * reader table lock is released, but the table slot stays tied to its thread
- * or \ref MDBX_txn. Use \ref mdbx_txn_abort() to discard a reset handle, and to
- * free its lock table slot if \ref MDBX_NOTLS is in use.
- *
- * Cursors opened within the transaction must not be used again after this
- * call, except with \ref mdbx_cursor_renew() and \ref mdbx_cursor_close().
- *
- * Reader locks generally don't interfere with writers, but they keep old
- * versions of database pages allocated. Thus they prevent the old pages from
- * being reused when writers commit new data, and so under heavy load the
- * database size may grow much more rapidly than otherwise.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_PANIC A fatal error occurred earlier and
- * the environment must be shut down.
- * \retval MDBX_BAD_TXN Transaction is already finished or never began.
- * \retval MDBX_EBADSIGN Transaction object has invalid signature,
- * e.g. transaction was already terminated
- * or memory was corrupted.
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL Transaction handle is NULL. */
-LIBMDBX_API int mdbx_txn_reset(MDBX_txn *txn);
-
-/** \brief Renew a read-only transaction.
- * \ingroup c_transactions
- *
- * This acquires a new reader lock for a transaction handle that had been
- * released by \ref mdbx_txn_reset(). It must be called before a reset
- * transaction may be used again.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_PANIC A fatal error occurred earlier and
- * the environment must be shut down.
- * \retval MDBX_BAD_TXN Transaction is already finished or never began.
- * \retval MDBX_EBADSIGN Transaction object has invalid signature,
- * e.g. transaction was already terminated
- * or memory was corrupted.
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL Transaction handle is NULL. */
-LIBMDBX_API int mdbx_txn_renew(MDBX_txn *txn);
-
-/** \brief The fours integers markers (aka "canary") associated with the
- * environment.
- * \ingroup c_crud
- * \see mdbx_canary_set()
- * \see mdbx_canary_get()
- *
- * The `x`, `y` and `z` values could be set by \ref mdbx_canary_put(), while the
- * 'v' will be always set to the transaction number. Updated values becomes
- * visible outside the current transaction only after it was committed. Current
- * values could be retrieved by \ref mdbx_canary_get(). */
-struct MDBX_canary {
- uint64_t x, y, z, v;
-};
-#ifndef __cplusplus
-/** \ingroup c_crud */
-typedef struct MDBX_canary MDBX_canary;
-#endif
-
-/** \brief Set integers markers (aka "canary") associated with the environment.
- * \ingroup c_crud
- * \see mdbx_canary_get()
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin()
- * \param [in] canary A optional pointer to \ref MDBX_canary structure for `x`,
- * `y` and `z` values from.
- * - If canary is NOT NULL then the `x`, `y` and `z` values will be
- * updated from given canary argument, but the 'v' be always set
- * to the current transaction number if at least one `x`, `y` or
- * `z` values have changed (i.e. if `x`, `y` and `z` have the same
- * values as currently present then nothing will be changes or
- * updated).
- * - if canary is NULL then the `v` value will be explicitly update
- * to the current transaction number without changes `x`, `y` nor
- * `z`.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_canary_put(MDBX_txn *txn, const MDBX_canary *canary);
-
-/** \brief Returns fours integers markers (aka "canary") associated with the
- * environment.
- * \ingroup c_crud
- * \see mdbx_canary_set()
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] canary The address of an MDBX_canary structure where the
- * information will be copied.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_canary_get(const MDBX_txn *txn, MDBX_canary *canary);
-
-/** \brief A callback function used to compare two keys in a database
- * \ingroup c_crud
- * \see mdbx_cmp() \see mdbx_get_keycmp()
- * \see mdbx_get_datacmp \see mdbx_dcmp()
- *
- * \anchor avoid_custom_comparators
- * It is recommend not using custom comparison functions, but instead
- * converting the keys to one of the forms that are suitable for built-in
- * comparators (for instance take look to the \ref value2key).
- * The reasons to not using custom comparators are:
- * - The order of records could not be validated without your code.
- * So `mdbx_chk` utility will reports "wrong order" errors
- * and the `-i` option is required to suppress ones.
- * - A records could not be ordered or sorted without your code.
- * So `mdbx_load` utility should be used with `-a` option to preserve
- * input data order.
- * - However, the custom comparators feature will never be removed.
- * You have been warned but still can use custom comparators knowing
- * about the issues noted above. In this case you should ignore `deprecated`
- * warnings or define `MDBX_DEPRECATED` macro to empty to avoid ones. */
-typedef int(MDBX_cmp_func)(const MDBX_val *a,
- const MDBX_val *b) MDBX_CXX17_NOEXCEPT;
-
-/** \brief Open or Create a database in the environment.
- * \ingroup c_dbi
- *
- * A database handle denotes the name and parameters of a database,
- * independently of whether such a database exists. The database handle may be
- * discarded by calling \ref mdbx_dbi_close(). The old database handle is
- * returned if the database was already open. The handle may only be closed
- * once.
- *
- * \note A notable difference between MDBX and LMDB is that MDBX make handles
- * opened for existing databases immediately available for other transactions,
- * regardless this transaction will be aborted or reset. The REASON for this is
- * to avoiding the requirement for multiple opening a same handles in
- * concurrent read transactions, and tracking of such open but hidden handles
- * until the completion of read transactions which opened them.
- *
- * Nevertheless, the handle for the NEWLY CREATED database will be invisible
- * for other transactions until the this write transaction is successfully
- * committed. If the write transaction is aborted the handle will be closed
- * automatically. After a successful commit the such handle will reside in the
- * shared environment, and may be used by other transactions.
- *
- * In contrast to LMDB, the MDBX allow this function to be called from multiple
- * concurrent transactions or threads in the same process.
- *
- * To use named database (with name != NULL), \ref mdbx_env_set_maxdbs()
- * must be called before opening the environment. Table names are
- * keys in the internal unnamed database, and may be read but not written.
- *
- * \param [in] txn transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] name The name of the database to open. If only a single
- * database is needed in the environment,
- * this value may be NULL.
- * \param [in] flags Special options for this database. This parameter must
- * be bitwise OR'ing together any of the constants
- * described here:
- *
- * - \ref MDBX_DB_DEFAULTS
- * Keys are arbitrary byte strings and compared from beginning to end.
- * - \ref MDBX_REVERSEKEY
- * Keys are arbitrary byte strings to be compared in reverse order,
- * from the end of the strings to the beginning.
- * - \ref MDBX_INTEGERKEY
- * Keys are binary integers in native byte order, either uint32_t or
- * uint64_t, and will be sorted as such. The keys must all be of the
- * same size and must be aligned while passing as arguments.
- * - \ref MDBX_DUPSORT
- * Duplicate keys may be used in the database. Or, from another point of
- * view, keys may have multiple data items, stored in sorted order. By
- * default keys must be unique and may have only a single data item.
- * - \ref MDBX_DUPFIXED
- * This flag may only be used in combination with \ref MDBX_DUPSORT. This
- * option tells the library that the data items for this database are
- * all the same size, which allows further optimizations in storage and
- * retrieval. When all data items are the same size, the
- * \ref MDBX_GET_MULTIPLE, \ref MDBX_NEXT_MULTIPLE and
- * \ref MDBX_PREV_MULTIPLE cursor operations may be used to retrieve
- * multiple items at once.
- * - \ref MDBX_INTEGERDUP
- * This option specifies that duplicate data items are binary integers,
- * similar to \ref MDBX_INTEGERKEY keys. The data values must all be of the
- * same size and must be aligned while passing as arguments.
- * - \ref MDBX_REVERSEDUP
- * This option specifies that duplicate data items should be compared as
- * strings in reverse order (the comparison is performed in the direction
- * from the last byte to the first).
- * - \ref MDBX_CREATE
- * Create the named database if it doesn't exist. This option is not
- * allowed in a read-only transaction or a read-only environment.
- *
- * \param [out] dbi Address where the new \ref MDBX_dbi handle
- * will be stored.
- *
- * For \ref mdbx_dbi_open_ex() additional arguments allow you to set custom
- * comparison functions for keys and values (for multimaps).
- * \see avoid_custom_comparators
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_NOTFOUND The specified database doesn't exist in the
- * environment and \ref MDBX_CREATE was not specified.
- * \retval MDBX_DBS_FULL Too many databases have been opened.
- * \see mdbx_env_set_maxdbs()
- * \retval MDBX_INCOMPATIBLE Database is incompatible with given flags,
- * i.e. the passed flags is different with which the
- * database was created, or the database was already
- * opened with a different comparison function(s).
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread. */
-LIBMDBX_API int mdbx_dbi_open(MDBX_txn *txn, const char *name,
- MDBX_db_flags_t flags, MDBX_dbi *dbi);
-
-/** \deprecated Please
- * \ref avoid_custom_comparators "avoid using custom comparators" and use
- * \ref mdbx_dbi_open() instead.
- *
- * \ingroup c_dbi
- *
- * \param [in] txn transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] name The name of the database to open. If only a single
- * database is needed in the environment,
- * this value may be NULL.
- * \param [in] flags Special options for this database.
- * \param [in] keycmp Optional custom key comparison function for a database.
- * \param [in] datacmp Optional custom data comparison function for a database.
- * \param [out] dbi Address where the new MDBX_dbi handle will be stored.
- * \returns A non-zero error value on failure and 0 on success. */
-MDBX_DEPRECATED LIBMDBX_API int
-mdbx_dbi_open_ex(MDBX_txn *txn, const char *name, MDBX_db_flags_t flags,
- MDBX_dbi *dbi, MDBX_cmp_func *keycmp, MDBX_cmp_func *datacmp);
-
-/** \defgroup value2key Value-to-Key functions
- * \brief Value-to-Key functions to
- * \ref avoid_custom_comparators "avoid using custom comparators"
- * \see key2value
- * @{
- *
- * The \ref mdbx_key_from_jsonInteger() build a keys which are comparable with
- * keys created by \ref mdbx_key_from_double(). So this allows mixing `int64_t`
- * and IEEE754 double values in one index for JSON-numbers with restriction for
- * integer numbers range corresponding to RFC-7159, i.e. \f$[-2^{53}+1,
- * 2^{53}-1]\f$. See bottom of page 6 at https://tools.ietf.org/html/rfc7159 */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API uint64_t
-mdbx_key_from_jsonInteger(const int64_t json_integer);
-
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API uint64_t
-mdbx_key_from_double(const double ieee754_64bit);
-
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API uint64_t
-mdbx_key_from_ptrdouble(const double *const ieee754_64bit);
-
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API uint32_t
-mdbx_key_from_float(const float ieee754_32bit);
-
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API uint32_t
-mdbx_key_from_ptrfloat(const float *const ieee754_32bit);
-
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_INLINE_API(uint64_t, mdbx_key_from_int64,
- (const int64_t i64)) {
- return UINT64_C(0x8000000000000000) + i64;
-}
-
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_INLINE_API(uint32_t, mdbx_key_from_int32,
- (const int32_t i32)) {
- return UINT32_C(0x80000000) + i32;
-}
-/** @} */
-
-/** \defgroup key2value Key-to-Value functions
- * \brief Key-to-Value functions to
- * \ref avoid_custom_comparators "avoid using custom comparators"
- * \see value2key
- * @{ */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int64_t
-mdbx_jsonInteger_from_key(const MDBX_val);
-
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API double
-mdbx_double_from_key(const MDBX_val);
-
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API float
-mdbx_float_from_key(const MDBX_val);
-
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int32_t
-mdbx_int32_from_key(const MDBX_val);
-
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int64_t
-mdbx_int64_from_key(const MDBX_val);
-/** @} */
-
-/** \brief Retrieve statistics for a database.
- * \ingroup c_statinfo
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [out] stat The address of an \ref MDBX_stat structure where
- * the statistics will be copied.
- * \param [in] bytes The size of \ref MDBX_stat.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_dbi_stat(MDBX_txn *txn, MDBX_dbi dbi, MDBX_stat *stat,
- size_t bytes);
-
-/** \brief Retrieve depth (bitmask) information of nested dupsort (multi-value)
- * B+trees for given database.
- * \ingroup c_statinfo
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [out] mask The address of an uint32_t value where the bitmask
- * will be stored.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL An invalid parameter was specified.
- * \retval MDBX_RESULT_TRUE The dbi isn't a dupsort (multi-value) database. */
-LIBMDBX_API int mdbx_dbi_dupsort_depthmask(MDBX_txn *txn, MDBX_dbi dbi,
- uint32_t *mask);
-
-/** \brief DBI state bits returted by \ref mdbx_dbi_flags_ex()
- * \ingroup c_statinfo
- * \see mdbx_dbi_flags_ex() */
-enum MDBX_dbi_state_t {
- /** DB was written in this txn */
- MDBX_DBI_DIRTY = 0x01,
- /** Named-DB record is older than txnID */
- MDBX_DBI_STALE = 0x02,
- /** Named-DB handle opened in this txn */
- MDBX_DBI_FRESH = 0x04,
- /** Named-DB handle created in this txn */
- MDBX_DBI_CREAT = 0x08,
-};
-#ifndef __cplusplus
-/** \ingroup c_statinfo */
-typedef enum MDBX_dbi_state_t MDBX_dbi_state_t;
-#else
-DEFINE_ENUM_FLAG_OPERATORS(MDBX_dbi_state_t)
-#endif
-
-/** \brief Retrieve the DB flags and status for a database handle.
- * \ingroup c_statinfo
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [out] flags Address where the flags will be returned.
- * \param [out] state Address where the state will be returned.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_dbi_flags_ex(MDBX_txn *txn, MDBX_dbi dbi, unsigned *flags,
- unsigned *state);
-/** \brief The shortcut to calling \ref mdbx_dbi_flags_ex() with `state=NULL`
- * for discarding it result.
- * \ingroup c_statinfo */
-LIBMDBX_INLINE_API(int, mdbx_dbi_flags,
- (MDBX_txn * txn, MDBX_dbi dbi, unsigned *flags)) {
- unsigned state;
- return mdbx_dbi_flags_ex(txn, dbi, flags, &state);
-}
-
-/** \brief Close a database handle. Normally unnecessary.
- * \ingroup c_dbi
- *
- * Closing a database handle is not necessary, but lets \ref mdbx_dbi_open()
- * reuse the handle value. Usually it's better to set a bigger
- * \ref mdbx_env_set_maxdbs(), unless that value would be large.
- *
- * \note Use with care.
- * This call is synchronized via mutex with \ref mdbx_dbi_close(), but NOT with
- * other transactions running by other threads. The "next" version of libmdbx
- * (\ref MithrilDB) will solve this issue.
- *
- * Handles should only be closed if no other threads are going to reference
- * the database handle or one of its cursors any further. Do not close a handle
- * if an existing transaction has modified its database. Doing so can cause
- * misbehavior from database corruption to errors like \ref MDBX_BAD_DBI
- * (since the DB name is gone).
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_dbi_close(MDBX_env *env, MDBX_dbi dbi);
-
-/** \brief Empty or delete and close a database.
- * \ingroup c_crud
- *
- * \see mdbx_dbi_close() \see mdbx_dbi_open()
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] del `false` to empty the DB, `true` to delete it
- * from the environment and close the DB handle.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_drop(MDBX_txn *txn, MDBX_dbi dbi, bool del);
-
-/** \brief Get items from a database.
- * \ingroup c_crud
- *
- * This function retrieves key/data pairs from the database. The address
- * and length of the data associated with the specified key are returned
- * in the structure to which data refers.
- * If the database supports duplicate keys (\ref MDBX_DUPSORT) then the
- * first data item for the key will be returned. Retrieval of other
- * items requires the use of \ref mdbx_cursor_get().
- *
- * \note The memory pointed to by the returned values is owned by the
- * database. The caller need not dispose of the memory, and may not
- * modify it in any way. For values returned in a read-only transaction
- * any modification attempts will cause a `SIGSEGV`.
- *
- * \note Values returned from the database are valid only until a
- * subsequent update operation, or the end of the transaction.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] key The key to search for in the database.
- * \param [in,out] data The data corresponding to the key.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_NOTFOUND The key was not in the database.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_get(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key,
- MDBX_val *data);
-
-/** \brief Get items from a database
- * and optionally number of data items for a given key.
- *
- * \ingroup c_crud
- *
- * Briefly this function does the same as \ref mdbx_get() with a few
- * differences:
- * 1. If values_count is NOT NULL, then returns the count
- * of multi-values/duplicates for a given key.
- * 2. Updates BOTH the key and the data for pointing to the actual key-value
- * pair inside the database.
- *
- * \param [in] txn A transaction handle returned
- * by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in,out] key The key to search for in the database.
- * \param [in,out] data The data corresponding to the key.
- * \param [out] values_count The optional address to return number of values
- * associated with given key:
- * = 0 - in case \ref MDBX_NOTFOUND error;
- * = 1 - exactly for databases
- * WITHOUT \ref MDBX_DUPSORT;
- * >= 1 for databases WITH \ref MDBX_DUPSORT.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_NOTFOUND The key was not in the database.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_get_ex(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key,
- MDBX_val *data, size_t *values_count);
-
-/** \brief Get equal or great item from a database.
- * \ingroup c_crud
- *
- * Briefly this function does the same as \ref mdbx_get() with a few
- * differences:
- * 1. Return equal or great (due comparison function) key-value
- * pair, but not only exactly matching with the key.
- * 2. On success return \ref MDBX_SUCCESS if key found exactly,
- * and \ref MDBX_RESULT_TRUE otherwise. Moreover, for databases with
- * \ref MDBX_DUPSORT flag the data argument also will be used to match over
- * multi-value/duplicates, and \ref MDBX_SUCCESS will be returned only when
- * BOTH the key and the data match exactly.
- * 3. Updates BOTH the key and the data for pointing to the actual key-value
- * pair inside the database.
- *
- * \param [in] txn A transaction handle returned
- * by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in,out] key The key to search for in the database.
- * \param [in,out] data The data corresponding to the key.
- *
- * \returns A non-zero error value on failure and \ref MDBX_RESULT_FALSE
- * or \ref MDBX_RESULT_TRUE on success (as described above).
- * Some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_NOTFOUND The key was not in the database.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_get_equal_or_great(MDBX_txn *txn, MDBX_dbi dbi,
- MDBX_val *key, MDBX_val *data);
-
-/** \brief Store items into a database.
- * \ingroup c_crud
- *
- * This function stores key/data pairs in the database. The default behavior
- * is to enter the new key/data pair, replacing any previously existing key
- * if duplicates are disallowed, or adding a duplicate data item if
- * duplicates are allowed (see \ref MDBX_DUPSORT).
- *
- * \param [in] txn A transaction handle returned
- * by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] key The key to store in the database.
- * \param [in,out] data The data to store.
- * \param [in] flags Special options for this operation.
- * This parameter must be set to 0 or by bitwise OR'ing
- * together one or more of the values described here:
- * - \ref MDBX_NODUPDATA
- * Enter the new key-value pair only if it does not already appear
- * in the database. This flag may only be specified if the database
- * was opened with \ref MDBX_DUPSORT. The function will return
- * \ref MDBX_KEYEXIST if the key/data pair already appears in the database.
- *
- * - \ref MDBX_NOOVERWRITE
- * Enter the new key/data pair only if the key does not already appear
- * in the database. The function will return \ref MDBX_KEYEXIST if the key
- * already appears in the database, even if the database supports
- * duplicates (see \ref MDBX_DUPSORT). The data parameter will be set
- * to point to the existing item.
- *
- * - \ref MDBX_CURRENT
- * Update an single existing entry, but not add new ones. The function will
- * return \ref MDBX_NOTFOUND if the given key not exist in the database.
- * In case multi-values for the given key, with combination of
- * the \ref MDBX_ALLDUPS will replace all multi-values,
- * otherwise return the \ref MDBX_EMULTIVAL.
- *
- * - \ref MDBX_RESERVE
- * Reserve space for data of the given size, but don't copy the given
- * data. Instead, return a pointer to the reserved space, which the
- * caller can fill in later - before the next update operation or the
- * transaction ends. This saves an extra memcpy if the data is being
- * generated later. MDBX does nothing else with this memory, the caller
- * is expected to modify all of the space requested. This flag must not
- * be specified if the database was opened with \ref MDBX_DUPSORT.
- *
- * - \ref MDBX_APPEND
- * Append the given key/data pair to the end of the database. This option
- * allows fast bulk loading when keys are already known to be in the
- * correct order. Loading unsorted keys with this flag will cause
- * a \ref MDBX_EKEYMISMATCH error.
- *
- * - \ref MDBX_APPENDDUP
- * As above, but for sorted dup data.
- *
- * - \ref MDBX_MULTIPLE
- * Store multiple contiguous data elements in a single request. This flag
- * may only be specified if the database was opened with
- * \ref MDBX_DUPFIXED. With combination the \ref MDBX_ALLDUPS
- * will replace all multi-values.
- * The data argument must be an array of two \ref MDBX_val. The `iov_len`
- * of the first \ref MDBX_val must be the size of a single data element.
- * The `iov_base` of the first \ref MDBX_val must point to the beginning
- * of the array of contiguous data elements which must be properly aligned
- * in case of database with \ref MDBX_INTEGERDUP flag.
- * The `iov_len` of the second \ref MDBX_val must be the count of the
- * number of data elements to store. On return this field will be set to
- * the count of the number of elements actually written. The `iov_base` of
- * the second \ref MDBX_val is unused.
- *
- * \see \ref c_crud_hints "Quick reference for Insert/Update/Delete operations"
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_KEYEXIST The key/value pair already exists in the database.
- * \retval MDBX_MAP_FULL The database is full, see \ref mdbx_env_set_mapsize().
- * \retval MDBX_TXN_FULL The transaction has too many dirty pages.
- * \retval MDBX_EACCES An attempt was made to write
- * in a read-only transaction.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_put(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key,
- MDBX_val *data, MDBX_put_flags_t flags);
-
-/** \brief Replace items in a database.
- * \ingroup c_crud
- *
- * This function allows to update or delete an existing value at the same time
- * as the previous value is retrieved. If the argument new_data equal is NULL
- * zero, the removal is performed, otherwise the update/insert.
- *
- * The current value may be in an already changed (aka dirty) page. In this
- * case, the page will be overwritten during the update, and the old value will
- * be lost. Therefore, an additional buffer must be passed via old_data
- * argument initially to copy the old value. If the buffer passed in is too
- * small, the function will return \ref MDBX_RESULT_TRUE by setting iov_len
- * field pointed by old_data argument to the appropriate value, without
- * performing any changes.
- *
- * For databases with non-unique keys (i.e. with \ref MDBX_DUPSORT flag),
- * another use case is also possible, when by old_data argument selects a
- * specific item from multi-value/duplicates with the same key for deletion or
- * update. To select this scenario in flags should simultaneously specify
- * \ref MDBX_CURRENT and \ref MDBX_NOOVERWRITE. This combination is chosen
- * because it makes no sense, and thus allows you to identify the request of
- * such a scenario.
- *
- * \param [in] txn A transaction handle returned
- * by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] key The key to store in the database.
- * \param [in] new_data The data to store, if NULL then deletion will
- * be performed.
- * \param [in,out] old_data The buffer for retrieve previous value as describe
- * above.
- * \param [in] flags Special options for this operation.
- * This parameter must be set to 0 or by bitwise
- * OR'ing together one or more of the values
- * described in \ref mdbx_put() description above,
- * and additionally
- * (\ref MDBX_CURRENT | \ref MDBX_NOOVERWRITE)
- * combination for selection particular item from
- * multi-value/duplicates.
- *
- * \see \ref c_crud_hints "Quick reference for Insert/Update/Delete operations"
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_replace(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key,
- MDBX_val *new_data, MDBX_val *old_data,
- MDBX_put_flags_t flags);
-
-typedef int (*MDBX_preserve_func)(void *context, MDBX_val *target,
- const void *src, size_t bytes);
-LIBMDBX_API int mdbx_replace_ex(MDBX_txn *txn, MDBX_dbi dbi,
- const MDBX_val *key, MDBX_val *new_data,
- MDBX_val *old_data, MDBX_put_flags_t flags,
- MDBX_preserve_func preserver,
- void *preserver_context);
-
-/** \brief Delete items from a database.
- * \ingroup c_crud
- *
- * This function removes key/data pairs from the database.
- *
- * \note The data parameter is NOT ignored regardless the database does
- * support sorted duplicate data items or not. If the data parameter
- * is non-NULL only the matching data item will be deleted. Otherwise, if data
- * parameter is NULL, any/all value(s) for specified key will be deleted.
- *
- * This function will return \ref MDBX_NOTFOUND if the specified key/data
- * pair is not in the database.
- *
- * \see \ref c_crud_hints "Quick reference for Insert/Update/Delete operations"
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] key The key to delete from the database.
- * \param [in] data The data to delete.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EACCES An attempt was made to write
- * in a read-only transaction.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_del(MDBX_txn *txn, MDBX_dbi dbi, const MDBX_val *key,
- const MDBX_val *data);
-
-/** \brief Create a cursor handle but not bind it to transaction nor DBI handle.
- * \ingroup c_cursors
- *
- * A cursor cannot be used when its database handle is closed. Nor when its
- * transaction has ended, except with \ref mdbx_cursor_bind() and \ref
- * mdbx_cursor_renew(). Also it can be discarded with \ref mdbx_cursor_close().
- *
- * A cursor must be closed explicitly always, before or after its transaction
- * ends. It can be reused with \ref mdbx_cursor_bind()
- * or \ref mdbx_cursor_renew() before finally closing it.
- *
- * \note In contrast to LMDB, the MDBX required that any opened cursors can be
- * reused and must be freed explicitly, regardless ones was opened in a
- * read-only or write transaction. The REASON for this is eliminates ambiguity
- * which helps to avoid errors such as: use-after-free, double-free, i.e.
- * memory corruption and segfaults.
- *
- * \param [in] context A pointer to application context to be associated with
- * created cursor and could be retrieved by
- * \ref mdbx_cursor_get_userctx() until cursor closed.
- *
- * \returns Created cursor handle or NULL in case out of memory. */
-LIBMDBX_API MDBX_cursor *mdbx_cursor_create(void *context);
-
-/** \brief Set application information associated with the \ref MDBX_cursor.
- * \ingroup c_cursors
- * \see mdbx_cursor_get_userctx()
- *
- * \param [in] cursor An cursor handle returned by \ref mdbx_cursor_create()
- * or \ref mdbx_cursor_open().
- * \param [in] ctx An arbitrary pointer for whatever the application needs.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_cursor_set_userctx(MDBX_cursor *cursor, void *ctx);
-
-/** \brief Get the application information associated with the MDBX_cursor.
- * \ingroup c_cursors
- * \see mdbx_cursor_set_userctx()
- *
- * \param [in] cursor An cursor handle returned by \ref mdbx_cursor_create()
- * or \ref mdbx_cursor_open().
- * \returns The pointer which was passed via the `context` parameter
- * of `mdbx_cursor_create()` or set by \ref mdbx_cursor_set_userctx(),
- * or `NULL` if something wrong. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API void *
-mdbx_cursor_get_userctx(const MDBX_cursor *cursor);
-
-/** \brief Bind cursor to specified transaction and DBI handle.
- * \ingroup c_cursors
- *
- * Using of the `mdbx_cursor_bind()` is equivalent to calling
- * \ref mdbx_cursor_renew() but with specifying an arbitrary dbi handle.
- *
- * A cursor may be associated with a new transaction, and referencing a new or
- * the same database handle as it was created with. This may be done whether the
- * previous transaction is live or dead.
- *
- * \note In contrast to LMDB, the MDBX required that any opened cursors can be
- * reused and must be freed explicitly, regardless ones was opened in a
- * read-only or write transaction. The REASON for this is eliminates ambiguity
- * which helps to avoid errors such as: use-after-free, double-free, i.e.
- * memory corruption and segfaults.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [out] cursor A cursor handle returned by \ref mdbx_cursor_create().
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_bind(MDBX_txn *txn, MDBX_cursor *cursor,
- MDBX_dbi dbi);
-
-/** \brief Create a cursor handle for the specified transaction and DBI handle.
- * \ingroup c_cursors
- *
- * Using of the `mdbx_cursor_open()` is equivalent to calling
- * \ref mdbx_cursor_create() and then \ref mdbx_cursor_bind() functions.
- *
- * A cursor cannot be used when its database handle is closed. Nor when its
- * transaction has ended, except with \ref mdbx_cursor_bind() and \ref
- * mdbx_cursor_renew(). Also it can be discarded with \ref mdbx_cursor_close().
- *
- * A cursor must be closed explicitly always, before or after its transaction
- * ends. It can be reused with \ref mdbx_cursor_bind()
- * or \ref mdbx_cursor_renew() before finally closing it.
- *
- * \note In contrast to LMDB, the MDBX required that any opened cursors can be
- * reused and must be freed explicitly, regardless ones was opened in a
- * read-only or write transaction. The REASON for this is eliminates ambiguity
- * which helps to avoid errors such as: use-after-free, double-free, i.e.
- * memory corruption and segfaults.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [out] cursor Address where the new \ref MDBX_cursor handle will be
- * stored.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_open(MDBX_txn *txn, MDBX_dbi dbi,
- MDBX_cursor **cursor);
-
-/** \brief Close a cursor handle.
- * \ingroup c_cursors
- *
- * The cursor handle will be freed and must not be used again after this call,
- * but its transaction may still be live.
- *
- * \note In contrast to LMDB, the MDBX required that any opened cursors can be
- * reused and must be freed explicitly, regardless ones was opened in a
- * read-only or write transaction. The REASON for this is eliminates ambiguity
- * which helps to avoid errors such as: use-after-free, double-free, i.e.
- * memory corruption and segfaults.
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open()
- * or \ref mdbx_cursor_create(). */
-LIBMDBX_API void mdbx_cursor_close(MDBX_cursor *cursor);
-
-/** \brief Renew a cursor handle.
- * \ingroup c_cursors
- *
- * The cursor may be associated with a new transaction, and referencing a new or
- * the same database handle as it was created with. This may be done whether the
- * previous transaction is live or dead.
- *
- * Using of the `mdbx_cursor_renew()` is equivalent to calling
- * \ref mdbx_cursor_bind() with the DBI handle that previously
- * the cursor was used with.
- *
- * \note In contrast to LMDB, the MDBX allow any cursor to be re-used by using
- * \ref mdbx_cursor_renew(), to avoid unnecessary malloc/free overhead until it
- * freed by \ref mdbx_cursor_close().
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_renew(MDBX_txn *txn, MDBX_cursor *cursor);
-
-/** \brief Return the cursor's transaction handle.
- * \ingroup c_cursors
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open(). */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API MDBX_txn *
-mdbx_cursor_txn(const MDBX_cursor *cursor);
-
-/** \brief Return the cursor's database handle.
- * \ingroup c_cursors
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open(). */
-LIBMDBX_API MDBX_dbi mdbx_cursor_dbi(const MDBX_cursor *cursor);
-
-/** \brief Copy cursor position and state.
- * \ingroup c_cursors
- *
- * \param [in] src A source cursor handle returned
- * by \ref mdbx_cursor_create() or \ref mdbx_cursor_open().
- *
- * \param [in,out] dest A destination cursor handle returned
- * by \ref mdbx_cursor_create() or \ref mdbx_cursor_open().
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_cursor_copy(const MDBX_cursor *src, MDBX_cursor *dest);
-
-/** \brief Retrieve by cursor.
- * \ingroup c_crud
- *
- * This function retrieves key/data pairs from the database. The address and
- * length of the key are returned in the object to which key refers (except
- * for the case of the \ref MDBX_SET option, in which the key object is
- * unchanged), and the address and length of the data are returned in the object
- * to which data refers.
- * \see mdbx_get()
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- * \param [in,out] key The key for a retrieved item.
- * \param [in,out] data The data of a retrieved item.
- * \param [in] op A cursor operation \ref MDBX_cursor_op.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_NOTFOUND No matching key found.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_get(MDBX_cursor *cursor, MDBX_val *key,
- MDBX_val *data, MDBX_cursor_op op);
-
-/** \brief Retrieve multiple non-dupsort key/value pairs by cursor.
- * \ingroup c_crud
- *
- * This function retrieves multiple key/data pairs from the database without
- * \ref MDBX_DUPSORT option. For `MDBX_DUPSORT` databases please
- * use \ref MDBX_GET_MULTIPLE and \ref MDBX_NEXT_MULTIPLE.
- *
- * The number of key and value items is returned in the `size_t count`
- * refers. The addresses and lengths of the keys and values are returned in the
- * array to which `pairs` refers.
- * \see mdbx_cursor_get()
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- * \param [out] count The number of key and value item returned, on success
- * it always be the even because the key-value
- * pairs are returned.
- * \param [in,out] pairs A pointer to the array of key value pairs.
- * \param [in] limit The size of pairs buffer as the number of items,
- * but not a pairs.
- * \param [in] op A cursor operation \ref MDBX_cursor_op (only
- * \ref MDBX_FIRST, \ref MDBX_NEXT, \ref MDBX_GET_CURRENT
- * are supported).
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_NOTFOUND No more key-value pairs are available.
- * \retval MDBX_ENODATA The cursor is already at the end of data.
- * \retval MDBX_RESULT_TRUE The specified limit is less than the available
- * key-value pairs on the current page/position
- * that the cursor points to.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_get_batch(MDBX_cursor *cursor, size_t *count,
- MDBX_val *pairs, size_t limit,
- MDBX_cursor_op op);
-
-/** \brief Store by cursor.
- * \ingroup c_crud
- *
- * This function stores key/data pairs into the database. The cursor is
- * positioned at the new item, or on failure usually near it.
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- * \param [in] key The key operated on.
- * \param [in,out] data The data operated on.
- * \param [in] flags Options for this operation. This parameter
- * must be set to 0 or by bitwise OR'ing together
- * one or more of the values described here:
- * - \ref MDBX_CURRENT
- * Replace the item at the current cursor position. The key parameter
- * must still be provided, and must match it, otherwise the function
- * return \ref MDBX_EKEYMISMATCH. With combination the
- * \ref MDBX_ALLDUPS will replace all multi-values.
- *
- * \note MDBX allows (unlike LMDB) you to change the size of the data and
- * automatically handles reordering for sorted duplicates
- * (see \ref MDBX_DUPSORT).
- *
- * - \ref MDBX_NODUPDATA
- * Enter the new key-value pair only if it does not already appear in the
- * database. This flag may only be specified if the database was opened
- * with \ref MDBX_DUPSORT. The function will return \ref MDBX_KEYEXIST
- * if the key/data pair already appears in the database.
- *
- * - \ref MDBX_NOOVERWRITE
- * Enter the new key/data pair only if the key does not already appear
- * in the database. The function will return \ref MDBX_KEYEXIST if the key
- * already appears in the database, even if the database supports
- * duplicates (\ref MDBX_DUPSORT).
- *
- * - \ref MDBX_RESERVE
- * Reserve space for data of the given size, but don't copy the given
- * data. Instead, return a pointer to the reserved space, which the
- * caller can fill in later - before the next update operation or the
- * transaction ends. This saves an extra memcpy if the data is being
- * generated later. This flag must not be specified if the database
- * was opened with \ref MDBX_DUPSORT.
- *
- * - \ref MDBX_APPEND
- * Append the given key/data pair to the end of the database. No key
- * comparisons are performed. This option allows fast bulk loading when
- * keys are already known to be in the correct order. Loading unsorted
- * keys with this flag will cause a \ref MDBX_KEYEXIST error.
- *
- * - \ref MDBX_APPENDDUP
- * As above, but for sorted dup data.
- *
- * - \ref MDBX_MULTIPLE
- * Store multiple contiguous data elements in a single request. This flag
- * may only be specified if the database was opened with
- * \ref MDBX_DUPFIXED. With combination the \ref MDBX_ALLDUPS
- * will replace all multi-values.
- * The data argument must be an array of two \ref MDBX_val. The `iov_len`
- * of the first \ref MDBX_val must be the size of a single data element.
- * The `iov_base` of the first \ref MDBX_val must point to the beginning
- * of the array of contiguous data elements which must be properly aligned
- * in case of database with \ref MDBX_INTEGERDUP flag.
- * The `iov_len` of the second \ref MDBX_val must be the count of the
- * number of data elements to store. On return this field will be set to
- * the count of the number of elements actually written. The `iov_base` of
- * the second \ref MDBX_val is unused.
- *
- * \see \ref c_crud_hints "Quick reference for Insert/Update/Delete operations"
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EKEYMISMATCH The given key value is mismatched to the current
- * cursor position
- * \retval MDBX_MAP_FULL The database is full,
- * see \ref mdbx_env_set_mapsize().
- * \retval MDBX_TXN_FULL The transaction has too many dirty pages.
- * \retval MDBX_EACCES An attempt was made to write in a read-only
- * transaction.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_put(MDBX_cursor *cursor, const MDBX_val *key,
- MDBX_val *data, MDBX_put_flags_t flags);
-
-/** \brief Delete current key/data pair.
- * \ingroup c_crud
- *
- * This function deletes the key/data pair to which the cursor refers. This
- * does not invalidate the cursor, so operations such as \ref MDBX_NEXT can
- * still be used on it. Both \ref MDBX_NEXT and \ref MDBX_GET_CURRENT will
- * return the same record after this operation.
- *
- * \param [in] cursor A cursor handle returned by mdbx_cursor_open().
- * \param [in] flags Options for this operation. This parameter must be set
- * to one of the values described here.
- *
- * - \ref MDBX_CURRENT Delete only single entry at current cursor position.
- * - \ref MDBX_ALLDUPS
- * or \ref MDBX_NODUPDATA (supported for compatibility)
- * Delete all of the data items for the current key. This flag has effect
- * only for database(s) was created with \ref MDBX_DUPSORT.
- *
- * \see \ref c_crud_hints "Quick reference for Insert/Update/Delete operations"
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_MAP_FULL The database is full,
- * see \ref mdbx_env_set_mapsize().
- * \retval MDBX_TXN_FULL The transaction has too many dirty pages.
- * \retval MDBX_EACCES An attempt was made to write in a read-only
- * transaction.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_del(MDBX_cursor *cursor, MDBX_put_flags_t flags);
-
-/** \brief Return count of duplicates for current key.
- * \ingroup c_crud
- *
- * This call is valid for all databases, but reasonable only for that support
- * sorted duplicate data items \ref MDBX_DUPSORT.
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- * \param [out] pcount Address where the count will be stored.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_THREAD_MISMATCH Given transaction is not owned
- * by current thread.
- * \retval MDBX_EINVAL Cursor is not initialized, or an invalid parameter
- * was specified. */
-LIBMDBX_API int mdbx_cursor_count(const MDBX_cursor *cursor, size_t *pcount);
-
-/** \brief Determines whether the cursor is pointed to a key-value pair or not,
- * i.e. was not positioned or points to the end of data.
- * \ingroup c_cursors
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- *
- * \returns A \ref MDBX_RESULT_TRUE or \ref MDBX_RESULT_FALSE value,
- * otherwise the error code:
- * \retval MDBX_RESULT_TRUE No more data available or cursor not
- * positioned
- * \retval MDBX_RESULT_FALSE A data is available
- * \retval Otherwise the error code */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int
-mdbx_cursor_eof(const MDBX_cursor *cursor);
-
-/** \brief Determines whether the cursor is pointed to the first key-value pair
- * or not.
- * \ingroup c_cursors
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- *
- * \returns A MDBX_RESULT_TRUE or MDBX_RESULT_FALSE value,
- * otherwise the error code:
- * \retval MDBX_RESULT_TRUE Cursor positioned to the first key-value pair
- * \retval MDBX_RESULT_FALSE Cursor NOT positioned to the first key-value
- * pair \retval Otherwise the error code */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int
-mdbx_cursor_on_first(const MDBX_cursor *cursor);
-
-/** \brief Determines whether the cursor is pointed to the last key-value pair
- * or not.
- * \ingroup c_cursors
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- *
- * \returns A \ref MDBX_RESULT_TRUE or \ref MDBX_RESULT_FALSE value,
- * otherwise the error code:
- * \retval MDBX_RESULT_TRUE Cursor positioned to the last key-value pair
- * \retval MDBX_RESULT_FALSE Cursor NOT positioned to the last key-value pair
- * \retval Otherwise the error code */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int
-mdbx_cursor_on_last(const MDBX_cursor *cursor);
-
-/** \addtogroup c_rqest
- * \details \note The estimation result varies greatly depending on the filling
- * of specific pages and the overall balance of the b-tree:
- *
- * 1. The number of items is estimated by analyzing the height and fullness of
- * the b-tree. The accuracy of the result directly depends on the balance of
- * the b-tree, which in turn is determined by the history of previous
- * insert/delete operations and the nature of the data (i.e. variability of
- * keys length and so on). Therefore, the accuracy of the estimation can vary
- * greatly in a particular situation.
- *
- * 2. To understand the potential spread of results, you should consider a
- * possible situations basing on the general criteria for splitting and merging
- * b-tree pages:
- * - the page is split into two when there is no space for added data;
- * - two pages merge if the result fits in half a page;
- * - thus, the b-tree can consist of an arbitrary combination of pages filled
- * both completely and only 1/4. Therefore, in the worst case, the result
- * can diverge 4 times for each level of the b-tree excepting the first and
- * the last.
- *
- * 3. In practice, the probability of extreme cases of the above situation is
- * close to zero and in most cases the error does not exceed a few percent. On
- * the other hand, it's just a chance you shouldn't overestimate. */
-
-/** \brief Estimates the distance between cursors as a number of elements.
- * \ingroup c_rqest
- *
- * This function performs a rough estimate based only on b-tree pages that are
- * common for the both cursor's stacks. The results of such estimation can be
- * used to build and/or optimize query execution plans.
- *
- * Please see notes on accuracy of the result in the details
- * of \ref c_rqest section.
- *
- * Both cursors must be initialized for the same database and the same
- * transaction.
- *
- * \param [in] first The first cursor for estimation.
- * \param [in] last The second cursor for estimation.
- * \param [out] distance_items The pointer to store estimated distance value,
- * i.e. `*distance_items = distance(first, last)`.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_estimate_distance(const MDBX_cursor *first,
- const MDBX_cursor *last,
- ptrdiff_t *distance_items);
-
-/** \brief Estimates the move distance.
- * \ingroup c_rqest
- *
- * This function performs a rough estimate distance between the current
- * cursor position and next position after the specified move-operation with
- * given key and data. The results of such estimation can be used to build
- * and/or optimize query execution plans. Current cursor position and state are
- * preserved.
- *
- * Please see notes on accuracy of the result in the details
- * of \ref c_rqest section.
- *
- * \param [in] cursor Cursor for estimation.
- * \param [in,out] key The key for a retrieved item.
- * \param [in,out] data The data of a retrieved item.
- * \param [in] move_op A cursor operation \ref MDBX_cursor_op.
- * \param [out] distance_items A pointer to store estimated move distance
- * as the number of elements.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_estimate_move(const MDBX_cursor *cursor, MDBX_val *key,
- MDBX_val *data, MDBX_cursor_op move_op,
- ptrdiff_t *distance_items);
-
-/** \brief Estimates the size of a range as a number of elements.
- * \ingroup c_rqest
- *
- * The results of such estimation can be used to build and/or optimize query
- * execution plans.
- *
- * Please see notes on accuracy of the result in the details
- * of \ref c_rqest section.
- *
- *
- * \param [in] txn A transaction handle returned
- * by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] begin_key The key of range beginning or NULL for explicit FIRST.
- * \param [in] begin_data Optional additional data to seeking among sorted
- * duplicates.
- * Only for \ref MDBX_DUPSORT, NULL otherwise.
- * \param [in] end_key The key of range ending or NULL for explicit LAST.
- * \param [in] end_data Optional additional data to seeking among sorted
- * duplicates.
- * Only for \ref MDBX_DUPSORT, NULL otherwise.
- * \param [out] distance_items A pointer to store range estimation result.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_estimate_range(MDBX_txn *txn, MDBX_dbi dbi,
- MDBX_val *begin_key, MDBX_val *begin_data,
- MDBX_val *end_key, MDBX_val *end_data,
- ptrdiff_t *distance_items);
-
-/** \brief The EPSILON value for mdbx_estimate_range()
- * \ingroup c_rqest */
-#define MDBX_EPSILON ((MDBX_val *)((ptrdiff_t)-1))
-
-/** \brief Determines whether the given address is on a dirty database page of
- * the transaction or not.
- * \ingroup c_statinfo
- *
- * Ultimately, this allows to avoid copy data from non-dirty pages.
- *
- * "Dirty" pages are those that have already been changed during a write
- * transaction. Accordingly, any further changes may result in such pages being
- * overwritten. Therefore, all functions libmdbx performing changes inside the
- * database as arguments should NOT get pointers to data in those pages. In
- * turn, "not dirty" pages before modification will be copied.
- *
- * In other words, data from dirty pages must either be copied before being
- * passed as arguments for further processing or rejected at the argument
- * validation stage. Thus, `mdbx_is_dirty()` allows you to get rid of
- * unnecessary copying, and perform a more complete check of the arguments.
- *
- * \note The address passed must point to the beginning of the data. This is
- * the only way to ensure that the actual page header is physically located in
- * the same memory page, including for multi-pages with long data.
- *
- * \note In rare cases the function may return a false positive answer
- * (\ref MDBX_RESULT_TRUE when data is NOT on a dirty page), but never a false
- * negative if the arguments are correct.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] ptr The address of data to check.
- *
- * \returns A MDBX_RESULT_TRUE or MDBX_RESULT_FALSE value,
- * otherwise the error code:
- * \retval MDBX_RESULT_TRUE Given address is on the dirty page.
- * \retval MDBX_RESULT_FALSE Given address is NOT on the dirty page.
- * \retval Otherwise the error code. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int mdbx_is_dirty(const MDBX_txn *txn,
- const void *ptr);
-
-/** \brief Sequence generation for a database.
- * \ingroup c_crud
- *
- * The function allows to create a linear sequence of unique positive integers
- * for each database. The function can be called for a read transaction to
- * retrieve the current sequence value, and the increment must be zero.
- * Sequence changes become visible outside the current write transaction after
- * it is committed, and discarded on abort.
- *
- * \param [in] txn A transaction handle returned
- * by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [out] result The optional address where the value of sequence
- * before the change will be stored.
- * \param [in] increment Value to increase the sequence,
- * must be 0 for read-only transactions.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_RESULT_TRUE Increasing the sequence has resulted in an
- * overflow and therefore cannot be executed. */
-LIBMDBX_API int mdbx_dbi_sequence(MDBX_txn *txn, MDBX_dbi dbi, uint64_t *result,
- uint64_t increment);
-
-/** \brief Compare two keys according to a particular database.
- * \ingroup c_crud
- * \see MDBX_cmp_func
- *
- * This returns a comparison as if the two data items were keys in the
- * specified database.
- *
- * \warning There ss a Undefined behavior if one of arguments is invalid.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] a The first item to compare.
- * \param [in] b The second item to compare.
- *
- * \returns < 0 if a < b, 0 if a == b, > 0 if a > b */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int mdbx_cmp(const MDBX_txn *txn,
- MDBX_dbi dbi,
- const MDBX_val *a,
- const MDBX_val *b);
-
-/** \brief Returns default internal key's comparator for given database flags.
- * \ingroup c_extra */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API MDBX_cmp_func *
-mdbx_get_keycmp(MDBX_db_flags_t flags);
-
-/** \brief Compare two data items according to a particular database.
- * \ingroup c_crud
- * \see MDBX_cmp_func
- *
- * This returns a comparison as if the two items were data items of the
- * specified database.
- *
- * \warning There ss a Undefined behavior if one of arguments is invalid.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] a The first item to compare.
- * \param [in] b The second item to compare.
- *
- * \returns < 0 if a < b, 0 if a == b, > 0 if a > b */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API int mdbx_dcmp(const MDBX_txn *txn,
- MDBX_dbi dbi,
- const MDBX_val *a,
- const MDBX_val *b);
-
-/** \brief Returns default internal data's comparator for given database flags
- * \ingroup c_extra */
-MDBX_NOTHROW_CONST_FUNCTION LIBMDBX_API MDBX_cmp_func *
-mdbx_get_datacmp(MDBX_db_flags_t flags);
-
-/** \brief A callback function used to enumerate the reader lock table.
- * \ingroup c_statinfo
- *
- * \param [in] ctx An arbitrary context pointer for the callback.
- * \param [in] num The serial number during enumeration,
- * starting from 1.
- * \param [in] slot The reader lock table slot number.
- * \param [in] txnid The ID of the transaction being read,
- * i.e. the MVCC-snapshot number.
- * \param [in] lag The lag from a recent MVCC-snapshot,
- * i.e. the number of committed write transactions
- * since the current read transaction started.
- * \param [in] pid The reader process ID.
- * \param [in] thread The reader thread ID.
- * \param [in] bytes_used The number of last used page
- * in the MVCC-snapshot which being read,
- * i.e. database file can't shrinked beyond this.
- * \param [in] bytes_retained The total size of the database pages that were
- * retired by committed write transactions after
- * the reader's MVCC-snapshot,
- * i.e. the space which would be freed after
- * the Reader releases the MVCC-snapshot
- * for reuse by completion read transaction.
- *
- * \returns < 0 on failure, >= 0 on success. \see mdbx_reader_list() */
-typedef int(MDBX_reader_list_func)(void *ctx, int num, int slot, mdbx_pid_t pid,
- mdbx_tid_t thread, uint64_t txnid,
- uint64_t lag, size_t bytes_used,
- size_t bytes_retained) MDBX_CXX17_NOEXCEPT;
-
-/** \brief Enumerate the entries in the reader lock table.
- *
- * \ingroup c_statinfo
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] func A \ref MDBX_reader_list_func function.
- * \param [in] ctx An arbitrary context pointer for the enumeration
- * function.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * or \ref MDBX_RESULT_TRUE if the reader lock table is empty. */
-LIBMDBX_API int mdbx_reader_list(const MDBX_env *env,
- MDBX_reader_list_func *func, void *ctx);
-
-/** \brief Check for stale entries in the reader lock table.
- * \ingroup c_extra
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [out] dead Number of stale slots that were cleared.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * or \ref MDBX_RESULT_TRUE if a dead reader(s) found or mutex was recovered. */
-LIBMDBX_API int mdbx_reader_check(MDBX_env *env, int *dead);
-
-/** \brief Returns a lag of the reading for the given transaction.
- * \ingroup c_statinfo
- *
- * Returns an information for estimate how much given read-only
- * transaction is lagging relative the to actual head.
- * \deprecated Please use \ref mdbx_txn_info() instead.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [out] percent Percentage of page allocation in the database.
- *
- * \returns Number of transactions committed after the given was started for
- * read, or negative value on failure. */
-MDBX_DEPRECATED LIBMDBX_API int mdbx_txn_straggler(const MDBX_txn *txn,
- int *percent);
-
-/** \brief Registers the current thread as a reader for the environment.
- * \ingroup c_extra
- *
- * To perform read operations without blocking, a reader slot must be assigned
- * for each thread. However, this assignment requires a short-term lock
- * acquisition which is performed automatically. This function allows you to
- * assign the reader slot in advance and thus avoid capturing the blocker when
- * the read transaction starts firstly from current thread.
- * \see mdbx_thread_unregister()
- *
- * \note Threads are registered automatically the first time a read transaction
- * starts. Therefore, there is no need to use this function, except in
- * special cases.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- *
- * \returns A non-zero error value on failure and 0 on success,
- * or \ref MDBX_RESULT_TRUE if thread is already registered. */
-LIBMDBX_API int mdbx_thread_register(const MDBX_env *env);
-
-/** \brief Unregisters the current thread as a reader for the environment.
- * \ingroup c_extra
- *
- * To perform read operations without blocking, a reader slot must be assigned
- * for each thread. However, the assigned reader slot will remain occupied until
- * the thread ends or the environment closes. This function allows you to
- * explicitly release the assigned reader slot.
- * \see mdbx_thread_register()
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- *
- * \returns A non-zero error value on failure and 0 on success, or
- * \ref MDBX_RESULT_TRUE if thread is not registered or already unregistered. */
-LIBMDBX_API int mdbx_thread_unregister(const MDBX_env *env);
-
-/** \brief A Handle-Slow-Readers callback function to resolve database
- * full/overflow issue due to a reader(s) which prevents the old data from being
- * recycled.
- * \ingroup c_err
- *
- * Read transactions prevent reuse of pages freed by newer write transactions,
- * thus the database can grow quickly. This callback will be called when there
- * is not enough space in the database (i.e. before increasing the database size
- * or before \ref MDBX_MAP_FULL error) and thus can be used to resolve issues
- * with a "long-lived" read transactions.
- * \see mdbx_env_set_hsr()
- * \see mdbx_env_get_hsr()
- * \see <a href="intro.html#long-lived-read">Long-lived read transactions</a>
- *
- * Using this callback you can choose how to resolve the situation:
- * - abort the write transaction with an error;
- * - wait for the read transaction(s) to complete;
- * - notify a thread performing a long-lived read transaction
- * and wait for an effect;
- * - kill the thread or whole process that performs the long-lived read
- * transaction;
- *
- * Depending on the arguments and needs, your implementation may wait,
- * terminate a process or thread that is performing a long read, or perform
- * some other action. In doing so it is important that the returned code always
- * corresponds to the performed action.
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- * \param [in] txn The current write transaction which internally at
- * the \ref MDBX_MAP_FULL condition.
- * \param [in] pid A pid of the reader process.
- * \param [in] tid A thread_id of the reader thread.
- * \param [in] laggard An oldest read transaction number on which stalled.
- * \param [in] gap A lag from the last committed txn.
- * \param [in] space A space that actually become available for reuse after
- * this reader finished. The callback function can take
- * this value into account to evaluate the impact that
- * a long-running transaction has.
- * \param [in] retry A retry number starting from 0.
- * If callback has returned 0 at least once, then at end
- * of current handling loop the callback function will be
- * called additionally with negative value to notify about
- * the end of loop. The callback function can use this value
- * to implement timeout logic while waiting for readers.
- *
- * \returns The RETURN CODE determines the further actions libmdbx and must
- * match the action which was executed by the callback:
- *
- * \retval -2 or less An error condition and the reader was not killed.
- *
- * \retval -1 The callback was unable to solve the problem and
- * agreed on \ref MDBX_MAP_FULL error;
- * libmdbx should increase the database size or
- * return \ref MDBX_MAP_FULL error.
- *
- * \retval 0 (zero) The callback solved the problem or just waited for
- * a while, libmdbx should rescan the reader lock table and
- * retry. This also includes a situation when corresponding
- * transaction terminated in normal way by
- * \ref mdbx_txn_abort() or \ref mdbx_txn_reset(),
- * and my be restarted. I.e. reader slot don't needed
- * to be cleaned from transaction.
- *
- * \retval 1 Transaction aborted asynchronous and reader slot
- * should be cleared immediately, i.e. read transaction
- * will not continue but \ref mdbx_txn_abort()
- * or \ref mdbx_txn_reset() will be called later.
- *
- * \retval 2 or great The reader process was terminated or killed,
- * and libmdbx should entirely reset reader registration.
- */
-typedef int(MDBX_hsr_func)(const MDBX_env *env, const MDBX_txn *txn,
- mdbx_pid_t pid, mdbx_tid_t tid, uint64_t laggard,
- unsigned gap, size_t space,
- int retry) MDBX_CXX17_NOEXCEPT;
-
-/** \brief Sets a Handle-Slow-Readers callback to resolve database full/overflow
- * issue due to a reader(s) which prevents the old data from being recycled.
- * \ingroup c_err
- *
- * The callback will only be triggered when the database is full due to a
- * reader(s) prevents the old data from being recycled.
- *
- * \see MDBX_hsr_func
- * \see mdbx_env_get_hsr()
- * \see <a href="intro.html#long-lived-read">Long-lived read transactions</a>
- *
- * \param [in] env An environment handle returned
- * by \ref mdbx_env_create().
- * \param [in] hsr_callback A \ref MDBX_hsr_func function
- * or NULL to disable.
- *
- * \returns A non-zero error value on failure and 0 on success. */
-LIBMDBX_API int mdbx_env_set_hsr(MDBX_env *env, MDBX_hsr_func *hsr_callback);
-
-/** \brief Gets current Handle-Slow-Readers callback used to resolve database
- * full/overflow issue due to a reader(s) which prevents the old data from being
- * recycled.
- * \see MDBX_hsr_func
- * \see mdbx_env_set_hsr()
- * \see <a href="intro.html#long-lived-read">Long-lived read transactions</a>
- *
- * \param [in] env An environment handle returned by \ref mdbx_env_create().
- *
- * \returns A MDBX_hsr_func function or NULL if disabled
- * or something wrong. */
-MDBX_NOTHROW_PURE_FUNCTION LIBMDBX_API MDBX_hsr_func *
-mdbx_env_get_hsr(const MDBX_env *env);
-
-/** \defgroup btree_traversal B-tree Traversal
- * This is internal API for mdbx_chk tool. You should avoid to use it, except
- * some extremal special cases.
- * \ingroup c_extra
- * @{ */
-
-/** \brief Page types for traverse the b-tree.
- * \see mdbx_env_pgwalk() \see MDBX_pgvisitor_func */
-enum MDBX_page_type_t {
- MDBX_page_broken,
- MDBX_page_meta,
- MDBX_page_large,
- MDBX_page_branch,
- MDBX_page_leaf,
- MDBX_page_dupfixed_leaf,
- MDBX_subpage_leaf,
- MDBX_subpage_dupfixed_leaf,
- MDBX_subpage_broken,
-};
-#ifndef __cplusplus
-typedef enum MDBX_page_type_t MDBX_page_type_t;
-#endif
-
-/** \brief Pseudo-name for MainDB */
-#define MDBX_PGWALK_MAIN ((const char *)((ptrdiff_t)0))
-/** \brief Pseudo-name for GarbageCollectorDB */
-#define MDBX_PGWALK_GC ((const char *)((ptrdiff_t)-1))
-/** \brief Pseudo-name for MetaPages */
-#define MDBX_PGWALK_META ((const char *)((ptrdiff_t)-2))
-
-/** \brief Callback function for traverse the b-tree. \see mdbx_env_pgwalk() */
-typedef int MDBX_pgvisitor_func(
- const uint64_t pgno, const unsigned number, void *const ctx, const int deep,
- const char *const dbi, const size_t page_size, const MDBX_page_type_t type,
- const MDBX_error_t err, const size_t nentries, const size_t payload_bytes,
- const size_t header_bytes, const size_t unused_bytes) MDBX_CXX17_NOEXCEPT;
-
-/** \brief B-tree traversal function. */
-LIBMDBX_API int mdbx_env_pgwalk(MDBX_txn *txn, MDBX_pgvisitor_func *visitor,
- void *ctx, bool dont_check_keys_ordering);
-
-/** \brief Open an environment instance using specific meta-page
- * for checking and recovery.
- *
- * This function mostly of internal API for `mdbx_chk` utility and subject to
- * change at any time. Do not use this function to avoid shooting your own
- * leg(s). */
-LIBMDBX_API int mdbx_env_open_for_recovery(MDBX_env *env, const char *pathname,
- unsigned target_meta,
- bool writeable);
-
-/** \brief Turn database to the specified meta-page.
- *
- * This function mostly of internal API for `mdbx_chk` utility and subject to
- * change at any time. Do not use this function to avoid shooting your own
- * leg(s). */
-LIBMDBX_API int mdbx_env_turn_for_recovery(MDBX_env *env, unsigned target_meta);
-
-/** @} B-tree Traversal */
-
-/**** Attribute support functions for Nexenta (scheduled for removal)
- * *****************************************************************/
-#if defined(MDBX_NEXENTA_ATTRS) || defined(DOXYGEN)
-/** \defgroup nexenta Attribute support functions for Nexenta
- * \ingroup c_crud
- * @{ */
-typedef uint_fast64_t mdbx_attr_t;
-
-/** Store by cursor with attribute.
- *
- * This function stores key/data pairs into the database. The cursor is
- * positioned at the new item, or on failure usually near it.
- *
- * \note Internally based on \ref MDBX_RESERVE feature,
- * therefore doesn't support \ref MDBX_DUPSORT.
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open()
- * \param [in] key The key operated on.
- * \param [in] data The data operated on.
- * \param [in] attr The attribute.
- * \param [in] flags Options for this operation. This parameter must be set
- * to 0 or one of the values described here:
- * - \ref MDBX_CURRENT
- * Replace the item at the current cursor position. The key parameter
- * must still be provided, and must match it, otherwise the function
- * return \ref MDBX_EKEYMISMATCH.
- *
- * - \ref MDBX_APPEND
- * Append the given key/data pair to the end of the database. No key
- * comparisons are performed. This option allows fast bulk loading when
- * keys are already known to be in the correct order. Loading unsorted
- * keys with this flag will cause a \ref MDBX_KEYEXIST error.
- *
- * \see \ref c_crud_hints "Quick reference for Insert/Update/Delete operations"
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_EKEYMISMATCH
- * \retval MDBX_MAP_FULL The database is full, see \ref mdbx_env_set_mapsize().
- * \retval MDBX_TXN_FULL The transaction has too many dirty pages.
- * \retval MDBX_EACCES An attempt was made to write in a read-only
- * transaction.
- * \retval MDBX_EINVAL an invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_put_attr(MDBX_cursor *cursor, MDBX_val *key,
- MDBX_val *data, mdbx_attr_t attr,
- MDBX_put_flags_t flags);
-
-/** Store items and attributes into a database.
- *
- * This function stores key/data pairs in the database. The default behavior
- * is to enter the new key/data pair, replacing any previously existing key
- * if duplicates are disallowed.
- *
- * \note Internally based on \ref MDBX_RESERVE feature,
- * therefore doesn't support \ref MDBX_DUPSORT.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] key The key to store in the database.
- * \param [in] attr The attribute to store in the database.
- * \param [in,out] data The data to store.
- * \param [in] flags Special options for this operation. This parameter
- * must be set to 0 or by bitwise OR'ing together one or
- * more of the values described here:
- * - \ref MDBX_NOOVERWRITE
- * Enter the new key/data pair only if the key does not already appear
- * in the database. The function will return \ref MDBX_KEYEXIST if the key
- * already appears in the database. The data parameter will be set to
- * point to the existing item.
- *
- * - \ref MDBX_CURRENT
- * Update an single existing entry, but not add new ones. The function
- * will return \ref MDBX_NOTFOUND if the given key not exist in the
- * database. Or the \ref MDBX_EMULTIVAL in case duplicates for the given
- * key.
- *
- * - \ref MDBX_APPEND
- * Append the given key/data pair to the end of the database. This option
- * allows fast bulk loading when keys are already known to be in the
- * correct order. Loading unsorted keys with this flag will cause
- * a \ref MDBX_EKEYMISMATCH error.
- *
- * \see \ref c_crud_hints "Quick reference for Insert/Update/Delete operations"
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_KEYEXIST
- * \retval MDBX_MAP_FULL The database is full, see \ref mdbx_env_set_mapsize().
- * \retval MDBX_TXN_FULL The transaction has too many dirty pages.
- * \retval MDBX_EACCES An attempt was made to write
- * in a read-only transaction.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_put_attr(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key,
- MDBX_val *data, mdbx_attr_t attr,
- MDBX_put_flags_t flags);
-
-/** Set items attribute from a database.
- *
- * This function stores key/data pairs attribute to the database.
- *
- * \note Internally based on \ref MDBX_RESERVE feature,
- * therefore doesn't support \ref MDBX_DUPSORT.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] key The key to search for in the database.
- * \param [in] data The data to be stored or NULL to save previous value.
- * \param [in] attr The attribute to be stored.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_NOTFOUND The key-value pair was not in the database.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_set_attr(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key,
- MDBX_val *data, mdbx_attr_t attr);
-
-/** Get items attribute from a database cursor.
- *
- * This function retrieves key/data pairs from the database. The address and
- * length of the key are returned in the object to which key refers (except
- * for the case of the \ref MDBX_SET option, in which the key object is
- * unchanged), and the address and length of the data are returned in the object
- * to which data refers.
- * \see mdbx_get()
- *
- * \param [in] cursor A cursor handle returned by \ref mdbx_cursor_open().
- * \param [in,out] key The key for a retrieved item.
- * \param [in,out] data The data of a retrieved item.
- * \param [out] pattr The pointer to retrieve attribute.
- * \param [in] op A cursor operation MDBX_cursor_op.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_NOTFOUND No matching key found.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_cursor_get_attr(MDBX_cursor *cursor, MDBX_val *key,
- MDBX_val *data, mdbx_attr_t *pattr,
- MDBX_cursor_op op);
-
-/** Get items attribute from a database.
- *
- * This function retrieves key/data pairs from the database. The address
- * and length of the data associated with the specified key are returned
- * in the structure to which data refers.
- * If the database supports duplicate keys (see \ref MDBX_DUPSORT) then the
- * first data item for the key will be returned. Retrieval of other
- * items requires the use of \ref mdbx_cursor_get().
- *
- * \note The memory pointed to by the returned values is owned by the
- * database. The caller need not dispose of the memory, and may not
- * modify it in any way. For values returned in a read-only transaction
- * any modification attempts will cause a `SIGSEGV`.
- *
- * \note Values returned from the database are valid only until a
- * subsequent update operation, or the end of the transaction.
- *
- * \param [in] txn A transaction handle returned by \ref mdbx_txn_begin().
- * \param [in] dbi A database handle returned by \ref mdbx_dbi_open().
- * \param [in] key The key to search for in the database.
- * \param [in,out] data The data corresponding to the key.
- * \param [out] pattr The pointer to retrieve attribute.
- *
- * \returns A non-zero error value on failure and 0 on success,
- * some possible errors are:
- * \retval MDBX_NOTFOUND The key was not in the database.
- * \retval MDBX_EINVAL An invalid parameter was specified. */
-LIBMDBX_API int mdbx_get_attr(MDBX_txn *txn, MDBX_dbi dbi, MDBX_val *key,
- MDBX_val *data, mdbx_attr_t *pattr);
-/** @} end of Attribute support functions for Nexenta */
-#endif /* MDBX_NEXENTA_ATTRS */
-
-/** @} end of C API */
-
-/*******************************************************************************
- * Workaround for mmaped-lookahead-cross-page-boundary bug
- * in an obsolete versions of Elbrus's libc and kernels. */
-#if defined(__e2k__) && defined(MDBX_E2K_MLHCPB_WORKAROUND) && \
- MDBX_E2K_MLHCPB_WORKAROUND
-LIBMDBX_API int mdbx_e2k_memcmp_bug_workaround(const void *s1, const void *s2,
- size_t n);
-LIBMDBX_API int mdbx_e2k_strcmp_bug_workaround(const char *s1, const char *s2);
-LIBMDBX_API int mdbx_e2k_strncmp_bug_workaround(const char *s1, const char *s2,
- size_t n);
-LIBMDBX_API size_t mdbx_e2k_strlen_bug_workaround(const char *s);
-LIBMDBX_API size_t mdbx_e2k_strnlen_bug_workaround(const char *s,
- size_t maxlen);
-#ifdef __cplusplus
-namespace std {
-inline int mdbx_e2k_memcmp_bug_workaround(const void *s1, const void *s2,
- size_t n) {
- return ::mdbx_e2k_memcmp_bug_workaround(s1, s2, n);
-}
-inline int mdbx_e2k_strcmp_bug_workaround(const char *s1, const char *s2) {
- return ::mdbx_e2k_strcmp_bug_workaround(s1, s2);
-}
-inline int mdbx_e2k_strncmp_bug_workaround(const char *s1, const char *s2,
- size_t n) {
- return ::mdbx_e2k_strncmp_bug_workaround(s1, s2, n);
-}
-inline size_t mdbx_e2k_strlen_bug_workaround(const char *s) {
- return ::mdbx_e2k_strlen_bug_workaround(s);
-}
-inline size_t mdbx_e2k_strnlen_bug_workaround(const char *s, size_t maxlen) {
- return ::mdbx_e2k_strnlen_bug_workaround(s, maxlen);
-}
-} // namespace std
-#endif /* __cplusplus */
-
-#include <string.h>
-#include <strings.h>
-#undef memcmp
-#define memcmp mdbx_e2k_memcmp_bug_workaround
-#undef bcmp
-#define bcmp mdbx_e2k_memcmp_bug_workaround
-#undef strcmp
-#define strcmp mdbx_e2k_strcmp_bug_workaround
-#undef strncmp
-#define strncmp mdbx_e2k_strncmp_bug_workaround
-#undef strlen
-#define strlen mdbx_e2k_strlen_bug_workaround
-#undef strnlen
-#define strnlen mdbx_e2k_strnlen_bug_workaround
-#endif /* MDBX_E2K_MLHCPB_WORKAROUND */
-
-#ifdef __cplusplus
-} /* extern "C" */
-#endif
-
-#endif /* LIBMDBX_H */
diff --git a/3rdparty/libmdbx/mdbx.h++ b/3rdparty/libmdbx/mdbx.h++
deleted file mode 100644
index 11d209e6948984d2a9724e65ec9f5008456c06c1..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx.h++
+++ /dev/null
@@ -1,5983 +0,0 @@
-/// \file mdbx.h++
-/// \brief The libmdbx C++ API header file.
-///
-/// \author Copyright (c) 2020-2022, Leonid Yuriev <leo@yuriev.ru>.
-/// \copyright SPDX-License-Identifier: Apache-2.0
-///
-/// Tested with:
-/// - Elbrus LCC >= 1.23 (http://www.mcst.ru/lcc);
-/// - GNU C++ >= 4.8;
-/// - clang >= 3.9;
-/// - MSVC >= 14.0 (Visual Studio 2015),
-/// but 19.2x could hang due optimizer bug;
-/// - AppleClang, but without C++20 concepts.
-///
-
-///
-/// The origin has been migrated to https://gitflic.ru/project/erthink/libmdbx
-/// since on 2022-04-15 the Github administration, without any warning nor
-/// explanation, deleted libmdbx along with a lot of other projects,
-/// simultaneously blocking access for many developers.
-/// For the same reason Github is blacklisted forever.
-///
-
-#pragma once
-
-/* Workaround for modern libstdc++ with CLANG < 4.x */
-#if defined(__SIZEOF_INT128__) && !defined(__GLIBCXX_TYPE_INT_N_0) && \
- defined(__clang__) && __clang_major__ < 4
-#define __GLIBCXX_BITSIZE_INT_N_0 128
-#define __GLIBCXX_TYPE_INT_N_0 __int128
-#endif /* Workaround for modern libstdc++ with CLANG < 4.x */
-
-#if !defined(__cplusplus) || __cplusplus < 201103L
-#if !defined(_MSC_VER) || _MSC_VER < 1900
-#error "C++11 compiler or better is required"
-#elif _MSC_VER >= 1910
-#error \
- "Please add `/Zc:__cplusplus` to MSVC compiler options to enforce it conform ISO C++"
-#endif /* MSVC is mad and don't define __cplusplus properly */
-#endif /* __cplusplus < 201103L */
-
-#if (defined(_WIN32) || defined(_WIN64)) && MDBX_WITHOUT_MSVC_CRT
-#error \
- "CRT is required for C++ API, the MDBX_WITHOUT_MSVC_CRT option must be disabled"
-#endif /* Windows */
-
-#ifndef __has_include
-#define __has_include(header) (0)
-#endif /* __has_include */
-
-#if __has_include(<version>)
-#include <version>
-#endif /* <version> */
-
-/* Disable min/max macros from C' headers */
-#ifndef NOMINMAX
-#define NOMINMAX
-#endif
-
-#include <algorithm> // for std::min/max
-#include <cassert> // for assert()
-#include <climits> // for CHAR_BIT
-#include <cstring> // for std::strlen, str:memcmp
-#include <exception> // for std::exception_ptr
-#include <ostream> // for std::ostream
-#include <sstream> // for std::ostringstream
-#include <stdexcept> // for std::invalid_argument
-#include <string> // for std::string
-#include <type_traits> // for std::is_pod<>, etc.
-#include <utility> // for std::make_pair
-#include <vector> // for std::vector<> as template args
-
-#if defined(__cpp_lib_memory_resource) && __cpp_lib_memory_resource >= 201603L
-#include <memory_resource>
-#endif
-
-#if defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L
-#include <string_view>
-#endif
-
-#if defined(__cpp_lib_filesystem) && __cpp_lib_filesystem >= 201703L
-#include <filesystem>
-#elif __has_include(<experimental/filesystem>)
-#include <experimental/filesystem>
-#endif
-
-#include "mdbx.h"
-
-#if (defined(__cpp_lib_bit_cast) && __cpp_lib_bit_cast >= 201806L) || \
- (defined(__cpp_lib_endian) && __cpp_lib_endian >= 201907L) || \
- (defined(__cpp_lib_bitops) && __cpp_lib_bitops >= 201907L) || \
- (defined(__cpp_lib_int_pow2) && __cpp_lib_int_pow2 >= 202002L)
-#include <bit>
-#elif !(defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && \
- defined(__ORDER_BIG_ENDIAN__))
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-#endif
-#endif
-#endif /* Byte Order */
-
-#if defined(DOXYGEN) || \
- defined(__cpp_constexpr) && __cpp_constexpr >= 201603L && \
- ((defined(_MSC_VER) && _MSC_VER >= 1915) || \
- (defined(__clang__) && __clang_major__ > 5) || \
- (defined(__GNUC__) && __GNUC__ > 7) || \
- (!defined(__GNUC__) && !defined(__clang__) && !defined(_MSC_VER)))
-#define MDBX_CXX17_CONSTEXPR constexpr
-#else
-#define MDBX_CXX17_CONSTEXPR inline
-#endif /* MDBX_CXX17_CONSTEXPR */
-
-#if defined(DOXYGEN) || defined(__cpp_lib_is_constant_evaluated) && \
- __cpp_lib_is_constant_evaluated >= 201811L && \
- defined(__cpp_lib_constexpr_string) && \
- __cpp_lib_constexpr_string >= 201907L
-#define MDBX_CXX20_CONSTEXPR constexpr
-#else
-#define MDBX_CXX20_CONSTEXPR inline
-#endif /* MDBX_CXX20_CONSTEXPR */
-
-#if defined(CONSTEXPR_ASSERT)
-#define MDBX_CONSTEXPR_ASSERT(expr) CONSTEXPR_ASSERT(expr)
-#elif defined NDEBUG
-#define MDBX_CONSTEXPR_ASSERT(expr) void(0)
-#else
-#define MDBX_CONSTEXPR_ASSERT(expr) \
- ((expr) ? void(0) : [] { assert(!#expr); }())
-#endif /* MDBX_CONSTEXPR_ASSERT */
-
-#ifndef MDBX_LIKELY
-#if defined(DOXYGEN) || \
- (defined(__GNUC__) || __has_builtin(__builtin_expect)) && \
- !defined(__COVERITY__)
-#define MDBX_LIKELY(cond) __builtin_expect(!!(cond), 1)
-#else
-#define MDBX_LIKELY(x) (x)
-#endif
-#endif /* MDBX_LIKELY */
-
-#ifndef MDBX_UNLIKELY
-#if defined(DOXYGEN) || \
- (defined(__GNUC__) || __has_builtin(__builtin_expect)) && \
- !defined(__COVERITY__)
-#define MDBX_UNLIKELY(cond) __builtin_expect(!!(cond), 0)
-#else
-#define MDBX_UNLIKELY(x) (x)
-#endif
-#endif /* MDBX_UNLIKELY */
-
-#if defined(__cpp_if_constexpr) && __cpp_if_constexpr >= 201606L
-#define MDBX_IF_CONSTEXPR constexpr
-#else
-#define MDBX_IF_CONSTEXPR
-#endif /* MDBX_IF_CONSTEXPR */
-
-#if defined(DOXYGEN) || \
- (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || \
- __cplusplus >= 201703L
-#define MDBX_CXX17_FALLTHROUGH [[fallthrough]]
-#else
-#define MDBX_CXX17_FALLTHROUGH
-#endif /* MDBX_CXX17_FALLTHROUGH */
-
-#if defined(DOXYGEN) || (__has_cpp_attribute(likely) >= 201803L && \
- (!defined(__GNUC__) || __GNUC__ > 9))
-#define MDBX_CXX20_LIKELY [[likely]]
-#else
-#define MDBX_CXX20_LIKELY
-#endif /* MDBX_CXX20_LIKELY */
-
-#ifndef MDBX_CXX20_UNLIKELY
-#if defined(DOXYGEN) || (__has_cpp_attribute(unlikely) >= 201803L && \
- (!defined(__GNUC__) || __GNUC__ > 9))
-#define MDBX_CXX20_UNLIKELY [[unlikely]]
-#else
-#define MDBX_CXX20_UNLIKELY
-#endif
-#endif /* MDBX_CXX20_UNLIKELY */
-
-#ifndef MDBX_HAVE_CXX20_CONCEPTS
-#if defined(DOXYGEN) || \
- (defined(__cpp_concepts) && __cpp_concepts >= 201907L && \
- (!defined(__clang__) || __has_include(<concepts>) || \
- (defined(__cpp_lib_concepts) && __cpp_lib_concepts >= 202002L)))
-#if __has_include(<concepts>) || \
- (defined(__cpp_lib_concepts) && __cpp_lib_concepts >= 202002L)
-#include <concepts>
-#endif /* <concepts> */
-#define MDBX_HAVE_CXX20_CONCEPTS 1
-#else
-#define MDBX_HAVE_CXX20_CONCEPTS 0
-#endif
-#endif /* MDBX_HAVE_CXX20_CONCEPTS */
-
-#ifndef MDBX_CXX20_CONCEPT
-#if MDBX_HAVE_CXX20_CONCEPTS
-#define MDBX_CXX20_CONCEPT(CONCEPT, NAME) CONCEPT NAME
-#else
-#define MDBX_CXX20_CONCEPT(CONCEPT, NAME) typename NAME
-#endif
-#endif /* MDBX_CXX20_CONCEPT */
-
-#ifndef MDBX_ASSERT_CXX20_CONCEPT_SATISFIED
-#if MDBX_HAVE_CXX20_CONCEPTS
-#define MDBX_ASSERT_CXX20_CONCEPT_SATISFIED(CONCEPT, TYPE) \
- static_assert(CONCEPT<TYPE>)
-#else
-#define MDBX_ASSERT_CXX20_CONCEPT_SATISFIED(CONCEPT, NAME) \
- static_assert(true, MDBX_STRINGIFY(CONCEPT) "<" MDBX_STRINGIFY(TYPE) ">")
-#endif
-#endif /* MDBX_ASSERT_CXX20_CONCEPT_SATISFIED */
-
-#ifdef _MSC_VER
-#pragma warning(push, 4)
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4251) /* 'std::FOO' needs to have dll-interface to \
- be used by clients of 'mdbx::BAR' */
-#pragma warning(disable : 4275) /* non dll-interface 'std::FOO' used as \
- base for dll-interface 'mdbx::BAR' */
-/* MSVC is mad and can generate this warning for its own intermediate
- * automatically generated code, which becomes unreachable after some kinds of
- * optimization (copy elision, etc). */
-#pragma warning(disable : 4702) /* unreachable code */
-#endif /* _MSC_VER (warnings) */
-
-//------------------------------------------------------------------------------
-/// \defgroup cxx_api C++ API
-/// @{
-
-namespace mdbx {
-
-// Functions whose signature depends on the `mdbx::byte` type
-// must be strictly defined as inline!
-#if defined(DOXYGEN) || (defined(__cpp_char8_t) && __cpp_char8_t >= 201811)
-// To enable all kinds of an compiler optimizations we use a byte-like type
-// that don't presumes aliases for pointers as does the `char` type and its
-// derivatives/typedefs.
-// Please see todo4recovery://erased_by_github/libmdbx/issues/263
-// for reasoning of the use of `char8_t` type and switching to `__restrict__`.
-using byte = char8_t;
-#else
-// Avoid `std::byte` since it doesn't add features but inconvenient
-// restrictions.
-using byte = unsigned char;
-#endif /* __cpp_char8_t >= 201811*/
-
-#if defined(__cpp_lib_endian) && __cpp_lib_endian >= 201907L
-using endian = ::std::endian;
-#elif defined(__BYTE_ORDER__) && defined(__ORDER_LITTLE_ENDIAN__) && \
- defined(__ORDER_BIG_ENDIAN__)
-enum class endian {
- little = __ORDER_LITTLE_ENDIAN__,
- big = __ORDER_BIG_ENDIAN__,
- native = __BYTE_ORDER__
-};
-#else
-#error \
- "Please use a C++ compiler provides byte order information or C++20 support"
-#endif /* Byte Order enum */
-
-/// \copydoc MDBX_version_info
-using version_info = ::MDBX_version_info;
-/// \brief Returns libmdbx version information.
-MDBX_CXX11_CONSTEXPR const version_info &get_version() noexcept;
-/// \copydoc MDBX_build_info
-using build_info = ::MDBX_build_info;
-/// \brief Returns libmdbx build information.
-MDBX_CXX11_CONSTEXPR const build_info &get_build() noexcept;
-
-/// \brief constexpr-compatible strlen().
-static MDBX_CXX17_CONSTEXPR size_t strlen(const char *c_str) noexcept;
-
-/// \brief Legacy default allocator
-/// but it is recommended to use \ref polymorphic_allocator.
-using legacy_allocator = ::std::string::allocator_type;
-
-struct slice;
-struct default_capacity_policy;
-template <class ALLOCATOR = legacy_allocator,
- class CAPACITY_POLICY = default_capacity_policy>
-class buffer;
-class env;
-class env_managed;
-class txn;
-class txn_managed;
-class cursor;
-class cursor_managed;
-
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_memory_resource) && \
- __cpp_lib_memory_resource >= 201603L && _GLIBCXX_USE_CXX11_ABI)
-/// \brief Default polymorphic allocator for modern code.
-using polymorphic_allocator = ::std::pmr::string::allocator_type;
-#endif /* __cpp_lib_memory_resource >= 201603L */
-
-/// \brief Default singe-byte string.
-template <class ALLOCATOR = legacy_allocator>
-using string = ::std::basic_string<char, ::std::char_traits<char>, ALLOCATOR>;
-
-using filehandle = ::mdbx_filehandle_t;
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_filesystem) && __cpp_lib_filesystem >= 201703L && \
- (!defined(__MAC_OS_X_VERSION_MIN_REQUIRED) || \
- __MAC_OS_X_VERSION_MIN_REQUIRED >= 101500) && \
- (!defined(__IPHONE_OS_VERSION_MIN_REQUIRED) || \
- __IPHONE_OS_VERSION_MIN_REQUIRED >= 130100))
-namespace filesystem = ::std::filesystem;
-#define MDBX_STD_FILESYSTEM_PATH ::mdbx::filesystem::path
-#elif defined(__cpp_lib_experimental_filesystem) && \
- __cpp_lib_experimental_filesystem >= 201406L
-namespace filesystem = ::std::experimental::filesystem;
-#define MDBX_STD_FILESYSTEM_PATH ::mdbx::filesystem::path
-#endif /* MDBX_STD_FILESYSTEM_PATH */
-
-#ifdef MDBX_STD_FILESYSTEM_PATH
-using path = MDBX_STD_FILESYSTEM_PATH;
-#elif defined(_WIN32) || defined(_WIN64)
-using path = ::std::wstring;
-#else
-using path = ::std::string;
-#endif /* mdbx::path */
-
-/// \brief Transfers C++ exceptions thru C callbacks.
-/// \details Implements saving exceptions before returning
-/// from an C++'s environment to the intermediate C code and re-throwing after
-/// returning from C to the C++'s environment.
-class LIBMDBX_API_TYPE exception_thunk {
- ::std::exception_ptr captured_;
-
-public:
- exception_thunk() noexcept = default;
- exception_thunk(const exception_thunk &) = delete;
- exception_thunk(exception_thunk &&) = delete;
- exception_thunk &operator=(const exception_thunk &) = delete;
- exception_thunk &operator=(exception_thunk &&) = delete;
- inline bool is_clean() const noexcept;
- inline void capture() noexcept;
- inline void rethrow_captured() const;
-};
-
-/// \brief Implements error information and throwing corresponding exceptions.
-class LIBMDBX_API_TYPE error {
- MDBX_error_t code_;
- inline error &operator=(MDBX_error_t error_code) noexcept;
-
-public:
- MDBX_CXX11_CONSTEXPR error(MDBX_error_t error_code) noexcept;
- error(const error &) = default;
- error(error &&) = default;
- error &operator=(const error &) = default;
- error &operator=(error &&) = default;
-
- MDBX_CXX11_CONSTEXPR friend bool operator==(const error &a,
- const error &b) noexcept;
- MDBX_CXX11_CONSTEXPR friend bool operator!=(const error &a,
- const error &b) noexcept;
-
- MDBX_CXX11_CONSTEXPR bool is_success() const noexcept;
- MDBX_CXX11_CONSTEXPR bool is_result_true() const noexcept;
- MDBX_CXX11_CONSTEXPR bool is_result_false() const noexcept;
- MDBX_CXX11_CONSTEXPR bool is_failure() const noexcept;
-
- /// \brief Returns error code.
- MDBX_CXX11_CONSTEXPR MDBX_error_t code() const noexcept;
-
- /// \brief Returns message for MDBX's errors only and "SYSTEM" for others.
- const char *what() const noexcept;
-
- /// \brief Returns message for any errors.
- ::std::string message() const;
-
- /// \brief Returns true for MDBX's errors.
- MDBX_CXX11_CONSTEXPR bool is_mdbx_error() const noexcept;
- /// \brief Panics on unrecoverable errors inside destructors etc.
- [[noreturn]] void panic(const char *context_where_when,
- const char *func_who_what) const noexcept;
- [[noreturn]] void throw_exception() const;
- [[noreturn]] static inline void throw_exception(int error_code);
- inline void throw_on_failure() const;
- inline void success_or_throw() const;
- inline void success_or_throw(const exception_thunk &) const;
- inline void panic_on_failure(const char *context_where,
- const char *func_who) const noexcept;
- inline void success_or_panic(const char *context_where,
- const char *func_who) const noexcept;
- static inline void throw_on_nullptr(const void *ptr, MDBX_error_t error_code);
- static inline void success_or_throw(MDBX_error_t error_code);
- static void success_or_throw(int error_code) {
- success_or_throw(static_cast<MDBX_error_t>(error_code));
- }
- static inline void throw_on_failure(int error_code);
- static inline bool boolean_or_throw(int error_code);
- static inline void success_or_throw(int error_code, const exception_thunk &);
- static inline void panic_on_failure(int error_code, const char *context_where,
- const char *func_who) noexcept;
- static inline void success_or_panic(int error_code, const char *context_where,
- const char *func_who) noexcept;
-};
-
-/// \brief Base class for all libmdbx's exceptions that are corresponds
-/// to libmdbx errors.
-///
-/// \see MDBX_error_t
-class LIBMDBX_API_TYPE exception : public ::std::runtime_error {
- using base = ::std::runtime_error;
- ::mdbx::error error_;
-
-public:
- exception(const ::mdbx::error &) noexcept;
- exception(const exception &) = default;
- exception(exception &&) = default;
- exception &operator=(const exception &) = default;
- exception &operator=(exception &&) = default;
- virtual ~exception() noexcept;
- const ::mdbx::error error() const noexcept { return error_; }
-};
-
-/// \brief Fatal exception that lead termination anyway
-/// in dangerous unrecoverable cases.
-class LIBMDBX_API_TYPE fatal : public exception {
- using base = exception;
-
-public:
- fatal(const ::mdbx::error &) noexcept;
- fatal(const exception &src) noexcept : fatal(src.error()) {}
- fatal(exception &&src) noexcept : fatal(src.error()) {}
- fatal(const fatal &src) noexcept : fatal(src.error()) {}
- fatal(fatal &&src) noexcept : fatal(src.error()) {}
- fatal &operator=(fatal &&) = default;
- fatal &operator=(const fatal &) = default;
- virtual ~fatal() noexcept;
-};
-
-#define MDBX_DECLARE_EXCEPTION(NAME) \
- struct LIBMDBX_API_TYPE NAME : public exception { \
- NAME(const ::mdbx::error &); \
- virtual ~NAME() noexcept; \
- }
-MDBX_DECLARE_EXCEPTION(bad_map_id);
-MDBX_DECLARE_EXCEPTION(bad_transaction);
-MDBX_DECLARE_EXCEPTION(bad_value_size);
-MDBX_DECLARE_EXCEPTION(db_corrupted);
-MDBX_DECLARE_EXCEPTION(db_full);
-MDBX_DECLARE_EXCEPTION(db_invalid);
-MDBX_DECLARE_EXCEPTION(db_too_large);
-MDBX_DECLARE_EXCEPTION(db_unable_extend);
-MDBX_DECLARE_EXCEPTION(db_version_mismatch);
-MDBX_DECLARE_EXCEPTION(db_wanna_write_for_recovery);
-MDBX_DECLARE_EXCEPTION(incompatible_operation);
-MDBX_DECLARE_EXCEPTION(internal_page_full);
-MDBX_DECLARE_EXCEPTION(internal_problem);
-MDBX_DECLARE_EXCEPTION(key_exists);
-MDBX_DECLARE_EXCEPTION(key_mismatch);
-MDBX_DECLARE_EXCEPTION(max_maps_reached);
-MDBX_DECLARE_EXCEPTION(max_readers_reached);
-MDBX_DECLARE_EXCEPTION(multivalue);
-MDBX_DECLARE_EXCEPTION(no_data);
-MDBX_DECLARE_EXCEPTION(not_found);
-MDBX_DECLARE_EXCEPTION(operation_not_permitted);
-MDBX_DECLARE_EXCEPTION(permission_denied_or_not_writeable);
-MDBX_DECLARE_EXCEPTION(reader_slot_busy);
-MDBX_DECLARE_EXCEPTION(remote_media);
-MDBX_DECLARE_EXCEPTION(something_busy);
-MDBX_DECLARE_EXCEPTION(thread_mismatch);
-MDBX_DECLARE_EXCEPTION(transaction_full);
-MDBX_DECLARE_EXCEPTION(transaction_overlapping);
-#undef MDBX_DECLARE_EXCEPTION
-
-[[noreturn]] LIBMDBX_API void throw_too_small_target_buffer();
-[[noreturn]] LIBMDBX_API void throw_max_length_exceeded();
-[[noreturn]] LIBMDBX_API void throw_out_range();
-[[noreturn]] LIBMDBX_API void throw_allocators_mismatch();
-static MDBX_CXX14_CONSTEXPR size_t check_length(size_t bytes);
-static MDBX_CXX14_CONSTEXPR size_t check_length(size_t headroom,
- size_t payload);
-static MDBX_CXX14_CONSTEXPR size_t check_length(size_t headroom, size_t payload,
- size_t tailroom);
-static MDBX_CXX17_CONSTEXPR size_t strlen(const char *c_str) noexcept;
-static MDBX_CXX20_CONSTEXPR void *memcpy(void *dest, const void *src,
- size_t bytes) noexcept;
-//------------------------------------------------------------------------------
-
-#if MDBX_HAVE_CXX20_CONCEPTS
-
-template <typename T>
-concept MutableByteProducer = requires(T a, char array[42]) {
- { a.is_empty() } -> std::same_as<bool>;
- { a.envisage_result_length() } -> std::same_as<size_t>;
- { a.write_bytes(&array[0], size_t(42)) } -> std::same_as<char *>;
-};
-
-template <typename T>
-concept ImmutableByteProducer = requires(const T &a, char array[42]) {
- { a.is_empty() } -> std::same_as<bool>;
- { a.envisage_result_length() } -> std::same_as<size_t>;
- { a.write_bytes(&array[0], size_t(42)) } -> std::same_as<char *>;
-};
-
-template <typename T>
-concept SliceTranscoder = ImmutableByteProducer<T> &&
- requires(const slice &source, const T &a) {
- T(source);
- { a.is_erroneous() } -> std::same_as<bool>;
-};
-
-#endif /* MDBX_HAVE_CXX20_CONCEPTS */
-
-template <class ALLOCATOR = legacy_allocator,
- typename CAPACITY_POLICY = default_capacity_policy,
- MDBX_CXX20_CONCEPT(MutableByteProducer, PRODUCER)>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-make_buffer(PRODUCER &producer, const ALLOCATOR &allocator = ALLOCATOR());
-
-template <class ALLOCATOR = legacy_allocator,
- typename CAPACITY_POLICY = default_capacity_policy,
- MDBX_CXX20_CONCEPT(ImmutableByteProducer, PRODUCER)>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-make_buffer(const PRODUCER &producer, const ALLOCATOR &allocator = ALLOCATOR());
-
-template <class ALLOCATOR = legacy_allocator,
- MDBX_CXX20_CONCEPT(MutableByteProducer, PRODUCER)>
-inline string<ALLOCATOR> make_string(PRODUCER &producer,
- const ALLOCATOR &allocator = ALLOCATOR());
-
-template <class ALLOCATOR = legacy_allocator,
- MDBX_CXX20_CONCEPT(ImmutableByteProducer, PRODUCER)>
-inline string<ALLOCATOR> make_string(const PRODUCER &producer,
- const ALLOCATOR &allocator = ALLOCATOR());
-
-/// \brief References a data located outside the slice.
-///
-/// The `slice` is similar in many ways to `std::string_view`, but it
-/// implements specific capabilities and manipulates with bytes but
-/// not a characters.
-///
-/// \copydetails MDBX_val
-struct LIBMDBX_API_TYPE slice : public ::MDBX_val {
- /// \todo slice& operator<<(slice&, ...) for reading
- /// \todo key-to-value (parse/unpack) functions
- /// \todo template<class X> key(X); for decoding keys while reading
-
- enum { max_length = MDBX_MAXDATASIZE };
-
- /// \brief Create an empty slice.
- MDBX_CXX11_CONSTEXPR slice() noexcept;
-
- /// \brief Create a slice that refers to [0,bytes-1] of memory bytes pointed
- /// by ptr.
- MDBX_CXX14_CONSTEXPR slice(const void *ptr, size_t bytes);
-
- /// \brief Create a slice that refers to [begin,end] of memory bytes.
- MDBX_CXX14_CONSTEXPR slice(const void *begin, const void *end);
-
- /// \brief Create a slice that refers to text[0,strlen(text)-1].
- template <size_t SIZE>
- MDBX_CXX14_CONSTEXPR slice(const char (&text)[SIZE]) : slice(text, SIZE - 1) {
- MDBX_CONSTEXPR_ASSERT(SIZE > 0 && text[SIZE - 1] == '\0');
- }
- /// \brief Create a slice that refers to c_str[0,strlen(c_str)-1].
- explicit MDBX_CXX17_CONSTEXPR slice(const char *c_str);
-
- /// \brief Create a slice that refers to the contents of "str".
- /// \note 'explicit' to avoid reference to the temporary std::string instance.
- template <class CHAR, class T, class A>
- explicit MDBX_CXX20_CONSTEXPR
- slice(const ::std::basic_string<CHAR, T, A> &str)
- : slice(str.data(), str.length() * sizeof(CHAR)) {}
-
- MDBX_CXX14_CONSTEXPR slice(const MDBX_val &src);
- MDBX_CXX11_CONSTEXPR slice(const slice &) noexcept = default;
- MDBX_CXX14_CONSTEXPR slice(MDBX_val &&src);
- MDBX_CXX14_CONSTEXPR slice(slice &&src) noexcept;
-
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L)
- /// \brief Create a slice that refers to the same contents as "string_view"
- template <class CHAR, class T>
- MDBX_CXX14_CONSTEXPR slice(const ::std::basic_string_view<CHAR, T> &sv)
- : slice(sv.data(), sv.data() + sv.length()) {}
-
- template <class CHAR, class T>
- slice(::std::basic_string_view<CHAR, T> &&sv) : slice(sv) {
- sv = {};
- }
-#endif /* __cpp_lib_string_view >= 201606L */
-
- template <size_t SIZE>
- static MDBX_CXX14_CONSTEXPR slice wrap(const char (&text)[SIZE]) {
- return slice(text);
- }
-
- template <typename POD>
- MDBX_CXX14_CONSTEXPR static slice wrap(const POD &pod) {
- static_assert(::std::is_standard_layout<POD>::value &&
- !::std::is_pointer<POD>::value,
- "Must be a standard layout type!");
- return slice(&pod, sizeof(pod));
- }
-
- inline slice &assign(const void *ptr, size_t bytes);
- inline slice &assign(const slice &src) noexcept;
- inline slice &assign(const ::MDBX_val &src);
- inline slice &assign(slice &&src) noexcept;
- inline slice &assign(::MDBX_val &&src);
- inline slice &assign(const void *begin, const void *end);
- template <class CHAR, class T, class ALLOCATOR>
- slice &assign(const ::std::basic_string<CHAR, T, ALLOCATOR> &str) {
- return assign(str.data(), str.length() * sizeof(CHAR));
- }
- inline slice &assign(const char *c_str);
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L)
- template <class CHAR, class T>
- slice &assign(const ::std::basic_string_view<CHAR, T> &view) {
- return assign(view.begin(), view.end());
- }
- template <class CHAR, class T>
- slice &assign(::std::basic_string_view<CHAR, T> &&view) {
- assign(view);
- view = {};
- return *this;
- }
-#endif /* __cpp_lib_string_view >= 201606L */
-
- slice &operator=(const slice &) noexcept = default;
- inline slice &operator=(slice &&src) noexcept;
- inline slice &operator=(::MDBX_val &&src);
-
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L)
- template <class CHAR, class T>
- slice &operator=(const ::std::basic_string_view<CHAR, T> &view) {
- return assign(view);
- }
-
- template <class CHAR, class T>
- slice &operator=(::std::basic_string_view<CHAR, T> &&view) {
- return assign(view);
- }
-
- /// \brief Return a string_view that references the same data as this slice.
- template <class CHAR = char, class T = ::std::char_traits<CHAR>>
- MDBX_CXX11_CONSTEXPR ::std::basic_string_view<CHAR, T>
- string_view() const noexcept {
- static_assert(sizeof(CHAR) == 1, "Must be single byte characters");
- return ::std::basic_string_view<CHAR, T>(char_ptr(), length());
- }
-
- /// \brief Return a string_view that references the same data as this slice.
- template <class CHAR, class T>
- MDBX_CXX11_CONSTEXPR explicit
- operator ::std::basic_string_view<CHAR, T>() const noexcept {
- return this->string_view<CHAR, T>();
- }
-#endif /* __cpp_lib_string_view >= 201606L */
-
- template <class CHAR = char, class T = ::std::char_traits<CHAR>,
- class ALLOCATOR = legacy_allocator>
- MDBX_CXX20_CONSTEXPR ::std::basic_string<CHAR, T, ALLOCATOR>
- as_string(const ALLOCATOR &allocator = ALLOCATOR()) const {
- static_assert(sizeof(CHAR) == 1, "Must be single byte characters");
- return ::std::basic_string<CHAR, T, ALLOCATOR>(char_ptr(), length(),
- allocator);
- }
-
- template <class CHAR, class T, class ALLOCATOR>
- MDBX_CXX20_CONSTEXPR explicit
- operator ::std::basic_string<CHAR, T, ALLOCATOR>() const {
- return as_string<CHAR, T, ALLOCATOR>();
- }
-
- /// \brief Returns a string with a hexadecimal dump of the slice content.
- template <class ALLOCATOR = legacy_allocator>
- inline string<ALLOCATOR>
- as_hex_string(bool uppercase = false, unsigned wrap_width = 0,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Returns a string with a
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump of the slice content.
- template <class ALLOCATOR = legacy_allocator>
- inline string<ALLOCATOR>
- as_base58_string(unsigned wrap_width = 0,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Returns a string with a
- /// [Base58](https://en.wikipedia.org/wiki/Base64) dump of the slice content.
- template <class ALLOCATOR = legacy_allocator>
- inline string<ALLOCATOR>
- as_base64_string(unsigned wrap_width = 0,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Returns a buffer with a hexadecimal dump of the slice content.
- template <class ALLOCATOR = legacy_allocator,
- class CAPACITY_POLICY = default_capacity_policy>
- inline buffer<ALLOCATOR, CAPACITY_POLICY>
- encode_hex(bool uppercase = false, unsigned wrap_width = 0,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Returns a buffer with a
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump of the slice content.
- template <class ALLOCATOR = legacy_allocator,
- class CAPACITY_POLICY = default_capacity_policy>
- inline buffer<ALLOCATOR, CAPACITY_POLICY>
- encode_base58(unsigned wrap_width = 0,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Returns a buffer with a
- /// [Base64](https://en.wikipedia.org/wiki/Base64) dump of the slice content.
- template <class ALLOCATOR = legacy_allocator,
- class CAPACITY_POLICY = default_capacity_policy>
- inline buffer<ALLOCATOR, CAPACITY_POLICY>
- encode_base64(unsigned wrap_width = 0,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Decodes hexadecimal dump from the slice content to returned buffer.
- template <class ALLOCATOR = legacy_allocator,
- class CAPACITY_POLICY = default_capacity_policy>
- inline buffer<ALLOCATOR, CAPACITY_POLICY>
- hex_decode(bool ignore_spaces = false,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Decodes [Base58](https://en.wikipedia.org/wiki/Base58) dump
- /// from the slice content to returned buffer.
- template <class ALLOCATOR = legacy_allocator,
- class CAPACITY_POLICY = default_capacity_policy>
- inline buffer<ALLOCATOR, CAPACITY_POLICY>
- base58_decode(bool ignore_spaces = false,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Decodes [Base64](https://en.wikipedia.org/wiki/Base64) dump
- /// from the slice content to returned buffer.
- template <class ALLOCATOR = legacy_allocator,
- class CAPACITY_POLICY = default_capacity_policy>
- inline buffer<ALLOCATOR, CAPACITY_POLICY>
- base64_decode(bool ignore_spaces = false,
- const ALLOCATOR &allocator = ALLOCATOR()) const;
-
- /// \brief Checks whether the content of the slice is printable.
- /// \param [in] disable_utf8 By default if `disable_utf8` is `false` function
- /// checks that content bytes are printable ASCII-7 characters or a valid UTF8
- /// sequences. Otherwise, if if `disable_utf8` is `true` function checks that
- /// content bytes are printable extended 8-bit ASCII codes.
- MDBX_NOTHROW_PURE_FUNCTION bool
- is_printable(bool disable_utf8 = false) const noexcept;
-
- /// \brief Checks whether the content of the slice is a hexadecimal dump.
- /// \param [in] ignore_spaces If `true` function will skips spaces surrounding
- /// (before, between and after) a encoded bytes. However, spaces should not
- /// break a pair of characters encoding a single byte.
- inline MDBX_NOTHROW_PURE_FUNCTION bool
- is_hex(bool ignore_spaces = false) const noexcept;
-
- /// \brief Checks whether the content of the slice is a
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump.
- /// \param [in] ignore_spaces If `true` function will skips spaces surrounding
- /// (before, between and after) a encoded bytes. However, spaces should not
- /// break a code group of characters.
- inline MDBX_NOTHROW_PURE_FUNCTION bool
- is_base58(bool ignore_spaces = false) const noexcept;
-
- /// \brief Checks whether the content of the slice is a
- /// [Base64](https://en.wikipedia.org/wiki/Base64) dump.
- /// \param [in] ignore_spaces If `true` function will skips spaces surrounding
- /// (before, between and after) a encoded bytes. However, spaces should not
- /// break a code group of characters.
- inline MDBX_NOTHROW_PURE_FUNCTION bool
- is_base64(bool ignore_spaces = false) const noexcept;
-
- inline void swap(slice &other) noexcept;
-
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L)
- template <class CHAR, class T>
- void swap(::std::basic_string_view<CHAR, T> &view) noexcept {
- static_assert(sizeof(CHAR) == 1, "Must be single byte characters");
- const auto temp = ::std::basic_string_view<CHAR, T>(*this);
- *this = view;
- view = temp;
- }
-#endif /* __cpp_lib_string_view >= 201606L */
-
- /// \brief Returns casted to pointer to byte an address of data.
- MDBX_CXX11_CONSTEXPR const byte *byte_ptr() const noexcept;
- MDBX_CXX11_CONSTEXPR byte *byte_ptr() noexcept;
-
- /// \brief Returns casted to pointer to byte an end of data.
- MDBX_CXX11_CONSTEXPR const byte *end_byte_ptr() const noexcept;
- MDBX_CXX11_CONSTEXPR byte *end_byte_ptr() noexcept;
-
- /// \brief Returns casted to pointer to char an address of data.
- MDBX_CXX11_CONSTEXPR const char *char_ptr() const noexcept;
- MDBX_CXX11_CONSTEXPR char *char_ptr() noexcept;
-
- /// \brief Returns casted to pointer to char an end of data.
- MDBX_CXX11_CONSTEXPR const char *end_char_ptr() const noexcept;
- MDBX_CXX11_CONSTEXPR char *end_char_ptr() noexcept;
-
- /// \brief Return a pointer to the beginning of the referenced data.
- MDBX_CXX11_CONSTEXPR const void *data() const noexcept;
- MDBX_CXX11_CONSTEXPR void *data() noexcept;
-
- /// \brief Return a pointer to the ending of the referenced data.
- MDBX_CXX11_CONSTEXPR const void *end() const noexcept;
- MDBX_CXX11_CONSTEXPR void *end() noexcept;
-
- /// \brief Returns the number of bytes.
- MDBX_CXX11_CONSTEXPR size_t length() const noexcept;
-
- /// \brief Set slice length.
- MDBX_CXX14_CONSTEXPR slice &set_length(size_t bytes);
-
- /// \brief Sets the length by specifying the end of the slice data.
- MDBX_CXX14_CONSTEXPR slice &set_end(const void *ptr);
-
- /// \brief Checks whether the slice is empty.
- MDBX_CXX11_CONSTEXPR bool empty() const noexcept;
-
- /// \brief Checks whether the slice data pointer is nullptr.
- MDBX_CXX11_CONSTEXPR bool is_null() const noexcept;
-
- /// \brief Returns the number of bytes.
- MDBX_CXX11_CONSTEXPR size_t size() const noexcept;
-
- /// \brief Returns true if slice is not empty.
- MDBX_CXX11_CONSTEXPR operator bool() const noexcept;
-
- /// \brief Depletes content of slice and make it invalid.
- MDBX_CXX14_CONSTEXPR void invalidate() noexcept;
-
- /// \brief Makes the slice empty and referencing to nothing.
- MDBX_CXX14_CONSTEXPR void clear() noexcept;
-
- /// \brief Drops the first "n" bytes from this slice.
- /// \pre REQUIRES: `n <= size()`
- inline void remove_prefix(size_t n) noexcept;
-
- /// \brief Drops the last "n" bytes from this slice.
- /// \pre REQUIRES: `n <= size()`
- inline void remove_suffix(size_t n) noexcept;
-
- /// \brief Drops the first "n" bytes from this slice.
- /// \throws std::out_of_range if `n > size()`
- inline void safe_remove_prefix(size_t n);
-
- /// \brief Drops the last "n" bytes from this slice.
- /// \throws std::out_of_range if `n > size()`
- inline void safe_remove_suffix(size_t n);
-
- /// \brief Checks if the data starts with the given prefix.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR bool
- starts_with(const slice &prefix) const noexcept;
-
- /// \brief Checks if the data ends with the given suffix.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR bool
- ends_with(const slice &suffix) const noexcept;
-
- /// \brief Returns the nth byte in the referenced data.
- /// \pre REQUIRES: `n < size()`
- MDBX_CXX11_CONSTEXPR byte operator[](size_t n) const noexcept;
-
- /// \brief Returns the nth byte in the referenced data with bounds checking.
- /// \throws std::out_of_range if `n >= size()`
- MDBX_CXX11_CONSTEXPR byte at(size_t n) const;
-
- /// \brief Returns the first "n" bytes of the slice.
- /// \pre REQUIRES: `n <= size()`
- MDBX_CXX14_CONSTEXPR slice head(size_t n) const noexcept;
-
- /// \brief Returns the last "n" bytes of the slice.
- /// \pre REQUIRES: `n <= size()`
- MDBX_CXX14_CONSTEXPR slice tail(size_t n) const noexcept;
-
- /// \brief Returns the middle "n" bytes of the slice.
- /// \pre REQUIRES: `from + n <= size()`
- MDBX_CXX14_CONSTEXPR slice middle(size_t from, size_t n) const noexcept;
-
- /// \brief Returns the first "n" bytes of the slice.
- /// \throws std::out_of_range if `n >= size()`
- MDBX_CXX14_CONSTEXPR slice safe_head(size_t n) const;
-
- /// \brief Returns the last "n" bytes of the slice.
- /// \throws std::out_of_range if `n >= size()`
- MDBX_CXX14_CONSTEXPR slice safe_tail(size_t n) const;
-
- /// \brief Returns the middle "n" bytes of the slice.
- /// \throws std::out_of_range if `from + n >= size()`
- MDBX_CXX14_CONSTEXPR slice safe_middle(size_t from, size_t n) const;
-
- /// \brief Returns the hash value of referenced data.
- /// \attention Function implementation and returned hash values may changed
- /// version to version, and in future the t1ha3 will be used here. Therefore
- /// values obtained from this function shouldn't be persisted anywhere.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR size_t
- hash_value() const noexcept;
-
- /// \brief Three-way fast non-lexicographically length-based comparison.
- /// \return value:
- /// == 0 if "a" == "b",
- /// < 0 if "a" shorter than "b",
- /// > 0 if "a" longer than "b",
- /// < 0 if "a" length-equal and lexicographically less than "b",
- /// > 0 if "a" length-equal and lexicographically great than "b".
- MDBX_NOTHROW_PURE_FUNCTION static MDBX_CXX14_CONSTEXPR intptr_t
- compare_fast(const slice &a, const slice &b) noexcept;
-
- /// \brief Three-way lexicographically comparison.
- /// \return value:
- /// < 0 if "a" < "b",
- /// == 0 if "a" == "b",
- /// > 0 if "a" > "b".
- MDBX_NOTHROW_PURE_FUNCTION static MDBX_CXX14_CONSTEXPR intptr_t
- compare_lexicographically(const slice &a, const slice &b) noexcept;
- friend MDBX_CXX14_CONSTEXPR bool operator==(const slice &a,
- const slice &b) noexcept;
- friend MDBX_CXX14_CONSTEXPR bool operator<(const slice &a,
- const slice &b) noexcept;
- friend MDBX_CXX14_CONSTEXPR bool operator>(const slice &a,
- const slice &b) noexcept;
- friend MDBX_CXX14_CONSTEXPR bool operator<=(const slice &a,
- const slice &b) noexcept;
- friend MDBX_CXX14_CONSTEXPR bool operator>=(const slice &a,
- const slice &b) noexcept;
- friend MDBX_CXX14_CONSTEXPR bool operator!=(const slice &a,
- const slice &b) noexcept;
-
- /// \brief Checks the slice is not refers to null address or has zero length.
- MDBX_CXX11_CONSTEXPR bool is_valid() const noexcept {
- return !(iov_base == nullptr && iov_len != 0);
- }
-
- /// \brief Build an invalid slice which non-zero length and refers to null
- /// address.
- MDBX_CXX14_CONSTEXPR static slice invalid() noexcept {
- return slice(size_t(-1));
- }
-
-protected:
- MDBX_CXX11_CONSTEXPR slice(size_t invalid_length) noexcept
- : ::MDBX_val({nullptr, invalid_length}) {}
-};
-
-//------------------------------------------------------------------------------
-
-namespace allocation_aware_details {
-
-template <typename A> constexpr bool allocator_is_always_equal() noexcept {
-#if defined(__cpp_lib_allocator_traits_is_always_equal) && \
- __cpp_lib_allocator_traits_is_always_equal >= 201411L
- return ::std::allocator_traits<A>::is_always_equal::value;
-#else
- return ::std::is_empty<A>::value;
-#endif /* __cpp_lib_allocator_traits_is_always_equal */
-}
-
-template <typename T, typename A = typename T::allocator_type,
- bool PoCMA = ::std::allocator_traits<
- A>::propagate_on_container_move_assignment::value>
-struct move_assign_alloc;
-
-template <typename T, typename A> struct move_assign_alloc<T, A, false> {
- static constexpr bool is_nothrow() noexcept {
- return allocator_is_always_equal<A>();
- }
- static MDBX_CXX20_CONSTEXPR bool is_moveable(T *target, T &source) noexcept {
- return allocator_is_always_equal<A>() ||
- target->get_allocator() == source.get_allocator();
- }
- static MDBX_CXX20_CONSTEXPR void propagate(T *target, T &source) noexcept {
- assert(target->get_allocator() != source.get_allocator());
- (void)target;
- (void)source;
- }
-};
-
-template <typename T, typename A> struct move_assign_alloc<T, A, true> {
- static constexpr bool is_nothrow() noexcept {
- return allocator_is_always_equal<A>() ||
- ::std::is_nothrow_move_assignable<A>::value;
- }
- static constexpr bool is_moveable(T *, T &) noexcept { return true; }
- static MDBX_CXX20_CONSTEXPR void propagate(T *target, T &source) {
- assert(target->get_allocator() != source.get_allocator());
- target->get_allocator() = ::std::move(source.get_allocator());
- }
-};
-
-template <typename T, typename A = typename T::allocator_type,
- bool PoCCA = ::std::allocator_traits<
- A>::propagate_on_container_copy_assignment::value>
-struct copy_assign_alloc;
-
-template <typename T, typename A> struct copy_assign_alloc<T, A, false> {
- static constexpr bool is_nothrow() noexcept { return false; }
- static MDBX_CXX20_CONSTEXPR void propagate(T *target,
- const T &source) noexcept {
- assert(target->get_allocator() != source.get_allocator());
- (void)target;
- (void)source;
- }
-};
-
-template <typename T, typename A> struct copy_assign_alloc<T, A, true> {
- static constexpr bool is_nothrow() noexcept {
- return allocator_is_always_equal<A>() ||
- ::std::is_nothrow_copy_assignable<A>::value;
- }
- static MDBX_CXX20_CONSTEXPR void
- propagate(T *target, const T &source) noexcept(is_nothrow()) {
- if MDBX_IF_CONSTEXPR (!allocator_is_always_equal<A>()) {
- if (MDBX_UNLIKELY(target->get_allocator() != source.get_allocator()))
- MDBX_CXX20_UNLIKELY target->get_allocator() =
- ::std::allocator_traits<A>::select_on_container_copy_construction(
- source.get_allocator());
- } else {
- /* gag for buggy compilers */
- (void)target;
- (void)source;
- }
- }
-};
-
-template <typename T, typename A = typename T::allocator_type,
- bool PoCS =
- ::std::allocator_traits<A>::propagate_on_container_swap::value>
-struct swap_alloc;
-
-template <typename T, typename A> struct swap_alloc<T, A, false> {
- static constexpr bool is_nothrow() noexcept {
- return allocator_is_always_equal<A>();
- }
- static MDBX_CXX20_CONSTEXPR void propagate(T *target,
- T &source) noexcept(is_nothrow()) {
- if MDBX_IF_CONSTEXPR (!allocator_is_always_equal<A>()) {
- if (MDBX_UNLIKELY(target->get_allocator() != source.get_allocator()))
- MDBX_CXX20_UNLIKELY throw_allocators_mismatch();
- } else {
- /* gag for buggy compilers */
- (void)target;
- (void)source;
- }
- }
-};
-
-template <typename T, typename A> struct swap_alloc<T, A, true> {
- static constexpr bool is_nothrow() noexcept {
- return allocator_is_always_equal<A>() ||
-#if defined(__cpp_lib_is_swappable) && __cpp_lib_is_swappable >= 201603L
- ::std::is_nothrow_swappable<A>() ||
-#endif /* __cpp_lib_is_swappable >= 201603L */
- (::std::is_nothrow_move_constructible<A>::value &&
- ::std::is_nothrow_move_assignable<A>::value);
- }
- static MDBX_CXX20_CONSTEXPR void propagate(T *target,
- T &source) noexcept(is_nothrow()) {
- if MDBX_IF_CONSTEXPR (!allocator_is_always_equal<A>()) {
- if (MDBX_UNLIKELY(target->get_allocator() != source.get_allocator()))
- MDBX_CXX20_UNLIKELY ::std::swap(*target, source);
- } else {
- /* gag for buggy compilers */
- (void)target;
- (void)source;
- }
- }
-};
-
-} // namespace allocation_aware_details
-
-struct default_capacity_policy {
- enum : size_t {
- extra_inplace_storage = 0,
- pettiness_threshold = 64,
- max_reserve = 65536
- };
-
- static MDBX_CXX11_CONSTEXPR size_t round(const size_t value) {
- static_assert((pettiness_threshold & (pettiness_threshold - 1)) == 0,
- "pettiness_threshold must be a power of 2");
- static_assert(pettiness_threshold % 2 == 0,
- "pettiness_threshold must be even");
- static_assert(pettiness_threshold >= sizeof(uint64_t),
- "pettiness_threshold must be > 7");
- constexpr const auto pettiness_mask = ~size_t(pettiness_threshold - 1);
- return (value + pettiness_threshold - 1) & pettiness_mask;
- }
-
- static MDBX_CXX11_CONSTEXPR size_t advise(const size_t current,
- const size_t wanna) {
- static_assert(max_reserve % pettiness_threshold == 0,
- "max_reserve must be a multiple of pettiness_threshold");
- static_assert(max_reserve / 3 > pettiness_threshold,
- "max_reserve must be > pettiness_threshold * 3");
- if (wanna > current)
- /* doubling capacity, but don't made reserve more than max_reserve */
- return round(wanna + ::std::min(size_t(max_reserve), current));
-
- if (current - wanna >
- /* shrink if reserve will more than half of current or max_reserve,
- * but not less than pettiness_threshold */
- ::std::min(wanna + pettiness_threshold, size_t(max_reserve)))
- return round(wanna);
-
- /* keep unchanged */
- return current;
- }
-};
-
-/// \brief Hexadecimal encoder which satisfy \ref SliceTranscoder concept.
-struct LIBMDBX_API to_hex {
- const slice source;
- const bool uppercase = false;
- const unsigned wrap_width = 0;
- MDBX_CXX11_CONSTEXPR to_hex(const slice &source, bool uppercase = false,
- unsigned wrap_width = 0) noexcept
- : source(source), uppercase(uppercase), wrap_width(wrap_width) {
- MDBX_ASSERT_CXX20_CONCEPT_SATISFIED(SliceTranscoder, to_hex);
- }
-
- /// \brief Returns a string with a hexadecimal dump of a passed slice.
- template <class ALLOCATOR = legacy_allocator>
- string<ALLOCATOR> as_string(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_string<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns a buffer with a hexadecimal dump of a passed slice.
- template <class ALLOCATOR = legacy_allocator,
- typename CAPACITY_POLICY = default_capacity_policy>
- buffer<ALLOCATOR, CAPACITY_POLICY>
- as_buffer(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_buffer<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns the buffer size in bytes needed for hexadecimal
- /// dump of a passed slice.
- MDBX_CXX11_CONSTEXPR size_t envisage_result_length() const noexcept {
- const size_t bytes = source.length() << 1;
- return wrap_width ? bytes + bytes / wrap_width : bytes;
- }
-
- /// \brief Fills the buffer by hexadecimal dump of a passed slice.
- /// \throws std::length_error if given buffer is too small.
- char *write_bytes(char *dest, size_t dest_size) const;
-
- /// \brief Output hexadecimal dump of passed slice to the std::ostream.
- /// \throws std::ios_base::failure corresponding to std::ostream::write()
- /// behaviour.
- ::std::ostream &output(::std::ostream &out) const;
-
- /// \brief Checks whether a passed slice is empty,
- /// and therefore there will be no output bytes.
- bool is_empty() const noexcept { return source.empty(); }
-
- /// \brief Checks whether the content of a passed slice is a valid data
- /// and could be encoded or unexpectedly not.
- bool is_erroneous() const noexcept { return false; }
-};
-
-/// \brief [Base58](https://en.wikipedia.org/wiki/Base58) encoder which satisfy
-/// \ref SliceTranscoder concept.
-struct LIBMDBX_API to_base58 {
- const slice source;
- const unsigned wrap_width = 0;
- MDBX_CXX11_CONSTEXPR
- to_base58(const slice &source, unsigned wrap_width = 0) noexcept
- : source(source), wrap_width(wrap_width) {
- MDBX_ASSERT_CXX20_CONCEPT_SATISFIED(SliceTranscoder, to_base58);
- }
-
- /// \brief Returns a string with a
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump of a passed slice.
- template <class ALLOCATOR = legacy_allocator>
- string<ALLOCATOR> as_string(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_string<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns a buffer with a
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump of a passed slice.
- template <class ALLOCATOR = legacy_allocator,
- typename CAPACITY_POLICY = default_capacity_policy>
- buffer<ALLOCATOR, CAPACITY_POLICY>
- as_buffer(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_buffer<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns the buffer size in bytes needed for
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump of passed slice.
- MDBX_CXX11_CONSTEXPR size_t envisage_result_length() const noexcept {
- const size_t bytes =
- source.length() / 8 * 11 + (source.length() % 8 * 43 + 31) / 32;
- return wrap_width ? bytes + bytes / wrap_width : bytes;
- }
-
- /// \brief Fills the buffer by [Base58](https://en.wikipedia.org/wiki/Base58)
- /// dump of passed slice.
- /// \throws std::length_error if given buffer is too small.
- char *write_bytes(char *dest, size_t dest_size) const;
-
- /// \brief Output [Base58](https://en.wikipedia.org/wiki/Base58)
- /// dump of passed slice to the std::ostream.
- /// \throws std::ios_base::failure corresponding to std::ostream::write()
- /// behaviour.
- ::std::ostream &output(::std::ostream &out) const;
-
- /// \brief Checks whether a passed slice is empty,
- /// and therefore there will be no output bytes.
- bool is_empty() const noexcept { return source.empty(); }
-
- /// \brief Checks whether the content of a passed slice is a valid data
- /// and could be encoded or unexpectedly not.
- bool is_erroneous() const noexcept { return false; }
-};
-
-/// \brief [Base64](https://en.wikipedia.org/wiki/Base64) encoder which satisfy
-/// \ref SliceTranscoder concept.
-struct LIBMDBX_API to_base64 {
- const slice source;
- const unsigned wrap_width = 0;
- MDBX_CXX11_CONSTEXPR
- to_base64(const slice &source, unsigned wrap_width = 0) noexcept
- : source(source), wrap_width(wrap_width) {
- MDBX_ASSERT_CXX20_CONCEPT_SATISFIED(SliceTranscoder, to_base64);
- }
-
- /// \brief Returns a string with a
- /// [Base64](https://en.wikipedia.org/wiki/Base64) dump of a passed slice.
- template <class ALLOCATOR = legacy_allocator>
- string<ALLOCATOR> as_string(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_string<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns a buffer with a
- /// [Base64](https://en.wikipedia.org/wiki/Base64) dump of a passed slice.
- template <class ALLOCATOR = legacy_allocator,
- typename CAPACITY_POLICY = default_capacity_policy>
- buffer<ALLOCATOR, CAPACITY_POLICY>
- as_buffer(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_buffer<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns the buffer size in bytes needed for
- /// [Base64](https://en.wikipedia.org/wiki/Base64) dump of passed slice.
- MDBX_CXX11_CONSTEXPR size_t envisage_result_length() const noexcept {
- const size_t bytes = (source.length() + 2) / 3 * 4;
- return wrap_width ? bytes + bytes / wrap_width : bytes;
- }
-
- /// \brief Fills the buffer by [Base64](https://en.wikipedia.org/wiki/Base64)
- /// dump of passed slice.
- /// \throws std::length_error if given buffer is too small.
- char *write_bytes(char *dest, size_t dest_size) const;
-
- /// \brief Output [Base64](https://en.wikipedia.org/wiki/Base64)
- /// dump of passed slice to the std::ostream.
- /// \throws std::ios_base::failure corresponding to std::ostream::write()
- /// behaviour.
- ::std::ostream &output(::std::ostream &out) const;
-
- /// \brief Checks whether a passed slice is empty,
- /// and therefore there will be no output bytes.
- bool is_empty() const noexcept { return source.empty(); }
-
- /// \brief Checks whether the content of a passed slice is a valid data
- /// and could be encoded or unexpectedly not.
- bool is_erroneous() const noexcept { return false; }
-};
-
-inline ::std::ostream &operator<<(::std::ostream &out, const to_hex &wrapper) {
- return wrapper.output(out);
-}
-inline ::std::ostream &operator<<(::std::ostream &out,
- const to_base58 &wrapper) {
- return wrapper.output(out);
-}
-inline ::std::ostream &operator<<(::std::ostream &out,
- const to_base64 &wrapper) {
- return wrapper.output(out);
-}
-
-/// \brief Hexadecimal decoder which satisfy \ref SliceTranscoder concept.
-struct LIBMDBX_API from_hex {
- const slice source;
- const bool ignore_spaces = false;
- MDBX_CXX11_CONSTEXPR from_hex(const slice &source,
- bool ignore_spaces = false) noexcept
- : source(source), ignore_spaces(ignore_spaces) {
- MDBX_ASSERT_CXX20_CONCEPT_SATISFIED(SliceTranscoder, from_hex);
- }
-
- /// \brief Decodes hexadecimal dump from a passed slice to returned string.
- template <class ALLOCATOR = legacy_allocator>
- string<ALLOCATOR> as_string(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_string<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Decodes hexadecimal dump from a passed slice to returned buffer.
- template <class ALLOCATOR = legacy_allocator,
- typename CAPACITY_POLICY = default_capacity_policy>
- buffer<ALLOCATOR, CAPACITY_POLICY>
- as_buffer(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_buffer<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns the number of bytes needed for conversion
- /// hexadecimal dump from a passed slice to decoded data.
- MDBX_CXX11_CONSTEXPR size_t envisage_result_length() const noexcept {
- return source.length() >> 1;
- }
-
- /// \brief Fills the destination with data decoded from hexadecimal dump
- /// from a passed slice.
- /// \throws std::length_error if given buffer is too small.
- char *write_bytes(char *dest, size_t dest_size) const;
-
- /// \brief Checks whether a passed slice is empty,
- /// and therefore there will be no output bytes.
- bool is_empty() const noexcept { return source.empty(); }
-
- /// \brief Checks whether the content of a passed slice is a valid hexadecimal
- /// dump, and therefore there could be decoded or not.
- bool is_erroneous() const noexcept;
-};
-
-/// \brief [Base58](https://en.wikipedia.org/wiki/Base58) decoder which satisfy
-/// \ref SliceTranscoder concept.
-struct LIBMDBX_API from_base58 {
- const slice source;
- const bool ignore_spaces = false;
- MDBX_CXX11_CONSTEXPR from_base58(const slice &source,
- bool ignore_spaces = false) noexcept
- : source(source), ignore_spaces(ignore_spaces) {
- MDBX_ASSERT_CXX20_CONCEPT_SATISFIED(SliceTranscoder, from_base58);
- }
-
- /// \brief Decodes [Base58](https://en.wikipedia.org/wiki/Base58) dump from a
- /// passed slice to returned string.
- template <class ALLOCATOR = legacy_allocator>
- string<ALLOCATOR> as_string(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_string<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Decodes [Base58](https://en.wikipedia.org/wiki/Base58) dump from a
- /// passed slice to returned buffer.
- template <class ALLOCATOR = legacy_allocator,
- typename CAPACITY_POLICY = default_capacity_policy>
- buffer<ALLOCATOR, CAPACITY_POLICY>
- as_buffer(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_buffer<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns the number of bytes needed for conversion
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump from a passed slice to
- /// decoded data.
- MDBX_CXX11_CONSTEXPR size_t envisage_result_length() const noexcept {
- return source.length() / 11 * 8 + source.length() % 11 * 32 / 43;
- }
-
- /// \brief Fills the destination with data decoded from
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump from a passed slice.
- /// \throws std::length_error if given buffer is too small.
- char *write_bytes(char *dest, size_t dest_size) const;
-
- /// \brief Checks whether a passed slice is empty,
- /// and therefore there will be no output bytes.
- bool is_empty() const noexcept { return source.empty(); }
-
- /// \brief Checks whether the content of a passed slice is a valid
- /// [Base58](https://en.wikipedia.org/wiki/Base58) dump, and therefore there
- /// could be decoded or not.
- bool is_erroneous() const noexcept;
-};
-
-/// \brief [Base64](https://en.wikipedia.org/wiki/Base64) decoder which satisfy
-/// \ref SliceTranscoder concept.
-struct LIBMDBX_API from_base64 {
- const slice source;
- const bool ignore_spaces = false;
- MDBX_CXX11_CONSTEXPR from_base64(const slice &source,
- bool ignore_spaces = false) noexcept
- : source(source), ignore_spaces(ignore_spaces) {
- MDBX_ASSERT_CXX20_CONCEPT_SATISFIED(SliceTranscoder, from_base64);
- }
-
- /// \brief Decodes [Base64](https://en.wikipedia.org/wiki/Base64) dump from a
- /// passed slice to returned string.
- template <class ALLOCATOR = legacy_allocator>
- string<ALLOCATOR> as_string(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_string<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Decodes [Base64](https://en.wikipedia.org/wiki/Base64) dump from a
- /// passed slice to returned buffer.
- template <class ALLOCATOR = legacy_allocator,
- typename CAPACITY_POLICY = default_capacity_policy>
- buffer<ALLOCATOR, CAPACITY_POLICY>
- as_buffer(const ALLOCATOR &allocator = ALLOCATOR()) const {
- return make_buffer<ALLOCATOR>(*this, allocator);
- }
-
- /// \brief Returns the number of bytes needed for conversion
- /// [Base64](https://en.wikipedia.org/wiki/Base64) dump from a passed slice to
- /// decoded data.
- MDBX_CXX11_CONSTEXPR size_t envisage_result_length() const noexcept {
- return (source.length() + 3) / 4 * 3;
- }
-
- /// \brief Fills the destination with data decoded from
- /// [Base64](https://en.wikipedia.org/wiki/Base64) dump from a passed slice.
- /// \throws std::length_error if given buffer is too small.
- char *write_bytes(char *dest, size_t dest_size) const;
-
- /// \brief Checks whether a passed slice is empty,
- /// and therefore there will be no output bytes.
- bool is_empty() const noexcept { return source.empty(); }
-
- /// \brief Checks whether the content of a passed slice is a valid
- /// [Base64](https://en.wikipedia.org/wiki/Base64) dump, and therefore there
- /// could be decoded or not.
- bool is_erroneous() const noexcept;
-};
-
-/// \brief The chunk of data stored inside the buffer or located outside it.
-template <class ALLOCATOR, typename CAPACITY_POLICY> class buffer {
-public:
-#if !defined(_MSC_VER) || _MSC_VER > 1900
- using allocator_type = typename ::std::allocator_traits<
- ALLOCATOR>::template rebind_alloc<uint64_t>;
-#else
- using allocator_type = typename ALLOCATOR::template rebind<uint64_t>::other;
-#endif /* MSVC is mad */
- using allocator_traits = ::std::allocator_traits<allocator_type>;
- using reservation_policy = CAPACITY_POLICY;
- enum : size_t {
- max_length = MDBX_MAXDATASIZE,
- max_capacity = (max_length / 3u * 4u + 1023u) & ~size_t(1023),
- extra_inplace_storage = reservation_policy::extra_inplace_storage,
- pettiness_threshold = reservation_policy::pettiness_threshold
- };
-
-private:
- friend class txn;
- struct silo;
- using move_assign_alloc =
- allocation_aware_details::move_assign_alloc<silo, allocator_type>;
- using copy_assign_alloc =
- allocation_aware_details::copy_assign_alloc<silo, allocator_type>;
- using swap_alloc = allocation_aware_details::swap_alloc<silo, allocator_type>;
- struct silo /* Empty Base Class Optimization */ : public allocator_type {
- MDBX_CXX20_CONSTEXPR const allocator_type &get_allocator() const noexcept {
- return *this;
- }
- MDBX_CXX20_CONSTEXPR allocator_type &get_allocator() noexcept {
- return *this;
- }
-
- using allocator_pointer = typename allocator_traits::pointer;
- using allocator_const_pointer = typename allocator_traits::const_pointer;
-
- MDBX_CXX20_CONSTEXPR ::std::pair<allocator_pointer, size_t>
- allocate_storage(size_t bytes) {
- assert(bytes >= sizeof(bin));
- constexpr size_t unit = sizeof(typename allocator_type::value_type);
- static_assert((unit & (unit - 1)) == 0,
- "size of ALLOCATOR::value_type should be a power of 2");
- static_assert(unit > 0, "size of ALLOCATOR::value_type must be > 0");
- const size_t n = (bytes + unit - 1) / unit;
- return ::std::make_pair(allocator_traits::allocate(get_allocator(), n),
- n * unit);
- }
- MDBX_CXX20_CONSTEXPR void deallocate_storage(allocator_pointer ptr,
- size_t bytes) {
- constexpr size_t unit = sizeof(typename allocator_type::value_type);
- assert(ptr && bytes >= sizeof(bin) && bytes >= unit && bytes % unit == 0);
- allocator_traits::deallocate(get_allocator(), ptr, bytes / unit);
- }
-
- static MDBX_CXX17_CONSTEXPR void *
- to_address(allocator_pointer ptr) noexcept {
-#if defined(__cpp_lib_to_address) && __cpp_lib_to_address >= 201711L
- return static_cast<void *>(::std::to_address(ptr));
-#else
- return static_cast<void *>(::std::addressof(*ptr));
-#endif /* __cpp_lib_to_address */
- }
- static MDBX_CXX17_CONSTEXPR const void *
- to_address(allocator_const_pointer ptr) noexcept {
-#if defined(__cpp_lib_to_address) && __cpp_lib_to_address >= 201711L
- return static_cast<const void *>(::std::to_address(ptr));
-#else
- return static_cast<const void *>(::std::addressof(*ptr));
-#endif /* __cpp_lib_to_address */
- }
-
- union bin {
- struct allocated {
- allocator_pointer ptr_;
- size_t capacity_bytes_;
- constexpr allocated(allocator_pointer ptr, size_t bytes) noexcept
- : ptr_(ptr), capacity_bytes_(bytes) {}
- constexpr allocated(const allocated &) noexcept = default;
- constexpr allocated(allocated &&) noexcept = default;
- MDBX_CXX17_CONSTEXPR allocated &
- operator=(const allocated &) noexcept = default;
- MDBX_CXX17_CONSTEXPR allocated &
- operator=(allocated &&) noexcept = default;
- };
-
- allocated allocated_;
- uint64_t align_hint_;
- byte inplace_[(sizeof(allocated) + extra_inplace_storage + 7u) &
- ~size_t(7)];
-
- static constexpr bool
- is_suitable_for_inplace(size_t capacity_bytes) noexcept {
- static_assert(sizeof(bin) == sizeof(inplace_), "WTF?");
- return capacity_bytes < sizeof(bin);
- }
-
- enum : byte {
- /* Little Endian:
- * last byte is the most significant byte of u_.allocated.cap,
- * so use higher bit of capacity as the inplace-flag */
- le_lastbyte_mask = 0x80,
- /* Big Endian:
- * last byte is the least significant byte of u_.allocated.cap,
- * so use lower bit of capacity as the inplace-flag. */
- be_lastbyte_mask = 0x01
- };
-
- static constexpr byte inplace_lastbyte_mask() noexcept {
- static_assert(
- endian::native == endian::little || endian::native == endian::big,
- "Only the little-endian or big-endian bytes order are supported");
- return (endian::native == endian::little) ? le_lastbyte_mask
- : be_lastbyte_mask;
- }
- constexpr byte lastbyte() const noexcept {
- return inplace_[sizeof(bin) - 1];
- }
- MDBX_CXX17_CONSTEXPR byte &lastbyte() noexcept {
- return inplace_[sizeof(bin) - 1];
- }
-
- constexpr bool is_inplace() const noexcept {
- return (lastbyte() & inplace_lastbyte_mask()) != 0;
- }
- constexpr bool is_allocated() const noexcept { return !is_inplace(); }
-
- template <bool destroy_ptr>
- MDBX_CXX17_CONSTEXPR byte *make_inplace() noexcept {
- if (destroy_ptr) {
- MDBX_CONSTEXPR_ASSERT(is_allocated());
- /* properly destroy allocator::pointer */
- allocated_.~allocated();
- }
- if (::std::is_trivial<allocator_pointer>::value)
- /* workaround for "uninitialized" warning from some compilers */
- ::std::memset(&allocated_.ptr_, 0, sizeof(allocated_.ptr_));
- lastbyte() = inplace_lastbyte_mask();
- MDBX_CONSTEXPR_ASSERT(is_inplace() && address() == inplace_ &&
- is_suitable_for_inplace(capacity()));
- return address();
- }
-
- template <bool construct_ptr>
- MDBX_CXX17_CONSTEXPR byte *
- make_allocated(allocator_pointer ptr, size_t capacity_bytes) noexcept {
- MDBX_CONSTEXPR_ASSERT(
- (capacity_bytes & be_lastbyte_mask) == 0 &&
- ((capacity_bytes >>
- (sizeof(allocated_.capacity_bytes_) - 1) * CHAR_BIT) &
- le_lastbyte_mask) == 0);
- if (construct_ptr)
- /* properly construct allocator::pointer */
- new (&allocated_) allocated(ptr, capacity_bytes);
- else {
- MDBX_CONSTEXPR_ASSERT(is_allocated());
- allocated_.ptr_ = ptr;
- allocated_.capacity_bytes_ = capacity_bytes;
- }
- MDBX_CONSTEXPR_ASSERT(is_allocated() && address() == to_address(ptr) &&
- capacity() == capacity_bytes);
- return address();
- }
-
- MDBX_CXX20_CONSTEXPR bin(size_t capacity_bytes = 0) noexcept {
- MDBX_CONSTEXPR_ASSERT(is_suitable_for_inplace(capacity_bytes));
- make_inplace<false>();
- (void)capacity_bytes;
- }
- MDBX_CXX20_CONSTEXPR bin(allocator_pointer ptr,
- size_t capacity_bytes) noexcept {
- MDBX_CONSTEXPR_ASSERT(!is_suitable_for_inplace(capacity_bytes));
- make_allocated<true>(ptr, capacity_bytes);
- }
- MDBX_CXX20_CONSTEXPR ~bin() {
- if (is_allocated())
- /* properly destroy allocator::pointer */
- allocated_.~allocated();
- }
- MDBX_CXX20_CONSTEXPR bin(bin &&ditto) noexcept {
- if (ditto.is_inplace()) {
- // micro-optimization: don't use make_inplace<> here
- // since memcpy() will copy the flag.
- memcpy(inplace_, ditto.inplace_, sizeof(inplace_));
- MDBX_CONSTEXPR_ASSERT(is_inplace());
- } else {
- new (&allocated_) allocated(::std::move(ditto.allocated_));
- ditto.make_inplace<true>();
- MDBX_CONSTEXPR_ASSERT(is_allocated());
- }
- }
-
- MDBX_CXX17_CONSTEXPR bin &operator=(const bin &ditto) noexcept {
- if (ditto.is_inplace()) {
- // micro-optimization: don't use make_inplace<> here
- // since memcpy() will copy the flag.
- if (is_allocated())
- /* properly destroy allocator::pointer */
- allocated_.~allocated();
- memcpy(inplace_, ditto.inplace_, sizeof(inplace_));
- MDBX_CONSTEXPR_ASSERT(is_inplace());
- } else if (is_inplace())
- make_allocated<true>(ditto.allocated_.ptr_,
- ditto.allocated_.capacity_bytes_);
- else
- make_allocated<false>(ditto.allocated_.ptr_,
- ditto.allocated_.capacity_bytes_);
- return *this;
- }
-
- MDBX_CXX17_CONSTEXPR bin &operator=(bin &&ditto) noexcept {
- operator=(const_cast<const bin &>(ditto));
- if (ditto.is_allocated())
- ditto.make_inplace<true>();
- return *this;
- }
-
- static MDBX_CXX20_CONSTEXPR size_t advise_capacity(const size_t current,
- const size_t wanna) {
- if (MDBX_UNLIKELY(wanna > max_capacity))
- MDBX_CXX20_UNLIKELY throw_max_length_exceeded();
-
- const size_t advised = reservation_policy::advise(current, wanna);
- assert(advised >= wanna);
- return ::std::min(size_t(max_capacity),
- ::std::max(sizeof(bin) - 1, advised));
- }
-
- constexpr const byte *address() const noexcept {
- return is_inplace()
- ? inplace_
- : static_cast<const byte *>(to_address(allocated_.ptr_));
- }
- MDBX_CXX17_CONSTEXPR byte *address() noexcept {
- return is_inplace() ? inplace_
- : static_cast<byte *>(to_address(allocated_.ptr_));
- }
- constexpr size_t capacity() const noexcept {
- return is_inplace() ? sizeof(bin) - 1 : allocated_.capacity_bytes_;
- }
- } bin_;
-
- MDBX_CXX20_CONSTEXPR void *init(size_t capacity) {
- capacity = bin::advise_capacity(0, capacity);
- if (bin_.is_suitable_for_inplace(capacity))
- new (&bin_) bin();
- else {
- const auto pair = allocate_storage(capacity);
- assert(pair.second >= capacity);
- new (&bin_) bin(pair.first, pair.second);
- }
- return bin_.address();
- }
-
- MDBX_CXX20_CONSTEXPR void release() noexcept {
- if (bin_.is_allocated()) {
- deallocate_storage(bin_.allocated_.ptr_,
- bin_.allocated_.capacity_bytes_);
- bin_.template make_inplace<true>();
- }
- }
-
- template <bool external_content>
- MDBX_CXX20_CONSTEXPR void *
- reshape(const size_t wanna_capacity, const size_t wanna_headroom,
- const void *const content, const size_t length) {
- assert(wanna_capacity >= wanna_headroom + length);
- const size_t old_capacity = bin_.capacity();
- const size_t new_capacity =
- bin::advise_capacity(old_capacity, wanna_capacity);
- assert(new_capacity >= wanna_capacity);
- if (MDBX_LIKELY(new_capacity == old_capacity))
- MDBX_CXX20_LIKELY {
- assert(bin_.is_inplace() ==
- bin::is_suitable_for_inplace(new_capacity));
- byte *const new_place = bin_.address() + wanna_headroom;
- if (MDBX_LIKELY(length))
- MDBX_CXX20_LIKELY {
- if (external_content)
- memcpy(new_place, content, length);
- else {
- const size_t old_headroom =
- bin_.address() - static_cast<const byte *>(content);
- assert(old_capacity >= old_headroom + length);
- if (MDBX_UNLIKELY(old_headroom != wanna_headroom))
- MDBX_CXX20_UNLIKELY ::std::memmove(new_place, content,
- length);
- }
- }
- return new_place;
- }
-
- if (bin::is_suitable_for_inplace(new_capacity)) {
- assert(bin_.is_allocated());
- const auto old_allocated = ::std::move(bin_.allocated_.ptr_);
- byte *const new_place =
- bin_.template make_inplace<true>() + wanna_headroom;
- if (MDBX_LIKELY(length))
- MDBX_CXX20_LIKELY memcpy(new_place, content, length);
- deallocate_storage(old_allocated, old_capacity);
- return new_place;
- }
-
- if (!bin_.is_allocated()) {
- const auto pair = allocate_storage(new_capacity);
- assert(pair.second >= new_capacity);
- byte *const new_place =
- static_cast<byte *>(to_address(pair.first)) + wanna_headroom;
- if (MDBX_LIKELY(length))
- MDBX_CXX20_LIKELY memcpy(new_place, content, length);
- bin_.template make_allocated<true>(pair.first, pair.second);
- return new_place;
- }
-
- const auto old_allocated = ::std::move(bin_.allocated_.ptr_);
- if (external_content)
- deallocate_storage(old_allocated, old_capacity);
- const auto pair = allocate_storage(new_capacity);
- assert(pair.second >= new_capacity);
- byte *const new_place =
- bin_.template make_allocated<false>(pair.first, pair.second) +
- wanna_headroom;
- if (MDBX_LIKELY(length))
- MDBX_CXX20_LIKELY memcpy(new_place, content, length);
- if (!external_content)
- deallocate_storage(old_allocated, old_capacity);
- return new_place;
- }
-
- MDBX_CXX20_CONSTEXPR const byte *get(size_t offset = 0) const noexcept {
- assert(capacity() >= offset);
- return bin_.address() + offset;
- }
- MDBX_CXX20_CONSTEXPR byte *get(size_t offset = 0) noexcept {
- assert(capacity() >= offset);
- return bin_.address() + offset;
- }
- MDBX_CXX20_CONSTEXPR byte *put(size_t offset, const void *ptr,
- size_t length) {
- assert(capacity() >= offset + length);
- return static_cast<byte *>(memcpy(get(offset), ptr, length));
- }
-
- //--------------------------------------------------------------------------
-
- MDBX_CXX20_CONSTEXPR
- silo() noexcept : allocator_type() { init(0); }
- MDBX_CXX20_CONSTEXPR
- silo(const allocator_type &alloc) noexcept : allocator_type(alloc) {
- init(0);
- }
- MDBX_CXX20_CONSTEXPR silo(size_t capacity) { init(capacity); }
- MDBX_CXX20_CONSTEXPR silo(size_t capacity, const allocator_type &alloc)
- : silo(alloc) {
- init(capacity);
- }
-
- MDBX_CXX20_CONSTEXPR silo(silo &&ditto) noexcept(
- ::std::is_nothrow_move_constructible<allocator_type>::value)
- : allocator_type(::std::move(ditto.get_allocator())),
- bin_(::std::move(ditto.bin_)) {}
-
- MDBX_CXX20_CONSTEXPR silo(size_t capacity, size_t headroom, const void *ptr,
- size_t length)
- : silo(capacity) {
- assert(capacity >= headroom + length);
- if (length)
- put(headroom, ptr, length);
- }
-
- // select_on_container_copy_construction()
- MDBX_CXX20_CONSTEXPR silo(size_t capacity, size_t headroom, const void *ptr,
- size_t length, const allocator_type &alloc)
- : silo(capacity, alloc) {
- assert(capacity >= headroom + length);
- if (length)
- put(headroom, ptr, length);
- }
-
- MDBX_CXX20_CONSTEXPR silo(const void *ptr, size_t length)
- : silo(length, 0, ptr, length) {}
- MDBX_CXX20_CONSTEXPR silo(const void *ptr, size_t length,
- const allocator_type &alloc)
- : silo(length, 0, ptr, length, alloc) {}
-
- ~silo() { release(); }
-
- //--------------------------------------------------------------------------
-
- MDBX_CXX20_CONSTEXPR void *assign(size_t headroom, const void *ptr,
- size_t length, size_t tailroom) {
- return reshape<true>(headroom + length + tailroom, headroom, ptr, length);
- }
- MDBX_CXX20_CONSTEXPR void *assign(const void *ptr, size_t length) {
- return assign(0, ptr, length, 0);
- }
-
- MDBX_CXX20_CONSTEXPR silo &assign(const silo &ditto, size_t headroom,
- slice &content) {
- assert(ditto.get() + headroom == content.byte_ptr());
- if MDBX_IF_CONSTEXPR (!allocation_aware_details::
- allocator_is_always_equal<allocator_type>()) {
- if (MDBX_UNLIKELY(get_allocator() != ditto.get_allocator()))
- MDBX_CXX20_UNLIKELY {
- release();
- allocation_aware_details::copy_assign_alloc<
- silo, allocator_type>::propagate(this, ditto);
- }
- }
- content.iov_base = reshape<true>(ditto.capacity(), headroom,
- content.data(), content.length());
- return *this;
- }
-
- MDBX_CXX20_CONSTEXPR silo &
- assign(silo &&ditto, size_t headroom, slice &content) noexcept(
- allocation_aware_details::move_assign_alloc<
- silo, allocator_type>::is_nothrow()) {
- assert(ditto.get() + headroom == content.byte_ptr());
- if (allocation_aware_details::move_assign_alloc<
- silo, allocator_type>::is_moveable(this, ditto)) {
- release();
- allocation_aware_details::move_assign_alloc<
- silo, allocator_type>::propagate(this, ditto);
- /* no reallocation nor copying required */
- bin_ = ::std::move(ditto.bin_);
- assert(get() + headroom == content.byte_ptr());
- } else {
- /* copy content since allocators are different */
- content.iov_base = reshape<true>(ditto.capacity(), headroom,
- content.data(), content.length());
- ditto.release();
- }
- return *this;
- }
-
- MDBX_CXX20_CONSTEXPR void clear() { reshape<true>(0, 0, nullptr, 0); }
- MDBX_CXX20_CONSTEXPR void resize(size_t capacity, size_t headroom,
- slice &content) {
- content.iov_base =
- reshape<false>(capacity, headroom, content.iov_base, content.iov_len);
- }
- MDBX_CXX20_CONSTEXPR void swap(silo &ditto) noexcept(
- allocation_aware_details::swap_alloc<silo,
- allocator_type>::is_nothrow()) {
- allocation_aware_details::swap_alloc<silo, allocator_type>::propagate(
- this, ditto);
- ::std::swap(bin_, ditto.bin_);
- }
-
- /* MDBX_CXX20_CONSTEXPR void shrink_to_fit() { TODO } */
-
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX11_CONSTEXPR size_t
- capacity() const noexcept {
- return bin_.capacity();
- }
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX11_CONSTEXPR const void *
- data(size_t offset = 0) const noexcept {
- return get(offset);
- }
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX11_CONSTEXPR void *
- data(size_t offset = 0) noexcept {
- return get(offset);
- }
- };
-
- silo silo_;
- ::mdbx::slice slice_;
-
- void insulate() {
- assert(is_reference());
- silo_.assign(slice_.char_ptr(), slice_.length());
- slice_.iov_base = silo_.data();
- }
-
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR const byte *
- silo_begin() const noexcept {
- return static_cast<const byte *>(silo_.data());
- }
-
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR const byte *
- silo_end() const noexcept {
- return silo_begin() + silo_.capacity();
- }
-
- struct data_preserver : public exception_thunk {
- buffer data;
- data_preserver(allocator_type &allocator) : data(allocator) {}
- static int callback(void *context, MDBX_val *target, const void *src,
- size_t bytes) noexcept {
- auto self = static_cast<data_preserver *>(context);
- assert(self->is_clean());
- assert(&self->data.slice_ == target);
- (void)target;
- try {
- self->data.assign(src, bytes, false);
- return MDBX_RESULT_FALSE;
- } catch (... /* capture any exception to rethrow it over C code */) {
- self->capture();
- return MDBX_RESULT_TRUE;
- }
- }
- MDBX_CXX11_CONSTEXPR operator MDBX_preserve_func() const noexcept {
- return callback;
- }
- MDBX_CXX11_CONSTEXPR operator const buffer &() const noexcept {
- return data;
- }
- MDBX_CXX11_CONSTEXPR operator buffer &() noexcept { return data; }
- };
-
-public:
- /// \todo buffer& operator<<(buffer&, ...) for writing
- /// \todo buffer& operator>>(buffer&, ...) for reading (delegated to slice)
- /// \todo template<class X> key(X) for encoding keys while writing
-
- /// \brief Returns the associated allocator.
- MDBX_CXX20_CONSTEXPR allocator_type get_allocator() const {
- return silo_.get_allocator();
- }
-
- /// \brief Checks whether data chunk stored inside the buffer, otherwise
- /// buffer just refers to data located outside the buffer.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR bool
- is_freestanding() const noexcept {
- static_assert(size_t(-long(max_length)) > max_length, "WTF?");
- return size_t(byte_ptr() - silo_begin()) < silo_.capacity();
- }
-
- /// \brief Checks whether the buffer just refers to data located outside
- /// the buffer, rather than stores it.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR bool
- is_reference() const noexcept {
- return !is_freestanding();
- }
-
- /// \brief Returns the number of bytes that can be held in currently allocated
- /// storage.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR size_t
- capacity() const noexcept {
- return is_freestanding() ? silo_.capacity() : 0;
- }
-
- /// \brief Returns the number of bytes that available in currently allocated
- /// storage ahead the currently beginning of data.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR size_t
- headroom() const noexcept {
- return is_freestanding() ? slice_.byte_ptr() - silo_begin() : 0;
- }
-
- /// \brief Returns the number of bytes that available in currently allocated
- /// storage after the currently data end.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR size_t
- tailroom() const noexcept {
- return is_freestanding() ? capacity() - headroom() - slice_.length() : 0;
- }
-
- /// \brief Returns casted to const pointer to byte an address of data.
- MDBX_CXX11_CONSTEXPR const byte *byte_ptr() const noexcept {
- return slice_.byte_ptr();
- }
-
- /// \brief Returns casted to const pointer to byte an end of data.
- MDBX_CXX11_CONSTEXPR const byte *end_byte_ptr() const noexcept {
- return slice_.end_byte_ptr();
- }
-
- /// \brief Returns casted to pointer to byte an address of data.
- /// \pre REQUIRES: The buffer should store data chunk, but not referenced to
- /// an external one.
- MDBX_CXX11_CONSTEXPR byte *byte_ptr() noexcept {
- MDBX_CONSTEXPR_ASSERT(is_freestanding());
- return const_cast<byte *>(slice_.byte_ptr());
- }
-
- /// \brief Returns casted to pointer to byte an end of data.
- /// \pre REQUIRES: The buffer should store data chunk, but not referenced to
- /// an external one.
- MDBX_CXX11_CONSTEXPR byte *end_byte_ptr() noexcept {
- MDBX_CONSTEXPR_ASSERT(is_freestanding());
- return const_cast<byte *>(slice_.end_byte_ptr());
- }
-
- /// \brief Returns casted to const pointer to char an address of data.
- MDBX_CXX11_CONSTEXPR const char *char_ptr() const noexcept {
- return slice_.char_ptr();
- }
-
- /// \brief Returns casted to const pointer to char an end of data.
- MDBX_CXX11_CONSTEXPR const char *end_char_ptr() const noexcept {
- return slice_.end_char_ptr();
- }
-
- /// \brief Returns casted to pointer to char an address of data.
- /// \pre REQUIRES: The buffer should store data chunk, but not referenced to
- /// an external one.
- MDBX_CXX11_CONSTEXPR char *char_ptr() noexcept {
- MDBX_CONSTEXPR_ASSERT(is_freestanding());
- return const_cast<char *>(slice_.char_ptr());
- }
-
- /// \brief Returns casted to pointer to char an end of data.
- /// \pre REQUIRES: The buffer should store data chunk, but not referenced to
- /// an external one.
- MDBX_CXX11_CONSTEXPR char *end_char_ptr() noexcept {
- MDBX_CONSTEXPR_ASSERT(is_freestanding());
- return const_cast<char *>(slice_.end_char_ptr());
- }
-
- /// \brief Return a const pointer to the beginning of the referenced data.
- MDBX_CXX11_CONSTEXPR const void *data() const noexcept {
- return slice_.data();
- }
-
- /// \brief Return a const pointer to the end of the referenced data.
- MDBX_CXX11_CONSTEXPR const void *end() const noexcept { return slice_.end(); }
-
- /// \brief Return a pointer to the beginning of the referenced data.
- /// \pre REQUIRES: The buffer should store data chunk, but not referenced to
- /// an external one.
- MDBX_CXX11_CONSTEXPR void *data() noexcept {
- MDBX_CONSTEXPR_ASSERT(is_freestanding());
- return const_cast<void *>(slice_.data());
- }
-
- /// \brief Return a pointer to the end of the referenced data.
- /// \pre REQUIRES: The buffer should store data chunk, but not referenced to
- /// an external one.
- MDBX_CXX11_CONSTEXPR void *end() noexcept {
- MDBX_CONSTEXPR_ASSERT(is_freestanding());
- return const_cast<void *>(slice_.end());
- }
-
- /// \brief Returns the number of bytes.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR size_t
- length() const noexcept {
- return MDBX_CONSTEXPR_ASSERT(is_reference() ||
- slice_.length() + headroom() <=
- silo_.capacity()),
- slice_.length();
- }
-
- /// \brief Set length of data.
- MDBX_CXX14_CONSTEXPR buffer &set_length(size_t bytes) {
- MDBX_CONSTEXPR_ASSERT(is_reference() ||
- bytes + headroom() <= silo_.capacity());
- slice_.set_length(bytes);
- return *this;
- }
-
- /// \brief Sets the length by specifying the end of the data.
- MDBX_CXX14_CONSTEXPR buffer &set_end(const void *ptr) {
- MDBX_CONSTEXPR_ASSERT(static_cast<const char *>(ptr) >= char_ptr());
- return set_length(static_cast<const char *>(ptr) - char_ptr());
- }
-
- /// \brief Makes buffer owning the data.
- /// \details If buffer refers to an external data, then makes it the owner
- /// of clone by allocating storage and copying the data.
- void make_freestanding() {
- if (is_reference())
- insulate();
- }
-
- MDBX_CXX20_CONSTEXPR buffer() noexcept = default;
- MDBX_CXX20_CONSTEXPR buffer(const allocator_type &allocator) noexcept
- : silo_(allocator) {}
-
- buffer(const struct slice &src, bool make_reference,
- const allocator_type &allocator = allocator_type())
- : silo_(allocator), slice_(src) {
- if (!make_reference)
- insulate();
- }
-
- buffer(const buffer &src, bool make_reference,
- const allocator_type &allocator = allocator_type())
- : buffer(src.slice_, make_reference, allocator) {}
-
- buffer(const void *ptr, size_t bytes, bool make_reference,
- const allocator_type &allocator = allocator_type())
- : buffer(::mdbx::slice(ptr, bytes), make_reference, allocator) {}
-
- template <class CHAR, class T, class A>
- buffer(const ::std::basic_string<CHAR, T, A> &) = delete;
- template <class CHAR, class T, class A>
- buffer(const ::std::basic_string<CHAR, T, A> &&) = delete;
-
- buffer(const char *c_str, bool make_reference,
- const allocator_type &allocator = allocator_type())
- : buffer(::mdbx::slice(c_str), make_reference, allocator) {}
-
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L)
- template <class CHAR, class T>
- buffer(const ::std::basic_string_view<CHAR, T> &view, bool make_reference,
- const allocator_type &allocator = allocator_type())
- : buffer(::mdbx::slice(view), make_reference, allocator) {}
-#endif /* __cpp_lib_string_view >= 201606L */
-
- MDBX_CXX20_CONSTEXPR
- buffer(const struct slice &src,
- const allocator_type &allocator = allocator_type())
- : silo_(src.data(), src.length(), allocator),
- slice_(silo_.data(), src.length()) {}
-
- MDBX_CXX20_CONSTEXPR
- buffer(const buffer &src, const allocator_type &allocator = allocator_type())
- : buffer(src.slice_, allocator) {}
-
- MDBX_CXX20_CONSTEXPR
- buffer(const void *ptr, size_t bytes,
- const allocator_type &allocator = allocator_type())
- : buffer(::mdbx::slice(ptr, bytes), allocator) {}
-
- template <class CHAR, class T, class A>
- MDBX_CXX20_CONSTEXPR
- buffer(const ::std::basic_string<CHAR, T, A> &str,
- const allocator_type &allocator = allocator_type())
- : buffer(::mdbx::slice(str), allocator) {}
-
- MDBX_CXX20_CONSTEXPR
- buffer(const char *c_str, const allocator_type &allocator = allocator_type())
- : buffer(::mdbx::slice(c_str), allocator) {}
-
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L)
- template <class CHAR, class T>
- MDBX_CXX20_CONSTEXPR
- buffer(const ::std::basic_string_view<CHAR, T> &view,
- const allocator_type &allocator = allocator_type())
- : buffer(::mdbx::slice(view), allocator) {}
-#endif /* __cpp_lib_string_view >= 201606L */
-
- buffer(size_t head_room, size_t tail_room,
- const allocator_type &allocator = allocator_type())
- : silo_(allocator) {
- slice_.iov_base = silo_.init(check_length(head_room, tail_room));
- assert(slice_.iov_len == 0);
- }
-
- buffer(size_t capacity, const allocator_type &allocator = allocator_type())
- : silo_(allocator) {
- slice_.iov_base = silo_.init(check_length(capacity));
- assert(slice_.iov_len == 0);
- }
-
- buffer(size_t head_room, const struct slice &src, size_t tail_room,
- const allocator_type &allocator = allocator_type())
- : silo_(allocator) {
- slice_.iov_base =
- silo_.init(check_length(head_room, src.length(), tail_room));
- slice_.iov_len = src.length();
- memcpy(slice_.iov_base, src.data(), src.length());
- }
-
- buffer(size_t head_room, const buffer &src, size_t tail_room,
- const allocator_type &allocator = allocator_type())
- : buffer(head_room, src.slice_, tail_room, allocator) {}
-
- inline buffer(const ::mdbx::txn &txn, const struct slice &src,
- const allocator_type &allocator = allocator_type());
-
- buffer(buffer &&src) noexcept(move_assign_alloc::is_nothrow())
- : silo_(::std::move(src.silo_)), slice_(::std::move(src.slice_)) {}
-
- MDBX_CXX11_CONSTEXPR const struct slice &slice() const noexcept {
- return slice_;
- }
-
- MDBX_CXX11_CONSTEXPR operator const struct slice &() const noexcept {
- return slice_;
- }
-
- template <typename POD>
- static buffer wrap(const POD &pod, bool make_reference = false,
- const allocator_type &allocator = allocator_type()) {
- return buffer(::mdbx::slice::wrap(pod), make_reference, allocator);
- }
-
- /// \brief Reserves storage space.
- void reserve(size_t wanna_headroom, size_t wanna_tailroom) {
- wanna_headroom = ::std::min(::std::max(headroom(), wanna_headroom),
- wanna_headroom + pettiness_threshold);
- wanna_tailroom = ::std::min(::std::max(tailroom(), wanna_tailroom),
- wanna_tailroom + pettiness_threshold);
- const size_t wanna_capacity =
- check_length(wanna_headroom, slice_.length(), wanna_tailroom);
- silo_.resize(wanna_capacity, wanna_headroom, slice_);
- assert(headroom() >= wanna_headroom &&
- headroom() <= wanna_headroom + pettiness_threshold);
- assert(tailroom() >= wanna_tailroom &&
- tailroom() <= wanna_tailroom + pettiness_threshold);
- }
-
- /// \brief Reserves space before the payload.
- void reserve_headroom(size_t wanna_headroom) { reserve(wanna_headroom, 0); }
-
- /// \brief Reserves space after the payload.
- void reserve_tailroom(size_t wanna_tailroom) { reserve(0, wanna_tailroom); }
-
- buffer &assign_reference(const void *ptr, size_t bytes) {
- silo_.clear();
- slice_.assign(ptr, bytes);
- return *this;
- }
-
- buffer &assign_freestanding(const void *ptr, size_t bytes) {
- silo_.assign(static_cast<const typename silo::value_type *>(ptr),
- check_length(bytes));
- slice_.assign(silo_.data(), bytes);
- return *this;
- }
-
- MDBX_CXX20_CONSTEXPR void
- swap(buffer &other) noexcept(swap_alloc::is_nothrow()) {
- silo_.swap(other.silo_);
- slice_.swap(other.slice_);
- }
-
- static buffer clone(const buffer &src,
- const allocator_type &allocator = allocator_type()) {
- return buffer(src.headroom(), src.slice_, src.tailroom(), allocator);
- }
-
- buffer &assign(const buffer &src, bool make_reference = false) {
- return assign(src.slice_, make_reference);
- }
-
- buffer &assign(const void *ptr, size_t bytes, bool make_reference = false) {
- return make_reference ? assign_reference(ptr, bytes)
- : assign_freestanding(ptr, bytes);
- }
-
- buffer &assign(const struct slice &src, bool make_reference = false) {
- return assign(src.data(), src.length(), make_reference);
- }
-
- buffer &assign(const ::MDBX_val &src, bool make_reference = false) {
- return assign(src.iov_base, src.iov_len, make_reference);
- }
-
- buffer &assign(struct slice &&src, bool make_reference = false) {
- assign(src.data(), src.length(), make_reference);
- src.invalidate();
- return *this;
- }
-
- buffer &assign(::MDBX_val &&src, bool make_reference = false) {
- assign(src.iov_base, src.iov_len, make_reference);
- src.iov_base = nullptr;
- return *this;
- }
-
- buffer &assign(const void *begin, const void *end,
- bool make_reference = false) {
- return assign(begin,
- static_cast<const byte *>(end) -
- static_cast<const byte *>(begin),
- make_reference);
- }
-
- template <class CHAR, class T, class A>
- buffer &assign(const ::std::basic_string<CHAR, T, A> &str,
- bool make_reference = false) {
- return assign(str.data(), str.length(), make_reference);
- }
-
- buffer &assign(const char *c_str, bool make_reference = false) {
- return assign(c_str, ::mdbx::strlen(c_str), make_reference);
- }
-
-#if defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L
- template <class CHAR, class T>
- buffer &assign(const ::std::basic_string_view<CHAR, T> &view,
- bool make_reference = false) {
- return assign(view.data(), view.length(), make_reference);
- }
-
- template <class CHAR, class T>
- buffer &assign(::std::basic_string_view<CHAR, T> &&view,
- bool make_reference = false) {
- assign(view.data(), view.length(), make_reference);
- view = {};
- return *this;
- }
-#endif /* __cpp_lib_string_view >= 201606L */
-
- buffer &operator=(const buffer &src) { return assign(src); }
-
- buffer &operator=(buffer &&src) noexcept(move_assign_alloc::is_nothrow()) {
- return assign(::std::move(src));
- }
-
- buffer &operator=(const struct slice &src) { return assign(src); }
-
- buffer &operator=(struct slice &&src) { return assign(::std::move(src)); }
-
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L)
- template <class CHAR, class T>
- buffer &operator=(const ::std::basic_string_view<CHAR, T> &view) noexcept {
- return assign(view);
- }
-
- /// \brief Return a string_view that references the data of this buffer.
- template <class CHAR = char, class T = ::std::char_traits<CHAR>>
- ::std::basic_string_view<CHAR, T> string_view() const noexcept {
- return slice_.string_view<CHAR, T>();
- }
-
- /// \brief Return a string_view that references the data of this buffer.
- template <class CHAR, class T>
- operator ::std::basic_string_view<CHAR, T>() const noexcept {
- return string_view<CHAR, T>();
- }
-#endif /* __cpp_lib_string_view >= 201606L */
-
- /// \brief Checks whether the string is empty.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR bool empty() const noexcept {
- return length() == 0;
- }
-
- /// \brief Checks whether the data pointer of the buffer is nullptr.
- MDBX_CXX11_CONSTEXPR bool is_null() const noexcept {
- return data() == nullptr;
- }
-
- /// \brief Returns the number of bytes.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX20_CONSTEXPR size_t size() const noexcept {
- return length();
- }
-
- /// \brief Returns the hash value of the data.
- /// \attention Function implementation and returned hash values may changed
- /// version to version, and in future the t1ha3 will be used here. Therefore
- /// values obtained from this function shouldn't be persisted anywhere.
- MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR size_t
- hash_value() const noexcept {
- return slice_.hash_value();
- }
-
- template <class CHAR = char, class T = ::std::char_traits<CHAR>,
- class A = legacy_allocator>
- MDBX_CXX20_CONSTEXPR ::std::basic_string<CHAR, T, A>
- as_string(const A &allocator = A()) const {
- return slice_.as_string<CHAR, T, A>(allocator);
- }
-
- template <class CHAR, class T, class A>
- MDBX_CXX20_CONSTEXPR explicit
- operator ::std::basic_string<CHAR, T, A>() const {
- return as_string<CHAR, T, A>();
- }
-
- /// \brief Checks if the data starts with the given prefix.
- MDBX_NOTHROW_PURE_FUNCTION bool
- starts_with(const struct slice &prefix) const noexcept {
- return slice_.starts_with(prefix);
- }
-
- /// \brief Checks if the data ends with the given suffix.
- MDBX_NOTHROW_PURE_FUNCTION bool
- ends_with(const struct slice &suffix) const noexcept {
- return slice_.ends_with(suffix);
- }
-
- /// \brief Clears the contents and storage.
- void clear() noexcept {
- slice_.clear();
- silo_.clear();
- }
-
- /// \brief Reduces memory usage by freeing unused storage space.
- void shrink_to_fit() { reserve(0, 0); }
-
- /// \brief Drops the first "n" bytes from the data chunk.
- /// \pre REQUIRES: `n <= size()`
- void remove_prefix(size_t n) noexcept { slice_.remove_prefix(n); }
-
- /// \brief Drops the last "n" bytes from the data chunk.
- /// \pre REQUIRES: `n <= size()`
- void remove_suffix(size_t n) noexcept { slice_.remove_suffix(n); }
-
- /// \brief Drops the first "n" bytes from the data chunk.
- /// \throws std::out_of_range if `n > size()`
- void safe_remove_prefix(size_t n) { slice_.safe_remove_prefix(n); }
-
- /// \brief Drops the last "n" bytes from the data chunk.
- /// \throws std::out_of_range if `n > size()`
- void safe_remove_suffix(size_t n) { slice_.safe_remove_suffix(n); }
-
- /// \brief Accesses the specified byte of data chunk.
- /// \pre REQUIRES: `n < size()`
- MDBX_CXX11_CONSTEXPR byte operator[](size_t n) const noexcept {
- MDBX_CONSTEXPR_ASSERT(n < size());
- return slice_[n];
- }
-
- /// \brief Accesses the specified byte of data chunk.
- /// \pre REQUIRES: `n < size()`
- MDBX_CXX11_CONSTEXPR byte &operator[](size_t n) noexcept {
- MDBX_CONSTEXPR_ASSERT(n < size());
- return byte_ptr()[n];
- }
-
- /// \brief Accesses the specified byte of data chunk with bounds checking.
- /// \throws std::out_of_range if `n >= size()`
- MDBX_CXX14_CONSTEXPR byte at(size_t n) const { return slice_.at(n); }
-
- /// \brief Accesses the specified byte of data chunk with bounds checking.
- /// \throws std::out_of_range if `n >= size()`
- MDBX_CXX14_CONSTEXPR byte &at(size_t n) {
- if (MDBX_UNLIKELY(n >= size()))
- MDBX_CXX20_UNLIKELY throw_out_range();
- return byte_ptr()[n];
- }
-
- /// \brief Returns the first "n" bytes of the data chunk.
- /// \pre REQUIRES: `n <= size()`
- MDBX_CXX14_CONSTEXPR struct slice head(size_t n) const noexcept {
- return slice_.head(n);
- }
-
- /// \brief Returns the last "n" bytes of the data chunk.
- /// \pre REQUIRES: `n <= size()`
- MDBX_CXX14_CONSTEXPR struct slice tail(size_t n) const noexcept {
- return slice_.tail(n);
- }
-
- /// \brief Returns the middle "n" bytes of the data chunk.
- /// \pre REQUIRES: `from + n <= size()`
- MDBX_CXX14_CONSTEXPR struct slice middle(size_t from,
- size_t n) const noexcept {
- return slice_.middle(from, n);
- }
-
- /// \brief Returns the first "n" bytes of the data chunk.
- /// \throws std::out_of_range if `n >= size()`
- MDBX_CXX14_CONSTEXPR struct slice safe_head(size_t n) const {
- return slice_.safe_head(n);
- }
-
- /// \brief Returns the last "n" bytes of the data chunk.
- /// \throws std::out_of_range if `n >= size()`
- MDBX_CXX14_CONSTEXPR struct slice safe_tail(size_t n) const {
- return slice_.safe_tail(n);
- }
-
- /// \brief Returns the middle "n" bytes of the data chunk.
- /// \throws std::out_of_range if `from + n >= size()`
- MDBX_CXX14_CONSTEXPR struct slice safe_middle(size_t from, size_t n) const {
- return slice_.safe_middle(from, n);
- }
-
- buffer &append(const void *src, size_t bytes) {
- if (MDBX_UNLIKELY(tailroom() < check_length(bytes)))
- MDBX_CXX20_UNLIKELY reserve_tailroom(bytes);
- memcpy(slice_.byte_ptr() + size(), src, bytes);
- slice_.iov_len += bytes;
- return *this;
- }
-
- buffer &append(const struct slice &chunk) {
- return append(chunk.data(), chunk.size());
- }
-
- buffer &add_header(const void *src, size_t bytes) {
- if (MDBX_UNLIKELY(headroom() < check_length(bytes)))
- MDBX_CXX20_UNLIKELY reserve_headroom(bytes);
- slice_.iov_base =
- memcpy(static_cast<char *>(slice_.iov_base) - bytes, src, bytes);
- slice_.iov_len += bytes;
- return *this;
- }
-
- buffer &add_header(const struct slice &chunk) {
- return add_header(chunk.data(), chunk.size());
- }
-
- template <MDBX_CXX20_CONCEPT(MutableByteProducer, PRODUCER)>
- buffer &append_producer(PRODUCER &producer) {
- const size_t wanna_bytes = producer.envisage_result_length();
- if (MDBX_UNLIKELY(tailroom() < check_length(wanna_bytes)))
- MDBX_CXX20_UNLIKELY reserve_tailroom(wanna_bytes);
- return set_end(producer.write_bytes(end_char_ptr(), tailroom()));
- }
-
- template <MDBX_CXX20_CONCEPT(ImmutableByteProducer, PRODUCER)>
- buffer &append_producer(const PRODUCER &producer) {
- const size_t wanna_bytes = producer.envisage_result_length();
- if (MDBX_UNLIKELY(tailroom() < check_length(wanna_bytes)))
- MDBX_CXX20_UNLIKELY reserve_tailroom(wanna_bytes);
- return set_end(producer.write_bytes(end_char_ptr(), tailroom()));
- }
-
- buffer &append_hex(const struct slice &data, bool uppercase = false,
- unsigned wrap_width = 0) {
- return append_producer(to_hex(data, uppercase, wrap_width));
- }
-
- buffer &append_base58(const struct slice &data, unsigned wrap_width = 0) {
- return append_producer(to_base58(data, wrap_width));
- }
-
- buffer &append_base64(const struct slice &data, unsigned wrap_width = 0) {
- return append_producer(to_base64(data, wrap_width));
- }
-
- buffer &append_decoded_hex(const struct slice &data,
- bool ignore_spaces = false) {
- return append_producer(from_hex(data, ignore_spaces));
- }
-
- buffer &append_decoded_base58(const struct slice &data,
- bool ignore_spaces = false) {
- return append_producer(from_base58(data, ignore_spaces));
- }
-
- buffer &append_decoded_base64(const struct slice &data,
- bool ignore_spaces = false) {
- return append_producer(from_base64(data, ignore_spaces));
- }
-
- //----------------------------------------------------------------------------
-
- template <size_t SIZE>
- static buffer key_from(const char (&text)[SIZE], bool make_reference = true) {
- return buffer(::mdbx::slice(text), make_reference);
- }
-
-#if defined(DOXYGEN) || \
- (defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L)
- template <class CHAR, class T>
- static buffer key_from(const ::std::basic_string_view<CHAR, T> &src,
- bool make_reference = false) {
- return buffer(src, make_reference);
- }
-#endif /* __cpp_lib_string_view >= 201606L */
-
- static buffer key_from(const char *src, bool make_reference = false) {
- return buffer(src, make_reference);
- }
-
- template <class CHAR, class T, class A>
- static buffer key_from(const ::std::basic_string<CHAR, T, A> &src,
- bool make_reference = false) {
- return buffer(src, make_reference);
- }
-
- static buffer key_from(const silo &&src) noexcept {
- return buffer(::std::move(src));
- }
-
- static buffer key_from(const double ieee754_64bit) {
- return wrap(::mdbx_key_from_double(ieee754_64bit));
- }
-
- static buffer key_from(const double *ieee754_64bit) {
- return wrap(::mdbx_key_from_ptrdouble(ieee754_64bit));
- }
-
- static buffer key_from(const uint64_t unsigned_int64) {
- return wrap(unsigned_int64);
- }
-
- static buffer key_from(const int64_t signed_int64) {
- return wrap(::mdbx_key_from_int64(signed_int64));
- }
-
- static buffer key_from_jsonInteger(const int64_t json_integer) {
- return wrap(::mdbx_key_from_jsonInteger(json_integer));
- }
-
- static buffer key_from(const float ieee754_32bit) {
- return wrap(::mdbx_key_from_float(ieee754_32bit));
- }
-
- static buffer key_from(const float *ieee754_32bit) {
- return wrap(::mdbx_key_from_ptrfloat(ieee754_32bit));
- }
-
- static buffer key_from(const uint32_t unsigned_int32) {
- return wrap(unsigned_int32);
- }
-
- static buffer key_from(const int32_t signed_int32) {
- return wrap(::mdbx_key_from_int32(signed_int32));
- }
-};
-
-template <class ALLOCATOR, class CAPACITY_POLICY,
- MDBX_CXX20_CONCEPT(MutableByteProducer, PRODUCER)>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-make_buffer(PRODUCER &producer, const ALLOCATOR &allocator) {
- if (MDBX_LIKELY(!producer.is_empty()))
- MDBX_CXX20_LIKELY {
- buffer<ALLOCATOR, CAPACITY_POLICY> result(
- producer.envisage_result_length(), allocator);
- result.set_end(
- producer.write_bytes(result.end_char_ptr(), result.tailroom()));
- return result;
- }
- return buffer<ALLOCATOR, CAPACITY_POLICY>(allocator);
-}
-
-template <class ALLOCATOR, class CAPACITY_POLICY,
- MDBX_CXX20_CONCEPT(ImmutableByteProducer, PRODUCER)>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-make_buffer(const PRODUCER &producer, const ALLOCATOR &allocator) {
- if (MDBX_LIKELY(!producer.is_empty()))
- MDBX_CXX20_LIKELY {
- buffer<ALLOCATOR, CAPACITY_POLICY> result(
- producer.envisage_result_length(), allocator);
- result.set_end(
- producer.write_bytes(result.end_char_ptr(), result.tailroom()));
- return result;
- }
- return buffer<ALLOCATOR, CAPACITY_POLICY>(allocator);
-}
-
-template <class ALLOCATOR, MDBX_CXX20_CONCEPT(MutableByteProducer, PRODUCER)>
-inline string<ALLOCATOR> make_string(PRODUCER &producer,
- const ALLOCATOR &allocator) {
- string<ALLOCATOR> result(allocator);
- if (MDBX_LIKELY(!producer.is_empty()))
- MDBX_CXX20_LIKELY {
- result.resize(producer.envisage_result_length());
- result.resize(producer.write_bytes(const_cast<char *>(result.data()),
- result.capacity()) -
- result.data());
- }
- return result;
-}
-
-template <class ALLOCATOR, MDBX_CXX20_CONCEPT(ImmutableByteProducer, PRODUCER)>
-inline string<ALLOCATOR> make_string(const PRODUCER &producer,
- const ALLOCATOR &allocator) {
- string<ALLOCATOR> result(allocator);
- if (MDBX_LIKELY(!producer.is_empty()))
- MDBX_CXX20_LIKELY {
- result.resize(producer.envisage_result_length());
- result.resize(producer.write_bytes(const_cast<char *>(result.data()),
- result.capacity()) -
- result.data());
- }
- return result;
-}
-
-/// \brief Combines data slice with boolean flag to represent result of certain
-/// operations.
-struct value_result {
- slice value;
- bool done;
- value_result(const slice &value, bool done) noexcept
- : value(value), done(done) {}
- value_result(const value_result &) noexcept = default;
- value_result &operator=(const value_result &) noexcept = default;
- MDBX_CXX14_CONSTEXPR operator bool() const noexcept {
- assert(!done || bool(value));
- return done;
- }
-};
-
-/// \brief Combines pair of slices for key and value to represent result of
-/// certain operations.
-struct pair {
- slice key, value;
- pair(const slice &key, const slice &value) noexcept
- : key(key), value(value) {}
- pair(const pair &) noexcept = default;
- pair &operator=(const pair &) noexcept = default;
- MDBX_CXX14_CONSTEXPR operator bool() const noexcept {
- assert(bool(key) == bool(value));
- return key;
- }
-};
-
-/// \brief Combines pair of slices for key and value with boolean flag to
-/// represent result of certain operations.
-struct pair_result : public pair {
- bool done;
- pair_result(const slice &key, const slice &value, bool done) noexcept
- : pair(key, value), done(done) {}
- pair_result(const pair_result &) noexcept = default;
- pair_result &operator=(const pair_result &) noexcept = default;
- MDBX_CXX14_CONSTEXPR operator bool() const noexcept {
- assert(!done || (bool(key) && bool(value)));
- return done;
- }
-};
-
-//------------------------------------------------------------------------------
-
-/// \brief Loop control constants for readers enumeration functor and other
-/// cases. \see env::enumerate_readers()
-enum loop_control { continue_loop = 0, exit_loop = INT32_MIN };
-
-/// \brief Kinds of the keys and corresponding modes of comparing it.
-enum class key_mode {
- usual = MDBX_DB_DEFAULTS, ///< Usual variable length keys with byte-by-byte
- ///< lexicographic comparison like `std::memcmp()`.
- reverse = MDBX_REVERSEKEY, ///< Variable length keys with byte-by-byte
- ///< lexicographic comparison in reverse order,
- ///< from the end of the keys to the beginning.
- ordinal = MDBX_INTEGERKEY, ///< Keys are binary integers in native byte order,
- ///< either `uint32_t` or `uint64_t`, and will be
- ///< sorted as such. The keys must all be of the
- ///< same size and must be aligned while passing
- ///< as arguments.
- msgpack = -1 ///< Keys are in [MessagePack](https://msgpack.org/)
- ///< format with appropriate comparison.
- ///< \note Not yet implemented and PRs are welcome.
-};
-
-/// \brief Kind of the values and sorted multi-values with corresponding
-/// comparison.
-enum class value_mode {
- single = MDBX_DB_DEFAULTS, ///< Usual single value for each key. In terms of
- ///< keys, they are unique.
- multi =
- MDBX_DUPSORT, ///< A more than one data value could be associated with
- ///< each key. Internally each key is stored once, and the
- ///< corresponding data values are sorted by byte-by-byte
- ///< lexicographic comparison like `std::memcmp()`.
- ///< In terms of keys, they are not unique, i.e. has
- ///< duplicates which are sorted by associated data values.
-#if CONSTEXPR_ENUM_FLAGS_OPERATIONS || defined(DOXYGEN)
- multi_reverse =
- MDBX_DUPSORT |
- MDBX_REVERSEDUP, ///< A more than one data value could be associated with
- ///< each key. Internally each key is stored once, and
- ///< the corresponding data values are sorted by
- ///< byte-by-byte lexicographic comparison in reverse
- ///< order, from the end of the keys to the beginning.
- ///< In terms of keys, they are not unique, i.e. has
- ///< duplicates which are sorted by associated data
- ///< values.
- multi_samelength =
- MDBX_DUPSORT |
- MDBX_DUPFIXED, ///< A more than one data value could be associated with
- ///< each key, and all data values must be same length.
- ///< Internally each key is stored once, and the
- ///< corresponding data values are sorted by byte-by-byte
- ///< lexicographic comparison like `std::memcmp()`. In
- ///< terms of keys, they are not unique, i.e. has
- ///< duplicates which are sorted by associated data values.
- multi_ordinal =
- MDBX_DUPSORT | MDBX_DUPFIXED |
- MDBX_INTEGERDUP, ///< A more than one data value could be associated with
- ///< each key, and all data values are binary integers in
- ///< native byte order, either `uint32_t` or `uint64_t`,
- ///< and will be sorted as such. Internally each key is
- ///< stored once, and the corresponding data values are
- ///< sorted. In terms of keys, they are not unique, i.e.
- ///< has duplicates which are sorted by associated data
- ///< values.
- multi_reverse_samelength =
- MDBX_DUPSORT | MDBX_REVERSEDUP |
- MDBX_DUPFIXED, ///< A more than one data value could be associated with
- ///< each key, and all data values must be same length.
- ///< Internally each key is stored once, and the
- ///< corresponding data values are sorted by byte-by-byte
- ///< lexicographic comparison in reverse order, from the
- ///< end of the keys to the beginning. In terms of keys,
- ///< they are not unique, i.e. has duplicates which are
- ///< sorted by associated data values.
- msgpack = -1 ///< A more than one data value could be associated with each
- ///< key. Values are in [MessagePack](https://msgpack.org/)
- ///< format with appropriate comparison. Internally each key is
- ///< stored once, and the corresponding data values are sorted.
- ///< In terms of keys, they are not unique, i.e. has duplicates
- ///< which are sorted by associated data values.
- ///< \note Not yet implemented and PRs are welcome.
-#else
- multi_reverse = uint32_t(MDBX_DUPSORT) | uint32_t(MDBX_REVERSEDUP),
- multi_samelength = uint32_t(MDBX_DUPSORT) | uint32_t(MDBX_DUPFIXED),
- multi_ordinal = uint32_t(MDBX_DUPSORT) | uint32_t(MDBX_DUPFIXED) |
- uint32_t(MDBX_INTEGERDUP),
- multi_reverse_samelength = uint32_t(MDBX_DUPSORT) |
- uint32_t(MDBX_REVERSEDUP) | uint32_t(MDBX_DUPFIXED)
-#endif
-};
-
-/// \brief A handle for an individual database (key-value spaces) in the
-/// environment.
-/// \see txn::open_map() \see txn::create_map()
-/// \see txn::clear_map() \see txn::drop_map()
-/// \see txn::get_handle_info() \see txn::get_map_stat()
-/// \see env::close_map()
-/// \see cursor::map()
-struct LIBMDBX_API_TYPE map_handle {
- MDBX_dbi dbi{0};
- MDBX_CXX11_CONSTEXPR map_handle() noexcept {}
- MDBX_CXX11_CONSTEXPR map_handle(MDBX_dbi dbi) noexcept : dbi(dbi) {}
- map_handle(const map_handle &) noexcept = default;
- map_handle &operator=(const map_handle &) noexcept = default;
- operator bool() const noexcept { return dbi != 0; }
-
- using flags = ::MDBX_db_flags_t;
- using state = ::MDBX_dbi_state_t;
- struct LIBMDBX_API_TYPE info {
- map_handle::flags flags;
- map_handle::state state;
- MDBX_CXX11_CONSTEXPR info(map_handle::flags flags,
- map_handle::state state) noexcept;
- info(const info &) noexcept = default;
- info &operator=(const info &) noexcept = default;
-#if CONSTEXPR_ENUM_FLAGS_OPERATIONS
- MDBX_CXX11_CONSTEXPR
-#else
- inline
-#endif
- ::mdbx::key_mode key_mode() const noexcept;
-#if CONSTEXPR_ENUM_FLAGS_OPERATIONS
- MDBX_CXX11_CONSTEXPR
-#else
- inline
-#endif
- ::mdbx::value_mode value_mode() const noexcept;
- };
-};
-
-/// \brief Key-value pairs put mode.
-enum put_mode {
- insert_unique = MDBX_NOOVERWRITE, ///< Insert only unique keys.
- upsert = MDBX_UPSERT, ///< Insert or update.
- update = MDBX_CURRENT, ///< Update existing, don't insert new.
-};
-
-/// \brief Unmanaged database environment.
-///
-/// Like other unmanaged classes, `env` allows copying and assignment for
-/// instances, but does not destroys the represented underlying object from the
-/// own class destructor.
-///
-/// An environment supports multiple key-value sub-databases (aka key-value
-/// spaces or tables), all residing in the same shared-memory map.
-class LIBMDBX_API_TYPE env {
- friend class txn;
-
-protected:
- MDBX_env *handle_{nullptr};
- MDBX_CXX11_CONSTEXPR env(MDBX_env *ptr) noexcept;
-
-public:
- MDBX_CXX11_CONSTEXPR env() noexcept = default;
- env(const env &) noexcept = default;
- inline env &operator=(env &&other) noexcept;
- inline env(env &&other) noexcept;
- inline ~env() noexcept;
-
- MDBX_CXX14_CONSTEXPR operator bool() const noexcept;
- MDBX_CXX14_CONSTEXPR operator const MDBX_env *() const;
- MDBX_CXX14_CONSTEXPR operator MDBX_env *();
- friend MDBX_CXX11_CONSTEXPR bool operator==(const env &a,
- const env &b) noexcept;
- friend MDBX_CXX11_CONSTEXPR bool operator!=(const env &a,
- const env &b) noexcept;
-
- //----------------------------------------------------------------------------
-
- /// Database geometry for size management.
- struct LIBMDBX_API_TYPE geometry {
- enum : int64_t {
- default_value = -1, ///< Means "keep current or use default"
- minimal_value = 0, ///< Means "minimal acceptable"
- maximal_value = INTPTR_MAX, ///< Means "maximal acceptable"
- kB = 1000, ///< \f$10^{3}\f$ bytes
- MB = kB * 1000, ///< \f$10^{6}\f$ bytes
- GB = MB * 1000, ///< \f$10^{9}\f$ bytes
- TB = GB * 1000, ///< \f$10^{12}\f$ bytes
- PB = TB * 1000, ///< \f$10^{15}\f$ bytes
- EB = PB * 1000, ///< \f$10^{18}\f$ bytes
- KiB = 1024, ///< \f$2^{10}\f$ bytes
- MiB = KiB << 10, ///< \f$2^{20}\f$ bytes
- GiB = MiB << 10, ///< \f$2^{30}\f$ bytes
- TiB = GiB << 10, ///< \f$2^{40}\f$ bytes
- PiB = TiB << 10, ///< \f$2^{50}\f$ bytes
- EiB = PiB << 10, ///< \f$2^{60}\f$ bytes
- };
-
- /// \brief Tagged type for output to std::ostream
- struct size {
- intptr_t bytes;
- MDBX_CXX11_CONSTEXPR size(intptr_t bytes) noexcept : bytes(bytes) {}
- MDBX_CXX11_CONSTEXPR operator intptr_t() const noexcept { return bytes; }
- };
-
- /// \brief The lower bound of database size in bytes.
- intptr_t size_lower{minimal_value};
-
- /// \brief The size in bytes to setup the database size for now.
- /// \details It is recommended always pass \ref default_value in this
- /// argument except some special cases.
- intptr_t size_now{default_value};
-
- /// \brief The upper bound of database size in bytes.
- /// \details It is recommended to avoid change upper bound while database is
- /// used by other processes or threaded (i.e. just pass \ref default_value
- /// in this argument except absolutely necessary). Otherwise you must be
- /// ready for \ref MDBX_UNABLE_EXTEND_MAPSIZE error(s), unexpected pauses
- /// during remapping and/or system errors like "address busy", and so on. In
- /// other words, there is no way to handle a growth of the upper bound
- /// robustly because there may be a lack of appropriate system resources
- /// (which are extremely volatile in a multi-process multi-threaded
- /// environment).
- intptr_t size_upper{maximal_value};
-
- /// \brief The growth step in bytes, must be greater than zero to allow the
- /// database to grow.
- intptr_t growth_step{default_value};
-
- /// \brief The shrink threshold in bytes, must be greater than zero to allow
- /// the database to shrink.
- intptr_t shrink_threshold{default_value};
-
- /// \brief The database page size for new database creation
- /// or \ref default_value otherwise.
- /// \details Must be power of 2 in the range between \ref MDBX_MIN_PAGESIZE
- /// and \ref MDBX_MAX_PAGESIZE.
- intptr_t pagesize{default_value};
-
- inline geometry &make_fixed(intptr_t size) noexcept;
- inline geometry &make_dynamic(intptr_t lower = minimal_value,
- intptr_t upper = maximal_value) noexcept;
- MDBX_CXX11_CONSTEXPR geometry() noexcept {}
- MDBX_CXX11_CONSTEXPR
- geometry(const geometry &) noexcept = default;
- MDBX_CXX11_CONSTEXPR geometry(intptr_t size_lower,
- intptr_t size_now = default_value,
- intptr_t size_upper = maximal_value,
- intptr_t growth_step = default_value,
- intptr_t shrink_threshold = default_value,
- intptr_t pagesize = default_value) noexcept
- : size_lower(size_lower), size_now(size_now), size_upper(size_upper),
- growth_step(growth_step), shrink_threshold(shrink_threshold),
- pagesize(pagesize) {}
- };
-
- /// \brief Operation mode.
- enum mode {
- readonly, ///< \copydoc MDBX_RDONLY
- write_file_io, // don't available on OpenBSD
- write_mapped_io ///< \copydoc MDBX_WRITEMAP
- };
-
- /// \brief Durability level.
- enum durability {
- robust_synchronous, ///< \copydoc MDBX_SYNC_DURABLE
- half_synchronous_weak_last, ///< \copydoc MDBX_NOMETASYNC
- lazy_weak_tail, ///< \copydoc MDBX_SAFE_NOSYNC
- whole_fragile ///< \copydoc MDBX_UTTERLY_NOSYNC
- };
-
- /// \brief Garbage reclaiming options.
- struct LIBMDBX_API_TYPE reclaiming_options {
- /// \copydoc MDBX_LIFORECLAIM
- bool lifo{false};
- /// \copydoc MDBX_COALESCE
- bool coalesce{false};
- MDBX_CXX11_CONSTEXPR reclaiming_options() noexcept {}
- MDBX_CXX11_CONSTEXPR
- reclaiming_options(const reclaiming_options &) noexcept = default;
- MDBX_CXX14_CONSTEXPR reclaiming_options &
- operator=(const reclaiming_options &) noexcept = default;
- reclaiming_options(MDBX_env_flags_t) noexcept;
- };
-
- /// \brief Operate options.
- struct LIBMDBX_API_TYPE operate_options {
- /// \copydoc MDBX_NOTLS
- bool orphan_read_transactions{false};
- bool nested_write_transactions{false};
- /// \copydoc MDBX_EXCLUSIVE
- bool exclusive{false};
- /// \copydoc MDBX_NORDAHEAD
- bool disable_readahead{false};
- /// \copydoc MDBX_NOMEMINIT
- bool disable_clear_memory{false};
- MDBX_CXX11_CONSTEXPR operate_options() noexcept {}
- MDBX_CXX11_CONSTEXPR
- operate_options(const operate_options &) noexcept = default;
- MDBX_CXX14_CONSTEXPR operate_options &
- operator=(const operate_options &) noexcept = default;
- operate_options(MDBX_env_flags_t) noexcept;
- };
-
- /// \brief Operate parameters.
- struct LIBMDBX_API_TYPE operate_parameters {
- /// \brief The maximum number of named databases for the environment.
- /// Zero means default value.
- unsigned max_maps{0};
- /// \brief The maximum number of threads/reader slots for the environment.
- /// Zero means default value.
- unsigned max_readers{0};
- env::mode mode{write_mapped_io};
- env::durability durability{robust_synchronous};
- env::reclaiming_options reclaiming;
- env::operate_options options;
-
- MDBX_CXX11_CONSTEXPR operate_parameters() noexcept {}
- MDBX_CXX11_CONSTEXPR
- operate_parameters(
- const unsigned max_maps, const unsigned max_readers = 0,
- const env::mode mode = env::mode::write_mapped_io,
- env::durability durability = env::durability::robust_synchronous,
- const env::reclaiming_options &reclaiming = env::reclaiming_options(),
- const env::operate_options &options = env::operate_options()) noexcept
- : max_maps(max_maps), max_readers(max_readers), mode(mode),
- durability(durability), reclaiming(reclaiming), options(options) {}
- MDBX_CXX11_CONSTEXPR
- operate_parameters(const operate_parameters &) noexcept = default;
- MDBX_CXX14_CONSTEXPR operate_parameters &
- operator=(const operate_parameters &) noexcept = default;
- MDBX_env_flags_t
- make_flags(bool accede = true, ///< \copydoc MDBX_ACCEDE
- bool use_subdirectory =
- false ///< use subdirectory to place the DB files
- ) const;
- static env::mode mode_from_flags(MDBX_env_flags_t) noexcept;
- static env::durability durability_from_flags(MDBX_env_flags_t) noexcept;
- inline static env::reclaiming_options
- reclaiming_from_flags(MDBX_env_flags_t flags) noexcept;
- inline static env::operate_options
- options_from_flags(MDBX_env_flags_t flags) noexcept;
- };
-
- /// \brief Returns current operation parameters.
- inline env::operate_parameters get_operation_parameters() const;
- /// \brief Returns current operation mode.
- inline env::mode get_mode() const;
- /// \brief Returns current durability mode.
- inline env::durability get_durability() const;
- /// \brief Returns current reclaiming options.
- inline env::reclaiming_options get_reclaiming() const;
- /// \brief Returns current operate options.
- inline env::operate_options get_options() const;
-
- /// \brief Returns `true` for a freshly created database,
- /// but `false` if at least one transaction was committed.
- bool is_pristine() const;
-
- /// \brief Checks whether the database is empty.
- bool is_empty() const;
-
- /// \brief Returns default page size for current system/platform.
- static size_t default_pagesize() noexcept {
- return ::mdbx_default_pagesize();
- }
-
- struct limits {
- limits() = delete;
- /// \brief Returns the minimal database page size in bytes.
- static inline size_t pagesize_min() noexcept;
- /// \brief Returns the maximal database page size in bytes.
- static inline size_t pagesize_max() noexcept;
- /// \brief Returns the minimal database size in bytes for specified page
- /// size.
- static inline size_t dbsize_min(intptr_t pagesize);
- /// \brief Returns the maximal database size in bytes for specified page
- /// size.
- static inline size_t dbsize_max(intptr_t pagesize);
- /// \brief Returns the minimal key size in bytes for specified database
- /// flags.
- static inline size_t key_min(MDBX_db_flags_t flags) noexcept;
- /// \brief Returns the minimal key size in bytes for specified keys mode.
- static inline size_t key_min(key_mode mode) noexcept;
- /// \brief Returns the maximal key size in bytes for specified page size and
- /// database flags.
- static inline size_t key_max(intptr_t pagesize, MDBX_db_flags_t flags);
- /// \brief Returns the maximal key size in bytes for specified page size and
- /// keys mode.
- static inline size_t key_max(intptr_t pagesize, key_mode mode);
- /// \brief Returns the maximal key size in bytes for given environment and
- /// database flags.
- static inline size_t key_max(const env &, MDBX_db_flags_t flags);
- /// \brief Returns the maximal key size in bytes for given environment and
- /// keys mode.
- static inline size_t key_max(const env &, key_mode mode);
- /// \brief Returns the minimal values size in bytes for specified database
- /// flags.
- static inline size_t value_min(MDBX_db_flags_t flags) noexcept;
- /// \brief Returns the minimal values size in bytes for specified values
- /// mode.
- static inline size_t value_min(value_mode) noexcept;
- /// \brief Returns the maximal value size in bytes for specified page size
- /// and database flags.
- static inline size_t value_max(intptr_t pagesize, MDBX_db_flags_t flags);
- /// \brief Returns the maximal value size in bytes for specified page size
- /// and values mode.
- static inline size_t value_max(intptr_t pagesize, value_mode);
- /// \brief Returns the maximal value size in bytes for given environment and
- /// database flags.
- static inline size_t value_max(const env &, MDBX_db_flags_t flags);
- /// \brief Returns the maximal value size in bytes for specified page size
- /// and values mode.
- static inline size_t value_max(const env &, value_mode);
- /// \brief Returns the maximal write transaction size (i.e. limit for
- /// summary volume of dirty pages) in bytes for specified page size.
- static inline size_t transaction_size_max(intptr_t pagesize);
- };
-
- /// \brief Returns the minimal database size in bytes for the environment.
- size_t dbsize_min() const { return limits::dbsize_min(this->get_pagesize()); }
- /// \brief Returns the maximal database size in bytes for the environment.
- size_t dbsize_max() const { return limits::dbsize_max(this->get_pagesize()); }
- /// \brief Returns the minimal key size in bytes for specified keys mode.
- size_t key_min(key_mode mode) const noexcept { return limits::key_min(mode); }
- /// \brief Returns the maximal key size in bytes for specified keys mode.
- size_t key_max(key_mode mode) const { return limits::key_max(*this, mode); }
- /// \brief Returns the minimal value size in bytes for specified values mode.
- size_t value_min(value_mode mode) const noexcept {
- return limits::value_min(mode);
- }
- /// \brief Returns the maximal value size in bytes for specified values mode.
- size_t value_max(value_mode mode) const {
- return limits::value_max(*this, mode);
- }
- /// \brief Returns the maximal write transaction size (i.e. limit for summary
- /// volume of dirty pages) in bytes.
- size_t transaction_size_max() const {
- return limits::transaction_size_max(this->get_pagesize());
- }
-
- /// \brief Make a copy (backup) of an existing environment to the specified
- /// path.
-#ifdef MDBX_STD_FILESYSTEM_PATH
- env ©(const MDBX_STD_FILESYSTEM_PATH &destination, bool compactify,
- bool force_dynamic_size = false);
-#endif /* MDBX_STD_FILESYSTEM_PATH */
-#if defined(_WIN32) || defined(_WIN64)
- env ©(const ::std::wstring &destination, bool compactify,
- bool force_dynamic_size = false);
-#endif /* Windows */
- env ©(const ::std::string &destination, bool compactify,
- bool force_dynamic_size = false);
-
- /// \brief Copy an environment to the specified file descriptor.
- env ©(filehandle fd, bool compactify, bool force_dynamic_size = false);
-
- /// \brief Deletion modes for \ref remove().
- enum remove_mode {
- /// \brief Just delete the environment's files and directory if any.
- /// \note On POSIX systems, processes already working with the database will
- /// continue to work without interference until it close the environment.
- /// \note On Windows, the behavior of `just_remove` is different
- /// because the system does not support deleting files that are currently
- /// memory mapped.
- just_remove = MDBX_ENV_JUST_DELETE,
- /// \brief Make sure that the environment is not being used by other
- /// processes, or return an error otherwise.
- ensure_unused = MDBX_ENV_ENSURE_UNUSED,
- /// \brief Wait until other processes closes the environment before
- /// deletion.
- wait_for_unused = MDBX_ENV_WAIT_FOR_UNUSED
- };
-
- /// \brief Removes the environment's files in a proper and multiprocess-safe
- /// way.
-#ifdef MDBX_STD_FILESYSTEM_PATH
- static bool remove(const MDBX_STD_FILESYSTEM_PATH &,
- const remove_mode mode = just_remove);
-#endif /* MDBX_STD_FILESYSTEM_PATH */
-#if defined(_WIN32) || defined(_WIN64)
- static bool remove(const ::std::wstring &,
- const remove_mode mode = just_remove);
-#endif /* Windows */
- static bool remove(const ::std::string &,
- const remove_mode mode = just_remove);
-
- /// \brief Statistics for a database in the MDBX environment.
- using stat = ::MDBX_stat;
-
- /// \brief Information about the environment.
- using info = ::MDBX_envinfo;
-
- /// \brief Returns snapshot statistics about the MDBX environment.
- inline stat get_stat() const;
-
- /// \brief Returns pagesize of this MDBX environment.
- size_t get_pagesize() const { return get_stat().ms_psize; }
-
- /// \brief Return snapshot information about the MDBX environment.
- inline info get_info() const;
-
- /// \brief Return statistics about the MDBX environment accordingly to the
- /// specified transaction.
- inline stat get_stat(const txn &) const;
-
- /// \brief Return information about the MDBX environment accordingly to the
- /// specified transaction.
- inline info get_info(const txn &) const;
-
- /// \brief Returns the file descriptor for the DXB file of MDBX environment.
- inline filehandle get_filehandle() const;
-
- /// \brief Return the path that was used for opening the environment.
- path get_path() const;
-
- /// Returns environment flags.
- inline MDBX_env_flags_t get_flags() const;
-
- /// \brief Returns the maximum number of threads/reader slots for the
- /// environment.
- inline unsigned max_readers() const;
-
- /// \brief Returns the maximum number of named databases for the environment.
- inline unsigned max_maps() const;
-
- /// \brief Returns the application context associated with the environment.
- inline void *get_context() const noexcept;
-
- /// \brief Sets the application context associated with the environment.
- inline env &set_context(void *);
-
- /// \brief Sets threshold to force flush the data buffers to disk, for
- /// non-sync durability modes.
- ///
- /// The threshold value affects all processes which operates with given
- /// environment until the last process close environment or a new value will
- /// be settled.
- /// Data is always written to disk when \ref txn_managed::commit() is called,
- /// but the operating system may keep it buffered. MDBX always flushes the OS
- /// buffers upon commit as well, unless the environment was opened with \ref
- /// whole_fragile, \ref lazy_weak_tail or in part \ref
- /// half_synchronous_weak_last. The default is 0, than mean no any threshold
- /// checked, and no additional flush will be made.
- ///
- inline env &set_sync_threshold(size_t bytes);
-
- /// \brief Sets relative period since the last unsteady commit to force flush
- /// the data buffers to disk, for non-sync durability modes.
- ///
- /// The relative period value affects all processes which operates with given
- /// environment until the last process close environment or a new value will
- /// be settled.
- /// Data is always written to disk when \ref txn_managed::commit() is called,
- /// but the operating system may keep it buffered. MDBX always flushes the OS
- /// buffers upon commit as well, unless the environment was opened with \ref
- /// whole_fragile, \ref lazy_weak_tail or in part \ref
- /// half_synchronous_weak_last. Settled period don't checked asynchronously,
- /// but only by the \ref txn_managed::commit() and \ref env::sync_to_disk()
- /// functions. Therefore, in cases where transactions are committed
- /// infrequently and/or irregularly, polling by \ref env::poll_sync_to_disk()
- /// may be a reasonable solution to timeout enforcement. The default is 0,
- /// than mean no any timeout checked, and no additional flush will be made.
- ///
- /// \param [in] seconds_16dot16 The period in 1/65536 of second when a
- /// synchronous flush would be made since the last unsteady commit.
- inline env &set_sync_period(unsigned seconds_16dot16);
-
- /// \brief Sets relative period since the last unsteady commit to force flush
- /// the data buffers to disk, for non-sync durability modes.
- ///
- /// The relative period value affects all processes which operates with given
- /// environment until the last process close environment or a new value will
- /// be settled.
- /// Data is always written to disk when \ref txn_managed::commit() is called,
- /// but the operating system may keep it buffered. MDBX always flushes the OS
- /// buffers upon commit as well, unless the environment was opened with \ref
- /// whole_fragile, \ref lazy_weak_tail or in part \ref
- /// half_synchronous_weak_last. Settled period don't checked asynchronously,
- /// but only by the \ref txn_managed::commit() and \ref env::sync_to_disk()
- /// functions. Therefore, in cases where transactions are committed
- /// infrequently and/or irregularly, polling by \ref env::poll_sync_to_disk()
- /// may be a reasonable solution to timeout enforcement. The default is 0,
- /// than mean no any timeout checked, and no additional flush will be made.
- ///
- /// \param [in] seconds The period in second when a synchronous flush would
- /// be made since the last unsteady commit.
- inline env &set_sync_period(double seconds);
-
- /// \brief Alter environment flags.
- inline env &alter_flags(MDBX_env_flags_t flags, bool on_off);
-
- /// \brief Set all size-related parameters of environment.
- inline env &set_geometry(const geometry &size);
-
- /// \brief Flush the environment data buffers.
- /// \return `True` if sync done or no data to sync, or `false` if the
- /// environment is busy by other thread or none of the thresholds are reached.
- inline bool sync_to_disk(bool force = true, bool nonblock = false);
-
- /// \brief Performs non-blocking polling of sync-to-disk thresholds.
- /// \return `True` if sync done or no data to sync, or `false` if the
- /// environment is busy by other thread or none of the thresholds are reached.
- bool poll_sync_to_disk() { return sync_to_disk(false, true); }
-
- /// \brief Close a key-value map (aka sub-database) handle. Normally
- /// unnecessary.
- ///
- /// Closing a database handle is not necessary, but lets \ref txn::open_map()
- /// reuse the handle value. Usually it's better to set a bigger
- /// \ref env::operate_parameters::max_maps, unless that value would be
- /// large.
- ///
- /// \note Use with care.
- /// This call is synchronized via mutex with other calls \ref close_map(), but
- /// NOT with other transactions running by other threads. The "next" version
- /// of libmdbx (\ref MithrilDB) will solve this issue.
- ///
- /// Handles should only be closed if no other threads are going to reference
- /// the database handle or one of its cursors any further. Do not close a
- /// handle if an existing transaction has modified its database. Doing so can
- /// cause misbehavior from database corruption to errors like
- /// \ref MDBX_BAD_DBI (since the DB name is gone).
- inline void close_map(const map_handle &);
-
- /// \brief Reader information
- struct reader_info {
- int slot; ///< The reader lock table slot number.
- mdbx_pid_t pid; ///< The reader process ID.
- mdbx_tid_t thread; ///< The reader thread ID.
- uint64_t transaction_id; ///< The ID of the transaction being read,
- ///< i.e. the MVCC-snapshot number.
- uint64_t transaction_lag; ///< The lag from a recent MVCC-snapshot,
- ///< i.e. the number of committed write
- /// transactions since the current read
- /// transaction started.
- size_t bytes_used; ///< The number of last used page in the MVCC-snapshot
- ///< which being read, i.e. database file can't shrinked
- ///< beyond this.
- size_t bytes_retained; ///< The total size of the database pages that
- ///< were retired by committed write transactions
- ///< after the reader's MVCC-snapshot, i.e. the space
- ///< which would be freed after the Reader releases
- ///< the MVCC-snapshot for reuse by completion read
- ///< transaction.
-
- MDBX_CXX11_CONSTEXPR reader_info(int slot, mdbx_pid_t pid,
- mdbx_tid_t thread, uint64_t txnid,
- uint64_t lag, size_t used,
- size_t retained) noexcept;
- };
-
- /// \brief Enumerate readers.
- ///
- /// The VISITOR class must have `int operator(const reader_info&, int serial)`
- /// which should return \ref continue_loop (zero) to continue enumeration,
- /// or any non-zero value to exit.
- ///
- /// \returns The last value returned from visitor' functor.
- template <typename VISITOR> inline int enumerate_readers(VISITOR &visitor);
-
- /// \brief Checks for stale readers in the lock table and
- /// return number of cleared slots.
- inline unsigned check_readers();
-
- /// \brief Sets a Handle-Slow-Readers callback to resolve database
- /// full/overflow issue due to a reader(s) which prevents the old data from
- /// being recycled.
- ///
- /// Such callback will be triggered in a case where there is not enough free
- /// space in the database due to long read transaction(s) which impedes
- /// reusing the pages of an old MVCC snapshot(s).
- ///
- /// Using this callback you can choose how to resolve the situation:
- /// - abort the write transaction with an error;
- /// - wait for the read transaction(s) to complete;
- /// - notify a thread performing a long-lived read transaction
- /// and wait for an effect;
- /// - kill the thread or whole process that performs the long-lived read
- /// transaction;
- ///
- /// \see long-lived-read
- inline env &set_HandleSlowReaders(MDBX_hsr_func *);
-
- /// \brief Returns the current Handle-Slow-Readers callback used to resolve
- /// database full/overflow issue due to a reader(s) which prevents the old
- /// data from being recycled.
- /// \see set_HandleSlowReaders()
- inline MDBX_hsr_func *get_HandleSlowReaders() const noexcept;
-
- /// \brief Starts read (read-only) transaction.
- inline txn_managed start_read() const;
-
- /// \brief Creates but not start read transaction.
- inline txn_managed prepare_read() const;
-
- /// \brief Starts write (read-write) transaction.
- inline txn_managed start_write(bool dont_wait = false);
-
- /// \brief Tries to start write (read-write) transaction without blocking.
- inline txn_managed try_start_write();
-};
-
-/// \brief Managed database environment.
-///
-/// As other managed classes, `env_managed` destroys the represented underlying
-/// object from the own class destructor, but disallows copying and assignment
-/// for instances.
-///
-/// An environment supports multiple key-value databases (aka key-value spaces
-/// or tables), all residing in the same shared-memory map.
-class LIBMDBX_API_TYPE env_managed : public env {
- using inherited = env;
- /// delegated constructor for RAII
- MDBX_CXX11_CONSTEXPR env_managed(MDBX_env *ptr) noexcept : inherited(ptr) {}
- void setup(unsigned max_maps, unsigned max_readers = 0);
-
-public:
- MDBX_CXX11_CONSTEXPR env_managed() noexcept = default;
-
- /// \brief Open existing database.
-#ifdef MDBX_STD_FILESYSTEM_PATH
- env_managed(const MDBX_STD_FILESYSTEM_PATH &, const operate_parameters &,
- bool accede = true);
-#endif /* MDBX_STD_FILESYSTEM_PATH */
-#if defined(_WIN32) || defined(_WIN64)
- env_managed(const ::std::wstring &, const operate_parameters &,
- bool accede = true);
-#endif /* Windows */
- env_managed(const ::std::string &, const operate_parameters &,
- bool accede = true);
-
- /// \brief Additional parameters for creating a new database.
- struct create_parameters {
- env::geometry geometry;
- mdbx_mode_t file_mode_bits{0640};
- bool use_subdirectory{false};
- MDBX_CXX11_CONSTEXPR create_parameters() noexcept = default;
- create_parameters(const create_parameters &) noexcept = default;
- };
-
- /// \brief Create new or open existing database.
-#ifdef MDBX_STD_FILESYSTEM_PATH
- env_managed(const MDBX_STD_FILESYSTEM_PATH &, const create_parameters &,
- const operate_parameters &, bool accede = true);
-#endif /* MDBX_STD_FILESYSTEM_PATH */
-#if defined(_WIN32) || defined(_WIN64)
- env_managed(const ::std::wstring &, const create_parameters &,
- const operate_parameters &, bool accede = true);
-#endif /* Windows */
- env_managed(const ::std::string &, const create_parameters &,
- const operate_parameters &, bool accede = true);
-
- /// \brief Explicitly closes the environment and release the memory map.
- ///
- /// Only a single thread may call this function. All transactions, databases,
- /// and cursors must already be closed before calling this function. Attempts
- /// to use any such handles after calling this function will cause a
- /// `SIGSEGV`. The environment handle will be freed and must not be used again
- /// after this call.
- ///
- /// \param [in] dont_sync A dont'sync flag, if non-zero the last checkpoint
- /// will be kept "as is" and may be still "weak" in the \ref lazy_weak_tail
- /// or \ref whole_fragile modes. Such "weak" checkpoint will be ignored
- /// on opening next time, and transactions since the last non-weak checkpoint
- /// (meta-page update) will rolledback for consistency guarantee.
- void close(bool dont_sync = false);
-
- env_managed(env_managed &&) = default;
- env_managed &operator=(env_managed &&other) {
- if (MDBX_UNLIKELY(handle_))
- MDBX_CXX20_UNLIKELY {
- assert(handle_ != other.handle_);
- close();
- }
- inherited::operator=(std::move(other));
- return *this;
- }
- env_managed(const env_managed &) = delete;
- env_managed &operator=(const env_managed &) = delete;
- virtual ~env_managed() noexcept;
-};
-
-/// \brief Unmanaged database transaction.
-///
-/// Like other unmanaged classes, `txn` allows copying and assignment for
-/// instances, but does not destroys the represented underlying object from the
-/// own class destructor.
-///
-/// All database operations require a transaction handle. Transactions may be
-/// read-only or read-write.
-class LIBMDBX_API_TYPE txn {
-protected:
- friend class cursor;
- MDBX_txn *handle_{nullptr};
- MDBX_CXX11_CONSTEXPR txn(MDBX_txn *ptr) noexcept;
-
-public:
- MDBX_CXX11_CONSTEXPR txn() noexcept = default;
- txn(const txn &) noexcept = default;
- inline txn &operator=(txn &&other) noexcept;
- inline txn(txn &&other) noexcept;
- inline ~txn() noexcept;
-
- MDBX_CXX14_CONSTEXPR operator bool() const noexcept;
- MDBX_CXX14_CONSTEXPR operator const MDBX_txn *() const;
- MDBX_CXX14_CONSTEXPR operator MDBX_txn *();
- friend MDBX_CXX11_CONSTEXPR bool operator==(const txn &a,
- const txn &b) noexcept;
- friend MDBX_CXX11_CONSTEXPR bool operator!=(const txn &a,
- const txn &b) noexcept;
-
- /// \brief Returns the transaction's environment.
- inline ::mdbx::env env() const noexcept;
- /// \brief Returns transaction's flags.
- inline MDBX_txn_flags_t flags() const;
- /// \brief Return the transaction's ID.
- inline uint64_t id() const;
-
- /// \brief Checks whether the given data is on a dirty page.
- inline bool is_dirty(const void *ptr) const;
-
- /// \brief Checks whether the transaction is read-only.
- bool is_readonly() const { return (flags() & MDBX_TXN_RDONLY) != 0; }
-
- /// \brief Checks whether the transaction is read-write.
- bool is_readwrite() const { return (flags() & MDBX_TXN_RDONLY) == 0; }
-
- using info = ::MDBX_txn_info;
- /// \brief Returns information about the MDBX transaction.
- inline info get_info(bool scan_reader_lock_table = false) const;
-
- /// \brief Returns maximal write transaction size (i.e. limit for summary
- /// volume of dirty pages) in bytes.
- size_t size_max() const { return env().transaction_size_max(); }
-
- /// \brief Returns current write transaction size (i.e.summary volume of dirty
- /// pages) in bytes.
- size_t size_current() const {
- assert(is_readwrite());
- return size_t(get_info().txn_space_dirty);
- }
-
- //----------------------------------------------------------------------------
-
- /// \brief Reset a read-only transaction.
- inline void reset_reading();
-
- /// \brief Renew a read-only transaction.
- inline void renew_reading();
-
- /// \brief Start nested write transaction.
- txn_managed start_nested();
-
- /// \brief Opens cursor for specified key-value map handle.
- inline cursor_managed open_cursor(map_handle map);
-
- /// \brief Open existing key-value map.
- inline map_handle open_map(
- const char *name,
- const ::mdbx::key_mode key_mode = ::mdbx::key_mode::usual,
- const ::mdbx::value_mode value_mode = ::mdbx::value_mode::single) const;
- /// \brief Open existing key-value map.
- inline map_handle open_map(
- const ::std::string &name,
- const ::mdbx::key_mode key_mode = ::mdbx::key_mode::usual,
- const ::mdbx::value_mode value_mode = ::mdbx::value_mode::single) const;
-
- /// \brief Create new or open existing key-value map.
- inline map_handle
- create_map(const char *name,
- const ::mdbx::key_mode key_mode = ::mdbx::key_mode::usual,
- const ::mdbx::value_mode value_mode = ::mdbx::value_mode::single);
- /// \brief Create new or open existing key-value map.
- inline map_handle
- create_map(const ::std::string &name,
- const ::mdbx::key_mode key_mode = ::mdbx::key_mode::usual,
- const ::mdbx::value_mode value_mode = ::mdbx::value_mode::single);
-
- /// \brief Drops key-value map using handle.
- inline void drop_map(map_handle map);
- /// \brief Drops key-value map using name.
- /// \return `True` if the key-value map existed and was deleted, either
- /// `false` if the key-value map did not exist and there is nothing to delete.
- bool drop_map(const char *name, bool throw_if_absent = false);
- /// \brief Drop key-value map.
- /// \return `True` if the key-value map existed and was deleted, either
- /// `false` if the key-value map did not exist and there is nothing to delete.
- inline bool drop_map(const ::std::string &name, bool throw_if_absent = false);
-
- /// \brief Clear key-value map.
- inline void clear_map(map_handle map);
- /// \return `True` if the key-value map existed and was cleared, either
- /// `false` if the key-value map did not exist and there is nothing to clear.
- bool clear_map(const char *name, bool throw_if_absent = false);
- /// \return `True` if the key-value map existed and was cleared, either
- /// `false` if the key-value map did not exist and there is nothing to clear.
- inline bool clear_map(const ::std::string &name,
- bool throw_if_absent = false);
-
- using map_stat = ::MDBX_stat;
- /// \brief Returns statistics for a sub-database.
- inline map_stat get_map_stat(map_handle map) const;
- /// \brief Returns depth (bitmask) information of nested dupsort (multi-value)
- /// B+trees for given database.
- inline uint32_t get_tree_deepmask(map_handle map) const;
- /// \brief Returns information about key-value map (aka sub-database) handle.
- inline map_handle::info get_handle_info(map_handle map) const;
-
- using canary = ::MDBX_canary;
- /// \brief Set integers markers (aka "canary") associated with the
- /// environment.
- inline txn &put_canary(const canary &);
- /// \brief Returns fours integers markers (aka "canary") associated with the
- /// environment.
- inline canary get_canary() const;
-
- /// Reads sequence generator associated with a key-value map (aka
- /// sub-database).
- inline uint64_t sequence(map_handle map) const;
- /// \brief Reads and increment sequence generator associated with a key-value
- /// map (aka sub-database).
- inline uint64_t sequence(map_handle map, uint64_t increment);
-
- /// \brief Compare two keys according to a particular key-value map (aka
- /// sub-database).
- inline int compare_keys(map_handle map, const slice &a,
- const slice &b) const noexcept;
- /// \brief Compare two values according to a particular key-value map (aka
- /// sub-database).
- inline int compare_values(map_handle map, const slice &a,
- const slice &b) const noexcept;
- /// \brief Compare keys of two pairs according to a particular key-value map
- /// (aka sub-database).
- inline int compare_keys(map_handle map, const pair &a,
- const pair &b) const noexcept;
- /// \brief Compare values of two pairs according to a particular key-value map
- /// (aka sub-database).
- inline int compare_values(map_handle map, const pair &a,
- const pair &b) const noexcept;
-
- /// \brief Get value by key from a key-value map (aka sub-database).
- inline slice get(map_handle map, const slice &key) const;
- /// \brief Get first of multi-value and values count by key from a key-value
- /// multimap (aka sub-database).
- inline slice get(map_handle map, slice key, size_t &values_count) const;
- /// \brief Get value by key from a key-value map (aka sub-database).
- inline slice get(map_handle map, const slice &key,
- const slice &value_at_absence) const;
- /// \brief Get first of multi-value and values count by key from a key-value
- /// multimap (aka sub-database).
- inline slice get(map_handle map, slice key, size_t &values_count,
- const slice &value_at_absence) const;
- /// \brief Get value for equal or great key from a database.
- /// \return Bundle of key-value pair and boolean flag,
- /// which will be `true` if the exact key was found and `false` otherwise.
- inline pair_result get_equal_or_great(map_handle map, const slice &key) const;
- /// \brief Get value for equal or great key from a database.
- /// \return Bundle of key-value pair and boolean flag,
- /// which will be `true` if the exact key was found and `false` otherwise.
- inline pair_result get_equal_or_great(map_handle map, const slice &key,
- const slice &value_at_absence) const;
-
- inline MDBX_error_t put(map_handle map, const slice &key, slice *value,
- MDBX_put_flags_t flags) noexcept;
- inline void put(map_handle map, const slice &key, slice value, put_mode mode);
- inline void insert(map_handle map, const slice &key, slice value);
- inline value_result try_insert(map_handle map, const slice &key, slice value);
- inline slice insert_reserve(map_handle map, const slice &key,
- size_t value_length);
- inline value_result try_insert_reserve(map_handle map, const slice &key,
- size_t value_length);
-
- inline void upsert(map_handle map, const slice &key, const slice &value);
- inline slice upsert_reserve(map_handle map, const slice &key,
- size_t value_length);
-
- inline void update(map_handle map, const slice &key, const slice &value);
- inline bool try_update(map_handle map, const slice &key, const slice &value);
- inline slice update_reserve(map_handle map, const slice &key,
- size_t value_length);
- inline value_result try_update_reserve(map_handle map, const slice &key,
- size_t value_length);
-
- /// \brief Removes all values for given key.
- inline bool erase(map_handle map, const slice &key);
-
- /// \brief Removes the particular multi-value entry of the key.
- inline bool erase(map_handle map, const slice &key, const slice &value);
-
- /// \brief Replaces the particular multi-value of the key with a new value.
- inline void replace(map_handle map, const slice &key, slice old_value,
- const slice &new_value);
-
- /// \brief Removes and return a value of the key.
- template <class ALLOCATOR, typename CAPACITY_POLICY>
- inline buffer<ALLOCATOR, CAPACITY_POLICY>
- extract(map_handle map, const slice &key,
- const typename buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type &
- allocator = buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type());
-
- /// \brief Replaces and returns a value of the key with new one.
- template <class ALLOCATOR, typename CAPACITY_POLICY>
- inline buffer<ALLOCATOR, CAPACITY_POLICY>
- replace(map_handle map, const slice &key, const slice &new_value,
- const typename buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type &
- allocator = buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type());
-
- template <class ALLOCATOR, typename CAPACITY_POLICY>
- inline buffer<ALLOCATOR, CAPACITY_POLICY> replace_reserve(
- map_handle map, const slice &key, slice &new_value,
- const typename buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type
- &allocator = buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type());
-
- /// \brief Adding a key-value pair, provided that ascending order of the keys
- /// and (optionally) values are preserved.
- ///
- /// Instead of splitting the full b+tree pages, the data will be placed on new
- /// ones. Thus appending is about two times faster than insertion, and the
- /// pages will be filled in completely mostly but not half as after splitting
- /// ones. On the other hand, any subsequent insertion or update with an
- /// increase in the length of the value will be twice as slow, since it will
- /// require splitting already filled pages.
- ///
- /// \param [in] map A map handle to append
- /// \param [in] key A key to be append
- /// \param [in] value A value to store with the key
- /// \param [in] multivalue_order_preserved
- /// If `multivalue_order_preserved == true` then the same rules applied for
- /// to pages of nested b+tree of multimap's values.
- inline void append(map_handle map, const slice &key, const slice &value,
- bool multivalue_order_preserved = true);
-
- size_t put_multiple(map_handle map, const slice &key,
- const size_t value_length, const void *values_array,
- size_t values_count, put_mode mode,
- bool allow_partial = false);
- template <typename VALUE>
- void put_multiple(map_handle map, const slice &key,
- const ::std::vector<VALUE> &vector, put_mode mode) {
- put_multiple(map, key, sizeof(VALUE), vector.data(), vector.size(), mode,
- false);
- }
-
- inline ptrdiff_t estimate(map_handle map, pair from, pair to) const;
- inline ptrdiff_t estimate(map_handle map, slice from, slice to) const;
- inline ptrdiff_t estimate_from_first(map_handle map, slice to) const;
- inline ptrdiff_t estimate_to_last(map_handle map, slice from) const;
-};
-
-/// \brief Managed database transaction.
-///
-/// As other managed classes, `txn_managed` destroys the represented underlying
-/// object from the own class destructor, but disallows copying and assignment
-/// for instances.
-///
-/// All database operations require a transaction handle. Transactions may be
-/// read-only or read-write.
-class LIBMDBX_API_TYPE txn_managed : public txn {
- using inherited = txn;
- friend class env;
- friend class txn;
- /// delegated constructor for RAII
- MDBX_CXX11_CONSTEXPR txn_managed(MDBX_txn *ptr) noexcept : inherited(ptr) {}
-
-public:
- MDBX_CXX11_CONSTEXPR txn_managed() noexcept = default;
- txn_managed(txn_managed &&) = default;
- txn_managed &operator=(txn_managed &&other) {
- if (MDBX_UNLIKELY(handle_))
- MDBX_CXX20_UNLIKELY {
- assert(handle_ != other.handle_);
- abort();
- }
- inherited::operator=(std::move(other));
- return *this;
- }
- txn_managed(const txn_managed &) = delete;
- txn_managed &operator=(const txn_managed &) = delete;
- ~txn_managed() noexcept;
-
- //----------------------------------------------------------------------------
-
- /// \brief Abandon all the operations of the transaction instead of saving
- /// them.
- void abort();
-
- /// \brief Commit all the operations of a transaction into the database.
- void commit();
-};
-
-/// \brief Unmanaged cursor.
-///
-/// Like other unmanaged classes, `cursor` allows copying and assignment for
-/// instances, but does not destroys the represented underlying object from the
-/// own class destructor.
-///
-/// \copydetails MDBX_cursor
-class LIBMDBX_API_TYPE cursor {
-protected:
- MDBX_cursor *handle_{nullptr};
- MDBX_CXX11_CONSTEXPR cursor(MDBX_cursor *ptr) noexcept;
-
-public:
- MDBX_CXX11_CONSTEXPR cursor() noexcept = default;
- cursor(const cursor &) noexcept = default;
- inline cursor &operator=(cursor &&other) noexcept;
- inline cursor(cursor &&other) noexcept;
- inline ~cursor() noexcept;
- MDBX_CXX14_CONSTEXPR operator bool() const noexcept;
- MDBX_CXX14_CONSTEXPR operator const MDBX_cursor *() const;
- MDBX_CXX14_CONSTEXPR operator MDBX_cursor *();
- friend MDBX_CXX11_CONSTEXPR bool operator==(const cursor &a,
- const cursor &b) noexcept;
- friend MDBX_CXX11_CONSTEXPR bool operator!=(const cursor &a,
- const cursor &b) noexcept;
-
- enum move_operation {
- first = MDBX_FIRST,
- last = MDBX_LAST,
- next = MDBX_NEXT,
- previous = MDBX_PREV,
- get_current = MDBX_GET_CURRENT,
-
- multi_prevkey_lastvalue = MDBX_PREV_NODUP,
- multi_currentkey_firstvalue = MDBX_FIRST_DUP,
- multi_currentkey_prevvalue = MDBX_PREV_DUP,
- multi_currentkey_nextvalue = MDBX_NEXT_DUP,
- multi_currentkey_lastvalue = MDBX_LAST_DUP,
- multi_nextkey_firstvalue = MDBX_NEXT_NODUP,
-
- multi_find_pair = MDBX_GET_BOTH,
- multi_exactkey_lowerboundvalue = MDBX_GET_BOTH_RANGE,
-
- find_key = MDBX_SET,
- key_exact = MDBX_SET_KEY,
- key_lowerbound = MDBX_SET_RANGE
- };
-
- struct move_result : public pair_result {
- inline move_result(const cursor &cursor, bool throw_notfound);
- inline move_result(cursor &cursor, move_operation operation,
- bool throw_notfound);
- inline move_result(cursor &cursor, move_operation operation,
- const slice &key, bool throw_notfound);
- inline move_result(cursor &cursor, move_operation operation,
- const slice &key, const slice &value,
- bool throw_notfound);
- move_result(const move_result &) noexcept = default;
- move_result &operator=(const move_result &) noexcept = default;
- };
-
-protected:
- inline bool move(move_operation operation, MDBX_val *key, MDBX_val *value,
- bool throw_notfound) const
- /* fake const, i.e. for some operations */;
- inline ptrdiff_t estimate(move_operation operation, MDBX_val *key,
- MDBX_val *value) const;
-
-public:
- inline move_result move(move_operation operation, bool throw_notfound);
- inline move_result to_first(bool throw_notfound = true);
- inline move_result to_previous(bool throw_notfound = true);
- inline move_result to_previous_last_multi(bool throw_notfound = true);
- inline move_result to_current_first_multi(bool throw_notfound = true);
- inline move_result to_current_prev_multi(bool throw_notfound = true);
- inline move_result current(bool throw_notfound = true) const;
- inline move_result to_current_next_multi(bool throw_notfound = true);
- inline move_result to_current_last_multi(bool throw_notfound = true);
- inline move_result to_next_first_multi(bool throw_notfound = true);
- inline move_result to_next(bool throw_notfound = true);
- inline move_result to_last(bool throw_notfound = true);
-
- inline move_result move(move_operation operation, const slice &key,
- bool throw_notfound);
- inline move_result find(const slice &key, bool throw_notfound = true);
- inline move_result lower_bound(const slice &key, bool throw_notfound = true);
-
- inline move_result move(move_operation operation, const slice &key,
- const slice &value, bool throw_notfound);
- inline move_result find_multivalue(const slice &key, const slice &value,
- bool throw_notfound = true);
- inline move_result lower_bound_multivalue(const slice &key,
- const slice &value,
- bool throw_notfound = false);
-
- inline bool seek(const slice &key);
- inline bool move(move_operation operation, slice &key, slice &value,
- bool throw_notfound);
-
- /// \brief Return count of duplicates for current key.
- inline size_t count_multivalue() const;
-
- inline bool eof() const;
- inline bool on_first() const;
- inline bool on_last() const;
- inline ptrdiff_t estimate(slice key, slice value) const;
- inline ptrdiff_t estimate(slice key) const;
- inline ptrdiff_t estimate(move_operation operation) const;
-
- //----------------------------------------------------------------------------
-
- /// \brief Renew/bind a cursor with a new transaction and previously used
- /// key-value map handle.
- inline void renew(::mdbx::txn &txn);
-
- /// \brief Bind/renew a cursor with a new transaction and specified key-value
- /// map handle.
- inline void bind(::mdbx::txn &txn, ::mdbx::map_handle map_handle);
-
- /// \brief Returns the cursor's transaction.
- inline ::mdbx::txn txn() const;
- inline map_handle map() const;
-
- inline operator ::mdbx::txn() const { return txn(); }
- inline operator ::mdbx::map_handle() const { return map(); }
-
- inline MDBX_error_t put(const slice &key, slice *value,
- MDBX_put_flags_t flags) noexcept;
- inline void insert(const slice &key, slice value);
- inline value_result try_insert(const slice &key, slice value);
- inline slice insert_reserve(const slice &key, size_t value_length);
- inline value_result try_insert_reserve(const slice &key, size_t value_length);
-
- inline void upsert(const slice &key, const slice &value);
- inline slice upsert_reserve(const slice &key, size_t value_length);
-
- inline void update(const slice &key, const slice &value);
- inline bool try_update(const slice &key, const slice &value);
- inline slice update_reserve(const slice &key, size_t value_length);
- inline value_result try_update_reserve(const slice &key, size_t value_length);
-
- /// \brief Removes single key-value pair or all multi-values at the current
- /// cursor position.
- inline bool erase(bool whole_multivalue = false);
-
- /// \brief Seeks and removes first value or whole multi-value of the given
- /// key.
- /// \return `True` if the key is found and a value(s) is removed.
- inline bool erase(const slice &key, bool whole_multivalue = true);
-
- /// \brief Seeks and removes the particular multi-value entry of the key.
- /// \return `True` if the given key-value pair is found and removed.
- inline bool erase(const slice &key, const slice &value);
-};
-
-/// \brief Managed cursor.
-///
-/// As other managed classes, `cursor_managed` destroys the represented
-/// underlying object from the own class destructor, but disallows copying and
-/// assignment for instances.
-///
-/// \copydetails MDBX_cursor
-class LIBMDBX_API_TYPE cursor_managed : public cursor {
- using inherited = cursor;
- friend class txn;
- /// delegated constructor for RAII
- MDBX_CXX11_CONSTEXPR cursor_managed(MDBX_cursor *ptr) noexcept
- : inherited(ptr) {}
-
-public:
- /// \brief Creates a new managed cursor with underlying object.
- cursor_managed() : cursor_managed(::mdbx_cursor_create(nullptr)) {
- if (MDBX_UNLIKELY(!handle_))
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_ENOMEM);
- }
-
- /// \brief Explicitly closes the cursor.
- void close();
-
- cursor_managed(cursor_managed &&) = default;
- cursor_managed &operator=(cursor_managed &&other) {
- if (MDBX_UNLIKELY(handle_))
- MDBX_CXX20_UNLIKELY {
- assert(handle_ != other.handle_);
- close();
- }
- inherited::operator=(std::move(other));
- return *this;
- }
-
- cursor_managed(const cursor_managed &) = delete;
- cursor_managed &operator=(const cursor_managed &) = delete;
- ~cursor_managed() noexcept { ::mdbx_cursor_close(handle_); }
-};
-
-//------------------------------------------------------------------------------
-
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &, const slice &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &, const pair &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &, const pair_result &);
-template <class ALLOCATOR, typename CAPACITY_POLICY>
-inline ::std::ostream &
-operator<<(::std::ostream &out, const buffer<ALLOCATOR, CAPACITY_POLICY> &it) {
- return (it.is_freestanding()
- ? out << "buf-" << it.headroom() << "." << it.tailroom()
- : out << "ref-")
- << it.slice();
-}
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &,
- const env::geometry::size &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &, const env::geometry &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &,
- const env::operate_parameters &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &, const env::mode &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &,
- const env::durability &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &,
- const env::reclaiming_options &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &,
- const env::operate_options &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &,
- const env_managed::create_parameters &);
-
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &,
- const MDBX_log_level_t &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &,
- const MDBX_debug_flags_t &);
-LIBMDBX_API ::std::ostream &operator<<(::std::ostream &, const error &);
-inline ::std::ostream &operator<<(::std::ostream &out,
- const MDBX_error_t &errcode) {
- return out << error(errcode);
-}
-
-//==============================================================================
-//
-// Inline body of the libmdbx C++ API
-//
-
-MDBX_CXX11_CONSTEXPR const version_info &get_version() noexcept {
- return ::mdbx_version;
-}
-MDBX_CXX11_CONSTEXPR const build_info &get_build() noexcept {
- return ::mdbx_build;
-}
-
-static MDBX_CXX17_CONSTEXPR size_t strlen(const char *c_str) noexcept {
-#if defined(__cpp_lib_is_constant_evaluated) && \
- __cpp_lib_is_constant_evaluated >= 201811L
- if (::std::is_constant_evaluated()) {
- for (size_t i = 0; c_str; ++i)
- if (!c_str[i])
- return i;
- return 0;
- }
-#endif /* __cpp_lib_is_constant_evaluated >= 201811 */
-#if defined(__cpp_lib_string_view) && __cpp_lib_string_view >= 201606L
- return c_str ? ::std::string_view(c_str).length() : 0;
-#else
- return c_str ? ::std::strlen(c_str) : 0;
-#endif
-}
-
-MDBX_MAYBE_UNUSED static MDBX_CXX20_CONSTEXPR void *
-memcpy(void *dest, const void *src, size_t bytes) noexcept {
-#if defined(__cpp_lib_is_constant_evaluated) && \
- __cpp_lib_is_constant_evaluated >= 201811L
- if (::std::is_constant_evaluated()) {
- for (size_t i = 0; i < bytes; ++i)
- static_cast<byte *>(dest)[i] = static_cast<const byte *>(src)[i];
- return dest;
- } else
-#endif /* __cpp_lib_is_constant_evaluated >= 201811 */
- return ::std::memcpy(dest, src, bytes);
-}
-
-static MDBX_CXX20_CONSTEXPR int memcmp(const void *a, const void *b,
- size_t bytes) noexcept {
-#if defined(__cpp_lib_is_constant_evaluated) && \
- __cpp_lib_is_constant_evaluated >= 201811L
- if (::std::is_constant_evaluated()) {
- for (size_t i = 0; i < bytes; ++i) {
- const int diff =
- static_cast<const byte *>(a)[i] - static_cast<const byte *>(b)[i];
- if (diff)
- return diff;
- }
- return 0;
- } else
-#endif /* __cpp_lib_is_constant_evaluated >= 201811 */
- return ::std::memcmp(a, b, bytes);
-}
-
-static MDBX_CXX14_CONSTEXPR size_t check_length(size_t bytes) {
- if (MDBX_UNLIKELY(bytes > size_t(MDBX_MAXDATASIZE)))
- MDBX_CXX20_UNLIKELY throw_max_length_exceeded();
- return bytes;
-}
-
-static MDBX_CXX14_CONSTEXPR size_t check_length(size_t headroom,
- size_t payload) {
- return check_length(check_length(headroom) + check_length(payload));
-}
-
-MDBX_MAYBE_UNUSED static MDBX_CXX14_CONSTEXPR size_t
-check_length(size_t headroom, size_t payload, size_t tailroom) {
- return check_length(check_length(headroom, payload) + check_length(tailroom));
-}
-
-inline bool exception_thunk::is_clean() const noexcept { return !captured_; }
-
-inline void exception_thunk::capture() noexcept {
- assert(is_clean());
- captured_ = ::std::current_exception();
-}
-
-inline void exception_thunk::rethrow_captured() const {
- if (captured_)
- MDBX_CXX20_UNLIKELY ::std::rethrow_exception(captured_);
-}
-
-//------------------------------------------------------------------------------
-
-MDBX_CXX11_CONSTEXPR error::error(MDBX_error_t error_code) noexcept
- : code_(error_code) {}
-
-inline error &error::operator=(MDBX_error_t error_code) noexcept {
- code_ = error_code;
- return *this;
-}
-
-MDBX_CXX11_CONSTEXPR bool operator==(const error &a, const error &b) noexcept {
- return a.code_ == b.code_;
-}
-
-MDBX_CXX11_CONSTEXPR bool operator!=(const error &a, const error &b) noexcept {
- return !(a == b);
-}
-
-MDBX_CXX11_CONSTEXPR bool error::is_success() const noexcept {
- return code_ == MDBX_SUCCESS;
-}
-
-MDBX_CXX11_CONSTEXPR bool error::is_result_true() const noexcept {
- return code_ == MDBX_RESULT_FALSE;
-}
-
-MDBX_CXX11_CONSTEXPR bool error::is_result_false() const noexcept {
- return code_ == MDBX_RESULT_TRUE;
-}
-
-MDBX_CXX11_CONSTEXPR bool error::is_failure() const noexcept {
- return code_ != MDBX_SUCCESS && code_ != MDBX_RESULT_TRUE;
-}
-
-MDBX_CXX11_CONSTEXPR MDBX_error_t error::code() const noexcept { return code_; }
-
-MDBX_CXX11_CONSTEXPR bool error::is_mdbx_error() const noexcept {
- return (code() >= MDBX_FIRST_LMDB_ERRCODE &&
- code() <= MDBX_LAST_LMDB_ERRCODE) ||
- (code() >= MDBX_FIRST_ADDED_ERRCODE &&
- code() <= MDBX_LAST_ADDED_ERRCODE);
-}
-
-inline void error::throw_exception(int error_code) {
- const error trouble(static_cast<MDBX_error_t>(error_code));
- trouble.throw_exception();
-}
-
-inline void error::throw_on_failure() const {
- if (MDBX_UNLIKELY(is_failure()))
- MDBX_CXX20_UNLIKELY throw_exception();
-}
-
-inline void error::success_or_throw() const {
- if (MDBX_UNLIKELY(!is_success()))
- MDBX_CXX20_UNLIKELY throw_exception();
-}
-
-inline void error::success_or_throw(const exception_thunk &thunk) const {
- assert(thunk.is_clean() || code() != MDBX_SUCCESS);
- if (MDBX_UNLIKELY(!is_success())) {
- MDBX_CXX20_UNLIKELY if (!thunk.is_clean()) thunk.rethrow_captured();
- else throw_exception();
- }
-}
-
-inline void error::panic_on_failure(const char *context_where,
- const char *func_who) const noexcept {
- if (MDBX_UNLIKELY(is_failure()))
- MDBX_CXX20_UNLIKELY panic(context_where, func_who);
-}
-
-inline void error::success_or_panic(const char *context_where,
- const char *func_who) const noexcept {
- if (MDBX_UNLIKELY(!is_success()))
- MDBX_CXX20_UNLIKELY panic(context_where, func_who);
-}
-
-inline void error::throw_on_nullptr(const void *ptr, MDBX_error_t error_code) {
- if (MDBX_UNLIKELY(ptr == nullptr))
- MDBX_CXX20_UNLIKELY error(error_code).throw_exception();
-}
-
-inline void error::throw_on_failure(int error_code) {
- error rc(static_cast<MDBX_error_t>(error_code));
- rc.throw_on_failure();
-}
-
-inline void error::success_or_throw(MDBX_error_t error_code) {
- error rc(error_code);
- rc.success_or_throw();
-}
-
-inline bool error::boolean_or_throw(int error_code) {
- switch (error_code) {
- case MDBX_RESULT_FALSE:
- return false;
- case MDBX_RESULT_TRUE:
- return true;
- default:
- MDBX_CXX20_UNLIKELY throw_exception(error_code);
- }
-}
-
-inline void error::success_or_throw(int error_code,
- const exception_thunk &thunk) {
- error rc(static_cast<MDBX_error_t>(error_code));
- rc.success_or_throw(thunk);
-}
-
-inline void error::panic_on_failure(int error_code, const char *context_where,
- const char *func_who) noexcept {
- error rc(static_cast<MDBX_error_t>(error_code));
- rc.panic_on_failure(context_where, func_who);
-}
-
-inline void error::success_or_panic(int error_code, const char *context_where,
- const char *func_who) noexcept {
- error rc(static_cast<MDBX_error_t>(error_code));
- rc.success_or_panic(context_where, func_who);
-}
-
-//------------------------------------------------------------------------------
-
-MDBX_CXX11_CONSTEXPR slice::slice() noexcept : ::MDBX_val({nullptr, 0}) {}
-
-MDBX_CXX14_CONSTEXPR slice::slice(const void *ptr, size_t bytes)
- : ::MDBX_val({const_cast<void *>(ptr), check_length(bytes)}) {}
-
-MDBX_CXX14_CONSTEXPR slice::slice(const void *begin, const void *end)
- : slice(begin, static_cast<const byte *>(end) -
- static_cast<const byte *>(begin)) {}
-
-MDBX_CXX17_CONSTEXPR slice::slice(const char *c_str)
- : slice(c_str, ::mdbx::strlen(c_str)) {}
-
-MDBX_CXX14_CONSTEXPR slice::slice(const MDBX_val &src)
- : slice(src.iov_base, src.iov_len) {}
-
-MDBX_CXX14_CONSTEXPR slice::slice(MDBX_val &&src) : slice(src) {
- src.iov_base = nullptr;
-}
-
-MDBX_CXX14_CONSTEXPR slice::slice(slice &&src) noexcept : slice(src) {
- src.invalidate();
-}
-
-inline slice &slice::assign(const void *ptr, size_t bytes) {
- iov_base = const_cast<void *>(ptr);
- iov_len = check_length(bytes);
- return *this;
-}
-
-inline slice &slice::assign(const slice &src) noexcept {
- iov_base = src.iov_base;
- iov_len = src.iov_len;
- return *this;
-}
-
-inline slice &slice::assign(const ::MDBX_val &src) {
- return assign(src.iov_base, src.iov_len);
-}
-
-slice &slice::assign(slice &&src) noexcept {
- assign(src);
- src.invalidate();
- return *this;
-}
-
-inline slice &slice::assign(::MDBX_val &&src) {
- assign(src.iov_base, src.iov_len);
- src.iov_base = nullptr;
- return *this;
-}
-
-inline slice &slice::assign(const void *begin, const void *end) {
- return assign(begin, static_cast<const byte *>(end) -
- static_cast<const byte *>(begin));
-}
-
-inline slice &slice::assign(const char *c_str) {
- return assign(c_str, ::mdbx::strlen(c_str));
-}
-
-inline slice &slice::operator=(slice &&src) noexcept {
- return assign(::std::move(src));
-}
-
-inline slice &slice::operator=(::MDBX_val &&src) {
- return assign(::std::move(src));
-}
-
-inline void slice::swap(slice &other) noexcept {
- const auto temp = *this;
- *this = other;
- other = temp;
-}
-
-MDBX_CXX11_CONSTEXPR const ::mdbx::byte *slice::byte_ptr() const noexcept {
- return static_cast<const byte *>(iov_base);
-}
-
-MDBX_CXX11_CONSTEXPR const ::mdbx::byte *slice::end_byte_ptr() const noexcept {
- return byte_ptr() + length();
-}
-
-MDBX_CXX11_CONSTEXPR ::mdbx::byte *slice::byte_ptr() noexcept {
- return static_cast<byte *>(iov_base);
-}
-
-MDBX_CXX11_CONSTEXPR ::mdbx::byte *slice::end_byte_ptr() noexcept {
- return byte_ptr() + length();
-}
-
-MDBX_CXX11_CONSTEXPR const char *slice::char_ptr() const noexcept {
- return static_cast<const char *>(iov_base);
-}
-
-MDBX_CXX11_CONSTEXPR const char *slice::end_char_ptr() const noexcept {
- return char_ptr() + length();
-}
-
-MDBX_CXX11_CONSTEXPR char *slice::char_ptr() noexcept {
- return static_cast<char *>(iov_base);
-}
-
-MDBX_CXX11_CONSTEXPR char *slice::end_char_ptr() noexcept {
- return char_ptr() + length();
-}
-
-MDBX_CXX11_CONSTEXPR const void *slice::data() const noexcept {
- return iov_base;
-}
-
-MDBX_CXX11_CONSTEXPR const void *slice::end() const noexcept {
- return static_cast<const void *>(end_byte_ptr());
-}
-
-MDBX_CXX11_CONSTEXPR void *slice::data() noexcept { return iov_base; }
-
-MDBX_CXX11_CONSTEXPR void *slice::end() noexcept {
- return static_cast<void *>(end_byte_ptr());
-}
-
-MDBX_CXX11_CONSTEXPR size_t slice::length() const noexcept { return iov_len; }
-
-MDBX_CXX14_CONSTEXPR slice &slice::set_length(size_t bytes) {
- iov_len = check_length(bytes);
- return *this;
-}
-
-MDBX_CXX14_CONSTEXPR slice &slice::set_end(const void *ptr) {
- MDBX_CONSTEXPR_ASSERT(static_cast<const char *>(ptr) >= char_ptr());
- return set_length(static_cast<const char *>(ptr) - char_ptr());
-}
-
-MDBX_CXX11_CONSTEXPR bool slice::empty() const noexcept {
- return length() == 0;
-}
-
-MDBX_CXX11_CONSTEXPR bool slice::is_null() const noexcept {
- return data() == nullptr;
-}
-
-MDBX_CXX11_CONSTEXPR size_t slice::size() const noexcept { return length(); }
-
-MDBX_CXX11_CONSTEXPR slice::operator bool() const noexcept {
- return !is_null();
-}
-
-MDBX_CXX14_CONSTEXPR void slice::invalidate() noexcept { iov_base = nullptr; }
-
-MDBX_CXX14_CONSTEXPR void slice::clear() noexcept {
- iov_base = nullptr;
- iov_len = 0;
-}
-
-inline void slice::remove_prefix(size_t n) noexcept {
- assert(n <= size());
- iov_base = static_cast<byte *>(iov_base) + n;
- iov_len -= n;
-}
-
-inline void slice::safe_remove_prefix(size_t n) {
- if (MDBX_UNLIKELY(n > size()))
- MDBX_CXX20_UNLIKELY throw_out_range();
- remove_prefix(n);
-}
-
-inline void slice::remove_suffix(size_t n) noexcept {
- assert(n <= size());
- iov_len -= n;
-}
-
-inline void slice::safe_remove_suffix(size_t n) {
- if (MDBX_UNLIKELY(n > size()))
- MDBX_CXX20_UNLIKELY throw_out_range();
- remove_suffix(n);
-}
-
-MDBX_CXX14_CONSTEXPR bool
-slice::starts_with(const slice &prefix) const noexcept {
- return length() >= prefix.length() &&
- memcmp(data(), prefix.data(), prefix.length()) == 0;
-}
-
-MDBX_CXX14_CONSTEXPR bool slice::ends_with(const slice &suffix) const noexcept {
- return length() >= suffix.length() &&
- memcmp(byte_ptr() + length() - suffix.length(), suffix.data(),
- suffix.length()) == 0;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR size_t
-slice::hash_value() const noexcept {
- size_t h = length() * 3977471;
- for (size_t i = 0; i < length(); ++i)
- h = (h ^ static_cast<const uint8_t *>(data())[i]) * 1664525 + 1013904223;
- return h ^ 3863194411 * (h >> 11);
-}
-
-MDBX_CXX11_CONSTEXPR byte slice::operator[](size_t n) const noexcept {
- MDBX_CONSTEXPR_ASSERT(n < size());
- return byte_ptr()[n];
-}
-
-MDBX_CXX11_CONSTEXPR byte slice::at(size_t n) const {
- if (MDBX_UNLIKELY(n >= size()))
- MDBX_CXX20_UNLIKELY throw_out_range();
- return byte_ptr()[n];
-}
-
-MDBX_CXX14_CONSTEXPR slice slice::head(size_t n) const noexcept {
- MDBX_CONSTEXPR_ASSERT(n <= size());
- return slice(data(), n);
-}
-
-MDBX_CXX14_CONSTEXPR slice slice::tail(size_t n) const noexcept {
- MDBX_CONSTEXPR_ASSERT(n <= size());
- return slice(char_ptr() + size() - n, n);
-}
-
-MDBX_CXX14_CONSTEXPR slice slice::middle(size_t from, size_t n) const noexcept {
- MDBX_CONSTEXPR_ASSERT(from + n <= size());
- return slice(char_ptr() + from, n);
-}
-
-MDBX_CXX14_CONSTEXPR slice slice::safe_head(size_t n) const {
- if (MDBX_UNLIKELY(n > size()))
- MDBX_CXX20_UNLIKELY throw_out_range();
- return head(n);
-}
-
-MDBX_CXX14_CONSTEXPR slice slice::safe_tail(size_t n) const {
- if (MDBX_UNLIKELY(n > size()))
- MDBX_CXX20_UNLIKELY throw_out_range();
- return tail(n);
-}
-
-MDBX_CXX14_CONSTEXPR slice slice::safe_middle(size_t from, size_t n) const {
- if (MDBX_UNLIKELY(n > max_length))
- MDBX_CXX20_UNLIKELY throw_max_length_exceeded();
- if (MDBX_UNLIKELY(from + n > size()))
- MDBX_CXX20_UNLIKELY throw_out_range();
- return middle(from, n);
-}
-
-MDBX_CXX14_CONSTEXPR intptr_t slice::compare_fast(const slice &a,
- const slice &b) noexcept {
- const intptr_t diff = intptr_t(a.length()) - intptr_t(b.length());
- return diff ? diff
- : MDBX_UNLIKELY(a.length() == 0 || a.data() == b.data())
- ? 0
- : memcmp(a.data(), b.data(), a.length());
-}
-
-MDBX_CXX14_CONSTEXPR intptr_t
-slice::compare_lexicographically(const slice &a, const slice &b) noexcept {
- const size_t shortest = ::std::min(a.length(), b.length());
- if (MDBX_LIKELY(shortest > 0))
- MDBX_CXX20_LIKELY {
- const intptr_t diff = memcmp(a.data(), b.data(), shortest);
- if (MDBX_LIKELY(diff != 0))
- MDBX_CXX20_LIKELY return diff;
- }
- return intptr_t(a.length()) - intptr_t(b.length());
-}
-
-MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR bool
-operator==(const slice &a, const slice &b) noexcept {
- return slice::compare_fast(a, b) == 0;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR bool
-operator<(const slice &a, const slice &b) noexcept {
- return slice::compare_lexicographically(a, b) < 0;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR bool
-operator>(const slice &a, const slice &b) noexcept {
- return slice::compare_lexicographically(a, b) > 0;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR bool
-operator<=(const slice &a, const slice &b) noexcept {
- return slice::compare_lexicographically(a, b) <= 0;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR bool
-operator>=(const slice &a, const slice &b) noexcept {
- return slice::compare_lexicographically(a, b) >= 0;
-}
-
-MDBX_NOTHROW_PURE_FUNCTION MDBX_CXX14_CONSTEXPR bool
-operator!=(const slice &a, const slice &b) noexcept {
- return slice::compare_fast(a, b) != 0;
-}
-
-template <class ALLOCATOR>
-inline string<ALLOCATOR>
-slice::as_hex_string(bool uppercase, unsigned wrap_width,
- const ALLOCATOR &allocator) const {
- return to_hex(*this, uppercase, wrap_width).as_string<ALLOCATOR>(allocator);
-}
-
-template <class ALLOCATOR>
-inline string<ALLOCATOR>
-slice::as_base58_string(unsigned wrap_width, const ALLOCATOR &allocator) const {
- return to_base58(*this, wrap_width).as_string<ALLOCATOR>(allocator);
-}
-
-template <class ALLOCATOR>
-inline string<ALLOCATOR>
-slice::as_base64_string(unsigned wrap_width, const ALLOCATOR &allocator) const {
- return to_base64(*this, wrap_width).as_string<ALLOCATOR>(allocator);
-}
-
-template <class ALLOCATOR, class CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-slice::encode_hex(bool uppercase, unsigned wrap_width,
- const ALLOCATOR &allocator) const {
- return to_hex(*this, uppercase, wrap_width)
- .as_buffer<ALLOCATOR, CAPACITY_POLICY>(allocator);
-}
-
-template <class ALLOCATOR, class CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-slice::encode_base58(unsigned wrap_width, const ALLOCATOR &allocator) const {
- return to_base58(*this, wrap_width)
- .as_buffer<ALLOCATOR, CAPACITY_POLICY>(allocator);
-}
-
-template <class ALLOCATOR, class CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-slice::encode_base64(unsigned wrap_width, const ALLOCATOR &allocator) const {
- return to_base64(*this, wrap_width)
- .as_buffer<ALLOCATOR, CAPACITY_POLICY>(allocator);
-}
-
-template <class ALLOCATOR, class CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-slice::hex_decode(bool ignore_spaces, const ALLOCATOR &allocator) const {
- return from_hex(*this, ignore_spaces)
- .as_buffer<ALLOCATOR, CAPACITY_POLICY>(allocator);
-}
-
-template <class ALLOCATOR, class CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-slice::base58_decode(bool ignore_spaces, const ALLOCATOR &allocator) const {
- return from_base58(*this, ignore_spaces)
- .as_buffer<ALLOCATOR, CAPACITY_POLICY>(allocator);
-}
-
-template <class ALLOCATOR, class CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-slice::base64_decode(bool ignore_spaces, const ALLOCATOR &allocator) const {
- return from_base64(*this, ignore_spaces)
- .as_buffer<ALLOCATOR, CAPACITY_POLICY>(allocator);
-}
-
-inline MDBX_NOTHROW_PURE_FUNCTION bool
-slice::is_hex(bool ignore_spaces) const noexcept {
- return !from_hex(*this, ignore_spaces).is_erroneous();
-}
-
-inline MDBX_NOTHROW_PURE_FUNCTION bool
-slice::is_base58(bool ignore_spaces) const noexcept {
- return !from_base58(*this, ignore_spaces).is_erroneous();
-}
-
-inline MDBX_NOTHROW_PURE_FUNCTION bool
-slice::is_base64(bool ignore_spaces) const noexcept {
- return !from_base64(*this, ignore_spaces).is_erroneous();
-}
-
-//------------------------------------------------------------------------------
-
-template <class ALLOCATOR, typename CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>::buffer(
- const txn &txn, const struct slice &src, const allocator_type &allocator)
- : buffer(src, !txn.is_dirty(src.data()), allocator) {}
-
-//------------------------------------------------------------------------------
-
-MDBX_CXX11_CONSTEXPR map_handle::info::info(map_handle::flags flags,
- map_handle::state state) noexcept
- : flags(flags), state(state) {}
-
-#if CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_CXX11_CONSTEXPR
-#endif
-::mdbx::key_mode map_handle::info::key_mode() const noexcept {
- return ::mdbx::key_mode(flags & (MDBX_REVERSEKEY | MDBX_INTEGERKEY));
-}
-
-#if CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_CXX11_CONSTEXPR
-#endif
-::mdbx::value_mode map_handle::info::value_mode() const noexcept {
- return ::mdbx::value_mode(flags & (MDBX_DUPSORT | MDBX_REVERSEDUP |
- MDBX_DUPFIXED | MDBX_INTEGERDUP));
-}
-
-//------------------------------------------------------------------------------
-
-MDBX_CXX11_CONSTEXPR env::env(MDBX_env *ptr) noexcept : handle_(ptr) {}
-
-inline env &env::operator=(env &&other) noexcept {
- handle_ = other.handle_;
- other.handle_ = nullptr;
- return *this;
-}
-
-inline env::env(env &&other) noexcept : handle_(other.handle_) {
- other.handle_ = nullptr;
-}
-
-inline env::~env() noexcept {
-#ifndef NDEBUG
- handle_ = reinterpret_cast<MDBX_env *>(uintptr_t(0xDeadBeef));
-#endif
-}
-
-MDBX_CXX14_CONSTEXPR env::operator bool() const noexcept {
- return handle_ != nullptr;
-}
-
-MDBX_CXX14_CONSTEXPR env::operator const MDBX_env *() const { return handle_; }
-
-MDBX_CXX14_CONSTEXPR env::operator MDBX_env *() { return handle_; }
-
-MDBX_CXX11_CONSTEXPR bool operator==(const env &a, const env &b) noexcept {
- return a.handle_ == b.handle_;
-}
-
-MDBX_CXX11_CONSTEXPR bool operator!=(const env &a, const env &b) noexcept {
- return a.handle_ != b.handle_;
-}
-
-inline env::geometry &env::geometry::make_fixed(intptr_t size) noexcept {
- size_lower = size_now = size_upper = size;
- growth_step = shrink_threshold = 0;
- return *this;
-}
-
-inline env::geometry &env::geometry::make_dynamic(intptr_t lower,
- intptr_t upper) noexcept {
- size_now = size_lower = lower;
- size_upper = upper;
- growth_step = shrink_threshold = default_value;
- return *this;
-}
-
-inline env::reclaiming_options env::operate_parameters::reclaiming_from_flags(
- MDBX_env_flags_t flags) noexcept {
- return reclaiming_options(flags);
-}
-
-inline env::operate_options
-env::operate_parameters::options_from_flags(MDBX_env_flags_t flags) noexcept {
- return operate_options(flags);
-}
-
-inline size_t env::limits::pagesize_min() noexcept { return MDBX_MIN_PAGESIZE; }
-
-inline size_t env::limits::pagesize_max() noexcept { return MDBX_MAX_PAGESIZE; }
-
-inline size_t env::limits::dbsize_min(intptr_t pagesize) {
- const intptr_t result = mdbx_limits_dbsize_min(pagesize);
- if (result < 0)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_EINVAL);
- return static_cast<size_t>(result);
-}
-
-inline size_t env::limits::dbsize_max(intptr_t pagesize) {
- const intptr_t result = mdbx_limits_dbsize_max(pagesize);
- if (result < 0)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_EINVAL);
- return static_cast<size_t>(result);
-}
-
-inline size_t env::limits::key_min(MDBX_db_flags_t flags) noexcept {
- return (flags & MDBX_INTEGERKEY) ? 4 : 0;
-}
-
-inline size_t env::limits::key_min(key_mode mode) noexcept {
- return key_min(MDBX_db_flags_t(mode));
-}
-
-inline size_t env::limits::key_max(intptr_t pagesize, MDBX_db_flags_t flags) {
- const intptr_t result = mdbx_limits_keysize_max(pagesize, flags);
- if (result < 0)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_EINVAL);
- return static_cast<size_t>(result);
-}
-
-inline size_t env::limits::key_max(intptr_t pagesize, key_mode mode) {
- return key_max(pagesize, MDBX_db_flags_t(mode));
-}
-
-inline size_t env::limits::key_max(const env &env, MDBX_db_flags_t flags) {
- const intptr_t result = mdbx_env_get_maxkeysize_ex(env, flags);
- if (result < 0)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_EINVAL);
- return static_cast<size_t>(result);
-}
-
-inline size_t env::limits::key_max(const env &env, key_mode mode) {
- return key_max(env, MDBX_db_flags_t(mode));
-}
-
-inline size_t env::limits::value_min(MDBX_db_flags_t flags) noexcept {
- return (flags & MDBX_INTEGERDUP) ? 4 : 0;
-}
-
-inline size_t env::limits::value_min(value_mode mode) noexcept {
- return value_min(MDBX_db_flags_t(mode));
-}
-
-inline size_t env::limits::value_max(intptr_t pagesize, MDBX_db_flags_t flags) {
- const intptr_t result = mdbx_limits_valsize_max(pagesize, flags);
- if (result < 0)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_EINVAL);
- return static_cast<size_t>(result);
-}
-
-inline size_t env::limits::value_max(intptr_t pagesize, value_mode mode) {
- return value_max(pagesize, MDBX_db_flags_t(mode));
-}
-
-inline size_t env::limits::value_max(const env &env, MDBX_db_flags_t flags) {
- const intptr_t result = mdbx_env_get_maxvalsize_ex(env, flags);
- if (result < 0)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_EINVAL);
- return static_cast<size_t>(result);
-}
-
-inline size_t env::limits::value_max(const env &env, value_mode mode) {
- return value_max(env, MDBX_db_flags_t(mode));
-}
-
-inline size_t env::limits::transaction_size_max(intptr_t pagesize) {
- const intptr_t result = mdbx_limits_txnsize_max(pagesize);
- if (result < 0)
- MDBX_CXX20_UNLIKELY error::throw_exception(MDBX_EINVAL);
- return static_cast<size_t>(result);
-}
-
-inline env::operate_parameters env::get_operation_parameters() const {
- const auto flags = get_flags();
- return operate_parameters(max_maps(), max_readers(),
- operate_parameters::mode_from_flags(flags),
- operate_parameters::durability_from_flags(flags),
- operate_parameters::reclaiming_from_flags(flags),
- operate_parameters::options_from_flags(flags));
-}
-
-inline env::mode env::get_mode() const {
- return operate_parameters::mode_from_flags(get_flags());
-}
-
-inline env::durability env::get_durability() const {
- return env::operate_parameters::durability_from_flags(get_flags());
-}
-
-inline env::reclaiming_options env::get_reclaiming() const {
- return env::operate_parameters::reclaiming_from_flags(get_flags());
-}
-
-inline env::operate_options env::get_options() const {
- return env::operate_parameters::options_from_flags(get_flags());
-}
-
-inline env::stat env::get_stat() const {
- env::stat r;
- error::success_or_throw(::mdbx_env_stat_ex(handle_, nullptr, &r, sizeof(r)));
- return r;
-}
-
-inline env::stat env::get_stat(const txn &txn) const {
- env::stat r;
- error::success_or_throw(::mdbx_env_stat_ex(handle_, txn, &r, sizeof(r)));
- return r;
-}
-
-inline env::info env::get_info() const {
- env::info r;
- error::success_or_throw(::mdbx_env_info_ex(handle_, nullptr, &r, sizeof(r)));
- return r;
-}
-
-inline env::info env::get_info(const txn &txn) const {
- env::info r;
- error::success_or_throw(::mdbx_env_info_ex(handle_, txn, &r, sizeof(r)));
- return r;
-}
-
-inline filehandle env::get_filehandle() const {
- filehandle fd;
- error::success_or_throw(::mdbx_env_get_fd(handle_, &fd));
- return fd;
-}
-
-inline MDBX_env_flags_t env::get_flags() const {
- unsigned bits;
- error::success_or_throw(::mdbx_env_get_flags(handle_, &bits));
- return MDBX_env_flags_t(bits);
-}
-
-inline unsigned env::max_readers() const {
- unsigned r;
- error::success_or_throw(::mdbx_env_get_maxreaders(handle_, &r));
- return r;
-}
-
-inline unsigned env::max_maps() const {
- unsigned r;
- error::success_or_throw(::mdbx_env_get_maxdbs(handle_, &r));
- return r;
-}
-
-inline void *env::get_context() const noexcept {
- return mdbx_env_get_userctx(handle_);
-}
-
-inline env &env::set_context(void *ptr) {
- error::success_or_throw(::mdbx_env_set_userctx(handle_, ptr));
- return *this;
-}
-
-inline env &env::set_sync_threshold(size_t bytes) {
- error::success_or_throw(::mdbx_env_set_syncbytes(handle_, bytes));
- return *this;
-}
-
-inline env &env::set_sync_period(unsigned seconds_16dot16) {
- error::success_or_throw(::mdbx_env_set_syncperiod(handle_, seconds_16dot16));
- return *this;
-}
-
-inline env &env::set_sync_period(double seconds) {
- return set_sync_period(unsigned(seconds * 65536));
-}
-
-inline env &env::alter_flags(MDBX_env_flags_t flags, bool on_off) {
- error::success_or_throw(::mdbx_env_set_flags(handle_, flags, on_off));
- return *this;
-}
-
-inline env &env::set_geometry(const geometry &geo) {
- error::success_or_throw(::mdbx_env_set_geometry(
- handle_, geo.size_lower, geo.size_now, geo.size_upper, geo.growth_step,
- geo.shrink_threshold, geo.pagesize));
- return *this;
-}
-
-inline bool env::sync_to_disk(bool force, bool nonblock) {
- const int err = ::mdbx_env_sync_ex(handle_, force, nonblock);
- switch (err) {
- case MDBX_SUCCESS /* flush done */:
- case MDBX_RESULT_TRUE /* no data pending for flush to disk */:
- return true;
- case MDBX_BUSY /* the environment is used by other thread */:
- return false;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline void env::close_map(const map_handle &handle) {
- error::success_or_throw(::mdbx_dbi_close(*this, handle.dbi));
-}
-
-MDBX_CXX11_CONSTEXPR
-env::reader_info::reader_info(int slot, mdbx_pid_t pid, mdbx_tid_t thread,
- uint64_t txnid, uint64_t lag, size_t used,
- size_t retained) noexcept
- : slot(slot), pid(pid), thread(thread), transaction_id(txnid),
- transaction_lag(lag), bytes_used(used), bytes_retained(retained) {}
-
-template <typename VISITOR>
-inline int env::enumerate_readers(VISITOR &visitor) {
- struct reader_visitor_thunk : public exception_thunk {
- VISITOR &visitor_;
- static int cb(void *ctx, int number, int slot, mdbx_pid_t pid,
- mdbx_tid_t thread, uint64_t txnid, uint64_t lag, size_t used,
- size_t retained) noexcept {
- reader_visitor_thunk *thunk = static_cast<reader_visitor_thunk *>(ctx);
- assert(thunk->is_clean());
- try {
- const reader_info info(slot, pid, thread, txnid, lag, used, retained);
- return loop_control(thunk->visitor_(info, number));
- } catch (... /* capture any exception to rethrow it over C code */) {
- thunk->capture();
- return loop_control::exit_loop;
- }
- }
- MDBX_CXX11_CONSTEXPR reader_visitor_thunk(VISITOR &visitor) noexcept
- : visitor_(visitor) {}
- };
- reader_visitor_thunk thunk(visitor);
- const auto rc = ::mdbx_reader_list(*this, thunk.cb, &thunk);
- thunk.rethrow_captured();
- return rc;
-}
-
-inline unsigned env::check_readers() {
- int dead_count;
- error::throw_on_failure(::mdbx_reader_check(*this, &dead_count));
- assert(dead_count >= 0);
- return static_cast<unsigned>(dead_count);
-}
-
-inline env &env::set_HandleSlowReaders(MDBX_hsr_func *cb) {
- error::success_or_throw(::mdbx_env_set_hsr(handle_, cb));
- return *this;
-}
-
-inline MDBX_hsr_func *env::get_HandleSlowReaders() const noexcept {
- return ::mdbx_env_get_hsr(handle_);
-}
-
-inline txn_managed env::start_read() const {
- ::MDBX_txn *ptr;
- error::success_or_throw(
- ::mdbx_txn_begin(handle_, nullptr, MDBX_TXN_RDONLY, &ptr));
- assert(ptr != nullptr);
- return txn_managed(ptr);
-}
-
-inline txn_managed env::prepare_read() const {
- ::MDBX_txn *ptr;
- error::success_or_throw(
- ::mdbx_txn_begin(handle_, nullptr, MDBX_TXN_RDONLY_PREPARE, &ptr));
- assert(ptr != nullptr);
- return txn_managed(ptr);
-}
-
-inline txn_managed env::start_write(bool dont_wait) {
- ::MDBX_txn *ptr;
- error::success_or_throw(::mdbx_txn_begin(
- handle_, nullptr, dont_wait ? MDBX_TXN_TRY : MDBX_TXN_READWRITE, &ptr));
- assert(ptr != nullptr);
- return txn_managed(ptr);
-}
-
-inline txn_managed env::try_start_write() { return start_write(true); }
-
-//------------------------------------------------------------------------------
-
-MDBX_CXX11_CONSTEXPR txn::txn(MDBX_txn *ptr) noexcept : handle_(ptr) {}
-
-inline txn &txn::operator=(txn &&other) noexcept {
- handle_ = other.handle_;
- other.handle_ = nullptr;
- return *this;
-}
-
-inline txn::txn(txn &&other) noexcept : handle_(other.handle_) {
- other.handle_ = nullptr;
-}
-
-inline txn::~txn() noexcept {
-#ifndef NDEBUG
- handle_ = reinterpret_cast<MDBX_txn *>(uintptr_t(0xDeadBeef));
-#endif
-}
-
-MDBX_CXX14_CONSTEXPR txn::operator bool() const noexcept {
- return handle_ != nullptr;
-}
-
-MDBX_CXX14_CONSTEXPR txn::operator const MDBX_txn *() const { return handle_; }
-
-MDBX_CXX14_CONSTEXPR txn::operator MDBX_txn *() { return handle_; }
-
-MDBX_CXX11_CONSTEXPR bool operator==(const txn &a, const txn &b) noexcept {
- return a.handle_ == b.handle_;
-}
-
-MDBX_CXX11_CONSTEXPR bool operator!=(const txn &a, const txn &b) noexcept {
- return a.handle_ != b.handle_;
-}
-
-inline bool txn::is_dirty(const void *ptr) const {
- int err = ::mdbx_is_dirty(handle_, ptr);
- switch (err) {
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- case MDBX_RESULT_TRUE:
- return true;
- case MDBX_RESULT_FALSE:
- return false;
- }
-}
-
-inline ::mdbx::env txn::env() const noexcept { return ::mdbx_txn_env(handle_); }
-
-inline MDBX_txn_flags_t txn::flags() const {
- const int bits = mdbx_txn_flags(handle_);
- error::throw_on_failure((bits != -1) ? MDBX_SUCCESS : MDBX_BAD_TXN);
- return static_cast<MDBX_txn_flags_t>(bits);
-}
-
-inline uint64_t txn::id() const {
- const uint64_t txnid = mdbx_txn_id(handle_);
- error::throw_on_failure(txnid ? MDBX_SUCCESS : MDBX_BAD_TXN);
- return txnid;
-}
-
-inline void txn::reset_reading() {
- error::success_or_throw(::mdbx_txn_reset(handle_));
-}
-
-inline void txn::renew_reading() {
- error::success_or_throw(::mdbx_txn_renew(handle_));
-}
-
-inline txn::info txn::get_info(bool scan_reader_lock_table) const {
- txn::info r;
- error::success_or_throw(::mdbx_txn_info(handle_, &r, scan_reader_lock_table));
- return r;
-}
-
-inline cursor_managed txn::open_cursor(map_handle map) {
- MDBX_cursor *ptr;
- error::success_or_throw(::mdbx_cursor_open(handle_, map.dbi, &ptr));
- return cursor_managed(ptr);
-}
-
-inline ::mdbx::map_handle
-txn::open_map(const char *name, const ::mdbx::key_mode key_mode,
- const ::mdbx::value_mode value_mode) const {
- ::mdbx::map_handle map;
- error::success_or_throw(::mdbx_dbi_open(
- handle_, name, MDBX_db_flags_t(key_mode) | MDBX_db_flags_t(value_mode),
- &map.dbi));
- assert(map.dbi != 0);
- return map;
-}
-
-inline ::mdbx::map_handle
-txn::open_map(const ::std::string &name, const ::mdbx::key_mode key_mode,
- const ::mdbx::value_mode value_mode) const {
- return open_map(name.c_str(), key_mode, value_mode);
-}
-
-inline ::mdbx::map_handle txn::create_map(const char *name,
- const ::mdbx::key_mode key_mode,
- const ::mdbx::value_mode value_mode) {
- ::mdbx::map_handle map;
- error::success_or_throw(::mdbx_dbi_open(
- handle_, name,
- MDBX_CREATE | MDBX_db_flags_t(key_mode) | MDBX_db_flags_t(value_mode),
- &map.dbi));
- assert(map.dbi != 0);
- return map;
-}
-
-inline ::mdbx::map_handle txn::create_map(const ::std::string &name,
- const ::mdbx::key_mode key_mode,
- const ::mdbx::value_mode value_mode) {
- return create_map(name.c_str(), key_mode, value_mode);
-}
-
-inline void txn::drop_map(map_handle map) {
- error::success_or_throw(::mdbx_drop(handle_, map.dbi, true));
-}
-
-inline bool txn::drop_map(const ::std::string &name, bool throw_if_absent) {
- return drop_map(name.c_str(), throw_if_absent);
-}
-
-inline void txn::clear_map(map_handle map) {
- error::success_or_throw(::mdbx_drop(handle_, map.dbi, false));
-}
-
-inline bool txn::clear_map(const ::std::string &name, bool throw_if_absent) {
- return clear_map(name.c_str(), throw_if_absent);
-}
-
-inline txn::map_stat txn::get_map_stat(map_handle map) const {
- txn::map_stat r;
- error::success_or_throw(::mdbx_dbi_stat(handle_, map.dbi, &r, sizeof(r)));
- return r;
-}
-
-inline uint32_t txn::get_tree_deepmask(map_handle map) const {
- uint32_t r;
- error::success_or_throw(::mdbx_dbi_dupsort_depthmask(handle_, map.dbi, &r));
- return r;
-}
-
-inline map_handle::info txn::get_handle_info(map_handle map) const {
- unsigned flags, state;
- error::success_or_throw(
- ::mdbx_dbi_flags_ex(handle_, map.dbi, &flags, &state));
- return map_handle::info(MDBX_db_flags_t(flags), MDBX_dbi_state_t(state));
-}
-
-inline txn &txn::put_canary(const txn::canary &canary) {
- error::success_or_throw(::mdbx_canary_put(handle_, &canary));
- return *this;
-}
-
-inline txn::canary txn::get_canary() const {
- txn::canary r;
- error::success_or_throw(::mdbx_canary_get(handle_, &r));
- return r;
-}
-
-inline uint64_t txn::sequence(map_handle map) const {
- uint64_t result;
- error::success_or_throw(::mdbx_dbi_sequence(handle_, map.dbi, &result, 0));
- return result;
-}
-
-inline uint64_t txn::sequence(map_handle map, uint64_t increment) {
- uint64_t result;
- error::success_or_throw(
- ::mdbx_dbi_sequence(handle_, map.dbi, &result, increment));
- return result;
-}
-
-inline int txn::compare_keys(map_handle map, const slice &a,
- const slice &b) const noexcept {
- return ::mdbx_cmp(handle_, map.dbi, &a, &b);
-}
-
-inline int txn::compare_values(map_handle map, const slice &a,
- const slice &b) const noexcept {
- return ::mdbx_dcmp(handle_, map.dbi, &a, &b);
-}
-
-inline int txn::compare_keys(map_handle map, const pair &a,
- const pair &b) const noexcept {
- return compare_keys(map, a.key, b.key);
-}
-
-inline int txn::compare_values(map_handle map, const pair &a,
- const pair &b) const noexcept {
- return compare_values(map, a.value, b.value);
-}
-
-inline slice txn::get(map_handle map, const slice &key) const {
- slice result;
- error::success_or_throw(::mdbx_get(handle_, map.dbi, &key, &result));
- return result;
-}
-
-inline slice txn::get(map_handle map, slice key, size_t &values_count) const {
- slice result;
- error::success_or_throw(
- ::mdbx_get_ex(handle_, map.dbi, &key, &result, &values_count));
- return result;
-}
-
-inline slice txn::get(map_handle map, const slice &key,
- const slice &value_at_absence) const {
- slice result;
- const int err = ::mdbx_get(handle_, map.dbi, &key, &result);
- switch (err) {
- case MDBX_SUCCESS:
- return result;
- case MDBX_NOTFOUND:
- return value_at_absence;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline slice txn::get(map_handle map, slice key, size_t &values_count,
- const slice &value_at_absence) const {
- slice result;
- const int err = ::mdbx_get_ex(handle_, map.dbi, &key, &result, &values_count);
- switch (err) {
- case MDBX_SUCCESS:
- return result;
- case MDBX_NOTFOUND:
- return value_at_absence;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline pair_result txn::get_equal_or_great(map_handle map,
- const slice &key) const {
- pair result(key, slice());
- bool exact = !error::boolean_or_throw(
- ::mdbx_get_equal_or_great(handle_, map.dbi, &result.key, &result.value));
- return pair_result(result.key, result.value, exact);
-}
-
-inline pair_result
-txn::get_equal_or_great(map_handle map, const slice &key,
- const slice &value_at_absence) const {
- pair result{key, slice()};
- const int err =
- ::mdbx_get_equal_or_great(handle_, map.dbi, &result.key, &result.value);
- switch (err) {
- case MDBX_SUCCESS:
- return pair_result{result.key, result.value, true};
- case MDBX_RESULT_TRUE:
- return pair_result{result.key, result.value, false};
- case MDBX_NOTFOUND:
- return pair_result{key, value_at_absence, false};
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline MDBX_error_t txn::put(map_handle map, const slice &key, slice *value,
- MDBX_put_flags_t flags) noexcept {
- return MDBX_error_t(::mdbx_put(handle_, map.dbi, &key, value, flags));
-}
-
-inline void txn::put(map_handle map, const slice &key, slice value,
- put_mode mode) {
- error::success_or_throw(put(map, key, &value, MDBX_put_flags_t(mode)));
-}
-
-inline void txn::insert(map_handle map, const slice &key, slice value) {
- error::success_or_throw(
- put(map, key, &value /* takes the present value in case MDBX_KEYEXIST */,
- MDBX_put_flags_t(put_mode::insert_unique)));
-}
-
-inline value_result txn::try_insert(map_handle map, const slice &key,
- slice value) {
- const int err =
- put(map, key, &value /* takes the present value in case MDBX_KEYEXIST */,
- MDBX_put_flags_t(put_mode::insert_unique));
- switch (err) {
- case MDBX_SUCCESS:
- return value_result{slice(), true};
- case MDBX_KEYEXIST:
- return value_result{value, false};
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline slice txn::insert_reserve(map_handle map, const slice &key,
- size_t value_length) {
- slice result(nullptr, value_length);
- error::success_or_throw(
- put(map, key, &result /* takes the present value in case MDBX_KEYEXIST */,
- MDBX_put_flags_t(put_mode::insert_unique) | MDBX_RESERVE));
- return result;
-}
-
-inline value_result txn::try_insert_reserve(map_handle map, const slice &key,
- size_t value_length) {
- slice result(nullptr, value_length);
- const int err =
- put(map, key, &result /* takes the present value in case MDBX_KEYEXIST */,
- MDBX_put_flags_t(put_mode::insert_unique) | MDBX_RESERVE);
- switch (err) {
- case MDBX_SUCCESS:
- return value_result{result, true};
- case MDBX_KEYEXIST:
- return value_result{result, false};
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline void txn::upsert(map_handle map, const slice &key, const slice &value) {
- error::success_or_throw(put(map, key, const_cast<slice *>(&value),
- MDBX_put_flags_t(put_mode::upsert)));
-}
-
-inline slice txn::upsert_reserve(map_handle map, const slice &key,
- size_t value_length) {
- slice result(nullptr, value_length);
- error::success_or_throw(put(
- map, key, &result, MDBX_put_flags_t(put_mode::upsert) | MDBX_RESERVE));
- return result;
-}
-
-inline void txn::update(map_handle map, const slice &key, const slice &value) {
- error::success_or_throw(put(map, key, const_cast<slice *>(&value),
- MDBX_put_flags_t(put_mode::update)));
-}
-
-inline bool txn::try_update(map_handle map, const slice &key,
- const slice &value) {
- const int err = put(map, key, const_cast<slice *>(&value),
- MDBX_put_flags_t(put_mode::update));
- switch (err) {
- case MDBX_SUCCESS:
- return true;
- case MDBX_NOTFOUND:
- return false;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline slice txn::update_reserve(map_handle map, const slice &key,
- size_t value_length) {
- slice result(nullptr, value_length);
- error::success_or_throw(put(
- map, key, &result, MDBX_put_flags_t(put_mode::update) | MDBX_RESERVE));
- return result;
-}
-
-inline value_result txn::try_update_reserve(map_handle map, const slice &key,
- size_t value_length) {
- slice result(nullptr, value_length);
- const int err =
- put(map, key, &result, MDBX_put_flags_t(put_mode::update) | MDBX_RESERVE);
- switch (err) {
- case MDBX_SUCCESS:
- return value_result{result, true};
- case MDBX_NOTFOUND:
- return value_result{slice(), false};
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline bool txn::erase(map_handle map, const slice &key) {
- const int err = ::mdbx_del(handle_, map.dbi, &key, nullptr);
- switch (err) {
- case MDBX_SUCCESS:
- return true;
- case MDBX_NOTFOUND:
- return false;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline bool txn::erase(map_handle map, const slice &key, const slice &value) {
- const int err = ::mdbx_del(handle_, map.dbi, &key, &value);
- switch (err) {
- case MDBX_SUCCESS:
- return true;
- case MDBX_NOTFOUND:
- return false;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline void txn::replace(map_handle map, const slice &key, slice old_value,
- const slice &new_value) {
- error::success_or_throw(::mdbx_replace_ex(
- handle_, map.dbi, &key, const_cast<slice *>(&new_value), &old_value,
- MDBX_CURRENT | MDBX_NOOVERWRITE, nullptr, nullptr));
-}
-
-template <class ALLOCATOR, typename CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-txn::extract(map_handle map, const slice &key,
- const typename buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type
- &allocator) {
- typename buffer<ALLOCATOR, CAPACITY_POLICY>::data_preserver result(allocator);
- error::success_or_throw(::mdbx_replace_ex(handle_, map.dbi, &key, nullptr,
- &result.slice_, MDBX_CURRENT,
- result, &result),
- result);
- return result;
-}
-
-template <class ALLOCATOR, typename CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY>
-txn::replace(map_handle map, const slice &key, const slice &new_value,
- const typename buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type
- &allocator) {
- typename buffer<ALLOCATOR, CAPACITY_POLICY>::data_preserver result(allocator);
- error::success_or_throw(
- ::mdbx_replace_ex(handle_, map.dbi, &key, const_cast<slice *>(&new_value),
- &result.slice_, MDBX_CURRENT, result, &result),
- result);
- return result;
-}
-
-template <class ALLOCATOR, typename CAPACITY_POLICY>
-inline buffer<ALLOCATOR, CAPACITY_POLICY> txn::replace_reserve(
- map_handle map, const slice &key, slice &new_value,
- const typename buffer<ALLOCATOR, CAPACITY_POLICY>::allocator_type
- &allocator) {
- typename buffer<ALLOCATOR, CAPACITY_POLICY>::data_preserver result(allocator);
- error::success_or_throw(
- ::mdbx_replace_ex(handle_, map.dbi, &key, &new_value, &result.slice_,
- MDBX_CURRENT | MDBX_RESERVE, result, &result),
- result);
- return result;
-}
-
-inline void txn::append(map_handle map, const slice &key, const slice &value,
- bool multivalue_order_preserved) {
- error::success_or_throw(::mdbx_put(
- handle_, map.dbi, const_cast<slice *>(&key), const_cast<slice *>(&value),
- multivalue_order_preserved ? (MDBX_APPEND | MDBX_APPENDDUP)
- : MDBX_APPEND));
-}
-
-inline size_t txn::put_multiple(map_handle map, const slice &key,
- const size_t value_length,
- const void *values_array, size_t values_count,
- put_mode mode, bool allow_partial) {
- MDBX_val args[2] = {{const_cast<void *>(values_array), value_length},
- {nullptr, values_count}};
- const int err = ::mdbx_put(handle_, map.dbi, const_cast<slice *>(&key), args,
- MDBX_put_flags_t(mode) | MDBX_MULTIPLE);
- switch (err) {
- case MDBX_SUCCESS:
- MDBX_CXX20_LIKELY break;
- case MDBX_KEYEXIST:
- if (allow_partial)
- break;
- mdbx_txn_break(handle_);
- MDBX_CXX17_FALLTHROUGH /* fallthrough */;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
- return args[1].iov_len /* done item count */;
-}
-
-inline ptrdiff_t txn::estimate(map_handle map, pair from, pair to) const {
- ptrdiff_t result;
- error::success_or_throw(mdbx_estimate_range(
- handle_, map.dbi, &from.key, &from.value, &to.key, &to.value, &result));
- return result;
-}
-
-inline ptrdiff_t txn::estimate(map_handle map, slice from, slice to) const {
- ptrdiff_t result;
- error::success_or_throw(mdbx_estimate_range(handle_, map.dbi, &from, nullptr,
- &to, nullptr, &result));
- return result;
-}
-
-inline ptrdiff_t txn::estimate_from_first(map_handle map, slice to) const {
- ptrdiff_t result;
- error::success_or_throw(mdbx_estimate_range(handle_, map.dbi, nullptr,
- nullptr, &to, nullptr, &result));
- return result;
-}
-
-inline ptrdiff_t txn::estimate_to_last(map_handle map, slice from) const {
- ptrdiff_t result;
- error::success_or_throw(mdbx_estimate_range(handle_, map.dbi, &from, nullptr,
- nullptr, nullptr, &result));
- return result;
-}
-
-//------------------------------------------------------------------------------
-
-MDBX_CXX11_CONSTEXPR cursor::cursor(MDBX_cursor *ptr) noexcept : handle_(ptr) {}
-
-inline cursor &cursor::operator=(cursor &&other) noexcept {
- handle_ = other.handle_;
- other.handle_ = nullptr;
- return *this;
-}
-
-inline cursor::cursor(cursor &&other) noexcept : handle_(other.handle_) {
- other.handle_ = nullptr;
-}
-
-inline cursor::~cursor() noexcept {
-#ifndef NDEBUG
- handle_ = reinterpret_cast<MDBX_cursor *>(uintptr_t(0xDeadBeef));
-#endif
-}
-
-MDBX_CXX14_CONSTEXPR cursor::operator bool() const noexcept {
- return handle_ != nullptr;
-}
-
-MDBX_CXX14_CONSTEXPR cursor::operator const MDBX_cursor *() const {
- return handle_;
-}
-
-MDBX_CXX14_CONSTEXPR cursor::operator MDBX_cursor *() { return handle_; }
-
-MDBX_CXX11_CONSTEXPR bool operator==(const cursor &a,
- const cursor &b) noexcept {
- return a.handle_ == b.handle_;
-}
-
-MDBX_CXX11_CONSTEXPR bool operator!=(const cursor &a,
- const cursor &b) noexcept {
- return a.handle_ != b.handle_;
-}
-
-inline cursor::move_result::move_result(const cursor &cursor,
- bool throw_notfound)
- : pair_result(key, value, false) {
- done = cursor.move(get_current, &key, &value, throw_notfound);
-}
-
-inline cursor::move_result::move_result(cursor &cursor,
- move_operation operation,
- bool throw_notfound)
- : pair_result(key, value, false) {
- done = cursor.move(operation, &key, &value, throw_notfound);
-}
-
-inline cursor::move_result::move_result(cursor &cursor,
- move_operation operation,
- const slice &key, bool throw_notfound)
- : pair_result(key, slice(), false) {
- this->done = cursor.move(operation, &this->key, &this->value, throw_notfound);
-}
-
-inline cursor::move_result::move_result(cursor &cursor,
- move_operation operation,
- const slice &key, const slice &value,
- bool throw_notfound)
- : pair_result(key, value, false) {
- this->done = cursor.move(operation, &this->key, &this->value, throw_notfound);
-}
-
-inline bool cursor::move(move_operation operation, MDBX_val *key,
- MDBX_val *value, bool throw_notfound) const {
- const int err =
- ::mdbx_cursor_get(handle_, key, value, MDBX_cursor_op(operation));
- switch (err) {
- case MDBX_SUCCESS:
- MDBX_CXX20_LIKELY return true;
- case MDBX_NOTFOUND:
- if (!throw_notfound)
- return false;
- MDBX_CXX17_FALLTHROUGH /* fallthrough */;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline ptrdiff_t cursor::estimate(move_operation operation, MDBX_val *key,
- MDBX_val *value) const {
- ptrdiff_t result;
- error::success_or_throw(::mdbx_estimate_move(
- *this, key, value, MDBX_cursor_op(operation), &result));
- return result;
-}
-
-inline ptrdiff_t estimate(const cursor &from, const cursor &to) {
- ptrdiff_t result;
- error::success_or_throw(mdbx_estimate_distance(from, to, &result));
- return result;
-}
-
-inline cursor::move_result cursor::move(move_operation operation,
- bool throw_notfound) {
- return move_result(*this, operation, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_first(bool throw_notfound) {
- return move(first, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_previous(bool throw_notfound) {
- return move(previous, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_previous_last_multi(bool throw_notfound) {
- return move(multi_prevkey_lastvalue, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_current_first_multi(bool throw_notfound) {
- return move(multi_currentkey_firstvalue, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_current_prev_multi(bool throw_notfound) {
- return move(multi_currentkey_prevvalue, throw_notfound);
-}
-
-inline cursor::move_result cursor::current(bool throw_notfound) const {
- return move_result(*this, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_current_next_multi(bool throw_notfound) {
- return move(multi_currentkey_nextvalue, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_current_last_multi(bool throw_notfound) {
- return move(multi_currentkey_lastvalue, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_next_first_multi(bool throw_notfound) {
- return move(multi_nextkey_firstvalue, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_next(bool throw_notfound) {
- return move(next, throw_notfound);
-}
-
-inline cursor::move_result cursor::to_last(bool throw_notfound) {
- return move(last, throw_notfound);
-}
-
-inline cursor::move_result cursor::move(move_operation operation,
- const slice &key, bool throw_notfound) {
- return move_result(*this, operation, key, throw_notfound);
-}
-
-inline cursor::move_result cursor::find(const slice &key, bool throw_notfound) {
- return move(key_exact, key, throw_notfound);
-}
-
-inline cursor::move_result cursor::lower_bound(const slice &key,
- bool throw_notfound) {
- return move(key_lowerbound, key, throw_notfound);
-}
-
-inline cursor::move_result cursor::move(move_operation operation,
- const slice &key, const slice &value,
- bool throw_notfound) {
- return move_result(*this, operation, key, value, throw_notfound);
-}
-
-inline cursor::move_result cursor::find_multivalue(const slice &key,
- const slice &value,
- bool throw_notfound) {
- return move(multi_find_pair, key, value, throw_notfound);
-}
-
-inline cursor::move_result cursor::lower_bound_multivalue(const slice &key,
- const slice &value,
- bool throw_notfound) {
- return move(multi_exactkey_lowerboundvalue, key, value, throw_notfound);
-}
-
-inline bool cursor::seek(const slice &key) {
- return move(find_key, const_cast<slice *>(&key), nullptr, false);
-}
-
-inline bool cursor::move(move_operation operation, slice &key, slice &value,
- bool throw_notfound) {
- return move(operation, &key, &value, throw_notfound);
-}
-
-inline size_t cursor::count_multivalue() const {
- size_t result;
- error::success_or_throw(::mdbx_cursor_count(*this, &result));
- return result;
-}
-
-inline bool cursor::eof() const {
- return error::boolean_or_throw(::mdbx_cursor_eof(*this));
-}
-
-inline bool cursor::on_first() const {
- return error::boolean_or_throw(::mdbx_cursor_on_first(*this));
-}
-
-inline bool cursor::on_last() const {
- return error::boolean_or_throw(::mdbx_cursor_on_last(*this));
-}
-
-inline ptrdiff_t cursor::estimate(slice key, slice value) const {
- return estimate(multi_exactkey_lowerboundvalue, &key, &value);
-}
-
-inline ptrdiff_t cursor::estimate(slice key) const {
- return estimate(key_lowerbound, &key, nullptr);
-}
-
-inline ptrdiff_t cursor::estimate(move_operation operation) const {
- slice unused_key;
- return estimate(operation, &unused_key, nullptr);
-}
-
-inline void cursor::renew(::mdbx::txn &txn) {
- error::success_or_throw(::mdbx_cursor_renew(txn, handle_));
-}
-
-inline void cursor::bind(::mdbx::txn &txn, ::mdbx::map_handle map_handle) {
- error::success_or_throw(::mdbx_cursor_bind(txn, handle_, map_handle.dbi));
-}
-
-inline txn cursor::txn() const {
- MDBX_txn *txn = ::mdbx_cursor_txn(handle_);
- error::throw_on_nullptr(txn, MDBX_EINVAL);
- return ::mdbx::txn(txn);
-}
-
-inline map_handle cursor::map() const {
- const MDBX_dbi dbi = ::mdbx_cursor_dbi(handle_);
- if (MDBX_UNLIKELY(dbi > MDBX_MAX_DBI))
- error::throw_exception(MDBX_EINVAL);
- return map_handle(dbi);
-}
-
-inline MDBX_error_t cursor::put(const slice &key, slice *value,
- MDBX_put_flags_t flags) noexcept {
- return MDBX_error_t(::mdbx_cursor_put(handle_, &key, value, flags));
-}
-
-inline void cursor::insert(const slice &key, slice value) {
- error::success_or_throw(
- put(key, &value /* takes the present value in case MDBX_KEYEXIST */,
- MDBX_put_flags_t(put_mode::insert_unique)));
-}
-
-inline value_result cursor::try_insert(const slice &key, slice value) {
- const int err =
- put(key, &value /* takes the present value in case MDBX_KEYEXIST */,
- MDBX_put_flags_t(put_mode::insert_unique));
- switch (err) {
- case MDBX_SUCCESS:
- return value_result{slice(), true};
- case MDBX_KEYEXIST:
- return value_result{value, false};
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline slice cursor::insert_reserve(const slice &key, size_t value_length) {
- slice result(nullptr, value_length);
- error::success_or_throw(
- put(key, &result /* takes the present value in case MDBX_KEYEXIST */,
- MDBX_put_flags_t(put_mode::insert_unique) | MDBX_RESERVE));
- return result;
-}
-
-inline value_result cursor::try_insert_reserve(const slice &key,
- size_t value_length) {
- slice result(nullptr, value_length);
- const int err =
- put(key, &result /* takes the present value in case MDBX_KEYEXIST */,
- MDBX_put_flags_t(put_mode::insert_unique) | MDBX_RESERVE);
- switch (err) {
- case MDBX_SUCCESS:
- return value_result{result, true};
- case MDBX_KEYEXIST:
- return value_result{result, false};
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline void cursor::upsert(const slice &key, const slice &value) {
- error::success_or_throw(put(key, const_cast<slice *>(&value),
- MDBX_put_flags_t(put_mode::upsert)));
-}
-
-inline slice cursor::upsert_reserve(const slice &key, size_t value_length) {
- slice result(nullptr, value_length);
- error::success_or_throw(
- put(key, &result, MDBX_put_flags_t(put_mode::upsert) | MDBX_RESERVE));
- return result;
-}
-
-inline void cursor::update(const slice &key, const slice &value) {
- error::success_or_throw(put(key, const_cast<slice *>(&value),
- MDBX_put_flags_t(put_mode::update)));
-}
-
-inline bool cursor::try_update(const slice &key, const slice &value) {
- const int err =
- put(key, const_cast<slice *>(&value), MDBX_put_flags_t(put_mode::update));
- switch (err) {
- case MDBX_SUCCESS:
- return true;
- case MDBX_NOTFOUND:
- return false;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline slice cursor::update_reserve(const slice &key, size_t value_length) {
- slice result(nullptr, value_length);
- error::success_or_throw(
- put(key, &result, MDBX_put_flags_t(put_mode::update) | MDBX_RESERVE));
- return result;
-}
-
-inline value_result cursor::try_update_reserve(const slice &key,
- size_t value_length) {
- slice result(nullptr, value_length);
- const int err =
- put(key, &result, MDBX_put_flags_t(put_mode::update) | MDBX_RESERVE);
- switch (err) {
- case MDBX_SUCCESS:
- return value_result{result, true};
- case MDBX_NOTFOUND:
- return value_result{slice(), false};
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline bool cursor::erase(bool whole_multivalue) {
- const int err = ::mdbx_cursor_del(handle_, whole_multivalue ? MDBX_ALLDUPS
- : MDBX_CURRENT);
- switch (err) {
- case MDBX_SUCCESS:
- MDBX_CXX20_LIKELY return true;
- case MDBX_NOTFOUND:
- return false;
- default:
- MDBX_CXX20_UNLIKELY error::throw_exception(err);
- }
-}
-
-inline bool cursor::erase(const slice &key, bool whole_multivalue) {
- bool found = seek(key);
- return found ? erase(whole_multivalue) : found;
-}
-
-inline bool cursor::erase(const slice &key, const slice &value) {
- move_result data = find_multivalue(key, value, false);
- return data.done && erase();
-}
-
-} // namespace mdbx
-
-//------------------------------------------------------------------------------
-
-namespace std {
-
-inline string to_string(const ::mdbx::slice &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-template <class ALLOCATOR, typename CAPACITY_POLICY>
-inline string
-to_string(const ::mdbx::buffer<ALLOCATOR, CAPACITY_POLICY> &buffer) {
- ostringstream out;
- out << buffer;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::pair &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::env::geometry &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::env::operate_parameters &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::env::mode &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::env::durability &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::env::reclaiming_options &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::env::operate_options &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::env_managed::create_parameters &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::MDBX_log_level_t &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::MDBX_debug_flags_t &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::mdbx::error &value) {
- ostringstream out;
- out << value;
- return out.str();
-}
-
-inline string to_string(const ::MDBX_error_t &errcode) {
- return to_string(::mdbx::error(errcode));
-}
-
-template <> struct hash<::mdbx::slice> {
- MDBX_CXX14_CONSTEXPR size_t
- operator()(::mdbx::slice const &slice) const noexcept {
- return slice.hash_value();
- }
-};
-
-} // namespace std
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-/// @} end of C++ API
diff --git a/3rdparty/libmdbx/mdbx_chk.c b/3rdparty/libmdbx/mdbx_chk.c
deleted file mode 100644
index 27506c7386a1a0d0f74098883fc5f3dd06aafc7a..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx_chk.c
+++ /dev/null
@@ -1,5264 +0,0 @@
-/* mdbx_chk.c - memory-mapped database check tool */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#ifdef _MSC_VER
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#pragma warning(disable : 4996) /* The POSIX name is deprecated... */
-#endif /* _MSC_VER (warnings) */
-
-#define xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#define MDBX_BUILD_SOURCERY c510b93d78b5848764f976ea67f06d278a7b99f141b4c98ceb095285f834f31c_v0_11_7_0_g40ec559c
-#ifdef MDBX_CONFIG_H
-#include MDBX_CONFIG_H
-#endif
-
-#define LIBMDBX_INTERNALS
-#ifdef xMDBX_TOOLS
-#define MDBX_DEPRECATED
-#endif /* xMDBX_TOOLS */
-
-#ifdef xMDBX_ALLOY
-/* Amalgamated build */
-#define MDBX_INTERNAL_FUNC static
-#define MDBX_INTERNAL_VAR static
-#else
-/* Non-amalgamated build */
-#define MDBX_INTERNAL_FUNC
-#define MDBX_INTERNAL_VAR extern
-#endif /* xMDBX_ALLOY */
-
-/** Disables using GNU/Linux libc extensions.
- * \ingroup build_option
- * \note This option couldn't be moved to the options.h since dependant
- * control macros/defined should be prepared before include the options.h */
-#ifndef MDBX_DISABLE_GNU_SOURCE
-#define MDBX_DISABLE_GNU_SOURCE 0
-#endif
-#if MDBX_DISABLE_GNU_SOURCE
-#undef _GNU_SOURCE
-#elif (defined(__linux__) || defined(__gnu_linux__)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif /* MDBX_DISABLE_GNU_SOURCE */
-
-/*----------------------------------------------------------------------------*/
-
-/* Should be defined before any includes */
-#if !defined(_FILE_OFFSET_BITS) && !defined(__ANDROID_API__) && \
- !defined(ANDROID)
-#define _FILE_OFFSET_BITS 64
-#endif
-
-#ifdef __APPLE__
-#define _DARWIN_C_SOURCE
-#endif
-
-#ifdef _MSC_VER
-#if _MSC_FULL_VER < 190024234
-/* Actually libmdbx was not tested with compilers older than 19.00.24234 (Visual
- * Studio 2015 Update 3). But you could remove this #error and try to continue
- * at your own risk. In such case please don't rise up an issues related ONLY to
- * old compilers.
- *
- * NOTE:
- * Unfortunately, there are several different builds of "Visual Studio" that
- * are called "Visual Studio 2015 Update 3".
- *
- * The 190024234 is used here because it is minimal version of Visual Studio
- * that was used for build and testing libmdbx in recent years. Soon this
- * value will be increased to 19.0.24241.7, since build and testing using
- * "Visual Studio 2015" will be performed only at https://ci.appveyor.com.
- *
- * Please ask Microsoft (but not us) for information about version differences
- * and how to and where you can obtain the latest "Visual Studio 2015" build
- * with all fixes.
- */
-#error \
- "At least \"Microsoft C/C++ Compiler\" version 19.00.24234 (Visual Studio 2015 Update 3) is required."
-#endif
-#ifndef _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#if _MSC_VER > 1913
-#pragma warning(disable : 5045) /* Compiler will insert Spectre mitigation... \
- */
-#endif
-#if _MSC_VER > 1914
-#pragma warning( \
- disable : 5105) /* winbase.h(9531): warning C5105: macro expansion \
- producing 'defined' has undefined behavior */
-#endif
-#pragma warning(disable : 4710) /* 'xyz': function not inlined */
-#pragma warning(disable : 4711) /* function 'xyz' selected for automatic \
- inline expansion */
-#pragma warning( \
- disable : 4201) /* nonstandard extension used : nameless struct / union */
-#pragma warning(disable : 4702) /* unreachable code */
-#pragma warning(disable : 4706) /* assignment within conditional expression */
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4324) /* 'xyz': structure was padded due to \
- alignment specifier */
-#pragma warning(disable : 4310) /* cast truncates constant value */
-#pragma warning( \
- disable : 4820) /* bytes padding added after data member for alignment */
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4366) /* the result of the unary '&' operator may be \
- unaligned */
-#pragma warning(disable : 4200) /* nonstandard extension used: zero-sized \
- array in struct/union */
-#pragma warning(disable : 4204) /* nonstandard extension used: non-constant \
- aggregate initializer */
-#pragma warning( \
- disable : 4505) /* unreferenced local function has been removed */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(__GNUC__) && __GNUC__ < 9
-#pragma GCC diagnostic ignored "-Wattributes"
-#endif /* GCC < 9 */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-#include "mdbx.h"
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#ifndef __GNUC_PREREQ
-# if defined(__GNUC__) && defined(__GNUC_MINOR__)
-# define __GNUC_PREREQ(maj, min) \
- ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GNUC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GNUC_PREREQ */
-
-#ifndef __CLANG_PREREQ
-# ifdef __clang__
-# define __CLANG_PREREQ(maj,min) \
- ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min))
-# else
-# define __CLANG_PREREQ(maj,min) (0)
-# endif
-#endif /* __CLANG_PREREQ */
-
-#ifndef __GLIBC_PREREQ
-# if defined(__GLIBC__) && defined(__GLIBC_MINOR__)
-# define __GLIBC_PREREQ(maj, min) \
- ((__GLIBC__ << 16) + __GLIBC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GLIBC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GLIBC_PREREQ */
-
-#ifndef __has_warning
-# define __has_warning(x) (0)
-#endif
-
-#ifndef __has_include
-# define __has_include(x) (0)
-#endif
-
-#if __has_feature(thread_sanitizer)
-# define __SANITIZE_THREAD__ 1
-#endif
-
-#if __has_feature(address_sanitizer)
-# define __SANITIZE_ADDRESS__ 1
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __extern_C
-# ifdef __cplusplus
-# define __extern_C extern "C"
-# else
-# define __extern_C
-# endif
-#endif /* __extern_C */
-
-#if !defined(nullptr) && !defined(__cplusplus) || (__cplusplus < 201103L && !defined(_MSC_VER))
-# define nullptr NULL
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __always_inline
-# if defined(__GNUC__) || __has_attribute(__always_inline__)
-# define __always_inline __inline __attribute__((__always_inline__))
-# elif defined(_MSC_VER)
-# define __always_inline __forceinline
-# else
-# define __always_inline
-# endif
-#endif /* __always_inline */
-
-#ifndef __noinline
-# if defined(__GNUC__) || __has_attribute(__noinline__)
-# define __noinline __attribute__((__noinline__))
-# elif defined(_MSC_VER)
-# define __noinline __declspec(noinline)
-# else
-# define __noinline
-# endif
-#endif /* __noinline */
-
-#ifndef __must_check_result
-# if defined(__GNUC__) || __has_attribute(__warn_unused_result__)
-# define __must_check_result __attribute__((__warn_unused_result__))
-# else
-# define __must_check_result
-# endif
-#endif /* __must_check_result */
-
-#if !defined(__noop) && !defined(_MSC_VER)
-# define __noop(...) do {} while(0)
-#endif /* __noop */
-
-#if defined(__fallthrough) && \
- (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__))
-#undef __fallthrough
-#endif /* __fallthrough workaround for MinGW */
-
-#ifndef __fallthrough
-# if defined(__cplusplus) && (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || __cplusplus >= 201703L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(8, 0) && defined(__cplusplus) && __cplusplus >= 201103L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(7, 0) && \
- (!defined(__LCC__) || (__LCC__ == 124 && __LCC_MINOR__ >= 12) || \
- (__LCC__ == 125 && __LCC_MINOR__ >= 5) || (__LCC__ >= 126))
-# define __fallthrough __attribute__((__fallthrough__))
-# elif defined(__clang__) && defined(__cplusplus) && __cplusplus >= 201103L &&\
- __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
-# define __fallthrough [[clang::fallthrough]]
-# else
-# define __fallthrough
-# endif
-#endif /* __fallthrough */
-
-#ifndef __unreachable
-# if __GNUC_PREREQ(4,5) || __has_builtin(__builtin_unreachable)
-# define __unreachable() __builtin_unreachable()
-# elif defined(_MSC_VER)
-# define __unreachable() __assume(0)
-# else
-# define __unreachable() __noop()
-# endif
-#endif /* __unreachable */
-
-#ifndef __prefetch
-# if defined(__GNUC__) || defined(__clang__) || __has_builtin(__builtin_prefetch)
-# define __prefetch(ptr) __builtin_prefetch(ptr)
-# else
-# define __prefetch(ptr) __noop(ptr)
-# endif
-#endif /* __prefetch */
-
-#ifndef __nothrow
-# if defined(__cplusplus)
-# if __cplusplus < 201703L
-# define __nothrow throw()
-# else
-# define __nothrow noexcept(true)
-# endif /* __cplusplus */
-# elif defined(__GNUC__) || __has_attribute(__nothrow__)
-# define __nothrow __attribute__((__nothrow__))
-# elif defined(_MSC_VER) && defined(__cplusplus)
-# define __nothrow __declspec(nothrow)
-# else
-# define __nothrow
-# endif
-#endif /* __nothrow */
-
-#ifndef __hidden
-# if defined(__GNUC__) || __has_attribute(__visibility__)
-# define __hidden __attribute__((__visibility__("hidden")))
-# else
-# define __hidden
-# endif
-#endif /* __hidden */
-
-#ifndef __optimize
-# if defined(__OPTIMIZE__)
-# if (defined(__GNUC__) && !defined(__clang__)) || __has_attribute(__optimize__)
-# define __optimize(ops) __attribute__((__optimize__(ops)))
-# else
-# define __optimize(ops)
-# endif
-# else
-# define __optimize(ops)
-# endif
-#endif /* __optimize */
-
-#ifndef __hot
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __hot __attribute__((__hot__)) __optimize(3)
-# elif defined(__clang__) && !__has_attribute(__hot_) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put frequently used functions in separate section */
-# define __hot __attribute__((__section__("text.hot"))) __optimize("O3")
-# elif defined(__GNUC__) || __has_attribute(__hot__)
-# define __hot __attribute__((__hot__)) __optimize("O3")
-# else
-# define __hot __optimize("O3")
-# endif
-# else
-# define __hot
-# endif
-#endif /* __hot */
-
-#ifndef __cold
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __cold __attribute__((__cold__)) __optimize(1)
-# elif defined(__clang__) && !__has_attribute(cold) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put infrequently used functions in separate section */
-# define __cold __attribute__((__section__("text.unlikely"))) __optimize("Os")
-# elif defined(__GNUC__) || __has_attribute(cold)
-# define __cold __attribute__((__cold__)) __optimize("Os")
-# else
-# define __cold __optimize("Os")
-# endif
-# else
-# define __cold
-# endif
-#endif /* __cold */
-
-#ifndef __flatten
-# if defined(__OPTIMIZE__) && (defined(__GNUC__) || __has_attribute(__flatten__))
-# define __flatten __attribute__((__flatten__))
-# else
-# define __flatten
-# endif
-#endif /* __flatten */
-
-#ifndef likely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define likely(cond) __builtin_expect(!!(cond), 1)
-# else
-# define likely(x) (!!(x))
-# endif
-#endif /* likely */
-
-#ifndef unlikely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define unlikely(cond) __builtin_expect(!!(cond), 0)
-# else
-# define unlikely(x) (!!(x))
-# endif
-#endif /* unlikely */
-
-#ifndef __anonymous_struct_extension__
-# if defined(__GNUC__)
-# define __anonymous_struct_extension__ __extension__
-# else
-# define __anonymous_struct_extension__
-# endif
-#endif /* __anonymous_struct_extension__ */
-
-#ifndef __Wpedantic_format_voidptr
- MDBX_MAYBE_UNUSED MDBX_PURE_FUNCTION static __inline const void*
- __Wpedantic_format_voidptr(const void* ptr) {return ptr;}
-# define __Wpedantic_format_voidptr(ARG) __Wpedantic_format_voidptr(ARG)
-#endif /* __Wpedantic_format_voidptr */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(MDBX_USE_VALGRIND)
-# include <valgrind/memcheck.h>
-# ifndef VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE
- /* LY: available since Valgrind 3.10 */
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# endif
-#elif !defined(RUNNING_ON_VALGRIND)
-# define VALGRIND_CREATE_MEMPOOL(h,r,z)
-# define VALGRIND_DESTROY_MEMPOOL(h)
-# define VALGRIND_MEMPOOL_TRIM(h,a,s)
-# define VALGRIND_MEMPOOL_ALLOC(h,a,s)
-# define VALGRIND_MEMPOOL_FREE(h,a)
-# define VALGRIND_MEMPOOL_CHANGE(h,a,b,s)
-# define VALGRIND_MAKE_MEM_NOACCESS(a,s)
-# define VALGRIND_MAKE_MEM_DEFINED(a,s)
-# define VALGRIND_MAKE_MEM_UNDEFINED(a,s)
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_CHECK_MEM_IS_ADDRESSABLE(a,s) (0)
-# define VALGRIND_CHECK_MEM_IS_DEFINED(a,s) (0)
-# define RUNNING_ON_VALGRIND (0)
-#endif /* MDBX_USE_VALGRIND */
-
-#ifdef __SANITIZE_ADDRESS__
-# include <sanitizer/asan_interface.h>
-#elif !defined(ASAN_POISON_MEMORY_REGION)
-# define ASAN_POISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-# define ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-#endif /* __SANITIZE_ADDRESS__ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef ARRAY_LENGTH
-# ifdef __cplusplus
- template <typename T, size_t N>
- char (&__ArraySizeHelper(T (&array)[N]))[N];
-# define ARRAY_LENGTH(array) (sizeof(::__ArraySizeHelper(array)))
-# else
-# define ARRAY_LENGTH(array) (sizeof(array) / sizeof(array[0]))
-# endif
-#endif /* ARRAY_LENGTH */
-
-#ifndef ARRAY_END
-# define ARRAY_END(array) (&array[ARRAY_LENGTH(array)])
-#endif /* ARRAY_END */
-
-#define CONCAT(a,b) a##b
-#define XCONCAT(a,b) CONCAT(a,b)
-
-#ifndef offsetof
-# define offsetof(type, member) __builtin_offsetof(type, member)
-#endif /* offsetof */
-
-#ifndef container_of
-# define container_of(ptr, type, member) \
- ((type *)((char *)(ptr) - offsetof(type, member)))
-#endif /* container_of */
-
-#define MDBX_TETRAD(a, b, c, d) \
- ((uint32_t)(a) << 24 | (uint32_t)(b) << 16 | (uint32_t)(c) << 8 | (d))
-
-#define MDBX_STRING_TETRAD(str) MDBX_TETRAD(str[0], str[1], str[2], str[3])
-
-#define FIXME "FIXME: " __FILE__ ", " MDBX_STRINGIFY(__LINE__)
-
-#ifndef STATIC_ASSERT_MSG
-# if defined(static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) static_assert(expr, msg)
-# elif defined(_STATIC_ASSERT)
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif defined(_MSC_VER)
-# include <crtdbg.h>
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) \
- || __has_feature(c_static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) _Static_assert(expr, msg)
-# else
-# define STATIC_ASSERT_MSG(expr, msg) switch (0) {case 0:case (expr):;}
-# endif
-#endif /* STATIC_ASSERT */
-
-#ifndef STATIC_ASSERT
-# define STATIC_ASSERT(expr) STATIC_ASSERT_MSG(expr, #expr)
-#endif
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__GNUC__) && !__GNUC_PREREQ(4, 2)
-/* Actually libmdbx was not tested with compilers older than GCC 4.2.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required GCC >= 4.2"
-#endif
-
-#if defined(__clang__) && !__CLANG_PREREQ(3, 8)
-/* Actually libmdbx was not tested with CLANG older than 3.8.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required CLANG >= 3.8"
-#endif
-
-#if defined(__GLIBC__) && !__GLIBC_PREREQ(2, 12)
-/* Actually libmdbx was not tested with something older than glibc 2.12.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old systems.
- */
-#warning "libmdbx was only tested with GLIBC >= 2.12."
-#endif
-
-#ifdef __SANITIZE_THREAD__
-#warning \
- "libmdbx don't compatible with ThreadSanitizer, you will get a lot of false-positive issues."
-#endif /* __SANITIZE_THREAD__ */
-
-#if __has_warning("-Wnested-anon-types")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wnested-anon-types"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wnested-anon-types"
-#else
-#pragma warning disable "nested-anon-types"
-#endif
-#endif /* -Wnested-anon-types */
-
-#if __has_warning("-Wconstant-logical-operand")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wconstant-logical-operand"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wconstant-logical-operand"
-#else
-#pragma warning disable "constant-logical-operand"
-#endif
-#endif /* -Wconstant-logical-operand */
-
-#if defined(__LCC__) && (__LCC__ <= 121)
-/* bug #2798 */
-#pragma diag_suppress alignment_reduction_ignored
-#elif defined(__ICC)
-#pragma warning(disable : 3453 1366)
-#elif __has_warning("-Walignment-reduction-ignored")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Walignment-reduction-ignored"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Walignment-reduction-ignored"
-#else
-#pragma warning disable "alignment-reduction-ignored"
-#endif
-#endif /* -Walignment-reduction-ignored */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(_WIN32) || defined(_WIN64)
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if !defined(_NO_CRT_STDIO_INLINE) && MDBX_BUILD_SHARED_LIBRARY && \
- !defined(xMDBX_TOOLS) && MDBX_WITHOUT_MSVC_CRT
-#define _NO_CRT_STDIO_INLINE
-#endif
-#elif !defined(_POSIX_C_SOURCE)
-#define _POSIX_C_SOURCE 200809L
-#endif /* Windows */
-
-/*----------------------------------------------------------------------------*/
-/* C99 includes */
-#include <inttypes.h>
-#include <stddef.h>
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <assert.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-/* C11' alignas() */
-#if __has_include(<stdalign.h>)
-#include <stdalign.h>
-#endif
-#if defined(alignas) || defined(__cplusplus)
-#define MDBX_ALIGNAS(N) alignas(N)
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
-#define MDBX_ALIGNAS(N) _Alignas(N)
-#elif defined(_MSC_VER)
-#define MDBX_ALIGNAS(N) __declspec(align(N))
-#elif __has_attribute(__aligned__) || defined(__GNUC__)
-#define MDBX_ALIGNAS(N) __attribute__((__aligned__(N)))
-#else
-#error "FIXME: Required alignas() or equivalent."
-#endif /* MDBX_ALIGNAS */
-
-/*----------------------------------------------------------------------------*/
-/* Systems includes */
-
-#ifdef __APPLE__
-#include <TargetConditionals.h>
-#endif /* Apple OSX & iOS */
-
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__)
-#include <sys/cdefs.h>
-#include <sys/mount.h>
-#include <sys/sysctl.h>
-#include <sys/types.h>
-#if defined(__FreeBSD__) || defined(__DragonFly__)
-#include <vm/vm_param.h>
-#elif defined(__OpenBSD__) || defined(__NetBSD__)
-#include <uvm/uvm_param.h>
-#else
-#define SYSCTL_LEGACY_NONCONST_MIB
-#endif
-#ifndef __MACH__
-#include <sys/vmmeter.h>
-#endif
-#else
-#include <malloc.h>
-#if !(defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(_WIN32) || defined(_WIN64))
-#include <mntent.h>
-#endif /* !Solaris */
-#endif /* !xBSD */
-
-#if defined(__FreeBSD__) || __has_include(<malloc_np.h>)
-#include <malloc_np.h>
-#endif
-
-#if defined(__APPLE__) || defined(__MACH__) || __has_include(<malloc/malloc.h>)
-#include <malloc/malloc.h>
-#endif /* MacOS */
-
-#if defined(__MACH__)
-#include <mach/host_info.h>
-#include <mach/mach_host.h>
-#include <mach/mach_port.h>
-#include <uuid/uuid.h>
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-#include <sched.h>
-#include <sys/sendfile.h>
-#include <sys/statfs.h>
-#endif /* Linux */
-
-#ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 0
-#endif
-
-#ifndef _XOPEN_SOURCE_EXTENDED
-#define _XOPEN_SOURCE_EXTENDED 0
-#else
-#include <utmpx.h>
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#include <kstat.h>
-#include <sys/mnttab.h>
-/* On Solaris, it's easier to add a missing prototype rather than find a
- * combination of #defines that break nothing. */
-__extern_C key_t ftok(const char *, int);
-#endif /* SunOS/Solaris */
-
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef _WIN32_WINNT
-#define _WIN32_WINNT 0x0601 /* Windows 7 */
-#elif _WIN32_WINNT < 0x0500
-#error At least 'Windows 2000' API is required for libmdbx.
-#endif /* _WIN32_WINNT */
-#if (defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* MinGW */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif /* WIN32_LEAN_AND_MEAN */
-#include <excpt.h>
-#include <tlhelp32.h>
-#include <windows.h>
-#include <winnt.h>
-#include <winternl.h>
-#define HAVE_SYS_STAT_H
-#define HAVE_SYS_TYPES_H
-typedef HANDLE mdbx_thread_t;
-typedef unsigned mdbx_thread_key_t;
-#define MAP_FAILED NULL
-#define HIGH_DWORD(v) ((DWORD)((sizeof(v) > 4) ? ((uint64_t)(v) >> 32) : 0))
-#define THREAD_CALL WINAPI
-#define THREAD_RESULT DWORD
-typedef struct {
- HANDLE mutex;
- HANDLE event[2];
-} mdbx_condpair_t;
-typedef CRITICAL_SECTION mdbx_fastmutex_t;
-
-#if !defined(_MSC_VER) && !defined(__try)
-/* *INDENT-OFF* */
-/* clang-format off */
-#define __try
-#define __except(COND) if(false)
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* stub for MSVC's __try/__except */
-
-#if MDBX_WITHOUT_MSVC_CRT
-
-#ifndef mdbx_malloc
-static inline void *mdbx_malloc(size_t bytes) {
- return HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_malloc */
-
-#ifndef mdbx_calloc
-static inline void *mdbx_calloc(size_t nelem, size_t size) {
- return HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, nelem * size);
-}
-#endif /* mdbx_calloc */
-
-#ifndef mdbx_realloc
-static inline void *mdbx_realloc(void *ptr, size_t bytes) {
- return ptr ? HeapReAlloc(GetProcessHeap(), 0, ptr, bytes)
- : HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_realloc */
-
-#ifndef mdbx_free
-static inline void mdbx_free(void *ptr) { HeapFree(GetProcessHeap(), 0, ptr); }
-#endif /* mdbx_free */
-
-#else /* MDBX_WITHOUT_MSVC_CRT */
-
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup _strdup
-
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-#ifndef snprintf
-#define snprintf _snprintf /* ntdll */
-#endif
-
-#ifndef vsnprintf
-#define vsnprintf _vsnprintf /* ntdll */
-#endif
-
-#else /*----------------------------------------------------------------------*/
-
-#include <unistd.h>
-#if !defined(_POSIX_MAPPED_FILES) || _POSIX_MAPPED_FILES < 1
-#error "libmdbx requires the _POSIX_MAPPED_FILES feature"
-#endif /* _POSIX_MAPPED_FILES */
-
-#include <pthread.h>
-#include <semaphore.h>
-#include <signal.h>
-#include <sys/file.h>
-#include <sys/ipc.h>
-#include <sys/mman.h>
-#include <sys/param.h>
-#include <sys/stat.h>
-#include <sys/statvfs.h>
-#include <sys/uio.h>
-typedef pthread_t mdbx_thread_t;
-typedef pthread_key_t mdbx_thread_key_t;
-#define INVALID_HANDLE_VALUE (-1)
-#define THREAD_CALL
-#define THREAD_RESULT void *
-typedef struct {
- pthread_mutex_t mutex;
- pthread_cond_t cond[2];
-} mdbx_condpair_t;
-typedef pthread_mutex_t mdbx_fastmutex_t;
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup strdup
-#endif /* Platform */
-
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
-/* malloc_usable_size() already provided */
-#elif defined(__APPLE__)
-#define malloc_usable_size(ptr) malloc_size(ptr)
-#elif defined(_MSC_VER) && !MDBX_WITHOUT_MSVC_CRT
-#define malloc_usable_size(ptr) _msize(ptr)
-#endif /* malloc_usable_size */
-
-#ifdef __ANDROID_API__
-#include <android/log.h>
-#if __ANDROID_API__ >= 21
-#include <sys/sendfile.h>
-#endif
-#endif /* Android */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(HAVE_SYS_STAT_H) || __has_include(<sys/stat.h>)
-#include <sys/stat.h>
-#endif
-#if defined(HAVE_SYS_TYPES_H) || __has_include(<sys/types.h>)
-#include <sys/types.h>
-#endif
-#if defined(HAVE_SYS_FILE_H) || __has_include(<sys/file.h>)
-#include <sys/file.h>
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-
-#ifndef SSIZE_MAX
-#define SSIZE_MAX INTPTR_MAX
-#endif
-
-#if !defined(MADV_DODUMP) && defined(MADV_CORE)
-#define MADV_DODUMP MADV_CORE
-#endif /* MADV_CORE -> MADV_DODUMP */
-
-#if !defined(MADV_DONTDUMP) && defined(MADV_NOCORE)
-#define MADV_DONTDUMP MADV_NOCORE
-#endif /* MADV_NOCORE -> MADV_DONTDUMP */
-
-#if defined(i386) || defined(__386) || defined(__i386) || defined(__i386__) || \
- defined(i486) || defined(__i486) || defined(__i486__) || \
- defined(i586) | defined(__i586) || defined(__i586__) || defined(i686) || \
- defined(__i686) || defined(__i686__) || defined(_M_IX86) || \
- defined(_X86_) || defined(__THW_INTEL__) || defined(__I86__) || \
- defined(__INTEL__) || defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64__) || defined(__amd64) || defined(_M_X64) || \
- defined(_M_AMD64) || defined(__IA32__) || defined(__INTEL__)
-#ifndef __ia32__
-/* LY: define neutral __ia32__ for x86 and x86-64 */
-#define __ia32__ 1
-#endif /* __ia32__ */
-#if !defined(__amd64__) && (defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64) || defined(_M_X64))
-/* LY: define trusty __amd64__ for all AMD64/x86-64 arch */
-#define __amd64__ 1
-#endif /* __amd64__ */
-#endif /* all x86 */
-
-#if (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
-#error \
- "Sanity checking failed: Two's complement, reasonably sized integer types"
-#endif
-
-#if UINTPTR_MAX > 0xffffFFFFul || ULONG_MAX > 0xffffFFFFul
-#define MDBX_WORDBITS 64
-#else
-#define MDBX_WORDBITS 32
-#endif /* MDBX_WORDBITS */
-
-#if defined(__ANDROID_API__) || defined(ANDROID)
-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "_FILE_OFFSET_BITS != MDBX_WORDBITS" (_FILE_OFFSET_BITS != MDBX_WORDBITS)
-#elif defined(__FILE_OFFSET_BITS) && __FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "__FILE_OFFSET_BITS != MDBX_WORDBITS" (__FILE_OFFSET_BITS != MDBX_WORDBITS)
-#endif
-#endif /* Android */
-
-/*----------------------------------------------------------------------------*/
-/* Compiler's includes for builtins/intrinsics */
-
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
-#include <intrin.h>
-#elif __GNUC_PREREQ(4, 4) || defined(__clang__)
-#if defined(__ia32__) || defined(__e2k__)
-#include <x86intrin.h>
-#endif /* __ia32__ */
-#if defined(__ia32__)
-#include <cpuid.h>
-#endif /* __ia32__ */
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
-#include <mbarrier.h>
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
-#include <machine/sys/inline.h>
-#elif defined(__IBMC__) && defined(__powerpc)
-#include <atomic.h>
-#elif defined(_AIX)
-#include <builtins.h>
-#include <sys/atomic_op.h>
-#elif (defined(__osf__) && defined(__DECC)) || defined(__alpha)
-#include <c_asm.h>
-#include <machine/builtins.h>
-#elif defined(__MWERKS__)
-/* CodeWarrior - troubles ? */
-#pragma gcc_extensions
-#elif defined(__SNC__)
-/* Sony PS3 - troubles ? */
-#elif defined(__hppa__) || defined(__hppa)
-#include <machine/inline.h>
-#else
-#error Unsupported C compiler, please use GNU C 4.4 or newer
-#endif /* Compiler */
-
-/*----------------------------------------------------------------------------*/
-/* Byteorder */
-
-#if !defined(__BYTE_ORDER__) || !defined(__ORDER_LITTLE_ENDIAN__) || \
- !defined(__ORDER_BIG_ENDIAN__)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(__GLIBC__) || defined(__GNU_LIBRARY__) || defined(__ANDROID_API__) || \
- defined(HAVE_ENDIAN_H) || __has_include(<endian.h>)
-#include <endian.h>
-#elif defined(__APPLE__) || defined(__MACH__) || defined(__OpenBSD__) || \
- defined(HAVE_MACHINE_ENDIAN_H) || __has_include(<machine/endian.h>)
-#include <machine/endian.h>
-#elif defined(HAVE_SYS_ISA_DEFS_H) || __has_include(<sys/isa_defs.h>)
-#include <sys/isa_defs.h>
-#elif (defined(HAVE_SYS_TYPES_H) && defined(HAVE_SYS_ENDIAN_H)) || \
- (__has_include(<sys/types.h>) && __has_include(<sys/endian.h>))
-#include <sys/endian.h>
-#include <sys/types.h>
-#elif defined(__bsdi__) || defined(__DragonFly__) || defined(__FreeBSD__) || \
- defined(__NetBSD__) || \
- defined(HAVE_SYS_PARAM_H) || __has_include(<sys/param.h>)
-#include <sys/param.h>
-#endif /* OS */
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-
-#else
-#error __BYTE_ORDER__ should be defined.
-#endif /* Arch */
-
-#endif
-#endif /* __BYTE_ORDER__ || __ORDER_LITTLE_ENDIAN__ || __ORDER_BIG_ENDIAN__ */
-
-/* Get the size of a memory page for the system.
- * This is the basic size that the platform's memory manager uses, and is
- * fundamental to the use of memory-mapped files. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline size_t
-mdbx_syspagesize(void) {
-#if defined(_WIN32) || defined(_WIN64)
- SYSTEM_INFO si;
- GetSystemInfo(&si);
- return si.dwPageSize;
-#else
- return sysconf(_SC_PAGE_SIZE);
-#endif
-}
-
-typedef struct mdbx_mmap_param {
- union {
- void *address;
- uint8_t *dxb;
- struct MDBX_lockinfo *lck;
- };
- mdbx_filehandle_t fd;
- size_t limit; /* mapping length, but NOT a size of file nor DB */
- size_t current; /* mapped region size, i.e. the size of file and DB */
- uint64_t filesize /* in-process cache of a file size */;
-#if defined(_WIN32) || defined(_WIN64)
- HANDLE section; /* memory-mapped section handle */
-#endif
-} mdbx_mmap_t;
-
-typedef union bin128 {
- __anonymous_struct_extension__ struct { uint64_t x, y; };
- __anonymous_struct_extension__ struct { uint32_t a, b, c, d; };
-} bin128_t;
-
-#if defined(_WIN32) || defined(_WIN64)
-typedef union MDBX_srwlock {
- __anonymous_struct_extension__ struct {
- long volatile readerCount;
- long volatile writerCount;
- };
- RTL_SRWLOCK native;
-} MDBX_srwlock;
-#endif /* Windows */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny);
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny);
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-#if defined(__cplusplus) && !defined(__STDC_NO_ATOMICS__) && (__has_include(<cstdatomic>) || __has_extension(cxx_atomic))
-#include <cstdatomic>
-#define MDBX_HAVE_C11ATOMICS
-#elif !defined(__cplusplus) && \
- (__STDC_VERSION__ >= 201112L || __has_extension(c_atomic)) && \
- !defined(__STDC_NO_ATOMICS__) && \
- (__GNUC_PREREQ(4, 9) || __CLANG_PREREQ(3, 8) || \
- !(defined(__GNUC__) || defined(__clang__)))
-#include <stdatomic.h>
-#define MDBX_HAVE_C11ATOMICS
-#elif defined(__GNUC__) || defined(__clang__)
-#elif defined(_MSC_VER)
-#pragma warning(disable : 4163) /* 'xyz': not available as an intrinsic */
-#pragma warning(disable : 4133) /* 'function': incompatible types - from \
- 'size_t' to 'LONGLONG' */
-#pragma warning(disable : 4244) /* 'return': conversion from 'LONGLONG' to \
- 'std::size_t', possible loss of data */
-#pragma warning(disable : 4267) /* 'function': conversion from 'size_t' to \
- 'long', possible loss of data */
-#pragma intrinsic(_InterlockedExchangeAdd, _InterlockedCompareExchange)
-#pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedCompareExchange64)
-#elif defined(__APPLE__)
-#include <libkern/OSAtomic.h>
-#else
-#error FIXME atomic-ops
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Memory/Compiler barriers, cache coherence */
-
-#if __has_include(<sys/cachectl.h>)
-#include <sys/cachectl.h>
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS should have explicit cache control */
-#include <sys/cachectl.h>
-#endif
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_compiler_barrier(void) {
-#if defined(__clang__) || defined(__GNUC__)
- __asm__ __volatile__("" ::: "memory");
-#elif defined(_MSC_VER)
- _ReadWriteBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
- __memory_barrier();
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __compiler_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_sched_fence(/* LY: no-arg meaning 'all expect ALU', e.g. 0x3D3D */);
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __fence();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_memory_barrier(void) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(memory_order_seq_cst);
-#elif defined(__ATOMIC_SEQ_CST)
-#ifdef __clang__
- __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
- __atomic_thread_fence(__ATOMIC_SEQ_CST);
-#endif
-#elif defined(__clang__) || defined(__GNUC__)
- __sync_synchronize();
-#elif defined(_WIN32) || defined(_WIN64)
- MemoryBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
-#if defined(__ia32__)
- _mm_mfence();
-#else
- __mf();
-#endif
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __machine_rw_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_mf();
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __lwsync();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* libc compatibility stuff */
-
-#if (!defined(__GLIBC__) && __GLIBC_PREREQ(2, 1)) && \
- (defined(_GNU_SOURCE) || defined(_BSD_SOURCE))
-#define mdbx_asprintf asprintf
-#define mdbx_vasprintf vasprintf
-#else
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC
- MDBX_PRINTF_ARGS(2, 3) int mdbx_asprintf(char **strp, const char *fmt, ...);
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt, va_list ap);
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* OS abstraction layer stuff */
-
-/* max bytes to write in one call */
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_WRITE UINT32_C(0x01000000)
-#else
-#define MAX_WRITE UINT32_C(0x3fff0000)
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-MDBX_INTERNAL_VAR uint32_t mdbx_linux_kernel_version;
-MDBX_INTERNAL_VAR bool mdbx_RunningOnWSL1 /* Windows Subsystem 1 for Linux */;
-#endif /* Linux */
-
-#ifndef mdbx_strdup
-LIBMDBX_API char *mdbx_strdup(const char *str);
-#endif
-
-MDBX_MAYBE_UNUSED static __inline int mdbx_get_errno(void) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD rc = GetLastError();
-#else
- int rc = errno;
-#endif
- return rc;
-}
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result);
-#endif
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr);
-#endif
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair, bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair);
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex);
-
-MDBX_INTERNAL_FUNC int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov,
- int iovcnt, uint64_t offset,
- size_t expected_written);
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t count,
- uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t count, uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t count);
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg);
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread);
-
-enum mdbx_syncmode_bits {
- MDBX_SYNC_NONE = 0,
- MDBX_SYNC_DATA = 1,
- MDBX_SYNC_SIZE = 2,
- MDBX_SYNC_IODQ = 4
-};
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- const enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length);
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos);
-MDBX_INTERNAL_FUNC int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length);
-
-enum mdbx_openfile_purpose {
- MDBX_OPEN_DXB_READ = 0,
- MDBX_OPEN_DXB_LAZY = 1,
- MDBX_OPEN_DXB_DSYNC = 2,
- MDBX_OPEN_LCK = 3,
- MDBX_OPEN_COPY = 4,
- MDBX_OPEN_DELETE = 5
-};
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait);
-
-#define MMAP_OPTION_TRUNCATE 1
-#define MMAP_OPTION_SEMAPHORE 2
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t must, const size_t limit,
- const unsigned options);
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map);
-#define MDBX_MRESIZE_MAY_MOVE 0x00000100
-#define MDBX_MRESIZE_MAY_UNMAP 0x00000200
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit);
-#if defined(_WIN32) || defined(_WIN64)
-typedef struct {
- unsigned limit, count;
- HANDLE handles[31];
-} mdbx_handle_array_t;
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array);
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array);
-#endif /* Windows */
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err);
-
-MDBX_MAYBE_UNUSED static __inline uint32_t mdbx_getpid(void) {
- STATIC_ASSERT(sizeof(mdbx_pid_t) <= sizeof(uint32_t));
-#if defined(_WIN32) || defined(_WIN64)
- return GetCurrentProcessId();
-#else
- STATIC_ASSERT(sizeof(pid_t) <= sizeof(uint32_t));
- return getpid();
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline uintptr_t mdbx_thread_self(void) {
- mdbx_tid_t thunk;
- STATIC_ASSERT(sizeof(uintptr_t) >= sizeof(thunk));
-#if defined(_WIN32) || defined(_WIN64)
- thunk = GetCurrentThreadId();
-#else
- thunk = pthread_self();
-#endif
- return (uintptr_t)thunk;
-}
-
-#if !defined(_WIN32) && !defined(_WIN64)
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void);
-#else
-static __inline int mdbx_check_tid4bionic(void) { return 0; }
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-MDBX_MAYBE_UNUSED static __inline int
-mdbx_pthread_mutex_lock(pthread_mutex_t *mutex) {
- int err = mdbx_check_tid4bionic();
- return unlikely(err) ? err : pthread_mutex_lock(mutex);
-}
-#endif /* !Windows */
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void);
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16);
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime);
-
-MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void);
-/*----------------------------------------------------------------------------*/
-/* lck stuff */
-
-/// \brief Initialization of synchronization primitives linked with MDBX_env
-/// instance both in LCK-file and within the current process.
-/// \param
-/// global_uniqueness_flag = true - denotes that there are no other processes
-/// working with DB and LCK-file. Thus the function MUST initialize
-/// shared synchronization objects in memory-mapped LCK-file.
-/// global_uniqueness_flag = false - denotes that at least one process is
-/// already working with DB and LCK-file, including the case when DB
-/// has already been opened in the current process. Thus the function
-/// MUST NOT initialize shared synchronization objects in memory-mapped
-/// LCK-file that are already in use.
-/// \return Error code or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag);
-
-/// \brief Disconnects from shared interprocess objects and destructs
-/// synchronization objects linked with MDBX_env instance
-/// within the current process.
-/// \param
-/// inprocess_neighbor = NULL - if the current process does not have other
-/// instances of MDBX_env linked with the DB being closed.
-/// Thus the function MUST check for other processes working with DB or
-/// LCK-file, and keep or destroy shared synchronization objects in
-/// memory-mapped LCK-file depending on the result.
-/// inprocess_neighbor = not-NULL - pointer to another instance of MDBX_env
-/// (anyone of there is several) working with DB or LCK-file within the
-/// current process. Thus the function MUST NOT try to acquire exclusive
-/// lock and/or try to destruct shared synchronization objects linked with
-/// DB or LCK-file. Moreover, the implementation MUST ensure correct work
-/// of other instances of MDBX_env within the current process, e.g.
-/// restore POSIX-fcntl locks after the closing of file descriptors.
-/// \return Error code (MDBX_PANIC) or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor);
-
-/// \brief Connects to shared interprocess locking objects and tries to acquire
-/// the maximum lock level (shared if exclusive is not available)
-/// Depending on implementation or/and platform (Windows) this function may
-/// acquire the non-OS super-level lock (e.g. for shared synchronization
-/// objects initialization), which will be downgraded to OS-exclusive or
-/// shared via explicit calling of mdbx_lck_downgrade().
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if an exclusive lock was acquired and thus
-/// the current process is the first and only after the last use of DB.
-/// MDBX_RESULT_FALSE (0) - if a shared lock was acquired and thus
-/// DB has already been opened and now is used by other processes.
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env);
-
-/// \brief Downgrades the level of initially acquired lock to
-/// operational level specified by argument. The reson for such downgrade:
-/// - unblocking of other processes that are waiting for access, i.e.
-/// if (env->me_flags & MDBX_EXCLUSIVE) != 0, then other processes
-/// should be made aware that access is unavailable rather than
-/// wait for it.
-/// - freeing locks that interfere file operation (especially for Windows)
-/// (env->me_flags & MDBX_EXCLUSIVE) == 0 - downgrade to shared lock.
-/// (env->me_flags & MDBX_EXCLUSIVE) != 0 - downgrade to exclusive
-/// operational lock.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env);
-
-/// \brief Locks LCK-file or/and table of readers for (de)registering.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env);
-
-/// \brief Unlocks LCK-file or/and table of readers after (de)registering.
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env);
-
-/// \brief Acquires lock for DB change (on writing transaction start)
-/// Reading transactions will not be blocked.
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-/// \return Error code or zero on success
-LIBMDBX_API int mdbx_txn_lock(MDBX_env *env, bool dont_wait);
-
-/// \brief Releases lock once DB changes is made (after writing transaction
-/// has finished).
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-LIBMDBX_API void mdbx_txn_unlock(MDBX_env *env);
-
-/// \brief Sets alive-flag of reader presence (indicative lock) for PID of
-/// the current process. The function does no more than needed for
-/// the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env);
-
-/// \brief Resets alive-flag of reader presence (indicative lock)
-/// for PID of the current process. The function does no more than needed
-/// for the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env);
-
-/// \brief Checks for reading process status with the given pid with help of
-/// alive-flag of presence (indicative lock) or using another way.
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if the reader process with the given PID is alive
-/// and working with DB (indicative lock is present).
-/// MDBX_RESULT_FALSE (0) - if the reader process with the given PID is absent
-/// or not working with DB (indicative lock is not present).
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid);
-
-#if defined(_WIN32) || defined(_WIN64)
-
-typedef void(WINAPI *MDBX_srwlock_function)(MDBX_srwlock *);
-MDBX_INTERNAL_VAR MDBX_srwlock_function mdbx_srwlock_Init,
- mdbx_srwlock_AcquireShared, mdbx_srwlock_ReleaseShared,
- mdbx_srwlock_AcquireExclusive, mdbx_srwlock_ReleaseExclusive;
-
-#if _WIN32_WINNT < 0x0600 /* prior to Windows Vista */
-typedef enum _FILE_INFO_BY_HANDLE_CLASS {
- FileBasicInfo,
- FileStandardInfo,
- FileNameInfo,
- FileRenameInfo,
- FileDispositionInfo,
- FileAllocationInfo,
- FileEndOfFileInfo,
- FileStreamInfo,
- FileCompressionInfo,
- FileAttributeTagInfo,
- FileIdBothDirectoryInfo,
- FileIdBothDirectoryRestartInfo,
- FileIoPriorityHintInfo,
- FileRemoteProtocolInfo,
- MaximumFileInfoByHandleClass
-} FILE_INFO_BY_HANDLE_CLASS,
- *PFILE_INFO_BY_HANDLE_CLASS;
-
-typedef struct _FILE_END_OF_FILE_INFO {
- LARGE_INTEGER EndOfFile;
-} FILE_END_OF_FILE_INFO, *PFILE_END_OF_FILE_INFO;
-
-#define REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK 0x00000001
-#define REMOTE_PROTOCOL_INFO_FLAG_OFFLINE 0x00000002
-
-typedef struct _FILE_REMOTE_PROTOCOL_INFO {
- USHORT StructureVersion;
- USHORT StructureSize;
- DWORD Protocol;
- USHORT ProtocolMajorVersion;
- USHORT ProtocolMinorVersion;
- USHORT ProtocolRevision;
- USHORT Reserved;
- DWORD Flags;
- struct {
- DWORD Reserved[8];
- } GenericReserved;
- struct {
- DWORD Reserved[16];
- } ProtocolSpecificReserved;
-} FILE_REMOTE_PROTOCOL_INFO, *PFILE_REMOTE_PROTOCOL_INFO;
-
-#endif /* _WIN32_WINNT < 0x0600 (prior to Windows Vista) */
-
-typedef BOOL(WINAPI *MDBX_GetFileInformationByHandleEx)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_GetFileInformationByHandleEx
- mdbx_GetFileInformationByHandleEx;
-
-typedef BOOL(WINAPI *MDBX_GetVolumeInformationByHandleW)(
- _In_ HANDLE hFile, _Out_opt_ LPWSTR lpVolumeNameBuffer,
- _In_ DWORD nVolumeNameSize, _Out_opt_ LPDWORD lpVolumeSerialNumber,
- _Out_opt_ LPDWORD lpMaximumComponentLength,
- _Out_opt_ LPDWORD lpFileSystemFlags,
- _Out_opt_ LPWSTR lpFileSystemNameBuffer, _In_ DWORD nFileSystemNameSize);
-MDBX_INTERNAL_VAR MDBX_GetVolumeInformationByHandleW
- mdbx_GetVolumeInformationByHandleW;
-
-typedef DWORD(WINAPI *MDBX_GetFinalPathNameByHandleW)(_In_ HANDLE hFile,
- _Out_ LPWSTR lpszFilePath,
- _In_ DWORD cchFilePath,
- _In_ DWORD dwFlags);
-MDBX_INTERNAL_VAR MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-
-typedef BOOL(WINAPI *MDBX_SetFileInformationByHandle)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_SetFileInformationByHandle
- mdbx_SetFileInformationByHandle;
-
-typedef NTSTATUS(NTAPI *MDBX_NtFsControlFile)(
- IN HANDLE FileHandle, IN OUT HANDLE Event,
- IN OUT PVOID /* PIO_APC_ROUTINE */ ApcRoutine, IN OUT PVOID ApcContext,
- OUT PIO_STATUS_BLOCK IoStatusBlock, IN ULONG FsControlCode,
- IN OUT PVOID InputBuffer, IN ULONG InputBufferLength,
- OUT OPTIONAL PVOID OutputBuffer, IN ULONG OutputBufferLength);
-MDBX_INTERNAL_VAR MDBX_NtFsControlFile mdbx_NtFsControlFile;
-
-typedef uint64_t(WINAPI *MDBX_GetTickCount64)(void);
-MDBX_INTERNAL_VAR MDBX_GetTickCount64 mdbx_GetTickCount64;
-
-#if !defined(_WIN32_WINNT_WIN8) || _WIN32_WINNT < _WIN32_WINNT_WIN8
-typedef struct _WIN32_MEMORY_RANGE_ENTRY {
- PVOID VirtualAddress;
- SIZE_T NumberOfBytes;
-} WIN32_MEMORY_RANGE_ENTRY, *PWIN32_MEMORY_RANGE_ENTRY;
-#endif /* Windows 8.x */
-
-typedef BOOL(WINAPI *MDBX_PrefetchVirtualMemory)(
- HANDLE hProcess, ULONG_PTR NumberOfEntries,
- PWIN32_MEMORY_RANGE_ENTRY VirtualAddresses, ULONG Flags);
-MDBX_INTERNAL_VAR MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-
-typedef enum _SECTION_INHERIT { ViewShare = 1, ViewUnmap = 2 } SECTION_INHERIT;
-
-typedef NTSTATUS(NTAPI *MDBX_NtExtendSection)(IN HANDLE SectionHandle,
- IN PLARGE_INTEGER NewSectionSize);
-MDBX_INTERNAL_VAR MDBX_NtExtendSection mdbx_NtExtendSection;
-
-static __inline bool mdbx_RunningUnderWine(void) {
- return !mdbx_NtExtendSection;
-}
-
-typedef LSTATUS(WINAPI *MDBX_RegGetValueA)(HKEY hkey, LPCSTR lpSubKey,
- LPCSTR lpValue, DWORD dwFlags,
- LPDWORD pdwType, PVOID pvData,
- LPDWORD pcbData);
-MDBX_INTERNAL_VAR MDBX_RegGetValueA mdbx_RegGetValueA;
-
-NTSYSAPI ULONG RtlRandomEx(PULONG Seed);
-
-#endif /* Windows */
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_MSC_VER) && _MSC_VER >= 1900
-/* LY: MSVC 2015/2017/2019 has buggy/inconsistent PRIuPTR/PRIxPTR macros
- * for internal format-args checker. */
-#undef PRIuPTR
-#undef PRIiPTR
-#undef PRIdPTR
-#undef PRIxPTR
-#define PRIuPTR "Iu"
-#define PRIiPTR "Ii"
-#define PRIdPTR "Id"
-#define PRIxPTR "Ix"
-#define PRIuSIZE "zu"
-#define PRIiSIZE "zi"
-#define PRIdSIZE "zd"
-#define PRIxSIZE "zx"
-#endif /* fix PRI*PTR for _MSC_VER */
-
-#ifndef PRIuSIZE
-#define PRIuSIZE PRIuPTR
-#define PRIiSIZE PRIiPTR
-#define PRIdSIZE PRIdPTR
-#define PRIxSIZE PRIxPTR
-#endif /* PRI*SIZE macros for MSVC */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-#define mdbx_sourcery_anchor XCONCAT(mdbx_sourcery_, MDBX_BUILD_SOURCERY)
-#if defined(xMDBX_TOOLS)
-extern LIBMDBX_API const char *const mdbx_sourcery_anchor;
-#endif
-
-/*******************************************************************************
- *******************************************************************************
- *******************************************************************************
- *
- *
- * #### ##### ##### # #### # # ####
- * # # # # # # # # ## # #
- * # # # # # # # # # # # ####
- * # # ##### # # # # # # # #
- * # # # # # # # # ## # #
- * #### # # # #### # # ####
- *
- *
- */
-
-/** \defgroup build_option Build options
- * The libmdbx build options.
- @{ */
-
-/** Using fcntl(F_FULLFSYNC) with 5-10 times slowdown */
-#define MDBX_OSX_WANNA_DURABILITY 0
-/** Using fsync() with chance of data lost on power failure */
-#define MDBX_OSX_WANNA_SPEED 1
-
-#ifndef MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-/** Choices \ref MDBX_OSX_WANNA_DURABILITY or \ref MDBX_OSX_WANNA_SPEED
- * for OSX & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY MDBX_OSX_WANNA_DURABILITY
-#endif /* MDBX_OSX_SPEED_INSTEADOF_DURABILITY */
-
-/** Controls checking PID against reuse DB environment after the fork() */
-#ifndef MDBX_ENV_CHECKPID
-#if defined(MADV_DONTFORK) || defined(_WIN32) || defined(_WIN64)
-/* PID check could be omitted:
- * - on Linux when madvise(MADV_DONTFORK) is available, i.e. after the fork()
- * mapped pages will not be available for child process.
- * - in Windows where fork() not available. */
-#define MDBX_ENV_CHECKPID 0
-#else
-#define MDBX_ENV_CHECKPID 1
-#endif
-#define MDBX_ENV_CHECKPID_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#else
-#define MDBX_ENV_CHECKPID_CONFIG MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#endif /* MDBX_ENV_CHECKPID */
-
-/** Controls checking transaction owner thread against misuse transactions from
- * other threads. */
-#ifndef MDBX_TXN_CHECKOWNER
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_TXN_CHECKOWNER_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#else
-#define MDBX_TXN_CHECKOWNER_CONFIG MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#endif /* MDBX_TXN_CHECKOWNER */
-
-/** Does a system have battery-backed Real-Time Clock or just a fake. */
-#ifndef MDBX_TRUST_RTC
-#if defined(__linux__) || defined(__gnu_linux__) || defined(__NetBSD__) || \
- defined(__OpenBSD__)
-#define MDBX_TRUST_RTC 0 /* a lot of embedded systems have a fake RTC */
-#else
-#define MDBX_TRUST_RTC 1
-#endif
-#define MDBX_TRUST_RTC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#else
-#define MDBX_TRUST_RTC_CONFIG MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#endif /* MDBX_TRUST_RTC */
-
-/** Controls online database auto-compactification during write-transactions. */
-#ifndef MDBX_ENABLE_REFUND
-#define MDBX_ENABLE_REFUND 1
-#elif !(MDBX_ENABLE_REFUND == 0 || MDBX_ENABLE_REFUND == 1)
-#error MDBX_ENABLE_REFUND must be defined as 0 or 1
-#endif /* MDBX_ENABLE_REFUND */
-
-/** Controls gathering statistics for page operations. */
-#ifndef MDBX_ENABLE_PGOP_STAT
-#define MDBX_ENABLE_PGOP_STAT 1
-#elif !(MDBX_ENABLE_PGOP_STAT == 0 || MDBX_ENABLE_PGOP_STAT == 1)
-#error MDBX_ENABLE_PGOP_STAT must be defined as 0 or 1
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-/** Controls use of POSIX madvise() hints and friends. */
-#ifndef MDBX_ENABLE_MADVISE
-#define MDBX_ENABLE_MADVISE 1
-#elif !(MDBX_ENABLE_MADVISE == 0 || MDBX_ENABLE_MADVISE == 1)
-#error MDBX_ENABLE_MADVISE must be defined as 0 or 1
-#endif /* MDBX_ENABLE_MADVISE */
-
-/** Disable some checks to reduce an overhead and detection probability of
- * database corruption to a values closer to the LMDB. */
-#ifndef MDBX_DISABLE_PAGECHECKS
-#define MDBX_DISABLE_PAGECHECKS 0
-#elif !(MDBX_DISABLE_PAGECHECKS == 0 || MDBX_DISABLE_PAGECHECKS == 1)
-#error MDBX_DISABLE_PAGECHECKS must be defined as 0 or 1
-#endif /* MDBX_DISABLE_PAGECHECKS */
-
-#ifndef MDBX_PNL_PREALLOC_FOR_RADIXSORT
-#define MDBX_PNL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_PNL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_PNL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_PNL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
-
-#ifndef MDBX_DPL_PREALLOC_FOR_RADIXSORT
-#define MDBX_DPL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_DPL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_DPL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_DPL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
-
-/* Basically, this build-option is for TODO. Guess it should be replaced
- * with MDBX_ENABLE_WRITEMAP_SPILLING with the three variants:
- * 0/OFF = Don't track dirty pages at all and don't spilling ones.
- * This should be by-default on Linux and may-be other systems
- * (not sure: Darwin/OSX, FreeBSD, Windows 10) where kernel provides
- * properly LRU tracking and async writing on-demand.
- * 1/ON = Lite tracking of dirty pages but with LRU labels and explicit
- * spilling with msync(MS_ASYNC). */
-#ifndef MDBX_FAKE_SPILL_WRITEMAP
-#if defined(__linux__) || defined(__gnu_linux__)
-#define MDBX_FAKE_SPILL_WRITEMAP 1 /* msync(MS_ASYNC) is no-op on Linux */
-#else
-#define MDBX_FAKE_SPILL_WRITEMAP 0
-#endif
-#elif !(MDBX_FAKE_SPILL_WRITEMAP == 0 || MDBX_FAKE_SPILL_WRITEMAP == 1)
-#error MDBX_FAKE_SPILL_WRITEMAP must be defined as 0 or 1
-#endif /* MDBX_FAKE_SPILL_WRITEMAP */
-
-/** Controls sort order of internal page number lists.
- * This mostly experimental/advanced option with not for regular MDBX users.
- * \warning The database format depend on this option and libmdbx builded with
- * different option value are incompatible. */
-#ifndef MDBX_PNL_ASCENDING
-#define MDBX_PNL_ASCENDING 0
-#elif !(MDBX_PNL_ASCENDING == 0 || MDBX_PNL_ASCENDING == 1)
-#error MDBX_PNL_ASCENDING must be defined as 0 or 1
-#endif /* MDBX_PNL_ASCENDING */
-
-/** Avoid dependence from MSVC CRT and use ntdll.dll instead. */
-#ifndef MDBX_WITHOUT_MSVC_CRT
-#define MDBX_WITHOUT_MSVC_CRT 1
-#elif !(MDBX_WITHOUT_MSVC_CRT == 0 || MDBX_WITHOUT_MSVC_CRT == 1)
-#error MDBX_WITHOUT_MSVC_CRT must be defined as 0 or 1
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-/** Size of buffer used during copying a environment/database file. */
-#ifndef MDBX_ENVCOPY_WRITEBUF
-#define MDBX_ENVCOPY_WRITEBUF 1048576u
-#elif MDBX_ENVCOPY_WRITEBUF < 65536u || MDBX_ENVCOPY_WRITEBUF > 1073741824u || \
- MDBX_ENVCOPY_WRITEBUF % 65536u
-#error MDBX_ENVCOPY_WRITEBUF must be defined in range 65536..1073741824 and be multiple of 65536
-#endif /* MDBX_ENVCOPY_WRITEBUF */
-
-/** Forces assertion checking */
-#ifndef MDBX_FORCE_ASSERTIONS
-#define MDBX_FORCE_ASSERTIONS 0
-#elif !(MDBX_FORCE_ASSERTIONS == 0 || MDBX_FORCE_ASSERTIONS == 1)
-#error MDBX_FORCE_ASSERTIONS must be defined as 0 or 1
-#endif /* MDBX_FORCE_ASSERTIONS */
-
-/** Presumed malloc size overhead for each allocation
- * to adjust allocations to be more aligned. */
-#ifndef MDBX_ASSUME_MALLOC_OVERHEAD
-#ifdef __SIZEOF_POINTER__
-#define MDBX_ASSUME_MALLOC_OVERHEAD (__SIZEOF_POINTER__ * 2u)
-#else
-#define MDBX_ASSUME_MALLOC_OVERHEAD (sizeof(void *) * 2u)
-#endif
-#elif MDBX_ASSUME_MALLOC_OVERHEAD < 0 || MDBX_ASSUME_MALLOC_OVERHEAD > 64 || \
- MDBX_ASSUME_MALLOC_OVERHEAD % 4
-#error MDBX_ASSUME_MALLOC_OVERHEAD must be defined in range 0..64 and be multiple of 4
-#endif /* MDBX_ASSUME_MALLOC_OVERHEAD */
-
-/** In case the MDBX_DEBUG is undefined set it corresponding to NDEBUG */
-#ifndef MDBX_DEBUG
-#ifdef NDEBUG
-#define MDBX_DEBUG 0
-#else
-#define MDBX_DEBUG 1
-#endif
-#endif /* MDBX_DEBUG */
-
-/** If defined then enables integration with Valgrind,
- * a memory analyzing tool. */
-#ifndef MDBX_USE_VALGRIND
-#endif /* MDBX_USE_VALGRIND */
-
-/** If defined then enables use C11 atomics,
- * otherwise detects ones availability automatically. */
-#ifndef MDBX_HAVE_C11ATOMICS
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-//------------------------------------------------------------------------------
-
-/** Win32 File Locking API for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_WIN32FILES -1
-
-/** SystemV IPC semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_SYSV 5
-
-/** POSIX-1 Shared anonymous semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX1988 1988
-
-/** POSIX-2001 Shared Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2001 2001
-
-/** POSIX-2008 Robust Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2008 2008
-
-/** BeOS Benaphores, aka Futexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_BENAPHORE 1995
-
-/** Advanced: Choices the locking implementation (autodetection by default). */
-#if defined(_WIN32) || defined(_WIN64)
-#define MDBX_LOCKING MDBX_LOCKING_WIN32FILES
-#else
-#ifndef MDBX_LOCKING
-#if defined(_POSIX_THREAD_PROCESS_SHARED) && \
- _POSIX_THREAD_PROCESS_SHARED >= 200112L && !defined(__FreeBSD__)
-
-/* Some platforms define the EOWNERDEAD error code even though they
- * don't support Robust Mutexes. If doubt compile with -MDBX_LOCKING=2001. */
-#if defined(EOWNERDEAD) && _POSIX_THREAD_PROCESS_SHARED >= 200809L && \
- ((defined(_POSIX_THREAD_ROBUST_PRIO_INHERIT) && \
- _POSIX_THREAD_ROBUST_PRIO_INHERIT > 0) || \
- (defined(_POSIX_THREAD_ROBUST_PRIO_PROTECT) && \
- _POSIX_THREAD_ROBUST_PRIO_PROTECT > 0) || \
- defined(PTHREAD_MUTEX_ROBUST) || defined(PTHREAD_MUTEX_ROBUST_NP)) && \
- (!defined(__GLIBC__) || \
- __GLIBC_PREREQ(2, 10) /* troubles with Robust mutexes before 2.10 */)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2008
-#else
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2001
-#endif
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX1988
-#else
-#define MDBX_LOCKING MDBX_LOCKING_SYSV
-#endif
-#define MDBX_LOCKING_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_LOCKING)
-#else
-#define MDBX_LOCKING_CONFIG MDBX_STRINGIFY(MDBX_LOCKING)
-#endif /* MDBX_LOCKING */
-#endif /* !Windows */
-
-/** Advanced: Using POSIX OFD-locks (autodetection by default). */
-#ifndef MDBX_USE_OFDLOCKS
-#if defined(F_OFD_SETLK) && defined(F_OFD_SETLKW) && defined(F_OFD_GETLK) && \
- !defined(MDBX_SAFE4QEMU) && \
- !defined(__sun) /* OFD-lock are broken on Solaris */
-#define MDBX_USE_OFDLOCKS 1
-#else
-#define MDBX_USE_OFDLOCKS 0
-#endif
-#define MDBX_USE_OFDLOCKS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#else
-#define MDBX_USE_OFDLOCKS_CONFIG MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#endif /* MDBX_USE_OFDLOCKS */
-
-/** Advanced: Using sendfile() syscall (autodetection by default). */
-#ifndef MDBX_USE_SENDFILE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 21)
-#define MDBX_USE_SENDFILE 1
-#else
-#define MDBX_USE_SENDFILE 0
-#endif
-#endif /* MDBX_USE_SENDFILE */
-
-/** Advanced: Using copy_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_COPYFILERANGE
-#if __GLIBC_PREREQ(2, 27) && defined(_GNU_SOURCE)
-#define MDBX_USE_COPYFILERANGE 1
-#else
-#define MDBX_USE_COPYFILERANGE 0
-#endif
-#endif /* MDBX_USE_COPYFILERANGE */
-
-/** Advanced: Using sync_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_SYNCFILERANGE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- defined(SYNC_FILE_RANGE_WRITE) && !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 26)
-#define MDBX_USE_SYNCFILERANGE 1
-#else
-#define MDBX_USE_SYNCFILERANGE 0
-#endif
-#endif /* MDBX_USE_SYNCFILERANGE */
-
-//------------------------------------------------------------------------------
-
-#ifndef MDBX_CPU_WRITEBACK_INCOHERENT
-#if defined(__ia32__) || defined(__e2k__) || defined(__hppa) || \
- defined(__hppa__) || defined(DOXYGEN)
-#define MDBX_CPU_WRITEBACK_INCOHERENT 0
-#else
-#define MDBX_CPU_WRITEBACK_INCOHERENT 1
-#endif
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-#ifndef MDBX_MMAP_INCOHERENT_FILE_WRITE
-#ifdef __OpenBSD__
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 1
-#else
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#ifndef MDBX_MMAP_INCOHERENT_CPU_CACHE
-#if defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS has cache coherency issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 1
-#else
-/* LY: assume no relevant mmap/dcache issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#ifndef MDBX_64BIT_ATOMIC
-#if MDBX_WORDBITS >= 64 || defined(DOXYGEN)
-#define MDBX_64BIT_ATOMIC 1
-#else
-#define MDBX_64BIT_ATOMIC 0
-#endif
-#define MDBX_64BIT_ATOMIC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#else
-#define MDBX_64BIT_ATOMIC_CONFIG MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#endif /* MDBX_64BIT_ATOMIC */
-
-#ifndef MDBX_64BIT_CAS
-#if defined(ATOMIC_LLONG_LOCK_FREE)
-#if ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__GCC_ATOMIC_LLONG_LOCK_FREE)
-#if __GCC_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__CLANG_ATOMIC_LLONG_LOCK_FREE)
-#if __CLANG_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(_MSC_VER) || defined(__APPLE__) || defined(DOXYGEN)
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS MDBX_64BIT_ATOMIC
-#endif
-#define MDBX_64BIT_CAS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#else
-#define MDBX_64BIT_CAS_CONFIG MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#endif /* MDBX_64BIT_CAS */
-
-#ifndef MDBX_UNALIGNED_OK
-#if defined(__ALIGNED__) || defined(__SANITIZE_UNDEFINED__)
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#elif defined(__ARM_FEATURE_UNALIGNED)
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#elif __CLANG_PREREQ(5, 0) || __GNUC_PREREQ(5, 0)
-/* expecting an optimization will well done, also this
- * hushes false-positives from UBSAN (undefined behaviour sanitizer) */
-#define MDBX_UNALIGNED_OK 0
-#elif defined(__e2k__) || defined(__elbrus__)
-#if __iset__ > 4
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#endif
-#elif defined(__ia32__)
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#endif
-#elif MDBX_UNALIGNED_OK == 1
-#undef MDBX_UNALIGNED_OK
-#define MDBX_UNALIGNED_OK 32 /* any unaligned access allowed */
-#endif /* MDBX_UNALIGNED_OK */
-
-#ifndef MDBX_CACHELINE_SIZE
-#if defined(SYSTEM_CACHE_ALIGNMENT_SIZE)
-#define MDBX_CACHELINE_SIZE SYSTEM_CACHE_ALIGNMENT_SIZE
-#elif defined(__ia64__) || defined(__ia64) || defined(_M_IA64)
-#define MDBX_CACHELINE_SIZE 128
-#else
-#define MDBX_CACHELINE_SIZE 64
-#endif
-#endif /* MDBX_CACHELINE_SIZE */
-
-/** @} end of build options */
-/*******************************************************************************
- *******************************************************************************
- ******************************************************************************/
-
-#ifdef DOXYGEN
-/* !!! Actually this is a fake definitions !!!
- * !!! for documentation generation by Doxygen !!! */
-
-/** Controls enabling of debugging features.
- *
- * - `MDBX_DEBUG = 0` (by default) Disables any debugging features at all,
- * including logging and assertion controls.
- * Logging level and corresponding debug flags changing
- * by \ref mdbx_setup_debug() will not have effect.
- * - `MDBX_DEBUG > 0` Enables code for the debugging features (logging,
- * assertions checking and internal audit).
- * Simultaneously sets the default logging level
- * to the `MDBX_DEBUG` value.
- * Also enables \ref MDBX_DBG_AUDIT if `MDBX_DEBUG >= 2`.
- *
- * \ingroup build_option */
-#define MDBX_DEBUG 0...7
-
-/** Disables using of GNU libc extensions. */
-#define MDBX_DISABLE_GNU_SOURCE 0 or 1
-
-#endif /* DOXYGEN */
-
-/* Undefine the NDEBUG if debugging is enforced by MDBX_DEBUG */
-#if MDBX_DEBUG
-#undef NDEBUG
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-enum MDBX_memory_order {
- mo_Relaxed,
- mo_AcquireRelease,
- mo_SequentialConsistency
-};
-
-typedef union {
- volatile uint32_t weak;
-#ifdef MDBX_HAVE_C11ATOMICS
- volatile _Atomic uint32_t c11a;
-#endif /* MDBX_HAVE_C11ATOMICS */
-} MDBX_atomic_uint32_t;
-
-typedef union {
- volatile uint64_t weak;
-#if defined(MDBX_HAVE_C11ATOMICS) && (MDBX_64BIT_CAS || MDBX_64BIT_ATOMIC)
- volatile _Atomic uint64_t c11a;
-#endif
-#if !defined(MDBX_HAVE_C11ATOMICS) || !MDBX_64BIT_CAS || !MDBX_64BIT_ATOMIC
- __anonymous_struct_extension__ struct {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- MDBX_atomic_uint32_t low, high;
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- MDBX_atomic_uint32_t high, low;
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- };
-#endif
-} MDBX_atomic_uint64_t;
-
-#ifdef MDBX_HAVE_C11ATOMICS
-
-/* Crutches for C11 atomic compiler's bugs */
-#if defined(__e2k__) && defined(__LCC__) && __LCC__ < /* FIXME */ 127
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->weak)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->weak)
-#elif defined(__clang__) && __clang__ < 8
-#define MDBX_c11a_ro(type, ptr) ((volatile _Atomic(type) *)&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#else
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#endif /* Crutches for C11 atomic compiler's bugs */
-
-static __always_inline memory_order mo_c11_store(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_release;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-
-static __always_inline memory_order mo_c11_load(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_acquire;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED static __always_inline void
-mdbx_memory_fence(enum MDBX_memory_order order, bool write) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(write ? mo_c11_store(order) : mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_compiler_barrier();
- if (write &&
- order > (MDBX_CPU_WRITEBACK_INCOHERENT ? mo_Relaxed : mo_AcquireRelease))
- mdbx_memory_barrier();
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_store32(MDBX_atomic_uint32_t *p, const uint32_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint32_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_load32(const MDBX_atomic_uint32_t *p, enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint32_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint32_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint32_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-/* Basic constants and types */
-
-/* A stamp that identifies a file as an MDBX file.
- * There's nothing special about this value other than that it is easily
- * recognizable, and it will reflect any byte order mismatches. */
-#define MDBX_MAGIC UINT64_C(/* 56-bit prime */ 0x59659DBDEF4C11)
-
-/* FROZEN: The version number for a database's datafile format. */
-#define MDBX_DATA_VERSION 3
-/* The version number for a database's lockfile format. */
-#define MDBX_LOCK_VERSION 4
-
-/* handle for the DB used to track free pages. */
-#define FREE_DBI 0
-/* handle for the default DB. */
-#define MAIN_DBI 1
-/* Number of DBs in metapage (free and main) - also hardcoded elsewhere */
-#define CORE_DBS 2
-
-/* Number of meta pages - also hardcoded elsewhere */
-#define NUM_METAS 3
-
-/* A page number in the database.
- *
- * MDBX uses 32 bit for page numbers. This limits database
- * size up to 2^44 bytes, in case of 4K pages. */
-typedef uint32_t pgno_t;
-typedef MDBX_atomic_uint32_t atomic_pgno_t;
-#define PRIaPGNO PRIu32
-#define MAX_PAGENO UINT32_C(0x7FFFffff)
-#define MIN_PAGENO NUM_METAS
-
-#define SAFE64_INVALID_THRESHOLD UINT64_C(0xffffFFFF00000000)
-
-/* A transaction ID. */
-typedef uint64_t txnid_t;
-typedef MDBX_atomic_uint64_t atomic_txnid_t;
-#define PRIaTXN PRIi64
-#define MIN_TXNID UINT64_C(1)
-#define MAX_TXNID (SAFE64_INVALID_THRESHOLD - 1)
-#define INITIAL_TXNID (MIN_TXNID + NUM_METAS - 1)
-#define INVALID_TXNID UINT64_MAX
-/* LY: for testing non-atomic 64-bit txnid on 32-bit arches.
- * #define xMDBX_TXNID_STEP (UINT32_MAX / 3) */
-#ifndef xMDBX_TXNID_STEP
-#if MDBX_64BIT_CAS
-#define xMDBX_TXNID_STEP 1u
-#else
-#define xMDBX_TXNID_STEP 2u
-#endif
-#endif /* xMDBX_TXNID_STEP */
-
-/* Used for offsets within a single page.
- * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
- * this is plenty. */
-typedef uint16_t indx_t;
-
-#define MEGABYTE ((size_t)1 << 20)
-
-/*----------------------------------------------------------------------------*/
-/* Core structures for database and shared memory (i.e. format definition) */
-#pragma pack(push, 4)
-
-/* Information about a single database in the environment. */
-typedef struct MDBX_db {
- uint16_t md_flags; /* see mdbx_dbi_open */
- uint16_t md_depth; /* depth of this tree */
- uint32_t md_xsize; /* key-size for MDBX_DUPFIXED (LEAF2 pages) */
- pgno_t md_root; /* the root page of this tree */
- pgno_t md_branch_pages; /* number of internal pages */
- pgno_t md_leaf_pages; /* number of leaf pages */
- pgno_t md_overflow_pages; /* number of overflow pages */
- uint64_t md_seq; /* table sequence counter */
- uint64_t md_entries; /* number of data items */
- uint64_t md_mod_txnid; /* txnid of last committed modification */
-} MDBX_db;
-
-/* database size-related parameters */
-typedef struct MDBX_geo {
- uint16_t grow_pv; /* datafile growth step as a 16-bit packed (exponential
- quantized) value */
- uint16_t shrink_pv; /* datafile shrink threshold as a 16-bit packed
- (exponential quantized) value */
- pgno_t lower; /* minimal size of datafile in pages */
- pgno_t upper; /* maximal size of datafile in pages */
- pgno_t now; /* current size of datafile in pages */
- pgno_t next; /* first unused page in the datafile,
- but actually the file may be shorter. */
-} MDBX_geo;
-
-/* Meta page content.
- * A meta page is the start point for accessing a database snapshot.
- * Pages 0-1 are meta pages. Transaction N writes meta page (N % 2). */
-typedef struct MDBX_meta {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with MDBX_DATA_VERSION. */
- uint32_t mm_magic_and_version[2];
-
- /* txnid that committed this page, the first of a two-phase-update pair */
- uint32_t mm_txnid_a[2];
-
- uint16_t mm_extra_flags; /* extra DB flags, zero (nothing) for now */
- uint8_t mm_validator_id; /* ID of checksum and page validation method,
- * zero (nothing) for now */
- uint8_t mm_extra_pagehdr; /* extra bytes in the page header,
- * zero (nothing) for now */
-
- MDBX_geo mm_geo; /* database size-related parameters */
-
- MDBX_db mm_dbs[CORE_DBS]; /* first is free space, 2nd is main db */
- /* The size of pages used in this DB */
-#define mm_psize mm_dbs[FREE_DBI].md_xsize
- MDBX_canary mm_canary;
-
-#define MDBX_DATASIGN_NONE 0u
-#define MDBX_DATASIGN_WEAK 1u
-#define SIGN_IS_STEADY(sign) ((sign) > MDBX_DATASIGN_WEAK)
-#define META_IS_STEADY(meta) \
- SIGN_IS_STEADY(unaligned_peek_u64_volatile(4, (meta)->mm_datasync_sign))
- uint32_t mm_datasync_sign[2];
-
- /* txnid that committed this page, the second of a two-phase-update pair */
- uint32_t mm_txnid_b[2];
-
- /* Number of non-meta pages which were put in GC after COW. May be 0 in case
- * DB was previously handled by libmdbx without corresponding feature.
- * This value in couple with mr_snapshot_pages_retired allows fast estimation
- * of "how much reader is restraining GC recycling". */
- uint32_t mm_pages_retired[2];
-
- /* The analogue /proc/sys/kernel/random/boot_id or similar to determine
- * whether the system was rebooted after the last use of the database files.
- * If there was no reboot, but there is no need to rollback to the last
- * steady sync point. Zeros mean that no relevant information is available
- * from the system. */
- bin128_t mm_bootid;
-
-} MDBX_meta;
-
-#pragma pack(1)
-
-/* Common header for all page types. The page type depends on mp_flags.
- *
- * P_BRANCH and P_LEAF pages have unsorted 'MDBX_node's at the end, with
- * sorted mp_ptrs[] entries referring to them. Exception: P_LEAF2 pages
- * omit mp_ptrs and pack sorted MDBX_DUPFIXED values after the page header.
- *
- * P_OVERFLOW records occupy one or more contiguous pages where only the
- * first has a page header. They hold the real data of F_BIGDATA nodes.
- *
- * P_SUBP sub-pages are small leaf "pages" with duplicate data.
- * A node with flag F_DUPDATA but not F_SUBDATA contains a sub-page.
- * (Duplicate data can also go in sub-databases, which use normal pages.)
- *
- * P_META pages contain MDBX_meta, the start point of an MDBX snapshot.
- *
- * Each non-metapage up to MDBX_meta.mm_last_pg is reachable exactly once
- * in the snapshot: Either used by a database or listed in a GC record. */
-typedef struct MDBX_page {
- union {
-#define IS_FROZEN(txn, p) ((p)->mp_txnid < (txn)->mt_txnid)
-#define IS_SPILLED(txn, p) ((p)->mp_txnid == (txn)->mt_txnid)
-#define IS_SHADOWED(txn, p) ((p)->mp_txnid > (txn)->mt_txnid)
-#define IS_VALID(txn, p) ((p)->mp_txnid <= (txn)->mt_front)
-#define IS_MODIFIABLE(txn, p) ((p)->mp_txnid == (txn)->mt_front)
- uint64_t mp_txnid;
- struct MDBX_page *mp_next; /* for in-memory list of freed pages */
- };
- uint16_t mp_leaf2_ksize; /* key size if this is a LEAF2 page */
-#define P_BRANCH 0x01 /* branch page */
-#define P_LEAF 0x02 /* leaf page */
-#define P_OVERFLOW 0x04 /* overflow page */
-#define P_META 0x08 /* meta page */
-#define P_BAD 0x10 /* explicit flag for invalid/bad page */
-#define P_LEAF2 0x20 /* for MDBX_DUPFIXED records */
-#define P_SUBP 0x40 /* for MDBX_DUPSORT sub-pages */
-#define P_SPILLED 0x2000 /* spilled in parent txn */
-#define P_LOOSE 0x4000 /* page was dirtied then freed, can be reused */
-#define P_FROZEN 0x8000 /* used for retire page with known status */
-#define P_ILL_BITS (~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_SPILLED))
- uint16_t mp_flags;
- union {
- uint32_t mp_pages; /* number of overflow pages */
- __anonymous_struct_extension__ struct {
- indx_t mp_lower; /* lower bound of free space */
- indx_t mp_upper; /* upper bound of free space */
- };
- };
- pgno_t mp_pgno; /* page number */
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- indx_t mp_ptrs[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_page;
-
-/* Size of the page header, excluding dynamic data at the end */
-#define PAGEHDRSZ ((unsigned)offsetof(MDBX_page, mp_ptrs))
-
-#pragma pack(pop)
-
-#if MDBX_ENABLE_PGOP_STAT
-/* Statistics of page operations overall of all (running, completed and aborted)
- * transactions */
-typedef struct {
- MDBX_atomic_uint64_t newly; /* Quantity of a new pages added */
- MDBX_atomic_uint64_t cow; /* Quantity of pages copied for update */
- MDBX_atomic_uint64_t clone; /* Quantity of parent's dirty pages clones
- for nested transactions */
- MDBX_atomic_uint64_t split; /* Page splits */
- MDBX_atomic_uint64_t merge; /* Page merges */
- MDBX_atomic_uint64_t spill; /* Quantity of spilled dirty pages */
- MDBX_atomic_uint64_t unspill; /* Quantity of unspilled/reloaded pages */
- MDBX_atomic_uint64_t
- wops; /* Number of explicit write operations (not a pages) to a disk */
-} MDBX_pgop_stat_t;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-#if MDBX_LOCKING == MDBX_LOCKING_WIN32FILES
-#define MDBX_CLOCK_SIGN UINT32_C(0xF10C)
-typedef void mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
-
-#define MDBX_CLOCK_SIGN UINT32_C(0xF18D)
-typedef mdbx_pid_t mdbx_ipclock_t;
-#ifndef EOWNERDEAD
-#define EOWNERDEAD MDBX_RESULT_TRUE
-#endif
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#define MDBX_CLOCK_SIGN UINT32_C(0x8017)
-typedef pthread_mutex_t mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-#define MDBX_CLOCK_SIGN UINT32_C(0xFC29)
-typedef sem_t mdbx_ipclock_t;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV && !defined(__cplusplus)
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc);
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc);
-#endif /* MDBX_LOCKING */
-
-/* Reader Lock Table
- *
- * Readers don't acquire any locks for their data access. Instead, they
- * simply record their transaction ID in the reader table. The reader
- * mutex is needed just to find an empty slot in the reader table. The
- * slot's address is saved in thread-specific data so that subsequent
- * read transactions started by the same thread need no further locking to
- * proceed.
- *
- * If MDBX_NOTLS is set, the slot address is not saved in thread-specific data.
- * No reader table is used if the database is on a read-only filesystem.
- *
- * Since the database uses multi-version concurrency control, readers don't
- * actually need any locking. This table is used to keep track of which
- * readers are using data from which old transactions, so that we'll know
- * when a particular old transaction is no longer in use. Old transactions
- * that have discarded any data pages can then have those pages reclaimed
- * for use by a later write transaction.
- *
- * The lock table is constructed such that reader slots are aligned with the
- * processor's cache line size. Any slot is only ever used by one thread.
- * This alignment guarantees that there will be no contention or cache
- * thrashing as threads update their own slot info, and also eliminates
- * any need for locking when accessing a slot.
- *
- * A writer thread will scan every slot in the table to determine the oldest
- * outstanding reader transaction. Any freed pages older than this will be
- * reclaimed by the writer. The writer doesn't use any locks when scanning
- * this table. This means that there's no guarantee that the writer will
- * see the most up-to-date reader info, but that's not required for correct
- * operation - all we need is to know the upper bound on the oldest reader,
- * we don't care at all about the newest reader. So the only consequence of
- * reading stale information here is that old pages might hang around a
- * while longer before being reclaimed. That's actually good anyway, because
- * the longer we delay reclaiming old pages, the more likely it is that a
- * string of contiguous pages can be found after coalescing old pages from
- * many old transactions together. */
-
-/* The actual reader record, with cacheline padding. */
-typedef struct MDBX_reader {
- /* Current Transaction ID when this transaction began, or (txnid_t)-1.
- * Multiple readers that start at the same time will probably have the
- * same ID here. Again, it's not important to exclude them from
- * anything; all we need to know is which version of the DB they
- * started from so we can avoid overwriting any data used in that
- * particular version. */
- MDBX_atomic_uint64_t /* txnid_t */ mr_txnid;
-
- /* The information we store in a single slot of the reader table.
- * In addition to a transaction ID, we also record the process and
- * thread ID that owns a slot, so that we can detect stale information,
- * e.g. threads or processes that went away without cleaning up.
- *
- * NOTE: We currently don't check for stale records.
- * We simply re-init the table when we know that we're the only process
- * opening the lock file. */
-
- /* The thread ID of the thread owning this txn. */
- MDBX_atomic_uint64_t mr_tid;
-
- /* The process ID of the process owning this reader txn. */
- MDBX_atomic_uint32_t mr_pid;
-
- /* The number of pages used in the reader's MVCC snapshot,
- * i.e. the value of meta->mm_geo.next and txn->mt_next_pgno */
- atomic_pgno_t mr_snapshot_pages_used;
- /* Number of retired pages at the time this reader starts transaction. So,
- * at any time the difference mm_pages_retired - mr_snapshot_pages_retired
- * will give the number of pages which this reader restraining from reuse. */
- MDBX_atomic_uint64_t mr_snapshot_pages_retired;
-} MDBX_reader;
-
-/* The header for the reader table (a memory-mapped lock file). */
-typedef struct MDBX_lockinfo {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with with MDBX_LOCK_VERSION. */
- uint64_t mti_magic_and_version;
-
- /* Format of this lock file. Must be set to MDBX_LOCK_FORMAT. */
- uint32_t mti_os_and_format;
-
- /* Flags which environment was opened. */
- MDBX_atomic_uint32_t mti_envmode;
-
- /* Threshold of un-synced-with-disk pages for auto-sync feature,
- * zero means no-threshold, i.e. auto-sync is disabled. */
- atomic_pgno_t mti_autosync_threshold;
-
- /* Low 32-bit of txnid with which meta-pages was synced,
- * i.e. for sync-polling in the MDBX_NOMETASYNC mode. */
- MDBX_atomic_uint32_t mti_meta_sync_txnid;
-
- /* Period for timed auto-sync feature, i.e. at the every steady checkpoint
- * the mti_unsynced_timeout sets to the current_time + mti_autosync_period.
- * The time value is represented in a suitable system-dependent form, for
- * example clock_gettime(CLOCK_BOOTTIME) or clock_gettime(CLOCK_MONOTONIC).
- * Zero means timed auto-sync is disabled. */
- MDBX_atomic_uint64_t mti_autosync_period;
-
- /* Marker to distinguish uniqueness of DB/CLK. */
- MDBX_atomic_uint64_t mti_bait_uniqueness;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
-#if MDBX_ENABLE_PGOP_STAT
- /* Statistics of costly ops of all (running, completed and aborted)
- * transactions */
- MDBX_pgop_stat_t mti_pgop_stat;
-#endif /* MDBX_ENABLE_PGOP_STAT*/
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Write transaction lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_wlock;
-#endif /* MDBX_LOCKING > 0 */
-
- atomic_txnid_t mti_oldest_reader;
-
- /* Timestamp of the last steady sync. Value is represented in a suitable
- * system-dependent form, for example clock_gettime(CLOCK_BOOTTIME) or
- * clock_gettime(CLOCK_MONOTONIC). */
- MDBX_atomic_uint64_t mti_sync_timestamp;
-
- /* Number un-synced-with-disk pages for auto-sync feature. */
- atomic_pgno_t mti_unsynced_pages;
-
- /* Number of page which was discarded last time by madvise(MADV_FREE). */
- atomic_pgno_t mti_discarded_tail;
-
- /* Timestamp of the last readers check. */
- MDBX_atomic_uint64_t mti_reader_check_timestamp;
-
- /* Shared anchor for tracking readahead edge and enabled/disabled status. */
- pgno_t mti_readahead_anchor;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Readeaders registration lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_rlock;
-#endif /* MDBX_LOCKING > 0 */
-
- /* The number of slots that have been used in the reader table.
- * This always records the maximum count, it is not decremented
- * when readers release their slots. */
- MDBX_atomic_uint32_t mti_numreaders;
- MDBX_atomic_uint32_t mti_readers_refresh_flag;
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
- MDBX_reader mti_readers[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_lockinfo;
-
-/* Lockfile format signature: version, features and field layout */
-#define MDBX_LOCK_FORMAT \
- (MDBX_CLOCK_SIGN * 27733 + (unsigned)sizeof(MDBX_reader) * 13 + \
- (unsigned)offsetof(MDBX_reader, mr_snapshot_pages_used) * 251 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_oldest_reader) * 83 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_numreaders) * 37 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_readers) * 29)
-
-#define MDBX_DATA_MAGIC \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + MDBX_DATA_VERSION)
-
-#define MDBX_DATA_MAGIC_LEGACY_COMPAT \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + 2)
-
-#define MDBX_DATA_MAGIC_LEGACY_DEVEL ((MDBX_MAGIC << 8) + 255)
-
-#define MDBX_LOCK_MAGIC ((MDBX_MAGIC << 8) + MDBX_LOCK_VERSION)
-
-/* The maximum size of a database page.
- *
- * It is 64K, but value-PAGEHDRSZ must fit in MDBX_page.mp_upper.
- *
- * MDBX will use database pages < OS pages if needed.
- * That causes more I/O in write transactions: The OS must
- * know (read) the whole page before writing a partial page.
- *
- * Note that we don't currently support Huge pages. On Linux,
- * regular data files cannot use Huge pages, and in general
- * Huge pages aren't actually pageable. We rely on the OS
- * demand-pager to read our data and page it out when memory
- * pressure from other processes is high. So until OSs have
- * actual paging support for Huge pages, they're not viable. */
-#define MAX_PAGESIZE MDBX_MAX_PAGESIZE
-#define MIN_PAGESIZE MDBX_MIN_PAGESIZE
-
-#define MIN_MAPSIZE (MIN_PAGESIZE * MIN_PAGENO)
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_MAPSIZE32 UINT32_C(0x38000000)
-#else
-#define MAX_MAPSIZE32 UINT32_C(0x7f000000)
-#endif
-#define MAX_MAPSIZE64 ((MAX_PAGENO + 1) * (uint64_t)MAX_PAGESIZE)
-
-#if MDBX_WORDBITS >= 64
-#define MAX_MAPSIZE MAX_MAPSIZE64
-#define MDBX_PGL_LIMIT ((size_t)MAX_PAGENO)
-#else
-#define MAX_MAPSIZE MAX_MAPSIZE32
-#define MDBX_PGL_LIMIT (MAX_MAPSIZE32 / MIN_PAGESIZE)
-#endif /* MDBX_WORDBITS */
-
-#define MDBX_READERS_LIMIT 32767
-#define MDBX_RADIXSORT_THRESHOLD 333
-
-/*----------------------------------------------------------------------------*/
-
-/* An PNL is an Page Number List, a sorted array of IDs.
- * The first element of the array is a counter for how many actual page-numbers
- * are in the list. By default PNLs are sorted in descending order, this allow
- * cut off a page with lowest pgno (at the tail) just truncating the list. The
- * sort order of PNLs is controlled by the MDBX_PNL_ASCENDING build option. */
-typedef pgno_t *MDBX_PNL;
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_ORDERED(first, last) ((first) < (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) >= (last))
-#else
-#define MDBX_PNL_ORDERED(first, last) ((first) > (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) <= (last))
-#endif
-
-/* List of txnid, only for MDBX_txn.tw.lifo_reclaimed */
-typedef txnid_t *MDBX_TXL;
-
-/* An Dirty-Page list item is an pgno/pointer pair. */
-typedef struct MDBX_dp {
- MDBX_page *ptr;
- pgno_t pgno;
- union {
- unsigned extra;
- __anonymous_struct_extension__ struct {
- unsigned multi : 1;
- unsigned lru : 31;
- };
- };
-} MDBX_dp;
-
-/* An DPL (dirty-page list) is a sorted array of MDBX_DPs. */
-typedef struct MDBX_dpl {
- unsigned sorted;
- unsigned length;
- unsigned detent; /* allocated size excluding the MDBX_DPL_RESERVE_GAP */
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_dp items[] /* dynamic size with holes at zero and after the last */;
-#endif
-} MDBX_dpl;
-
-/* PNL sizes */
-#define MDBX_PNL_GRANULATE 1024
-#define MDBX_PNL_INITIAL \
- (MDBX_PNL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(pgno_t))
-
-#define MDBX_TXL_GRANULATE 32
-#define MDBX_TXL_INITIAL \
- (MDBX_TXL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-#define MDBX_TXL_MAX \
- ((1u << 17) - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-
-#define MDBX_PNL_ALLOCLEN(pl) ((pl)[-1])
-#define MDBX_PNL_SIZE(pl) ((pl)[0])
-#define MDBX_PNL_FIRST(pl) ((pl)[1])
-#define MDBX_PNL_LAST(pl) ((pl)[MDBX_PNL_SIZE(pl)])
-#define MDBX_PNL_BEGIN(pl) (&(pl)[1])
-#define MDBX_PNL_END(pl) (&(pl)[MDBX_PNL_SIZE(pl) + 1])
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_FIRST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_LAST(pl)
-#else
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_LAST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_FIRST(pl)
-#endif
-
-#define MDBX_PNL_SIZEOF(pl) ((MDBX_PNL_SIZE(pl) + 1) * sizeof(pgno_t))
-#define MDBX_PNL_IS_EMPTY(pl) (MDBX_PNL_SIZE(pl) == 0)
-
-/*----------------------------------------------------------------------------*/
-/* Internal structures */
-
-/* Auxiliary DB info.
- * The information here is mostly static/read-only. There is
- * only a single copy of this record in the environment. */
-typedef struct MDBX_dbx {
- MDBX_val md_name; /* name of the database */
- MDBX_cmp_func *md_cmp; /* function for comparing keys */
- MDBX_cmp_func *md_dcmp; /* function for comparing data items */
- size_t md_klen_min, md_klen_max; /* min/max key length for the database */
- size_t md_vlen_min,
- md_vlen_max; /* min/max value/data length for the database */
-} MDBX_dbx;
-
-/* A database transaction.
- * Every operation requires a transaction handle. */
-struct MDBX_txn {
-#define MDBX_MT_SIGNATURE UINT32_C(0x93D53A31)
- uint32_t mt_signature;
-
- /* Transaction Flags */
- /* mdbx_txn_begin() flags */
-#define MDBX_TXN_RO_BEGIN_FLAGS (MDBX_TXN_RDONLY | MDBX_TXN_RDONLY_PREPARE)
-#define MDBX_TXN_RW_BEGIN_FLAGS \
- (MDBX_TXN_NOMETASYNC | MDBX_TXN_NOSYNC | MDBX_TXN_TRY)
- /* Additional flag for mdbx_sync_locked() */
-#define MDBX_SHRINK_ALLOWED UINT32_C(0x40000000)
-
- /* internal txn flags */
-#define MDBX_TXN_FINISHED 0x01 /* txn is finished or never began */
-#define MDBX_TXN_ERROR 0x02 /* txn is unusable after an error */
-#define MDBX_TXN_DIRTY 0x04 /* must write, even if dirty list is empty */
-#define MDBX_TXN_SPILLS 0x08 /* txn or a parent has spilled pages */
-#define MDBX_TXN_HAS_CHILD 0x10 /* txn has an MDBX_txn.mt_child */
- /* most operations on the txn are currently illegal */
-#define MDBX_TXN_BLOCKED \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_HAS_CHILD)
-
-#define TXN_FLAGS \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_DIRTY | MDBX_TXN_SPILLS | \
- MDBX_TXN_HAS_CHILD)
-
-#if (TXN_FLAGS & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS)) || \
- ((MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS | TXN_FLAGS) & \
- MDBX_SHRINK_ALLOWED)
-#error "Oops, some flags overlapped or wrong"
-#endif
- uint32_t mt_flags;
-
- MDBX_txn *mt_parent; /* parent of a nested txn */
- /* Nested txn under this txn, set together with flag MDBX_TXN_HAS_CHILD */
- MDBX_txn *mt_child;
- MDBX_geo mt_geo;
- /* next unallocated page */
-#define mt_next_pgno mt_geo.next
- /* corresponding to the current size of datafile */
-#define mt_end_pgno mt_geo.now
-
- /* The ID of this transaction. IDs are integers incrementing from 1.
- * Only committed write transactions increment the ID. If a transaction
- * aborts, the ID may be re-used by the next writer. */
- txnid_t mt_txnid;
- txnid_t mt_front;
-
- MDBX_env *mt_env; /* the DB environment */
- /* Array of records for each DB known in the environment. */
- MDBX_dbx *mt_dbxs;
- /* Array of MDBX_db records for each known DB */
- MDBX_db *mt_dbs;
- /* Array of sequence numbers for each DB handle */
- unsigned *mt_dbiseqs;
-
- /* Transaction DBI Flags */
-#define DBI_DIRTY MDBX_DBI_DIRTY /* DB was written in this txn */
-#define DBI_STALE MDBX_DBI_STALE /* Named-DB record is older than txnID */
-#define DBI_FRESH MDBX_DBI_FRESH /* Named-DB handle opened in this txn */
-#define DBI_CREAT MDBX_DBI_CREAT /* Named-DB handle created in this txn */
-#define DBI_VALID 0x10 /* DB handle is valid, see also DB_VALID */
-#define DBI_USRVALID 0x20 /* As DB_VALID, but not set for FREE_DBI */
-#define DBI_AUDITED 0x40 /* Internal flag for accounting during audit */
- /* Array of flags for each DB */
- uint8_t *mt_dbistate;
- /* Number of DB records in use, or 0 when the txn is finished.
- * This number only ever increments until the txn finishes; we
- * don't decrement it when individual DB handles are closed. */
- MDBX_dbi mt_numdbs;
- size_t mt_owner; /* thread ID that owns this transaction */
- MDBX_canary mt_canary;
- void *mt_userctx; /* User-settable context */
- MDBX_cursor **mt_cursors;
-
- union {
- struct {
- /* For read txns: This thread/txn's reader table slot, or NULL. */
- MDBX_reader *reader;
- } to;
- struct {
- /* In write txns, array of cursors for each DB */
- pgno_t *reclaimed_pglist; /* Reclaimed GC pages */
- txnid_t last_reclaimed; /* ID of last used record */
-#if MDBX_ENABLE_REFUND
- pgno_t loose_refund_wl /* FIXME: describe */;
-#endif /* MDBX_ENABLE_REFUND */
- /* dirtylist room: Dirty array size - dirty pages visible to this txn.
- * Includes ancestor txns' dirty pages not hidden by other txns'
- * dirty/spilled pages. Thus commit(nested txn) has room to merge
- * dirtylist into mt_parent after freeing hidden mt_parent pages. */
- unsigned dirtyroom;
- /* a sequence to spilling dirty page with LRU policy */
- unsigned dirtylru;
- /* For write txns: Modified pages. Sorted when not MDBX_WRITEMAP. */
- MDBX_dpl *dirtylist;
- /* The list of reclaimed txns from GC */
- MDBX_TXL lifo_reclaimed;
- /* The list of pages that became unused during this transaction. */
- MDBX_PNL retired_pages;
- /* The list of loose pages that became unused and may be reused
- * in this transaction, linked through `mp_next`. */
- MDBX_page *loose_pages;
- /* Number of loose pages (tw.loose_pages) */
- unsigned loose_count;
- /* The sorted list of dirty pages we temporarily wrote to disk
- * because the dirty list was full. page numbers in here are
- * shifted left by 1, deleted slots have the LSB set. */
- MDBX_PNL spill_pages;
- unsigned spill_least_removed;
- } tw;
- };
-};
-
-#if MDBX_WORDBITS >= 64
-#define CURSOR_STACK 32
-#else
-#define CURSOR_STACK 24
-#endif
-
-struct MDBX_xcursor;
-
-/* Cursors are used for all DB operations.
- * A cursor holds a path of (page pointer, key index) from the DB
- * root to a position in the DB, plus other state. MDBX_DUPSORT
- * cursors include an xcursor to the current data item. Write txns
- * track their cursors and keep them up to date when data moves.
- * Exception: An xcursor's pointer to a P_SUBP page can be stale.
- * (A node with F_DUPDATA but no F_SUBDATA contains a subpage). */
-struct MDBX_cursor {
-#define MDBX_MC_LIVE UINT32_C(0xFE05D5B1)
-#define MDBX_MC_READY4CLOSE UINT32_C(0x2817A047)
-#define MDBX_MC_WAIT4EOT UINT32_C(0x90E297A7)
- uint32_t mc_signature;
- /* The database handle this cursor operates on */
- MDBX_dbi mc_dbi;
- /* Next cursor on this DB in this txn */
- MDBX_cursor *mc_next;
- /* Backup of the original cursor if this cursor is a shadow */
- MDBX_cursor *mc_backup;
- /* Context used for databases with MDBX_DUPSORT, otherwise NULL */
- struct MDBX_xcursor *mc_xcursor;
- /* The transaction that owns this cursor */
- MDBX_txn *mc_txn;
- /* The database record for this cursor */
- MDBX_db *mc_db;
- /* The database auxiliary record for this cursor */
- MDBX_dbx *mc_dbx;
- /* The mt_dbistate for this database */
- uint8_t *mc_dbistate;
- unsigned mc_snum; /* number of pushed pages */
- unsigned mc_top; /* index of top page, normally mc_snum-1 */
-
- /* Cursor state flags. */
-#define C_INITIALIZED 0x01 /* cursor has been initialized and is valid */
-#define C_EOF 0x02 /* No more data */
-#define C_SUB 0x04 /* Cursor is a sub-cursor */
-#define C_DEL 0x08 /* last op was a cursor_del */
-#define C_UNTRACK 0x10 /* Un-track cursor when closing */
-#define C_RECLAIMING 0x20 /* GC lookup is prohibited */
-#define C_GCFREEZE 0x40 /* reclaimed_pglist must not be updated */
-
- /* Cursor checking flags. */
-#define C_COPYING 0x100 /* skip key-value length check (copying simplify) */
-#define C_UPDATING 0x200 /* update/rebalance pending */
-#define C_RETIRING 0x400 /* refs to child pages may be invalid */
-#define C_SKIPORD 0x800 /* don't check keys ordering */
-
- unsigned mc_flags; /* see mdbx_cursor */
- MDBX_page *mc_pg[CURSOR_STACK]; /* stack of pushed pages */
- indx_t mc_ki[CURSOR_STACK]; /* stack of page indices */
-};
-
-/* Context for sorted-dup records.
- * We could have gone to a fully recursive design, with arbitrarily
- * deep nesting of sub-databases. But for now we only handle these
- * levels - main DB, optional sub-DB, sorted-duplicate DB. */
-typedef struct MDBX_xcursor {
- /* A sub-cursor for traversing the Dup DB */
- MDBX_cursor mx_cursor;
- /* The database record for this Dup DB */
- MDBX_db mx_db;
- /* The auxiliary DB record for this Dup DB */
- MDBX_dbx mx_dbx;
-} MDBX_xcursor;
-
-typedef struct MDBX_cursor_couple {
- MDBX_cursor outer;
- void *mc_userctx; /* User-settable context */
- MDBX_xcursor inner;
-} MDBX_cursor_couple;
-
-/* The database environment. */
-struct MDBX_env {
- /* ----------------------------------------------------- mostly static part */
-#define MDBX_ME_SIGNATURE UINT32_C(0x9A899641)
- MDBX_atomic_uint32_t me_signature;
- /* Failed to update the meta page. Probably an I/O error. */
-#define MDBX_FATAL_ERROR UINT32_C(0x80000000)
- /* Some fields are initialized. */
-#define MDBX_ENV_ACTIVE UINT32_C(0x20000000)
- /* me_txkey is set */
-#define MDBX_ENV_TXKEY UINT32_C(0x10000000)
- /* Legacy MDBX_MAPASYNC (prior v0.9) */
-#define MDBX_DEPRECATED_MAPASYNC UINT32_C(0x100000)
-#define ENV_INTERNAL_FLAGS (MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE | MDBX_ENV_TXKEY)
- uint32_t me_flags;
- mdbx_mmap_t me_dxb_mmap; /* The main data file */
-#define me_map me_dxb_mmap.dxb
-#define me_lazy_fd me_dxb_mmap.fd
- mdbx_filehandle_t me_dsync_fd;
- mdbx_mmap_t me_lck_mmap; /* The lock file */
-#define me_lfd me_lck_mmap.fd
- struct MDBX_lockinfo *me_lck;
-
- unsigned me_psize; /* DB page size, initialized from me_os_psize */
- unsigned me_leaf_nodemax; /* max size of a leaf-node */
- uint8_t me_psize2log; /* log2 of DB page size */
- int8_t me_stuck_meta; /* recovery-only: target meta page or less that zero */
- uint16_t me_merge_threshold,
- me_merge_threshold_gc; /* pages emptier than this are candidates for
- merging */
- unsigned me_os_psize; /* OS page size, from mdbx_syspagesize() */
- unsigned me_maxreaders; /* size of the reader table */
- MDBX_dbi me_maxdbs; /* size of the DB table */
- uint32_t me_pid; /* process ID of this env */
- mdbx_thread_key_t me_txkey; /* thread-key for readers */
- char *me_pathname; /* path to the DB files */
- void *me_pbuf; /* scratch area for DUPSORT put() */
- MDBX_txn *me_txn0; /* preallocated write transaction */
-
- MDBX_dbx *me_dbxs; /* array of static DB info */
- uint16_t *me_dbflags; /* array of flags from MDBX_db.md_flags */
- unsigned *me_dbiseqs; /* array of dbi sequence numbers */
- unsigned
- me_maxgc_ov1page; /* Number of pgno_t fit in a single overflow page */
- uint32_t me_live_reader; /* have liveness lock in reader table */
- void *me_userctx; /* User-settable context */
- MDBX_hsr_func *me_hsr_callback; /* Callback for kicking laggard readers */
-
- struct {
- unsigned dp_reserve_limit;
- unsigned rp_augment_limit;
- unsigned dp_limit;
- unsigned dp_initial;
- uint8_t dp_loose_limit;
- uint8_t spill_max_denominator;
- uint8_t spill_min_denominator;
- uint8_t spill_parent4child_denominator;
- unsigned merge_threshold_16dot16_percent;
- union {
- unsigned all;
- /* tracks options with non-auto values but tuned by user */
- struct {
- unsigned dp_limit : 1;
- } non_auto;
- } flags;
- } me_options;
-
- /* struct me_dbgeo used for accepting db-geo params from user for the new
- * database creation, i.e. when mdbx_env_set_geometry() was called before
- * mdbx_env_open(). */
- struct {
- size_t lower; /* minimal size of datafile */
- size_t upper; /* maximal size of datafile */
- size_t now; /* current size of datafile */
- size_t grow; /* step to grow datafile */
- size_t shrink; /* threshold to shrink datafile */
- } me_dbgeo;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- union {
- key_t key;
- int semid;
- } me_sysv_ipc;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
- MDBX_env *me_lcklist_next;
-
- /* --------------------------------------------------- mostly volatile part */
-
- MDBX_txn *me_txn; /* current write transaction */
- mdbx_fastmutex_t me_dbi_lock;
- MDBX_dbi me_numdbs; /* number of DBs opened */
-
- MDBX_page *me_dp_reserve; /* list of malloc'ed blocks for re-use */
- unsigned me_dp_reserve_len;
- /* PNL of pages that became unused in a write txn */
- MDBX_PNL me_retired_pages;
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_srwlock me_remap_guard;
- /* Workaround for LockFileEx and WriteFile multithread bug */
- CRITICAL_SECTION me_windowsbug_lock;
-#else
- mdbx_fastmutex_t me_remap_guard;
-#endif
-
- /* -------------------------------------------------------------- debugging */
-
-#if MDBX_DEBUG
- MDBX_assert_func *me_assert_func; /* Callback for assertion failures */
-#endif
-#ifdef MDBX_USE_VALGRIND
- int me_valgrind_handle;
-#endif
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- pgno_t me_poison_edge;
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-#ifndef xMDBX_DEBUG_SPILLING
-#define xMDBX_DEBUG_SPILLING 0
-#endif
-#if xMDBX_DEBUG_SPILLING == 2
- unsigned debug_dirtied_est, debug_dirtied_act;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- /* ------------------------------------------------- stub for lck-less mode */
- MDBX_atomic_uint64_t
- x_lckless_stub[(sizeof(MDBX_lockinfo) + MDBX_CACHELINE_SIZE - 1) /
- sizeof(MDBX_atomic_uint64_t)];
-};
-
-#ifndef __cplusplus
-/*----------------------------------------------------------------------------*/
-/* Debug and Logging stuff */
-
-#define MDBX_RUNTIME_FLAGS_INIT \
- ((MDBX_DEBUG) > 0) * MDBX_DBG_ASSERT + ((MDBX_DEBUG) > 1) * MDBX_DBG_AUDIT
-
-extern uint8_t mdbx_runtime_flags;
-extern uint8_t mdbx_loglevel;
-extern MDBX_debug_func *mdbx_debug_logger;
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny) {
-#if MDBX_DEBUG
- if (MDBX_DBG_JITTER & mdbx_runtime_flags)
- mdbx_osal_jitter(tiny);
-#else
- (void)tiny;
-#endif
-}
-
-MDBX_INTERNAL_FUNC void MDBX_PRINTF_ARGS(4, 5)
- mdbx_debug_log(int level, const char *function, int line, const char *fmt,
- ...) MDBX_PRINTF_ARGS(4, 5);
-MDBX_INTERNAL_FUNC void mdbx_debug_log_va(int level, const char *function,
- int line, const char *fmt,
- va_list args);
-
-#if MDBX_DEBUG
-#define mdbx_log_enabled(msg) unlikely(msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() unlikely((mdbx_runtime_flags & MDBX_DBG_AUDIT))
-#else /* MDBX_DEBUG */
-#define mdbx_log_enabled(msg) (msg < MDBX_LOG_VERBOSE && msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() (0)
-#endif /* MDBX_DEBUG */
-
-#if MDBX_FORCE_ASSERTIONS
-#define mdbx_assert_enabled() (1)
-#elif MDBX_DEBUG
-#define mdbx_assert_enabled() likely((mdbx_runtime_flags & MDBX_DBG_ASSERT))
-#else
-#define mdbx_assert_enabled() (0)
-#endif /* assertions */
-
-#if !MDBX_DEBUG && defined(__ANDROID_API__)
-#define mdbx_assert_fail(env, msg, func, line) \
- __android_log_assert(msg, "mdbx", "%s:%u", func, line)
-#else
-void mdbx_assert_fail(const MDBX_env *env, const char *msg, const char *func,
- int line);
-#endif
-
-#define mdbx_debug_extra(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, __func__, __LINE__, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug_extra_print(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, NULL, 0, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_trace(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_TRACE)) \
- mdbx_debug_log(MDBX_LOG_TRACE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_DEBUG)) \
- mdbx_debug_log(MDBX_LOG_DEBUG, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_verbose(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_VERBOSE)) \
- mdbx_debug_log(MDBX_LOG_VERBOSE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_notice(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_NOTICE)) \
- mdbx_debug_log(MDBX_LOG_NOTICE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_warning(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_WARN)) \
- mdbx_debug_log(MDBX_LOG_WARN, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_error(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) \
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_fatal(fmt, ...) \
- mdbx_debug_log(MDBX_LOG_FATAL, __func__, __LINE__, fmt "\n", __VA_ARGS__);
-
-#define mdbx_ensure_msg(env, expr, msg) \
- do { \
- if (unlikely(!(expr))) \
- mdbx_assert_fail(env, msg, __func__, __LINE__); \
- } while (0)
-
-#define mdbx_ensure(env, expr) mdbx_ensure_msg(env, expr, #expr)
-
-/* assert(3) variant in environment context */
-#define mdbx_assert(env, expr) \
- do { \
- if (mdbx_assert_enabled()) \
- mdbx_ensure(env, expr); \
- } while (0)
-
-/* assert(3) variant in cursor context */
-#define mdbx_cassert(mc, expr) mdbx_assert((mc)->mc_txn->mt_env, expr)
-
-/* assert(3) variant in transaction context */
-#define mdbx_tassert(txn, expr) mdbx_assert((txn)->mt_env, expr)
-
-#ifndef xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-#undef assert
-#define assert(expr) mdbx_assert(NULL, expr)
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Cache coherence and mmap invalidation */
-
-#if MDBX_CPU_WRITEBACK_INCOHERENT
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_memory_barrier()
-#else
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_compiler_barrier()
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-MDBX_MAYBE_UNUSED static __inline void
-mdbx_flush_incoherent_mmap(void *addr, size_t nbytes, const intptr_t pagesize) {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- char *const begin = (char *)(-pagesize & (intptr_t)addr);
- char *const end =
- (char *)(-pagesize & (intptr_t)((char *)addr + nbytes + pagesize - 1));
- int err = msync(begin, end - begin, MS_SYNC | MS_INVALIDATE) ? errno : 0;
- mdbx_assert(nullptr, err == 0);
- (void)err;
-#else
- (void)pagesize;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#if MDBX_MMAP_INCOHERENT_CPU_CACHE
-#ifdef DCACHE
- /* MIPS has cache coherency issues.
- * Note: for any nbytes >= on-chip cache size, entire is flushed. */
- cacheflush(addr, nbytes, DCACHE);
-#else
-#error "Oops, cacheflush() not available"
-#endif /* DCACHE */
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#if !MDBX_MMAP_INCOHERENT_FILE_WRITE && !MDBX_MMAP_INCOHERENT_CPU_CACHE
- (void)addr;
- (void)nbytes;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* Internal prototypes */
-
-MDBX_INTERNAL_FUNC int mdbx_cleanup_dead_readers(MDBX_env *env, int rlocked,
- int *dead);
-MDBX_INTERNAL_FUNC int mdbx_rthc_alloc(mdbx_thread_key_t *key,
- MDBX_reader *begin, MDBX_reader *end);
-MDBX_INTERNAL_FUNC void mdbx_rthc_remove(const mdbx_thread_key_t key);
-
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_init(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_dtor(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_thread_dtor(void *ptr);
-
-#endif /* !__cplusplus */
-
-#define MDBX_IS_ERROR(rc) \
- ((rc) != MDBX_RESULT_TRUE && (rc) != MDBX_RESULT_FALSE)
-
-/* Internal error codes, not exposed outside libmdbx */
-#define MDBX_NO_ROOT (MDBX_LAST_ADDED_ERRCODE + 10)
-
-/* Debugging output value of a cursor DBI: Negative in a sub-cursor. */
-#define DDBI(mc) \
- (((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
-
-/* Key size which fits in a DKBUF (debug key buffer). */
-#define DKBUF_MAX 511
-#define DKBUF char _kbuf[DKBUF_MAX * 4 + 2]
-#define DKEY(x) mdbx_dump_val(x, _kbuf, DKBUF_MAX * 2 + 1)
-#define DVAL(x) mdbx_dump_val(x, _kbuf + DKBUF_MAX * 2 + 1, DKBUF_MAX * 2 + 1)
-
-#if MDBX_DEBUG
-#define DKBUF_DEBUG DKBUF
-#define DKEY_DEBUG(x) DKEY(x)
-#define DVAL_DEBUG(x) DVAL(x)
-#else
-#define DKBUF_DEBUG ((void)(0))
-#define DKEY_DEBUG(x) ("-")
-#define DVAL_DEBUG(x) ("-")
-#endif
-
-/* An invalid page number.
- * Mainly used to denote an empty tree. */
-#define P_INVALID (~(pgno_t)0)
-
-/* Test if the flags f are set in a flag word w. */
-#define F_ISSET(w, f) (((w) & (f)) == (f))
-
-/* Round n up to an even number. */
-#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
-
-/* Default size of memory map.
- * This is certainly too small for any actual applications. Apps should
- * always set the size explicitly using mdbx_env_set_geometry(). */
-#define DEFAULT_MAPSIZE MEGABYTE
-
-/* Number of slots in the reader table.
- * This value was chosen somewhat arbitrarily. The 61 is a prime number,
- * and such readers plus a couple mutexes fit into single 4KB page.
- * Applications should set the table size using mdbx_env_set_maxreaders(). */
-#define DEFAULT_READERS 61
-
-/* Test if a page is a leaf page */
-#define IS_LEAF(p) (((p)->mp_flags & P_LEAF) != 0)
-/* Test if a page is a LEAF2 page */
-#define IS_LEAF2(p) unlikely(((p)->mp_flags & P_LEAF2) != 0)
-/* Test if a page is a branch page */
-#define IS_BRANCH(p) (((p)->mp_flags & P_BRANCH) != 0)
-/* Test if a page is an overflow page */
-#define IS_OVERFLOW(p) unlikely(((p)->mp_flags & P_OVERFLOW) != 0)
-/* Test if a page is a sub page */
-#define IS_SUBP(p) (((p)->mp_flags & P_SUBP) != 0)
-
-#define PAGETYPE(p) ((p)->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW))
-
-/* Header for a single key/data pair within a page.
- * Used in pages of type P_BRANCH and P_LEAF without P_LEAF2.
- * We guarantee 2-byte alignment for 'MDBX_node's.
- *
- * Leaf node flags describe node contents. F_BIGDATA says the node's
- * data part is the page number of an overflow page with actual data.
- * F_DUPDATA and F_SUBDATA can be combined giving duplicate data in
- * a sub-page/sub-database, and named databases (just F_SUBDATA). */
-typedef struct MDBX_node {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- union {
- uint32_t mn_dsize;
- uint32_t mn_pgno32;
- };
- uint8_t mn_flags; /* see mdbx_node flags */
- uint8_t mn_extra;
- uint16_t mn_ksize; /* key size */
-#else
- uint16_t mn_ksize; /* key size */
- uint8_t mn_extra;
- uint8_t mn_flags; /* see mdbx_node flags */
- union {
- uint32_t mn_pgno32;
- uint32_t mn_dsize;
- };
-#endif /* __BYTE_ORDER__ */
-
- /* mdbx_node Flags */
-#define F_BIGDATA 0x01 /* data put on overflow page */
-#define F_SUBDATA 0x02 /* data is a sub-database */
-#define F_DUPDATA 0x04 /* data has duplicates */
-
- /* valid flags for mdbx_node_add() */
-#define NODE_ADD_FLAGS (F_DUPDATA | F_SUBDATA | MDBX_RESERVE | MDBX_APPEND)
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- uint8_t mn_data[] /* key and data are appended here */;
-#endif /* C99 */
-} MDBX_node;
-
-#define DB_PERSISTENT_FLAGS \
- (MDBX_REVERSEKEY | MDBX_DUPSORT | MDBX_INTEGERKEY | MDBX_DUPFIXED | \
- MDBX_INTEGERDUP | MDBX_REVERSEDUP)
-
-/* mdbx_dbi_open() flags */
-#define DB_USABLE_FLAGS (DB_PERSISTENT_FLAGS | MDBX_CREATE | MDBX_DB_ACCEDE)
-
-#define DB_VALID 0x8000 /* DB handle is valid, for me_dbflags */
-#define DB_INTERNAL_FLAGS DB_VALID
-
-#if DB_INTERNAL_FLAGS & DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-#if DB_PERSISTENT_FLAGS & ~DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-
-/* max number of pages to commit in one writev() call */
-#define MDBX_COMMIT_PAGES 64
-#if defined(IOV_MAX) && IOV_MAX < MDBX_COMMIT_PAGES /* sysconf(_SC_IOV_MAX) */
-#undef MDBX_COMMIT_PAGES
-#define MDBX_COMMIT_PAGES IOV_MAX
-#endif
-
-/*
- * /
- * | -1, a < b
- * CMP2INT(a,b) = < 0, a == b
- * | 1, a > b
- * \
- */
-#ifndef __e2k__
-/* LY: fast enough on most systems */
-#define CMP2INT(a, b) (((b) > (a)) ? -1 : (a) > (b))
-#else
-/* LY: more parallelable on VLIW Elbrus */
-#define CMP2INT(a, b) (((a) > (b)) - ((b) > (a)))
-#endif
-
-/* Do not spill pages to disk if txn is getting full, may fail instead */
-#define MDBX_NOSPILL 0x8000
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-int64pgno(int64_t i64) {
- if (likely(i64 >= (int64_t)MIN_PAGENO && i64 <= (int64_t)MAX_PAGENO + 1))
- return (pgno_t)i64;
- return (i64 < (int64_t)MIN_PAGENO) ? MIN_PAGENO : MAX_PAGENO;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_add(size_t base, size_t augend) {
- assert(base <= MAX_PAGENO + 1 && augend < MAX_PAGENO);
- return int64pgno(base + augend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_sub(size_t base, size_t subtrahend) {
- assert(base >= MIN_PAGENO && base <= MAX_PAGENO + 1 &&
- subtrahend < MAX_PAGENO);
- return int64pgno(base - subtrahend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline bool
-is_powerof2(size_t x) {
- return (x & (x - 1)) == 0;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-floor_powerof2(size_t value, size_t granularity) {
- assert(is_powerof2(granularity));
- return value & ~(granularity - 1);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-ceil_powerof2(size_t value, size_t granularity) {
- return floor_powerof2(value + granularity - 1, granularity);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static unsigned
-log2n_powerof2(size_t value) {
- assert(value > 0 && value < INT32_MAX && is_powerof2(value));
- assert((value & -(int32_t)value) == value);
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_ctzl)
- return __builtin_ctzl(value);
-#elif defined(_MSC_VER)
- unsigned long index;
- _BitScanForward(&index, (unsigned long)value);
- return index;
-#else
- static const uint8_t debruijn_ctz32[32] = {
- 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
- 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9};
- return debruijn_ctz32[(uint32_t)(value * 0x077CB531u) >> 27];
-#endif
-}
-
-/* Only a subset of the mdbx_env flags can be changed
- * at runtime. Changing other flags requires closing the
- * environment and re-opening it with the new flags. */
-#define ENV_CHANGEABLE_FLAGS \
- (MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC | MDBX_DEPRECATED_MAPASYNC | \
- MDBX_NOMEMINIT | MDBX_COALESCE | MDBX_PAGEPERTURB | MDBX_ACCEDE)
-#define ENV_CHANGELESS_FLAGS \
- (MDBX_NOSUBDIR | MDBX_RDONLY | MDBX_WRITEMAP | MDBX_NOTLS | MDBX_NORDAHEAD | \
- MDBX_LIFORECLAIM | MDBX_EXCLUSIVE)
-#define ENV_USABLE_FLAGS (ENV_CHANGEABLE_FLAGS | ENV_CHANGELESS_FLAGS)
-
-#if !defined(__cplusplus) || CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_MAYBE_UNUSED static void static_checks(void) {
- STATIC_ASSERT_MSG(INT16_MAX - CORE_DBS == MDBX_MAX_DBI,
- "Oops, MDBX_MAX_DBI or CORE_DBS?");
- STATIC_ASSERT_MSG((unsigned)(MDBX_DB_ACCEDE | MDBX_CREATE) ==
- ((DB_USABLE_FLAGS | DB_INTERNAL_FLAGS) &
- (ENV_USABLE_FLAGS | ENV_INTERNAL_FLAGS)),
- "Oops, some flags overlapped or wrong");
- STATIC_ASSERT_MSG((ENV_INTERNAL_FLAGS & ENV_USABLE_FLAGS) == 0,
- "Oops, some flags overlapped or wrong");
-}
-#endif /* Disabled for MSVC 19.0 (VisualStudio 2015) */
-
-#ifdef __cplusplus
-}
-#endif
-
-#define MDBX_ASAN_POISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("POISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_POISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#define MDBX_ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("UNPOISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_UNPOISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-typedef struct flagbit {
- int bit;
- const char *name;
-} flagbit;
-
-const flagbit dbflags[] = {{MDBX_DUPSORT, "dupsort"},
- {MDBX_INTEGERKEY, "integerkey"},
- {MDBX_REVERSEKEY, "reversekey"},
- {MDBX_DUPFIXED, "dupfixed"},
- {MDBX_REVERSEDUP, "reversedup"},
- {MDBX_INTEGERDUP, "integerdup"},
- {0, nullptr}};
-
-#if defined(_WIN32) || defined(_WIN64)
-/*
- * POSIX getopt for Windows
- *
- * AT&T Public License
- *
- * Code given out at the 1985 UNIFORUM conference in Dallas.
- */
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif
-#endif /* _MSC_VER (warnings) */
-
-#include <stdio.h>
-#include <string.h>
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-/*----------------------------------------------------------------------------*/
-
-#ifndef NULL
-#define NULL 0
-#endif
-
-#ifndef EOF
-#define EOF (-1)
-#endif
-
-int optind = 1;
-int optopt;
-char *optarg;
-
-int getopt(int argc, char *const argv[], const char *opts) {
- static int sp = 1;
- int c;
- const char *cp;
-
- if (sp == 1) {
- if (optind >= argc || argv[optind][0] != '-' || argv[optind][1] == '\0')
- return EOF;
- else if (strcmp(argv[optind], "--") == 0) {
- optind++;
- return EOF;
- }
- }
- optopt = c = argv[optind][sp];
- if (c == ':' || (cp = strchr(opts, c)) == NULL) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "illegal option", c);
- if (argv[optind][++sp] == '\0') {
- optind++;
- sp = 1;
- }
- return '?';
- }
- if (*++cp == ':') {
- if (argv[optind][sp + 1] != '\0')
- optarg = &argv[optind++][sp + 1];
- else if (++optind >= argc) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "option requires an argument",
- c);
- sp = 1;
- return '?';
- } else
- optarg = argv[optind++];
- sp = 1;
- } else {
- if (argv[optind][++sp] == '\0') {
- sp = 1;
- optind++;
- }
- optarg = NULL;
- }
- return c;
-}
-
-static volatile BOOL user_break;
-static BOOL WINAPI ConsoleBreakHandlerRoutine(DWORD dwCtrlType) {
- (void)dwCtrlType;
- user_break = 1;
- return true;
-}
-
-static uint64_t GetMilliseconds(void) {
- LARGE_INTEGER Counter, Frequency;
- return (QueryPerformanceFrequency(&Frequency) &&
- QueryPerformanceCounter(&Counter))
- ? Counter.QuadPart * 1000ul / Frequency.QuadPart
- : 0;
-}
-
-#else /* WINDOWS */
-
-static volatile sig_atomic_t user_break;
-static void signal_handler(int sig) {
- (void)sig;
- user_break = 1;
-}
-
-#endif /* !WINDOWS */
-
-#define EXIT_INTERRUPTED (EXIT_FAILURE + 4)
-#define EXIT_FAILURE_SYS (EXIT_FAILURE + 3)
-#define EXIT_FAILURE_MDBX (EXIT_FAILURE + 2)
-#define EXIT_FAILURE_CHECK_MAJOR (EXIT_FAILURE + 1)
-#define EXIT_FAILURE_CHECK_MINOR EXIT_FAILURE
-
-typedef struct {
- const char *name;
- struct {
- uint64_t branch, large_count, large_volume, leaf;
- uint64_t subleaf_dupsort, leaf_dupfixed, subleaf_dupfixed;
- uint64_t total, empty, other;
- } pages;
- uint64_t payload_bytes;
- uint64_t lost_bytes;
-} walk_dbi_t;
-
-struct {
- short *pagemap;
- uint64_t total_payload_bytes;
- uint64_t pgcount;
- walk_dbi_t
- dbi[MDBX_MAX_DBI + CORE_DBS + /* account pseudo-entry for meta */ 1];
-} walk;
-
-#define dbi_free walk.dbi[FREE_DBI]
-#define dbi_main walk.dbi[MAIN_DBI]
-#define dbi_meta walk.dbi[CORE_DBS]
-
-int envflags = MDBX_RDONLY | MDBX_EXCLUSIVE;
-MDBX_env *env;
-MDBX_txn *txn;
-MDBX_envinfo envinfo;
-size_t userdb_count, skipped_subdb;
-uint64_t total_unused_bytes, reclaimable_pages, gc_pages, alloc_pages,
- unused_pages, backed_pages;
-unsigned verbose;
-bool ignore_wrong_order, quiet, dont_traversal;
-const char *only_subdb;
-int stuck_meta = -1;
-
-struct problem {
- struct problem *pr_next;
- size_t count;
- const char *caption;
-};
-
-struct problem *problems_list;
-unsigned total_problems, data_tree_problems, gc_tree_problems;
-
-static void MDBX_PRINTF_ARGS(1, 2) print(const char *msg, ...) {
- if (!quiet) {
- va_list args;
-
- fflush(stderr);
- va_start(args, msg);
- vfprintf(stdout, msg, args);
- va_end(args);
- }
-}
-
-static void va_log(MDBX_log_level_t level, const char *msg, va_list args) {
- static const char *const prefixes[] = {
- "!!!fatal: ", " ! " /* error */, " ~ " /* warning */,
- " " /* notice */, " // " /* verbose */, " //// " /* debug */,
- " ////// " /* trace */
- };
-
- FILE *out = stdout;
- if (level <= MDBX_LOG_ERROR) {
- total_problems++;
- out = stderr;
- }
-
- if (!quiet && verbose + 1 >= (unsigned)level &&
- (unsigned)level < ARRAY_LENGTH(prefixes)) {
- fflush(nullptr);
- fputs(prefixes[level], out);
- vfprintf(out, msg, args);
- if (msg[strlen(msg) - 1] != '\n')
- fputc('\n', out);
- fflush(nullptr);
- }
-
- if (level == MDBX_LOG_FATAL) {
- exit(EXIT_FAILURE_MDBX);
- abort();
- }
-}
-
-static void MDBX_PRINTF_ARGS(1, 2) error(const char *msg, ...) {
- va_list args;
- va_start(args, msg);
- va_log(MDBX_LOG_ERROR, msg, args);
- va_end(args);
-}
-
-static void logger(MDBX_log_level_t level, const char *function, int line,
- const char *msg, va_list args) {
- (void)line;
- (void)function;
- if (level < MDBX_LOG_EXTRA)
- va_log(level, msg, args);
-}
-
-static int check_user_break(void) {
- switch (user_break) {
- case 0:
- return MDBX_SUCCESS;
- case 1:
- print(" - interrupted by signal\n");
- fflush(nullptr);
- user_break = 2;
- }
- return MDBX_EINTR;
-}
-
-static void pagemap_cleanup(void) {
- for (size_t i = CORE_DBS + /* account pseudo-entry for meta */ 1;
- i < ARRAY_LENGTH(walk.dbi); ++i) {
- if (walk.dbi[i].name) {
- mdbx_free((void *)walk.dbi[i].name);
- walk.dbi[i].name = nullptr;
- }
- }
-
- mdbx_free(walk.pagemap);
- walk.pagemap = nullptr;
-}
-
-static walk_dbi_t *pagemap_lookup_dbi(const char *dbi_name, bool silent) {
- static walk_dbi_t *last;
-
- if (dbi_name == MDBX_PGWALK_MAIN)
- return &dbi_main;
- if (dbi_name == MDBX_PGWALK_GC)
- return &dbi_free;
- if (dbi_name == MDBX_PGWALK_META)
- return &dbi_meta;
-
- if (last && strcmp(last->name, dbi_name) == 0)
- return last;
-
- walk_dbi_t *dbi = walk.dbi + CORE_DBS + /* account pseudo-entry for meta */ 1;
- for (; dbi < ARRAY_END(walk.dbi) && dbi->name; ++dbi) {
- if (strcmp(dbi->name, dbi_name) == 0)
- return last = dbi;
- }
-
- if (verbose > 0 && !silent) {
- print(" - found '%s' area\n", dbi_name);
- fflush(nullptr);
- }
-
- if (dbi == ARRAY_END(walk.dbi))
- return nullptr;
-
- dbi->name = mdbx_strdup(dbi_name);
- return last = dbi;
-}
-
-static void MDBX_PRINTF_ARGS(4, 5)
- problem_add(const char *object, uint64_t entry_number, const char *msg,
- const char *extra, ...) {
- total_problems++;
-
- if (!quiet) {
- int need_fflush = 0;
- struct problem *p;
-
- for (p = problems_list; p; p = p->pr_next)
- if (p->caption == msg)
- break;
-
- if (!p) {
- p = mdbx_calloc(1, sizeof(*p));
- if (unlikely(!p))
- return;
- p->caption = msg;
- p->pr_next = problems_list;
- problems_list = p;
- need_fflush = 1;
- }
-
- p->count++;
- if (verbose > 1) {
- print(" %s #%" PRIu64 ": %s", object, entry_number, msg);
- if (extra) {
- va_list args;
- printf(" (");
- va_start(args, extra);
- vfprintf(stdout, extra, args);
- va_end(args);
- printf(")");
- }
- printf("\n");
- if (need_fflush)
- fflush(nullptr);
- }
- }
-}
-
-static struct problem *problems_push(void) {
- struct problem *p = problems_list;
- problems_list = nullptr;
- return p;
-}
-
-static size_t problems_pop(struct problem *list) {
- size_t count = 0;
-
- if (problems_list) {
- int i;
-
- print(" - problems: ");
- for (i = 0; problems_list; ++i) {
- struct problem *p = problems_list->pr_next;
- count += problems_list->count;
- print("%s%s (%" PRIuPTR ")", i ? ", " : "", problems_list->caption,
- problems_list->count);
- mdbx_free(problems_list);
- problems_list = p;
- }
- print("\n");
- fflush(nullptr);
- }
-
- problems_list = list;
- return count;
-}
-
-static int pgvisitor(const uint64_t pgno, const unsigned pgnumber,
- void *const ctx, const int deep,
- const char *const dbi_name_or_tag, const size_t page_size,
- const MDBX_page_type_t pagetype, const MDBX_error_t err,
- const size_t nentries, const size_t payload_bytes,
- const size_t header_bytes, const size_t unused_bytes) {
- (void)ctx;
- const bool is_gc_tree = dbi_name_or_tag == MDBX_PGWALK_GC;
- if (deep > 42) {
- problem_add("deep", deep, "too large", nullptr);
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- return MDBX_CORRUPTED /* avoid infinite loop/recursion */;
- }
-
- walk_dbi_t *dbi = pagemap_lookup_dbi(dbi_name_or_tag, false);
- if (!dbi) {
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- return MDBX_ENOMEM;
- }
-
- const size_t page_bytes = payload_bytes + header_bytes + unused_bytes;
- walk.pgcount += pgnumber;
-
- const char *pagetype_caption;
- bool branch = false;
- switch (pagetype) {
- default:
- problem_add("page", pgno, "unknown page-type", "type %u, deep %i",
- (unsigned)pagetype, deep);
- pagetype_caption = "unknown";
- dbi->pages.other += pgnumber;
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- break;
- case MDBX_page_broken:
- pagetype_caption = "broken";
- dbi->pages.other += pgnumber;
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- break;
- case MDBX_subpage_broken:
- pagetype_caption = "broken-subpage";
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- break;
- case MDBX_page_meta:
- pagetype_caption = "meta";
- dbi->pages.other += pgnumber;
- break;
- case MDBX_page_large:
- pagetype_caption = "large";
- dbi->pages.large_volume += pgnumber;
- dbi->pages.large_count += 1;
- break;
- case MDBX_page_branch:
- pagetype_caption = "branch";
- dbi->pages.branch += pgnumber;
- branch = true;
- break;
- case MDBX_page_leaf:
- pagetype_caption = "leaf";
- dbi->pages.leaf += pgnumber;
- break;
- case MDBX_page_dupfixed_leaf:
- pagetype_caption = "leaf-dupfixed";
- dbi->pages.leaf_dupfixed += pgnumber;
- break;
- case MDBX_subpage_leaf:
- pagetype_caption = "subleaf-dupsort";
- dbi->pages.subleaf_dupsort += 1;
- break;
- case MDBX_subpage_dupfixed_leaf:
- pagetype_caption = "subleaf-dupfixed";
- dbi->pages.subleaf_dupfixed += 1;
- break;
- }
-
- if (pgnumber) {
- if (verbose > 3 && (!only_subdb || strcmp(only_subdb, dbi->name) == 0)) {
- if (pgnumber == 1)
- print(" %s-page %" PRIu64, pagetype_caption, pgno);
- else
- print(" %s-span %" PRIu64 "[%u]", pagetype_caption, pgno, pgnumber);
- print(" of %s: header %" PRIiPTR ", %s %" PRIiPTR ", payload %" PRIiPTR
- ", unused %" PRIiPTR ", deep %i\n",
- dbi->name, header_bytes,
- (pagetype == MDBX_page_branch) ? "keys" : "entries", nentries,
- payload_bytes, unused_bytes, deep);
- }
-
- bool already_used = false;
- for (unsigned n = 0; n < pgnumber; ++n) {
- uint64_t spanpgno = pgno + n;
- if (spanpgno >= alloc_pages) {
- problem_add("page", spanpgno, "wrong page-no",
- "%s-page: %" PRIu64 " > %" PRIu64 ", deep %i",
- pagetype_caption, spanpgno, alloc_pages, deep);
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- } else if (walk.pagemap[spanpgno]) {
- walk_dbi_t *coll_dbi = &walk.dbi[walk.pagemap[spanpgno] - 1];
- problem_add("page", spanpgno,
- (branch && coll_dbi == dbi) ? "loop" : "already used",
- "%s-page: by %s, deep %i", pagetype_caption, coll_dbi->name,
- deep);
- already_used = true;
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- } else {
- walk.pagemap[spanpgno] = (short)(dbi - walk.dbi + 1);
- dbi->pages.total += 1;
- }
- }
-
- if (already_used)
- return branch ? MDBX_RESULT_TRUE /* avoid infinite loop/recursion */
- : MDBX_SUCCESS;
- }
-
- if (MDBX_IS_ERROR(err)) {
- problem_add("page", pgno, "invalid/corrupted", "%s-page", pagetype_caption);
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- } else {
- if (unused_bytes > page_size) {
- problem_add("page", pgno, "illegal unused-bytes",
- "%s-page: %u < %" PRIuPTR " < %u", pagetype_caption, 0,
- unused_bytes, envinfo.mi_dxb_pagesize);
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- }
-
- if (header_bytes < (int)sizeof(long) ||
- (size_t)header_bytes >= envinfo.mi_dxb_pagesize - sizeof(long)) {
- problem_add("page", pgno, "illegal header-length",
- "%s-page: %" PRIuPTR " < %" PRIuPTR " < %" PRIuPTR,
- pagetype_caption, sizeof(long), header_bytes,
- envinfo.mi_dxb_pagesize - sizeof(long));
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- }
- if (payload_bytes < 1) {
- if (nentries > 1) {
- problem_add("page", pgno, "zero size-of-entry",
- "%s-page: payload %" PRIuPTR " bytes, %" PRIuPTR " entries",
- pagetype_caption, payload_bytes, nentries);
- /* if ((size_t)header_bytes + unused_bytes < page_size) {
- // LY: hush a misuse error
- page_bytes = page_size;
- } */
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- } else {
- problem_add("page", pgno, "empty",
- "%s-page: payload %" PRIuPTR " bytes, %" PRIuPTR
- " entries, deep %i",
- pagetype_caption, payload_bytes, nentries, deep);
- dbi->pages.empty += 1;
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- }
- }
-
- if (pgnumber) {
- if (page_bytes != page_size) {
- problem_add("page", pgno, "misused",
- "%s-page: %" PRIuPTR " != %" PRIuPTR " (%" PRIuPTR
- "h + %" PRIuPTR "p + %" PRIuPTR "u), deep %i",
- pagetype_caption, page_size, page_bytes, header_bytes,
- payload_bytes, unused_bytes, deep);
- if (page_size > page_bytes)
- dbi->lost_bytes += page_size - page_bytes;
- data_tree_problems += !is_gc_tree;
- gc_tree_problems += is_gc_tree;
- } else {
- dbi->payload_bytes += payload_bytes + header_bytes;
- walk.total_payload_bytes += payload_bytes + header_bytes;
- }
- }
- }
-
- return check_user_break();
-}
-
-typedef int(visitor)(const uint64_t record_number, const MDBX_val *key,
- const MDBX_val *data);
-static int process_db(MDBX_dbi dbi_handle, char *dbi_name, visitor *handler,
- bool silent);
-
-static int handle_userdb(const uint64_t record_number, const MDBX_val *key,
- const MDBX_val *data) {
- (void)record_number;
- (void)key;
- (void)data;
- return check_user_break();
-}
-
-static int handle_freedb(const uint64_t record_number, const MDBX_val *key,
- const MDBX_val *data) {
- char *bad = "";
- pgno_t *iptr = data->iov_base;
-
- if (key->iov_len != sizeof(txnid_t))
- problem_add("entry", record_number, "wrong txn-id size",
- "key-size %" PRIiPTR, key->iov_len);
- else {
- txnid_t txnid;
- memcpy(&txnid, key->iov_base, sizeof(txnid));
- if (txnid < 1 || txnid > envinfo.mi_recent_txnid)
- problem_add("entry", record_number, "wrong txn-id", "%" PRIaTXN, txnid);
- else {
- if (data->iov_len < sizeof(pgno_t) || data->iov_len % sizeof(pgno_t))
- problem_add("entry", txnid, "wrong idl size", "%" PRIuPTR,
- data->iov_len);
- size_t number = (data->iov_len >= sizeof(pgno_t)) ? *iptr++ : 0;
- if (number < 1 || number > MDBX_PGL_LIMIT)
- problem_add("entry", txnid, "wrong idl length", "%" PRIuPTR, number);
- else if ((number + 1) * sizeof(pgno_t) > data->iov_len) {
- problem_add("entry", txnid, "trimmed idl",
- "%" PRIuSIZE " > %" PRIuSIZE " (corruption)",
- (number + 1) * sizeof(pgno_t), data->iov_len);
- number = data->iov_len / sizeof(pgno_t) - 1;
- } else if (data->iov_len - (number + 1) * sizeof(pgno_t) >=
- /* LY: allow gap up to one page. it is ok
- * and better than shink-and-retry inside mdbx_update_gc() */
- envinfo.mi_dxb_pagesize)
- problem_add("entry", txnid, "extra idl space",
- "%" PRIuSIZE " < %" PRIuSIZE " (minor, not a trouble)",
- (number + 1) * sizeof(pgno_t), data->iov_len);
-
- gc_pages += number;
- if (envinfo.mi_latter_reader_txnid > txnid)
- reclaimable_pages += number;
-
- pgno_t prev = MDBX_PNL_ASCENDING ? NUM_METAS - 1 : txn->mt_next_pgno;
- pgno_t span = 1;
- for (unsigned i = 0; i < number; ++i) {
- if (check_user_break())
- return MDBX_EINTR;
- const pgno_t pgno = iptr[i];
- if (pgno < NUM_METAS)
- problem_add("entry", txnid, "wrong idl entry",
- "pgno %" PRIaPGNO " < meta-pages %u", pgno, NUM_METAS);
- else if (pgno >= backed_pages)
- problem_add("entry", txnid, "wrong idl entry",
- "pgno %" PRIaPGNO " > backed-pages %" PRIu64, pgno,
- backed_pages);
- else if (pgno >= alloc_pages)
- problem_add("entry", txnid, "wrong idl entry",
- "pgno %" PRIaPGNO " > alloc-pages %" PRIu64, pgno,
- alloc_pages - 1);
- else {
- if (MDBX_PNL_DISORDERED(prev, pgno)) {
- bad = " [bad sequence]";
- problem_add("entry", txnid, "bad sequence",
- "%" PRIaPGNO " %c [%u].%" PRIaPGNO, prev,
- (prev == pgno) ? '=' : (MDBX_PNL_ASCENDING ? '>' : '<'),
- i, pgno);
- }
- if (walk.pagemap) {
- int idx = walk.pagemap[pgno];
- if (idx == 0)
- walk.pagemap[pgno] = -1;
- else if (idx > 0)
- problem_add("page", pgno, "already used", "by %s",
- walk.dbi[idx - 1].name);
- else
- problem_add("page", pgno, "already listed in GC", nullptr);
- }
- }
- prev = pgno;
- while (i + span < number &&
- iptr[i + span] == (MDBX_PNL_ASCENDING ? pgno_add(pgno, span)
- : pgno_sub(pgno, span)))
- ++span;
- }
- if (verbose > 3 && !only_subdb) {
- print(" transaction %" PRIaTXN ", %" PRIuPTR
- " pages, maxspan %" PRIaPGNO "%s\n",
- txnid, number, span, bad);
- if (verbose > 4) {
- for (unsigned i = 0; i < number; i += span) {
- const pgno_t pgno = iptr[i];
- for (span = 1;
- i + span < number &&
- iptr[i + span] == (MDBX_PNL_ASCENDING ? pgno_add(pgno, span)
- : pgno_sub(pgno, span));
- ++span)
- ;
- if (span > 1) {
- print(" %9" PRIaPGNO "[%" PRIaPGNO "]\n", pgno, span);
- } else
- print(" %9" PRIaPGNO "\n", pgno);
- }
- }
- }
- }
- }
-
- return check_user_break();
-}
-
-static int equal_or_greater(const MDBX_val *a, const MDBX_val *b) {
- return (a->iov_len == b->iov_len &&
- memcmp(a->iov_base, b->iov_base, a->iov_len) == 0)
- ? 0
- : 1;
-}
-
-static int handle_maindb(const uint64_t record_number, const MDBX_val *key,
- const MDBX_val *data) {
- char *name;
- int rc;
- size_t i;
-
- name = key->iov_base;
- for (i = 0; i < key->iov_len; ++i) {
- if (name[i] < ' ')
- return handle_userdb(record_number, key, data);
- }
-
- name = mdbx_malloc(key->iov_len + 1);
- if (unlikely(!name))
- return MDBX_ENOMEM;
- memcpy(name, key->iov_base, key->iov_len);
- name[key->iov_len] = '\0';
- userdb_count++;
-
- rc = process_db(~0u, name, handle_userdb, false);
- mdbx_free(name);
- if (rc != MDBX_INCOMPATIBLE)
- return rc;
-
- return handle_userdb(record_number, key, data);
-}
-
-static const char *db_flags2keymode(unsigned flags) {
- flags &= (MDBX_REVERSEKEY | MDBX_INTEGERKEY);
- switch (flags) {
- case 0:
- return "usual";
- case MDBX_REVERSEKEY:
- return "reserve";
- case MDBX_INTEGERKEY:
- return "ordinal";
- case MDBX_REVERSEKEY | MDBX_INTEGERKEY:
- return "msgpack";
- default:
- assert(false);
- __unreachable();
- }
-}
-
-static const char *db_flags2valuemode(unsigned flags) {
- flags &= (MDBX_DUPSORT | MDBX_REVERSEDUP | MDBX_DUPFIXED | MDBX_INTEGERDUP);
- switch (flags) {
- case 0:
- return "single";
- case MDBX_DUPSORT:
- return "multi";
- case MDBX_REVERSEDUP:
- case MDBX_DUPSORT | MDBX_REVERSEDUP:
- return "multi-reverse";
- case MDBX_DUPFIXED:
- case MDBX_DUPSORT | MDBX_DUPFIXED:
- return "multi-samelength";
- case MDBX_DUPFIXED | MDBX_REVERSEDUP:
- case MDBX_DUPSORT | MDBX_DUPFIXED | MDBX_REVERSEDUP:
- return "multi-reverse-samelength";
- case MDBX_INTEGERDUP:
- case MDBX_DUPSORT | MDBX_INTEGERDUP:
- case MDBX_DUPSORT | MDBX_DUPFIXED | MDBX_INTEGERDUP:
- case MDBX_DUPFIXED | MDBX_INTEGERDUP:
- return "multi-ordinal";
- case MDBX_INTEGERDUP | MDBX_REVERSEDUP:
- case MDBX_DUPSORT | MDBX_INTEGERDUP | MDBX_REVERSEDUP:
- return "multi-msgpack";
- case MDBX_DUPFIXED | MDBX_INTEGERDUP | MDBX_REVERSEDUP:
- case MDBX_DUPSORT | MDBX_DUPFIXED | MDBX_INTEGERDUP | MDBX_REVERSEDUP:
- return "reserved";
- default:
- assert(false);
- __unreachable();
- }
-}
-
-static int process_db(MDBX_dbi dbi_handle, char *dbi_name, visitor *handler,
- bool silent) {
- MDBX_cursor *mc;
- MDBX_stat ms;
- MDBX_val key, data;
- MDBX_val prev_key, prev_data;
- unsigned flags;
- int rc, i;
- struct problem *saved_list;
- uint64_t problems_count;
-
- uint64_t record_count = 0, dups = 0;
- uint64_t key_bytes = 0, data_bytes = 0;
-
- if ((MDBX_TXN_FINISHED | MDBX_TXN_ERROR) & mdbx_txn_flags(txn)) {
- print(" ! abort processing '%s' due to a previous error\n",
- dbi_name ? dbi_name : "@MAIN");
- return MDBX_BAD_TXN;
- }
-
- if (dbi_handle == ~0u) {
- rc = mdbx_dbi_open_ex(
- txn, dbi_name, MDBX_DB_ACCEDE, &dbi_handle,
- (dbi_name && ignore_wrong_order) ? equal_or_greater : nullptr,
- (dbi_name && ignore_wrong_order) ? equal_or_greater : nullptr);
- if (rc) {
- if (!dbi_name ||
- rc !=
- MDBX_INCOMPATIBLE) /* LY: mainDB's record is not a user's DB. */ {
- error("mdbx_dbi_open('%s') failed, error %d %s\n",
- dbi_name ? dbi_name : "main", rc, mdbx_strerror(rc));
- }
- return rc;
- }
- }
-
- if (dbi_handle >= CORE_DBS && dbi_name && only_subdb &&
- strcmp(only_subdb, dbi_name) != 0) {
- if (verbose) {
- print("Skip processing '%s'...\n", dbi_name);
- fflush(nullptr);
- }
- skipped_subdb++;
- return MDBX_SUCCESS;
- }
-
- if (!silent && verbose) {
- print("Processing '%s'...\n", dbi_name ? dbi_name : "@MAIN");
- fflush(nullptr);
- }
-
- rc = mdbx_dbi_flags(txn, dbi_handle, &flags);
- if (rc) {
- error("mdbx_dbi_flags() failed, error %d %s\n", rc, mdbx_strerror(rc));
- return rc;
- }
-
- rc = mdbx_dbi_stat(txn, dbi_handle, &ms, sizeof(ms));
- if (rc) {
- error("mdbx_dbi_stat() failed, error %d %s\n", rc, mdbx_strerror(rc));
- return rc;
- }
-
- if (!silent && verbose) {
- print(" - key-value kind: %s-key => %s-value", db_flags2keymode(flags),
- db_flags2valuemode(flags));
- if (verbose > 1) {
- print(", flags:");
- if (!flags)
- print(" none");
- else {
- for (i = 0; dbflags[i].bit; i++)
- if (flags & dbflags[i].bit)
- print(" %s", dbflags[i].name);
- }
- if (verbose > 2)
- print(" (0x%02X), dbi-id %d", flags, dbi_handle);
- }
- print("\n");
- if (ms.ms_mod_txnid)
- print(" - last modification txn#%" PRIu64 "\n", ms.ms_mod_txnid);
- if (verbose > 1) {
- print(" - page size %u, entries %" PRIu64 "\n", ms.ms_psize,
- ms.ms_entries);
- print(" - b-tree depth %u, pages: branch %" PRIu64 ", leaf %" PRIu64
- ", overflow %" PRIu64 "\n",
- ms.ms_depth, ms.ms_branch_pages, ms.ms_leaf_pages,
- ms.ms_overflow_pages);
- }
- }
-
- walk_dbi_t *dbi = (dbi_handle < CORE_DBS)
- ? &walk.dbi[dbi_handle]
- : pagemap_lookup_dbi(dbi_name, true);
- if (!dbi) {
- error("too many DBIs or out of memory\n");
- return MDBX_ENOMEM;
- }
- if (!dont_traversal) {
- const uint64_t subtotal_pages =
- ms.ms_branch_pages + ms.ms_leaf_pages + ms.ms_overflow_pages;
- if (subtotal_pages != dbi->pages.total)
- error("%s pages mismatch (%" PRIu64 " != walked %" PRIu64 ")\n",
- "subtotal", subtotal_pages, dbi->pages.total);
- if (ms.ms_branch_pages != dbi->pages.branch)
- error("%s pages mismatch (%" PRIu64 " != walked %" PRIu64 ")\n", "branch",
- ms.ms_branch_pages, dbi->pages.branch);
- const uint64_t allleaf_pages = dbi->pages.leaf + dbi->pages.leaf_dupfixed;
- if (ms.ms_leaf_pages != allleaf_pages)
- error("%s pages mismatch (%" PRIu64 " != walked %" PRIu64 ")\n",
- "all-leaf", ms.ms_leaf_pages, allleaf_pages);
- if (ms.ms_overflow_pages != dbi->pages.large_volume)
- error("%s pages mismatch (%" PRIu64 " != walked %" PRIu64 ")\n",
- "large/overlow", ms.ms_overflow_pages, dbi->pages.large_volume);
- }
- rc = mdbx_cursor_open(txn, dbi_handle, &mc);
- if (rc) {
- error("mdbx_cursor_open() failed, error %d %s\n", rc, mdbx_strerror(rc));
- return rc;
- }
-
- if (ignore_wrong_order) { /* for debugging with enabled assertions */
- mc->mc_flags |= C_SKIPORD;
- if (mc->mc_xcursor)
- mc->mc_xcursor->mx_cursor.mc_flags |= C_SKIPORD;
- }
-
- const size_t maxkeysize = mdbx_env_get_maxkeysize_ex(env, flags);
- saved_list = problems_push();
- prev_key.iov_base = nullptr;
- prev_key.iov_len = 0;
- prev_data.iov_base = nullptr;
- prev_data.iov_len = 0;
- rc = mdbx_cursor_get(mc, &key, &data, MDBX_FIRST);
- while (rc == MDBX_SUCCESS) {
- rc = check_user_break();
- if (rc)
- goto bailout;
-
- bool bad_key = false;
- if (key.iov_len > maxkeysize) {
- problem_add("entry", record_count, "key length exceeds max-key-size",
- "%" PRIuPTR " > %" PRIuPTR, key.iov_len, maxkeysize);
- bad_key = true;
- } else if ((flags & MDBX_INTEGERKEY) && key.iov_len != sizeof(uint64_t) &&
- key.iov_len != sizeof(uint32_t)) {
- problem_add("entry", record_count, "wrong key length",
- "%" PRIuPTR " != 4or8", key.iov_len);
- bad_key = true;
- }
-
- bool bad_data = false;
- if ((flags & MDBX_INTEGERDUP) && data.iov_len != sizeof(uint64_t) &&
- data.iov_len != sizeof(uint32_t)) {
- problem_add("entry", record_count, "wrong data length",
- "%" PRIuPTR " != 4or8", data.iov_len);
- bad_data = true;
- }
-
- if (prev_key.iov_base) {
- if (prev_data.iov_base && !bad_data && (flags & MDBX_DUPFIXED) &&
- prev_data.iov_len != data.iov_len) {
- problem_add("entry", record_count, "different data length",
- "%" PRIuPTR " != %" PRIuPTR, prev_data.iov_len,
- data.iov_len);
- bad_data = true;
- }
-
- if (!bad_key) {
- int cmp = mdbx_cmp(txn, dbi_handle, &key, &prev_key);
- if (cmp == 0) {
- ++dups;
- if ((flags & MDBX_DUPSORT) == 0) {
- problem_add("entry", record_count, "duplicated entries", nullptr);
- if (prev_data.iov_base && data.iov_len == prev_data.iov_len &&
- memcmp(data.iov_base, prev_data.iov_base, data.iov_len) == 0) {
- problem_add("entry", record_count, "complete duplicate", nullptr);
- }
- } else if (!bad_data && prev_data.iov_base) {
- cmp = mdbx_dcmp(txn, dbi_handle, &data, &prev_data);
- if (cmp == 0) {
- problem_add("entry", record_count, "complete duplicate", nullptr);
- } else if (cmp < 0 && !ignore_wrong_order) {
- problem_add("entry", record_count, "wrong order of multi-values",
- nullptr);
- }
- }
- } else if (cmp < 0 && !ignore_wrong_order) {
- problem_add("entry", record_count, "wrong order of entries", nullptr);
- }
- }
- }
-
- if (handler) {
- rc = handler(record_count, &key, &data);
- if (MDBX_IS_ERROR(rc))
- goto bailout;
- }
-
- record_count++;
- key_bytes += key.iov_len;
- data_bytes += data.iov_len;
-
- if (!bad_key) {
- if (verbose && (flags & MDBX_INTEGERKEY) && !prev_key.iov_base)
- print(" - fixed key-size %" PRIuPTR "\n", key.iov_len);
- prev_key = key;
- }
- if (!bad_data) {
- if (verbose && (flags & (MDBX_INTEGERDUP | MDBX_DUPFIXED)) &&
- !prev_data.iov_base)
- print(" - fixed data-size %" PRIuPTR "\n", data.iov_len);
- prev_data = data;
- }
- rc = mdbx_cursor_get(mc, &key, &data, MDBX_NEXT);
- }
- if (rc != MDBX_NOTFOUND)
- error("mdbx_cursor_get() failed, error %d %s\n", rc, mdbx_strerror(rc));
- else
- rc = 0;
-
- if (record_count != ms.ms_entries)
- problem_add("entry", record_count, "different number of entries",
- "%" PRIu64 " != %" PRIu64, record_count, ms.ms_entries);
-bailout:
- problems_count = problems_pop(saved_list);
- if (!silent && verbose) {
- print(" - summary: %" PRIu64 " records, %" PRIu64 " dups, %" PRIu64
- " key's bytes, %" PRIu64 " data's "
- "bytes, %" PRIu64 " problems\n",
- record_count, dups, key_bytes, data_bytes, problems_count);
- fflush(nullptr);
- }
-
- mdbx_cursor_close(mc);
- return (rc || problems_count) ? MDBX_RESULT_TRUE : MDBX_SUCCESS;
-}
-
-static void usage(char *prog) {
- fprintf(stderr,
- "usage: %s [-V] [-v] [-q] [-c] [-0|1|2] [-w] [-d] [-i] [-s subdb] "
- "dbpath\n"
- " -V\t\tprint version and exit\n"
- " -v\t\tmore verbose, could be used multiple times\n"
- " -q\t\tbe quiet\n"
- " -c\t\tforce cooperative mode (don't try exclusive)\n"
- " -w\t\twrite-mode checking\n"
- " -d\t\tdisable page-by-page traversal of B-tree\n"
- " -i\t\tignore wrong order errors (for custom comparators case)\n"
- " -s subdb\tprocess a specific subdatabase only\n"
- " -0|1|2\tforce using specific meta-page 0, or 2 for checking\n"
- " -t\t\tturn to a specified meta-page on successful check\n"
- " -T\t\tturn to a specified meta-page EVEN ON UNSUCCESSFUL CHECK!\n",
- prog);
- exit(EXIT_INTERRUPTED);
-}
-
-static bool meta_ot(txnid_t txn_a, uint64_t sign_a, txnid_t txn_b,
- uint64_t sign_b, const bool wanna_steady) {
- if (txn_a == txn_b)
- return SIGN_IS_STEADY(sign_b);
-
- if (wanna_steady && SIGN_IS_STEADY(sign_a) != SIGN_IS_STEADY(sign_b))
- return SIGN_IS_STEADY(sign_b);
-
- return txn_a < txn_b;
-}
-
-static bool meta_eq(txnid_t txn_a, uint64_t sign_a, txnid_t txn_b,
- uint64_t sign_b) {
- if (!txn_a || txn_a != txn_b)
- return false;
-
- if (SIGN_IS_STEADY(sign_a) != SIGN_IS_STEADY(sign_b))
- return false;
-
- return true;
-}
-
-static int meta_recent(const bool wanna_steady) {
- if (meta_ot(envinfo.mi_meta0_txnid, envinfo.mi_meta0_sign,
- envinfo.mi_meta1_txnid, envinfo.mi_meta1_sign, wanna_steady))
- return meta_ot(envinfo.mi_meta2_txnid, envinfo.mi_meta2_sign,
- envinfo.mi_meta1_txnid, envinfo.mi_meta1_sign, wanna_steady)
- ? 1
- : 2;
- else
- return meta_ot(envinfo.mi_meta0_txnid, envinfo.mi_meta0_sign,
- envinfo.mi_meta2_txnid, envinfo.mi_meta2_sign, wanna_steady)
- ? 2
- : 0;
-}
-
-static int meta_tail(int head) {
- switch (head) {
- case 0:
- return meta_ot(envinfo.mi_meta1_txnid, envinfo.mi_meta1_sign,
- envinfo.mi_meta2_txnid, envinfo.mi_meta2_sign, true)
- ? 1
- : 2;
- case 1:
- return meta_ot(envinfo.mi_meta0_txnid, envinfo.mi_meta0_sign,
- envinfo.mi_meta2_txnid, envinfo.mi_meta2_sign, true)
- ? 0
- : 2;
- case 2:
- return meta_ot(envinfo.mi_meta0_txnid, envinfo.mi_meta0_sign,
- envinfo.mi_meta1_txnid, envinfo.mi_meta1_sign, true)
- ? 0
- : 1;
- default:
- assert(false);
- return -1;
- }
-}
-
-static int meta_head(void) { return meta_recent(false); }
-
-void verbose_meta(int num, txnid_t txnid, uint64_t sign, uint64_t bootid_x,
- uint64_t bootid_y) {
- const bool have_bootid = (bootid_x | bootid_y) != 0;
- const bool bootid_match = bootid_x == envinfo.mi_bootid.current.x &&
- bootid_y == envinfo.mi_bootid.current.y;
-
- print(" - meta-%d: ", num);
- switch (sign) {
- case MDBX_DATASIGN_NONE:
- print("no-sync/legacy");
- break;
- case MDBX_DATASIGN_WEAK:
- print("weak-%s", bootid_match ? (have_bootid ? "intact (same boot-id)"
- : "unknown (no boot-id")
- : "dead");
- break;
- default:
- print("steady");
- break;
- }
- print(" txn#%" PRIu64, txnid);
-
- const int head = meta_head();
- if (num == head)
- print(", head");
- else if (num == meta_tail(head))
- print(", tail");
- else
- print(", stay");
-
- if (stuck_meta >= 0) {
- if (num == stuck_meta)
- print(", forced for checking");
- } else if (txnid > envinfo.mi_recent_txnid &&
- (envflags & (MDBX_EXCLUSIVE | MDBX_RDONLY)) == MDBX_EXCLUSIVE)
- print(", rolled-back %" PRIu64 " (%" PRIu64 " >>> %" PRIu64 ")",
- txnid - envinfo.mi_recent_txnid, txnid, envinfo.mi_recent_txnid);
- print("\n");
-}
-
-static uint64_t get_meta_txnid(const unsigned meta_id) {
- switch (meta_id) {
- default:
- assert(false);
- error("unexpected meta_id %u\n", meta_id);
- return 0;
- case 0:
- return envinfo.mi_meta0_txnid;
- case 1:
- return envinfo.mi_meta1_txnid;
- case 2:
- return envinfo.mi_meta2_txnid;
- }
-}
-
-static void print_size(const char *prefix, const uint64_t value,
- const char *suffix) {
- const char sf[] =
- "KMGTPEZY"; /* LY: Kilo, Mega, Giga, Tera, Peta, Exa, Zetta, Yotta! */
- double k = 1024.0;
- size_t i;
- for (i = 0; sf[i + 1] && value / k > 1000.0; ++i)
- k *= 1024;
- print("%s%" PRIu64 " (%.2f %cb)%s", prefix, value, value / k, sf[i], suffix);
-}
-
-int main(int argc, char *argv[]) {
- int rc;
- char *prog = argv[0];
- char *envname;
- unsigned problems_maindb = 0, problems_freedb = 0, problems_meta = 0;
- bool write_locked = false;
- bool turn_meta = false;
- bool force_turn_meta = false;
-
- double elapsed;
-#if defined(_WIN32) || defined(_WIN64)
- uint64_t timestamp_start, timestamp_finish;
- timestamp_start = GetMilliseconds();
-#else
- struct timespec timestamp_start, timestamp_finish;
- if (clock_gettime(CLOCK_MONOTONIC, ×tamp_start)) {
- rc = errno;
- error("clock_gettime() failed, error %d %s\n", rc, mdbx_strerror(rc));
- return EXIT_FAILURE_SYS;
- }
-#endif
-
- dbi_meta.name = "@META";
- dbi_free.name = "@GC";
- dbi_main.name = "@MAIN";
- atexit(pagemap_cleanup);
-
- if (argc < 2)
- usage(prog);
-
- for (int i; (i = getopt(argc, argv,
- "0"
- "1"
- "2"
- "T"
- "V"
- "v"
- "q"
- "n"
- "w"
- "c"
- "t"
- "d"
- "i"
- "s:")) != EOF;) {
- switch (i) {
- case 'V':
- printf("mdbx_chk version %d.%d.%d.%d\n"
- " - source: %s %s, commit %s, tree %s\n"
- " - anchor: %s\n"
- " - build: %s for %s by %s\n"
- " - flags: %s\n"
- " - options: %s\n",
- mdbx_version.major, mdbx_version.minor, mdbx_version.release,
- mdbx_version.revision, mdbx_version.git.describe,
- mdbx_version.git.datetime, mdbx_version.git.commit,
- mdbx_version.git.tree, mdbx_sourcery_anchor, mdbx_build.datetime,
- mdbx_build.target, mdbx_build.compiler, mdbx_build.flags,
- mdbx_build.options);
- return EXIT_SUCCESS;
- case 'v':
- verbose++;
- break;
- case '0':
- stuck_meta = 0;
- break;
- case '1':
- stuck_meta = 1;
- break;
- case '2':
- stuck_meta = 2;
- break;
- case 't':
- turn_meta = true;
- break;
- case 'T':
- turn_meta = force_turn_meta = true;
- quiet = false;
- if (verbose < 2)
- verbose = 2;
- break;
- case 'q':
- quiet = true;
- break;
- case 'n':
- break;
- case 'w':
- envflags &= ~MDBX_RDONLY;
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- /* Temporary `workaround` for OpenBSD kernel's flaw.
- * See todo4recovery://erased_by_github/libmdbx/issues/67 */
- envflags |= MDBX_WRITEMAP;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
- break;
- case 'c':
- envflags = (envflags & ~MDBX_EXCLUSIVE) | MDBX_ACCEDE;
- break;
- case 'd':
- dont_traversal = true;
- break;
- case 's':
- if (only_subdb && strcmp(only_subdb, optarg))
- usage(prog);
- only_subdb = optarg;
- break;
- case 'i':
- ignore_wrong_order = true;
- break;
- default:
- usage(prog);
- }
- }
-
- if (optind != argc - 1)
- usage(prog);
-
- rc = MDBX_SUCCESS;
- if (stuck_meta >= 0 && (envflags & MDBX_EXCLUSIVE) == 0) {
- error("exclusive mode is required to using specific meta-page(%d) for "
- "checking.\n",
- stuck_meta);
- rc = EXIT_INTERRUPTED;
- }
- if (turn_meta) {
- if (stuck_meta < 0) {
- error("meta-page must be specified (by -0, -1 or -2 options) to turn to "
- "it.\n");
- rc = EXIT_INTERRUPTED;
- }
- if (envflags & MDBX_RDONLY) {
- error("write-mode must be enabled to turn to the specified meta-page.\n");
- rc = EXIT_INTERRUPTED;
- }
- if (only_subdb || dont_traversal) {
- error("whole database checking with tree-traversal are required to turn "
- "to the specified meta-page.\n");
- rc = EXIT_INTERRUPTED;
- }
- }
- if (rc)
- exit(rc);
-
-#if defined(_WIN32) || defined(_WIN64)
- SetConsoleCtrlHandler(ConsoleBreakHandlerRoutine, true);
-#else
-#ifdef SIGPIPE
- signal(SIGPIPE, signal_handler);
-#endif
-#ifdef SIGHUP
- signal(SIGHUP, signal_handler);
-#endif
- signal(SIGINT, signal_handler);
- signal(SIGTERM, signal_handler);
-#endif /* !WINDOWS */
-
- envname = argv[optind];
- print("mdbx_chk %s (%s, T-%s)\nRunning for %s in 'read-%s' mode...\n",
- mdbx_version.git.describe, mdbx_version.git.datetime,
- mdbx_version.git.tree, envname,
- (envflags & MDBX_RDONLY) ? "only" : "write");
- fflush(nullptr);
- mdbx_setup_debug((verbose < MDBX_LOG_TRACE - 1)
- ? (MDBX_log_level_t)(verbose + 1)
- : MDBX_LOG_TRACE,
- MDBX_DBG_LEGACY_OVERLAP | MDBX_DBG_DONT_UPGRADE, logger);
-
- rc = mdbx_env_create(&env);
- if (rc) {
- error("mdbx_env_create() failed, error %d %s\n", rc, mdbx_strerror(rc));
- return rc < 0 ? EXIT_FAILURE_MDBX : EXIT_FAILURE_SYS;
- }
-
- rc = mdbx_env_set_maxdbs(env, MDBX_MAX_DBI);
- if (rc) {
- error("mdbx_env_set_maxdbs() failed, error %d %s\n", rc, mdbx_strerror(rc));
- goto bailout;
- }
-
- if (stuck_meta >= 0) {
- rc = mdbx_env_open_for_recovery(env, envname, stuck_meta,
- (envflags & MDBX_RDONLY) ? false : true);
- } else {
- rc = mdbx_env_open(env, envname, envflags, 0);
- if ((envflags & MDBX_EXCLUSIVE) &&
- (rc == MDBX_BUSY ||
-#if defined(_WIN32) || defined(_WIN64)
- rc == ERROR_LOCK_VIOLATION || rc == ERROR_SHARING_VIOLATION
-#else
- rc == EBUSY || rc == EAGAIN
-#endif
- )) {
- envflags &= ~MDBX_EXCLUSIVE;
- rc = mdbx_env_open(env, envname, envflags | MDBX_ACCEDE, 0);
- }
- }
-
- if (rc) {
- error("mdbx_env_open() failed, error %d %s\n", rc, mdbx_strerror(rc));
- if (rc == MDBX_WANNA_RECOVERY && (envflags & MDBX_RDONLY))
- print("Please run %s in the read-write mode (with '-w' option).\n", prog);
- goto bailout;
- }
- if (verbose)
- print(" - %s mode\n",
- (envflags & MDBX_EXCLUSIVE) ? "monopolistic" : "cooperative");
-
- if ((envflags & (MDBX_RDONLY | MDBX_EXCLUSIVE)) == 0) {
- rc = mdbx_txn_lock(env, false);
- if (rc != MDBX_SUCCESS) {
- error("mdbx_txn_lock() failed, error %d %s\n", rc, mdbx_strerror(rc));
- goto bailout;
- }
- write_locked = true;
- }
-
- rc = mdbx_txn_begin(env, nullptr, MDBX_TXN_RDONLY, &txn);
- if (rc) {
- error("mdbx_txn_begin() failed, error %d %s\n", rc, mdbx_strerror(rc));
- goto bailout;
- }
-
- rc = mdbx_env_info_ex(env, txn, &envinfo, sizeof(envinfo));
- if (rc) {
- error("mdbx_env_info_ex() failed, error %d %s\n", rc, mdbx_strerror(rc));
- goto bailout;
- }
- if (verbose) {
- print(" - current boot-id ");
- if (envinfo.mi_bootid.current.x | envinfo.mi_bootid.current.y)
- print("%016" PRIx64 "-%016" PRIx64 "\n", envinfo.mi_bootid.current.x,
- envinfo.mi_bootid.current.y);
- else
- print("unavailable\n");
- }
-
- mdbx_filehandle_t dxb_fd;
- rc = mdbx_env_get_fd(env, &dxb_fd);
- if (rc) {
- error("mdbx_env_get_fd() failed, error %d %s\n", rc, mdbx_strerror(rc));
- goto bailout;
- }
-
- uint64_t dxb_filesize = 0;
-#if defined(_WIN32) || defined(_WIN64)
- {
- BY_HANDLE_FILE_INFORMATION info;
- if (!GetFileInformationByHandle(dxb_fd, &info))
- rc = GetLastError();
- else
- dxb_filesize = info.nFileSizeLow | (uint64_t)info.nFileSizeHigh << 32;
- }
-#else
- {
- struct stat st;
- STATIC_ASSERT_MSG(sizeof(off_t) <= sizeof(uint64_t),
- "libmdbx requires 64-bit file I/O on 64-bit systems");
- if (fstat(dxb_fd, &st))
- rc = errno;
- else
- dxb_filesize = st.st_size;
- }
-#endif
- if (rc) {
- error("mdbx_filesize() failed, error %d %s\n", rc, mdbx_strerror(rc));
- goto bailout;
- }
-
- errno = 0;
- const uint64_t dxbfile_pages = dxb_filesize / envinfo.mi_dxb_pagesize;
- alloc_pages = txn->mt_next_pgno;
- backed_pages = envinfo.mi_geo.current / envinfo.mi_dxb_pagesize;
- if (backed_pages > dxbfile_pages) {
- print(" ! backed-pages %" PRIu64 " > file-pages %" PRIu64 "\n",
- backed_pages, dxbfile_pages);
- ++problems_meta;
- }
- if (dxbfile_pages < NUM_METAS)
- print(" ! file-pages %" PRIu64 " < %u\n", dxbfile_pages, NUM_METAS);
- if (backed_pages < NUM_METAS)
- print(" ! backed-pages %" PRIu64 " < %u\n", backed_pages, NUM_METAS);
- if (backed_pages < NUM_METAS || dxbfile_pages < NUM_METAS)
- goto bailout;
- if (backed_pages > MAX_PAGENO + 1) {
- print(" ! backed-pages %" PRIu64 " > max-pages %" PRIaPGNO "\n",
- backed_pages, MAX_PAGENO + 1);
- ++problems_meta;
- backed_pages = MAX_PAGENO + 1;
- }
-
- if ((envflags & (MDBX_EXCLUSIVE | MDBX_RDONLY)) != MDBX_RDONLY) {
- if (backed_pages > dxbfile_pages) {
- print(" ! backed-pages %" PRIu64 " > file-pages %" PRIu64 "\n",
- backed_pages, dxbfile_pages);
- ++problems_meta;
- backed_pages = dxbfile_pages;
- }
- if (alloc_pages > backed_pages) {
- print(" ! alloc-pages %" PRIu64 " > backed-pages %" PRIu64 "\n",
- alloc_pages, backed_pages);
- ++problems_meta;
- alloc_pages = backed_pages;
- }
- } else {
- /* LY: DB may be shrinked by writer down to the allocated pages. */
- if (alloc_pages > backed_pages) {
- print(" ! alloc-pages %" PRIu64 " > backed-pages %" PRIu64 "\n",
- alloc_pages, backed_pages);
- ++problems_meta;
- alloc_pages = backed_pages;
- }
- if (alloc_pages > dxbfile_pages) {
- print(" ! alloc-pages %" PRIu64 " > file-pages %" PRIu64 "\n",
- alloc_pages, dxbfile_pages);
- ++problems_meta;
- alloc_pages = dxbfile_pages;
- }
- if (backed_pages > dxbfile_pages)
- backed_pages = dxbfile_pages;
- }
-
- if (verbose) {
- print(" - pagesize %u (%u system), max keysize %d..%d"
- ", max readers %u\n",
- envinfo.mi_dxb_pagesize, envinfo.mi_sys_pagesize,
- mdbx_env_get_maxkeysize_ex(env, MDBX_DUPSORT),
- mdbx_env_get_maxkeysize_ex(env, 0), envinfo.mi_maxreaders);
- print_size(" - mapsize ", envinfo.mi_mapsize, "\n");
- if (envinfo.mi_geo.lower == envinfo.mi_geo.upper)
- print_size(" - fixed datafile: ", envinfo.mi_geo.current, "");
- else {
- print_size(" - dynamic datafile: ", envinfo.mi_geo.lower, "");
- print_size(" .. ", envinfo.mi_geo.upper, ", ");
- print_size("+", envinfo.mi_geo.grow, ", ");
- print_size("-", envinfo.mi_geo.shrink, "\n");
- print_size(" - current datafile: ", envinfo.mi_geo.current, "");
- }
- printf(", %" PRIu64 " pages\n",
- envinfo.mi_geo.current / envinfo.mi_dxb_pagesize);
-#if defined(_WIN32) || defined(_WIN64)
- if (envinfo.mi_geo.shrink && envinfo.mi_geo.current != envinfo.mi_geo.upper)
- print(
- " WARNING: Due Windows system limitations a "
- "file couldn't\n be truncated while the database "
- "is opened. So, the size\n database file "
- "of may by large than the database itself,\n "
- "until it will be closed or reopened in read-write mode.\n");
-#endif
- verbose_meta(0, envinfo.mi_meta0_txnid, envinfo.mi_meta0_sign,
- envinfo.mi_bootid.meta0.x, envinfo.mi_bootid.meta0.y);
- verbose_meta(1, envinfo.mi_meta1_txnid, envinfo.mi_meta1_sign,
- envinfo.mi_bootid.meta1.x, envinfo.mi_bootid.meta1.y);
- verbose_meta(2, envinfo.mi_meta2_txnid, envinfo.mi_meta2_sign,
- envinfo.mi_bootid.meta2.x, envinfo.mi_bootid.meta2.y);
- }
-
- if (stuck_meta >= 0) {
- if (verbose) {
- print(" - skip checking meta-pages since the %u"
- " is selected for verification\n",
- stuck_meta);
- print(" - transactions: recent %" PRIu64
- ", selected for verification %" PRIu64 ", lag %" PRIi64 "\n",
- envinfo.mi_recent_txnid, get_meta_txnid(stuck_meta),
- envinfo.mi_recent_txnid - get_meta_txnid(stuck_meta));
- }
- } else {
- if (verbose > 1)
- print(" - performs check for meta-pages clashes\n");
- if (meta_eq(envinfo.mi_meta0_txnid, envinfo.mi_meta0_sign,
- envinfo.mi_meta1_txnid, envinfo.mi_meta1_sign)) {
- print(" ! meta-%d and meta-%d are clashed\n", 0, 1);
- ++problems_meta;
- }
- if (meta_eq(envinfo.mi_meta1_txnid, envinfo.mi_meta1_sign,
- envinfo.mi_meta2_txnid, envinfo.mi_meta2_sign)) {
- print(" ! meta-%d and meta-%d are clashed\n", 1, 2);
- ++problems_meta;
- }
- if (meta_eq(envinfo.mi_meta2_txnid, envinfo.mi_meta2_sign,
- envinfo.mi_meta0_txnid, envinfo.mi_meta0_sign)) {
- print(" ! meta-%d and meta-%d are clashed\n", 2, 0);
- ++problems_meta;
- }
-
- const unsigned steady_meta_id = meta_recent(true);
- const uint64_t steady_meta_txnid = get_meta_txnid(steady_meta_id);
- const unsigned weak_meta_id = meta_recent(false);
- const uint64_t weak_meta_txnid = get_meta_txnid(weak_meta_id);
- if (envflags & MDBX_EXCLUSIVE) {
- if (verbose > 1)
- print(" - performs full check recent-txn-id with meta-pages\n");
- if (steady_meta_txnid != envinfo.mi_recent_txnid) {
- print(" ! steady meta-%d txn-id mismatch recent-txn-id (%" PRIi64
- " != %" PRIi64 ")\n",
- steady_meta_id, steady_meta_txnid, envinfo.mi_recent_txnid);
- ++problems_meta;
- }
- } else if (write_locked) {
- if (verbose > 1)
- print(" - performs lite check recent-txn-id with meta-pages (not a "
- "monopolistic mode)\n");
- if (weak_meta_txnid != envinfo.mi_recent_txnid) {
- print(" ! weak meta-%d txn-id mismatch recent-txn-id (%" PRIi64
- " != %" PRIi64 ")\n",
- weak_meta_id, weak_meta_txnid, envinfo.mi_recent_txnid);
- ++problems_meta;
- }
- } else if (verbose) {
- print(" - skip check recent-txn-id with meta-pages (monopolistic or "
- "read-write mode only)\n");
- }
- total_problems += problems_meta;
-
- if (verbose)
- print(" - transactions: recent %" PRIu64 ", latter reader %" PRIu64
- ", lag %" PRIi64 "\n",
- envinfo.mi_recent_txnid, envinfo.mi_latter_reader_txnid,
- envinfo.mi_recent_txnid - envinfo.mi_latter_reader_txnid);
- }
-
- if (!dont_traversal) {
- struct problem *saved_list;
- size_t traversal_problems;
- uint64_t empty_pages, lost_bytes;
-
- print("Traversal b-tree by txn#%" PRIaTXN "...\n", txn->mt_txnid);
- fflush(nullptr);
- walk.pagemap = mdbx_calloc((size_t)backed_pages, sizeof(*walk.pagemap));
- if (!walk.pagemap) {
- rc = errno ? errno : MDBX_ENOMEM;
- error("calloc() failed, error %d %s\n", rc, mdbx_strerror(rc));
- goto bailout;
- }
-
- saved_list = problems_push();
- rc = mdbx_env_pgwalk(txn, pgvisitor, nullptr,
- true /* always skip key ordering checking to avoid
- MDBX_CORRUPTED when using custom comparators */);
- traversal_problems = problems_pop(saved_list);
-
- if (rc) {
- if (rc != MDBX_EINTR || !check_user_break())
- error("mdbx_env_pgwalk() failed, error %d %s\n", rc, mdbx_strerror(rc));
- goto bailout;
- }
-
- for (uint64_t n = 0; n < alloc_pages; ++n)
- if (!walk.pagemap[n])
- unused_pages += 1;
-
- empty_pages = lost_bytes = 0;
- for (walk_dbi_t *dbi = &dbi_main; dbi < ARRAY_END(walk.dbi) && dbi->name;
- ++dbi) {
- empty_pages += dbi->pages.empty;
- lost_bytes += dbi->lost_bytes;
- }
-
- if (verbose) {
- uint64_t total_page_bytes = walk.pgcount * envinfo.mi_dxb_pagesize;
- print(" - pages: walked %" PRIu64 ", left/unused %" PRIu64 "\n",
- walk.pgcount, unused_pages);
- if (verbose > 1) {
- for (walk_dbi_t *dbi = walk.dbi; dbi < ARRAY_END(walk.dbi) && dbi->name;
- ++dbi) {
- print(" %s: subtotal %" PRIu64, dbi->name, dbi->pages.total);
- if (dbi->pages.other && dbi->pages.other != dbi->pages.total)
- print(", other %" PRIu64, dbi->pages.other);
- if (dbi->pages.branch)
- print(", branch %" PRIu64, dbi->pages.branch);
- if (dbi->pages.large_count)
- print(", large %" PRIu64, dbi->pages.large_count);
- uint64_t all_leaf = dbi->pages.leaf + dbi->pages.leaf_dupfixed;
- if (all_leaf) {
- print(", leaf %" PRIu64, all_leaf);
- if (verbose > 2 &&
- (dbi->pages.subleaf_dupsort | dbi->pages.leaf_dupfixed |
- dbi->pages.subleaf_dupfixed))
- print(" (usual %" PRIu64 ", sub-dupsort %" PRIu64
- ", dupfixed %" PRIu64 ", sub-dupfixed %" PRIu64 ")",
- dbi->pages.leaf, dbi->pages.subleaf_dupsort,
- dbi->pages.leaf_dupfixed, dbi->pages.subleaf_dupfixed);
- }
- print("\n");
- }
- }
-
- if (verbose > 1)
- print(" - usage: total %" PRIu64 " bytes, payload %" PRIu64
- " (%.1f%%), unused "
- "%" PRIu64 " (%.1f%%)\n",
- total_page_bytes, walk.total_payload_bytes,
- walk.total_payload_bytes * 100.0 / total_page_bytes,
- total_page_bytes - walk.total_payload_bytes,
- (total_page_bytes - walk.total_payload_bytes) * 100.0 /
- total_page_bytes);
- if (verbose > 2) {
- for (walk_dbi_t *dbi = walk.dbi; dbi < ARRAY_END(walk.dbi) && dbi->name;
- ++dbi)
- if (dbi->pages.total) {
- uint64_t dbi_bytes = dbi->pages.total * envinfo.mi_dxb_pagesize;
- print(" %s: subtotal %" PRIu64 " bytes (%.1f%%),"
- " payload %" PRIu64 " (%.1f%%), unused %" PRIu64 " (%.1f%%)",
- dbi->name, dbi_bytes, dbi_bytes * 100.0 / total_page_bytes,
- dbi->payload_bytes, dbi->payload_bytes * 100.0 / dbi_bytes,
- dbi_bytes - dbi->payload_bytes,
- (dbi_bytes - dbi->payload_bytes) * 100.0 / dbi_bytes);
- if (dbi->pages.empty)
- print(", %" PRIu64 " empty pages", dbi->pages.empty);
- if (dbi->lost_bytes)
- print(", %" PRIu64 " bytes lost", dbi->lost_bytes);
- print("\n");
- } else
- print(" %s: empty\n", dbi->name);
- }
- print(" - summary: average fill %.1f%%",
- walk.total_payload_bytes * 100.0 / total_page_bytes);
- if (empty_pages)
- print(", %" PRIu64 " empty pages", empty_pages);
- if (lost_bytes)
- print(", %" PRIu64 " bytes lost", lost_bytes);
- print(", %" PRIuPTR " problems\n", traversal_problems);
- }
- } else if (verbose) {
- print("Skipping b-tree walk...\n");
- fflush(nullptr);
- }
-
- if (!verbose)
- print("Iterating DBIs...\n");
- if (data_tree_problems) {
- print("Skip processing %s since tree is corrupted (%u problems)\n", "@MAIN",
- data_tree_problems);
- problems_maindb = data_tree_problems;
- } else
- problems_maindb = process_db(~0u, /* MAIN_DBI */ nullptr, nullptr, false);
-
- if (gc_tree_problems) {
- print("Skip processing %s since tree is corrupted (%u problems)\n", "@GC",
- gc_tree_problems);
- problems_freedb = gc_tree_problems;
- } else
- problems_freedb = process_db(FREE_DBI, "@GC", handle_freedb, false);
-
- if (verbose) {
- uint64_t value = envinfo.mi_mapsize / envinfo.mi_dxb_pagesize;
- double percent = value / 100.0;
- print(" - space: %" PRIu64 " total pages", value);
- print(", backed %" PRIu64 " (%.1f%%)", backed_pages,
- backed_pages / percent);
- print(", allocated %" PRIu64 " (%.1f%%)", alloc_pages,
- alloc_pages / percent);
-
- if (verbose > 1) {
- value = envinfo.mi_mapsize / envinfo.mi_dxb_pagesize - alloc_pages;
- print(", remained %" PRIu64 " (%.1f%%)", value, value / percent);
-
- value = dont_traversal ? alloc_pages - gc_pages : walk.pgcount;
- print(", used %" PRIu64 " (%.1f%%)", value, value / percent);
-
- print(", gc %" PRIu64 " (%.1f%%)", gc_pages, gc_pages / percent);
-
- value = gc_pages - reclaimable_pages;
- print(", detained %" PRIu64 " (%.1f%%)", value, value / percent);
-
- print(", reclaimable %" PRIu64 " (%.1f%%)", reclaimable_pages,
- reclaimable_pages / percent);
- }
-
- value = envinfo.mi_mapsize / envinfo.mi_dxb_pagesize - alloc_pages +
- reclaimable_pages;
- print(", available %" PRIu64 " (%.1f%%)\n", value, value / percent);
- }
-
- if (problems_maindb == 0 && problems_freedb == 0) {
- if (!dont_traversal &&
- (envflags & (MDBX_EXCLUSIVE | MDBX_RDONLY)) != MDBX_RDONLY) {
- if (walk.pgcount != alloc_pages - gc_pages) {
- error("used pages mismatch (%" PRIu64 "(walked) != %" PRIu64
- "(allocated - GC))\n",
- walk.pgcount, alloc_pages - gc_pages);
- }
- if (unused_pages != gc_pages) {
- error("gc pages mismatch (%" PRIu64 "(expected) != %" PRIu64 "(GC))\n",
- unused_pages, gc_pages);
- }
- } else if (verbose) {
- print(" - skip check used and gc pages (btree-traversal with "
- "monopolistic or read-write mode only)\n");
- }
-
- if (!process_db(MAIN_DBI, nullptr, handle_maindb, true)) {
- if (!userdb_count && verbose)
- print(" - does not contain multiple databases\n");
- }
- }
-
- if (rc == 0 && total_problems == 1 && problems_meta == 1 && !dont_traversal &&
- (envflags & MDBX_RDONLY) == 0 && !only_subdb && stuck_meta < 0 &&
- get_meta_txnid(meta_recent(true)) < envinfo.mi_recent_txnid) {
- print("Perform sync-to-disk for make steady checkpoint at txn-id #%" PRIi64
- "\n",
- envinfo.mi_recent_txnid);
- fflush(nullptr);
- if (write_locked) {
- mdbx_txn_unlock(env);
- write_locked = false;
- }
- rc = mdbx_env_sync_ex(env, true, false);
- if (rc != MDBX_SUCCESS)
- error("mdbx_env_pgwalk() failed, error %d %s\n", rc, mdbx_strerror(rc));
- else {
- total_problems -= 1;
- problems_meta -= 1;
- }
- }
-
- if (turn_meta && stuck_meta >= 0 && !dont_traversal && !only_subdb &&
- (envflags & (MDBX_RDONLY | MDBX_EXCLUSIVE)) == MDBX_EXCLUSIVE) {
- const bool successful_check = (rc | total_problems | problems_meta) == 0;
- if (successful_check || force_turn_meta) {
- fflush(nullptr);
- print(" = Performing turn to the specified meta-page (%d) due to %s!\n",
- stuck_meta,
- successful_check ? "successful check" : "the -T option was given");
- fflush(nullptr);
- rc = mdbx_env_turn_for_recovery(env, stuck_meta);
- if (rc != MDBX_SUCCESS)
- error("mdbx_env_turn_for_recovery() failed, error %d %s\n", rc,
- mdbx_strerror(rc));
- } else {
- print(" = Skipping turn to the specified meta-page (%d) due to "
- "unsuccessful check!\n",
- stuck_meta);
- }
- }
-
-bailout:
- if (txn)
- mdbx_txn_abort(txn);
- if (write_locked) {
- mdbx_txn_unlock(env);
- write_locked = false;
- }
- if (env) {
- const bool dont_sync = rc != 0 || total_problems;
- mdbx_env_close_ex(env, dont_sync);
- }
- fflush(nullptr);
- if (rc) {
- if (rc < 0)
- return user_break ? EXIT_INTERRUPTED : EXIT_FAILURE_SYS;
- return EXIT_FAILURE_MDBX;
- }
-
-#if defined(_WIN32) || defined(_WIN64)
- timestamp_finish = GetMilliseconds();
- elapsed = (timestamp_finish - timestamp_start) * 1e-3;
-#else
- if (clock_gettime(CLOCK_MONOTONIC, ×tamp_finish)) {
- rc = errno;
- error("clock_gettime() failed, error %d %s\n", rc, mdbx_strerror(rc));
- return EXIT_FAILURE_SYS;
- }
- elapsed = timestamp_finish.tv_sec - timestamp_start.tv_sec +
- (timestamp_finish.tv_nsec - timestamp_start.tv_nsec) * 1e-9;
-#endif /* !WINDOWS */
-
- if (total_problems) {
- print("Total %u error%s detected, elapsed %.3f seconds.\n", total_problems,
- (total_problems > 1) ? "s are" : " is", elapsed);
- if (problems_meta || problems_maindb || problems_freedb)
- return EXIT_FAILURE_CHECK_MAJOR;
- return EXIT_FAILURE_CHECK_MINOR;
- }
- print("No error is detected, elapsed %.3f seconds\n", elapsed);
- return EXIT_SUCCESS;
-}
diff --git a/3rdparty/libmdbx/mdbx_copy.c b/3rdparty/libmdbx/mdbx_copy.c
deleted file mode 100644
index 8cfc4f5cfff71f0135b10d9c9df78d9feda35a37..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx_copy.c
+++ /dev/null
@@ -1,3660 +0,0 @@
-/* mdbx_copy.c - memory-mapped database backup tool */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#ifdef _MSC_VER
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#pragma warning(disable : 4996) /* The POSIX name is deprecated... */
-#endif /* _MSC_VER (warnings) */
-
-#define xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#define MDBX_BUILD_SOURCERY c510b93d78b5848764f976ea67f06d278a7b99f141b4c98ceb095285f834f31c_v0_11_7_0_g40ec559c
-#ifdef MDBX_CONFIG_H
-#include MDBX_CONFIG_H
-#endif
-
-#define LIBMDBX_INTERNALS
-#ifdef xMDBX_TOOLS
-#define MDBX_DEPRECATED
-#endif /* xMDBX_TOOLS */
-
-#ifdef xMDBX_ALLOY
-/* Amalgamated build */
-#define MDBX_INTERNAL_FUNC static
-#define MDBX_INTERNAL_VAR static
-#else
-/* Non-amalgamated build */
-#define MDBX_INTERNAL_FUNC
-#define MDBX_INTERNAL_VAR extern
-#endif /* xMDBX_ALLOY */
-
-/** Disables using GNU/Linux libc extensions.
- * \ingroup build_option
- * \note This option couldn't be moved to the options.h since dependant
- * control macros/defined should be prepared before include the options.h */
-#ifndef MDBX_DISABLE_GNU_SOURCE
-#define MDBX_DISABLE_GNU_SOURCE 0
-#endif
-#if MDBX_DISABLE_GNU_SOURCE
-#undef _GNU_SOURCE
-#elif (defined(__linux__) || defined(__gnu_linux__)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif /* MDBX_DISABLE_GNU_SOURCE */
-
-/*----------------------------------------------------------------------------*/
-
-/* Should be defined before any includes */
-#if !defined(_FILE_OFFSET_BITS) && !defined(__ANDROID_API__) && \
- !defined(ANDROID)
-#define _FILE_OFFSET_BITS 64
-#endif
-
-#ifdef __APPLE__
-#define _DARWIN_C_SOURCE
-#endif
-
-#ifdef _MSC_VER
-#if _MSC_FULL_VER < 190024234
-/* Actually libmdbx was not tested with compilers older than 19.00.24234 (Visual
- * Studio 2015 Update 3). But you could remove this #error and try to continue
- * at your own risk. In such case please don't rise up an issues related ONLY to
- * old compilers.
- *
- * NOTE:
- * Unfortunately, there are several different builds of "Visual Studio" that
- * are called "Visual Studio 2015 Update 3".
- *
- * The 190024234 is used here because it is minimal version of Visual Studio
- * that was used for build and testing libmdbx in recent years. Soon this
- * value will be increased to 19.0.24241.7, since build and testing using
- * "Visual Studio 2015" will be performed only at https://ci.appveyor.com.
- *
- * Please ask Microsoft (but not us) for information about version differences
- * and how to and where you can obtain the latest "Visual Studio 2015" build
- * with all fixes.
- */
-#error \
- "At least \"Microsoft C/C++ Compiler\" version 19.00.24234 (Visual Studio 2015 Update 3) is required."
-#endif
-#ifndef _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#if _MSC_VER > 1913
-#pragma warning(disable : 5045) /* Compiler will insert Spectre mitigation... \
- */
-#endif
-#if _MSC_VER > 1914
-#pragma warning( \
- disable : 5105) /* winbase.h(9531): warning C5105: macro expansion \
- producing 'defined' has undefined behavior */
-#endif
-#pragma warning(disable : 4710) /* 'xyz': function not inlined */
-#pragma warning(disable : 4711) /* function 'xyz' selected for automatic \
- inline expansion */
-#pragma warning( \
- disable : 4201) /* nonstandard extension used : nameless struct / union */
-#pragma warning(disable : 4702) /* unreachable code */
-#pragma warning(disable : 4706) /* assignment within conditional expression */
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4324) /* 'xyz': structure was padded due to \
- alignment specifier */
-#pragma warning(disable : 4310) /* cast truncates constant value */
-#pragma warning( \
- disable : 4820) /* bytes padding added after data member for alignment */
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4366) /* the result of the unary '&' operator may be \
- unaligned */
-#pragma warning(disable : 4200) /* nonstandard extension used: zero-sized \
- array in struct/union */
-#pragma warning(disable : 4204) /* nonstandard extension used: non-constant \
- aggregate initializer */
-#pragma warning( \
- disable : 4505) /* unreferenced local function has been removed */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(__GNUC__) && __GNUC__ < 9
-#pragma GCC diagnostic ignored "-Wattributes"
-#endif /* GCC < 9 */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-#include "mdbx.h"
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#ifndef __GNUC_PREREQ
-# if defined(__GNUC__) && defined(__GNUC_MINOR__)
-# define __GNUC_PREREQ(maj, min) \
- ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GNUC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GNUC_PREREQ */
-
-#ifndef __CLANG_PREREQ
-# ifdef __clang__
-# define __CLANG_PREREQ(maj,min) \
- ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min))
-# else
-# define __CLANG_PREREQ(maj,min) (0)
-# endif
-#endif /* __CLANG_PREREQ */
-
-#ifndef __GLIBC_PREREQ
-# if defined(__GLIBC__) && defined(__GLIBC_MINOR__)
-# define __GLIBC_PREREQ(maj, min) \
- ((__GLIBC__ << 16) + __GLIBC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GLIBC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GLIBC_PREREQ */
-
-#ifndef __has_warning
-# define __has_warning(x) (0)
-#endif
-
-#ifndef __has_include
-# define __has_include(x) (0)
-#endif
-
-#if __has_feature(thread_sanitizer)
-# define __SANITIZE_THREAD__ 1
-#endif
-
-#if __has_feature(address_sanitizer)
-# define __SANITIZE_ADDRESS__ 1
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __extern_C
-# ifdef __cplusplus
-# define __extern_C extern "C"
-# else
-# define __extern_C
-# endif
-#endif /* __extern_C */
-
-#if !defined(nullptr) && !defined(__cplusplus) || (__cplusplus < 201103L && !defined(_MSC_VER))
-# define nullptr NULL
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __always_inline
-# if defined(__GNUC__) || __has_attribute(__always_inline__)
-# define __always_inline __inline __attribute__((__always_inline__))
-# elif defined(_MSC_VER)
-# define __always_inline __forceinline
-# else
-# define __always_inline
-# endif
-#endif /* __always_inline */
-
-#ifndef __noinline
-# if defined(__GNUC__) || __has_attribute(__noinline__)
-# define __noinline __attribute__((__noinline__))
-# elif defined(_MSC_VER)
-# define __noinline __declspec(noinline)
-# else
-# define __noinline
-# endif
-#endif /* __noinline */
-
-#ifndef __must_check_result
-# if defined(__GNUC__) || __has_attribute(__warn_unused_result__)
-# define __must_check_result __attribute__((__warn_unused_result__))
-# else
-# define __must_check_result
-# endif
-#endif /* __must_check_result */
-
-#if !defined(__noop) && !defined(_MSC_VER)
-# define __noop(...) do {} while(0)
-#endif /* __noop */
-
-#if defined(__fallthrough) && \
- (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__))
-#undef __fallthrough
-#endif /* __fallthrough workaround for MinGW */
-
-#ifndef __fallthrough
-# if defined(__cplusplus) && (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || __cplusplus >= 201703L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(8, 0) && defined(__cplusplus) && __cplusplus >= 201103L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(7, 0) && \
- (!defined(__LCC__) || (__LCC__ == 124 && __LCC_MINOR__ >= 12) || \
- (__LCC__ == 125 && __LCC_MINOR__ >= 5) || (__LCC__ >= 126))
-# define __fallthrough __attribute__((__fallthrough__))
-# elif defined(__clang__) && defined(__cplusplus) && __cplusplus >= 201103L &&\
- __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
-# define __fallthrough [[clang::fallthrough]]
-# else
-# define __fallthrough
-# endif
-#endif /* __fallthrough */
-
-#ifndef __unreachable
-# if __GNUC_PREREQ(4,5) || __has_builtin(__builtin_unreachable)
-# define __unreachable() __builtin_unreachable()
-# elif defined(_MSC_VER)
-# define __unreachable() __assume(0)
-# else
-# define __unreachable() __noop()
-# endif
-#endif /* __unreachable */
-
-#ifndef __prefetch
-# if defined(__GNUC__) || defined(__clang__) || __has_builtin(__builtin_prefetch)
-# define __prefetch(ptr) __builtin_prefetch(ptr)
-# else
-# define __prefetch(ptr) __noop(ptr)
-# endif
-#endif /* __prefetch */
-
-#ifndef __nothrow
-# if defined(__cplusplus)
-# if __cplusplus < 201703L
-# define __nothrow throw()
-# else
-# define __nothrow noexcept(true)
-# endif /* __cplusplus */
-# elif defined(__GNUC__) || __has_attribute(__nothrow__)
-# define __nothrow __attribute__((__nothrow__))
-# elif defined(_MSC_VER) && defined(__cplusplus)
-# define __nothrow __declspec(nothrow)
-# else
-# define __nothrow
-# endif
-#endif /* __nothrow */
-
-#ifndef __hidden
-# if defined(__GNUC__) || __has_attribute(__visibility__)
-# define __hidden __attribute__((__visibility__("hidden")))
-# else
-# define __hidden
-# endif
-#endif /* __hidden */
-
-#ifndef __optimize
-# if defined(__OPTIMIZE__)
-# if (defined(__GNUC__) && !defined(__clang__)) || __has_attribute(__optimize__)
-# define __optimize(ops) __attribute__((__optimize__(ops)))
-# else
-# define __optimize(ops)
-# endif
-# else
-# define __optimize(ops)
-# endif
-#endif /* __optimize */
-
-#ifndef __hot
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __hot __attribute__((__hot__)) __optimize(3)
-# elif defined(__clang__) && !__has_attribute(__hot_) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put frequently used functions in separate section */
-# define __hot __attribute__((__section__("text.hot"))) __optimize("O3")
-# elif defined(__GNUC__) || __has_attribute(__hot__)
-# define __hot __attribute__((__hot__)) __optimize("O3")
-# else
-# define __hot __optimize("O3")
-# endif
-# else
-# define __hot
-# endif
-#endif /* __hot */
-
-#ifndef __cold
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __cold __attribute__((__cold__)) __optimize(1)
-# elif defined(__clang__) && !__has_attribute(cold) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put infrequently used functions in separate section */
-# define __cold __attribute__((__section__("text.unlikely"))) __optimize("Os")
-# elif defined(__GNUC__) || __has_attribute(cold)
-# define __cold __attribute__((__cold__)) __optimize("Os")
-# else
-# define __cold __optimize("Os")
-# endif
-# else
-# define __cold
-# endif
-#endif /* __cold */
-
-#ifndef __flatten
-# if defined(__OPTIMIZE__) && (defined(__GNUC__) || __has_attribute(__flatten__))
-# define __flatten __attribute__((__flatten__))
-# else
-# define __flatten
-# endif
-#endif /* __flatten */
-
-#ifndef likely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define likely(cond) __builtin_expect(!!(cond), 1)
-# else
-# define likely(x) (!!(x))
-# endif
-#endif /* likely */
-
-#ifndef unlikely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define unlikely(cond) __builtin_expect(!!(cond), 0)
-# else
-# define unlikely(x) (!!(x))
-# endif
-#endif /* unlikely */
-
-#ifndef __anonymous_struct_extension__
-# if defined(__GNUC__)
-# define __anonymous_struct_extension__ __extension__
-# else
-# define __anonymous_struct_extension__
-# endif
-#endif /* __anonymous_struct_extension__ */
-
-#ifndef __Wpedantic_format_voidptr
- MDBX_MAYBE_UNUSED MDBX_PURE_FUNCTION static __inline const void*
- __Wpedantic_format_voidptr(const void* ptr) {return ptr;}
-# define __Wpedantic_format_voidptr(ARG) __Wpedantic_format_voidptr(ARG)
-#endif /* __Wpedantic_format_voidptr */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(MDBX_USE_VALGRIND)
-# include <valgrind/memcheck.h>
-# ifndef VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE
- /* LY: available since Valgrind 3.10 */
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# endif
-#elif !defined(RUNNING_ON_VALGRIND)
-# define VALGRIND_CREATE_MEMPOOL(h,r,z)
-# define VALGRIND_DESTROY_MEMPOOL(h)
-# define VALGRIND_MEMPOOL_TRIM(h,a,s)
-# define VALGRIND_MEMPOOL_ALLOC(h,a,s)
-# define VALGRIND_MEMPOOL_FREE(h,a)
-# define VALGRIND_MEMPOOL_CHANGE(h,a,b,s)
-# define VALGRIND_MAKE_MEM_NOACCESS(a,s)
-# define VALGRIND_MAKE_MEM_DEFINED(a,s)
-# define VALGRIND_MAKE_MEM_UNDEFINED(a,s)
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_CHECK_MEM_IS_ADDRESSABLE(a,s) (0)
-# define VALGRIND_CHECK_MEM_IS_DEFINED(a,s) (0)
-# define RUNNING_ON_VALGRIND (0)
-#endif /* MDBX_USE_VALGRIND */
-
-#ifdef __SANITIZE_ADDRESS__
-# include <sanitizer/asan_interface.h>
-#elif !defined(ASAN_POISON_MEMORY_REGION)
-# define ASAN_POISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-# define ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-#endif /* __SANITIZE_ADDRESS__ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef ARRAY_LENGTH
-# ifdef __cplusplus
- template <typename T, size_t N>
- char (&__ArraySizeHelper(T (&array)[N]))[N];
-# define ARRAY_LENGTH(array) (sizeof(::__ArraySizeHelper(array)))
-# else
-# define ARRAY_LENGTH(array) (sizeof(array) / sizeof(array[0]))
-# endif
-#endif /* ARRAY_LENGTH */
-
-#ifndef ARRAY_END
-# define ARRAY_END(array) (&array[ARRAY_LENGTH(array)])
-#endif /* ARRAY_END */
-
-#define CONCAT(a,b) a##b
-#define XCONCAT(a,b) CONCAT(a,b)
-
-#ifndef offsetof
-# define offsetof(type, member) __builtin_offsetof(type, member)
-#endif /* offsetof */
-
-#ifndef container_of
-# define container_of(ptr, type, member) \
- ((type *)((char *)(ptr) - offsetof(type, member)))
-#endif /* container_of */
-
-#define MDBX_TETRAD(a, b, c, d) \
- ((uint32_t)(a) << 24 | (uint32_t)(b) << 16 | (uint32_t)(c) << 8 | (d))
-
-#define MDBX_STRING_TETRAD(str) MDBX_TETRAD(str[0], str[1], str[2], str[3])
-
-#define FIXME "FIXME: " __FILE__ ", " MDBX_STRINGIFY(__LINE__)
-
-#ifndef STATIC_ASSERT_MSG
-# if defined(static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) static_assert(expr, msg)
-# elif defined(_STATIC_ASSERT)
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif defined(_MSC_VER)
-# include <crtdbg.h>
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) \
- || __has_feature(c_static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) _Static_assert(expr, msg)
-# else
-# define STATIC_ASSERT_MSG(expr, msg) switch (0) {case 0:case (expr):;}
-# endif
-#endif /* STATIC_ASSERT */
-
-#ifndef STATIC_ASSERT
-# define STATIC_ASSERT(expr) STATIC_ASSERT_MSG(expr, #expr)
-#endif
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__GNUC__) && !__GNUC_PREREQ(4, 2)
-/* Actually libmdbx was not tested with compilers older than GCC 4.2.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required GCC >= 4.2"
-#endif
-
-#if defined(__clang__) && !__CLANG_PREREQ(3, 8)
-/* Actually libmdbx was not tested with CLANG older than 3.8.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required CLANG >= 3.8"
-#endif
-
-#if defined(__GLIBC__) && !__GLIBC_PREREQ(2, 12)
-/* Actually libmdbx was not tested with something older than glibc 2.12.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old systems.
- */
-#warning "libmdbx was only tested with GLIBC >= 2.12."
-#endif
-
-#ifdef __SANITIZE_THREAD__
-#warning \
- "libmdbx don't compatible with ThreadSanitizer, you will get a lot of false-positive issues."
-#endif /* __SANITIZE_THREAD__ */
-
-#if __has_warning("-Wnested-anon-types")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wnested-anon-types"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wnested-anon-types"
-#else
-#pragma warning disable "nested-anon-types"
-#endif
-#endif /* -Wnested-anon-types */
-
-#if __has_warning("-Wconstant-logical-operand")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wconstant-logical-operand"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wconstant-logical-operand"
-#else
-#pragma warning disable "constant-logical-operand"
-#endif
-#endif /* -Wconstant-logical-operand */
-
-#if defined(__LCC__) && (__LCC__ <= 121)
-/* bug #2798 */
-#pragma diag_suppress alignment_reduction_ignored
-#elif defined(__ICC)
-#pragma warning(disable : 3453 1366)
-#elif __has_warning("-Walignment-reduction-ignored")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Walignment-reduction-ignored"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Walignment-reduction-ignored"
-#else
-#pragma warning disable "alignment-reduction-ignored"
-#endif
-#endif /* -Walignment-reduction-ignored */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(_WIN32) || defined(_WIN64)
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if !defined(_NO_CRT_STDIO_INLINE) && MDBX_BUILD_SHARED_LIBRARY && \
- !defined(xMDBX_TOOLS) && MDBX_WITHOUT_MSVC_CRT
-#define _NO_CRT_STDIO_INLINE
-#endif
-#elif !defined(_POSIX_C_SOURCE)
-#define _POSIX_C_SOURCE 200809L
-#endif /* Windows */
-
-/*----------------------------------------------------------------------------*/
-/* C99 includes */
-#include <inttypes.h>
-#include <stddef.h>
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <assert.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-/* C11' alignas() */
-#if __has_include(<stdalign.h>)
-#include <stdalign.h>
-#endif
-#if defined(alignas) || defined(__cplusplus)
-#define MDBX_ALIGNAS(N) alignas(N)
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
-#define MDBX_ALIGNAS(N) _Alignas(N)
-#elif defined(_MSC_VER)
-#define MDBX_ALIGNAS(N) __declspec(align(N))
-#elif __has_attribute(__aligned__) || defined(__GNUC__)
-#define MDBX_ALIGNAS(N) __attribute__((__aligned__(N)))
-#else
-#error "FIXME: Required alignas() or equivalent."
-#endif /* MDBX_ALIGNAS */
-
-/*----------------------------------------------------------------------------*/
-/* Systems includes */
-
-#ifdef __APPLE__
-#include <TargetConditionals.h>
-#endif /* Apple OSX & iOS */
-
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__)
-#include <sys/cdefs.h>
-#include <sys/mount.h>
-#include <sys/sysctl.h>
-#include <sys/types.h>
-#if defined(__FreeBSD__) || defined(__DragonFly__)
-#include <vm/vm_param.h>
-#elif defined(__OpenBSD__) || defined(__NetBSD__)
-#include <uvm/uvm_param.h>
-#else
-#define SYSCTL_LEGACY_NONCONST_MIB
-#endif
-#ifndef __MACH__
-#include <sys/vmmeter.h>
-#endif
-#else
-#include <malloc.h>
-#if !(defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(_WIN32) || defined(_WIN64))
-#include <mntent.h>
-#endif /* !Solaris */
-#endif /* !xBSD */
-
-#if defined(__FreeBSD__) || __has_include(<malloc_np.h>)
-#include <malloc_np.h>
-#endif
-
-#if defined(__APPLE__) || defined(__MACH__) || __has_include(<malloc/malloc.h>)
-#include <malloc/malloc.h>
-#endif /* MacOS */
-
-#if defined(__MACH__)
-#include <mach/host_info.h>
-#include <mach/mach_host.h>
-#include <mach/mach_port.h>
-#include <uuid/uuid.h>
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-#include <sched.h>
-#include <sys/sendfile.h>
-#include <sys/statfs.h>
-#endif /* Linux */
-
-#ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 0
-#endif
-
-#ifndef _XOPEN_SOURCE_EXTENDED
-#define _XOPEN_SOURCE_EXTENDED 0
-#else
-#include <utmpx.h>
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#include <kstat.h>
-#include <sys/mnttab.h>
-/* On Solaris, it's easier to add a missing prototype rather than find a
- * combination of #defines that break nothing. */
-__extern_C key_t ftok(const char *, int);
-#endif /* SunOS/Solaris */
-
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef _WIN32_WINNT
-#define _WIN32_WINNT 0x0601 /* Windows 7 */
-#elif _WIN32_WINNT < 0x0500
-#error At least 'Windows 2000' API is required for libmdbx.
-#endif /* _WIN32_WINNT */
-#if (defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* MinGW */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif /* WIN32_LEAN_AND_MEAN */
-#include <excpt.h>
-#include <tlhelp32.h>
-#include <windows.h>
-#include <winnt.h>
-#include <winternl.h>
-#define HAVE_SYS_STAT_H
-#define HAVE_SYS_TYPES_H
-typedef HANDLE mdbx_thread_t;
-typedef unsigned mdbx_thread_key_t;
-#define MAP_FAILED NULL
-#define HIGH_DWORD(v) ((DWORD)((sizeof(v) > 4) ? ((uint64_t)(v) >> 32) : 0))
-#define THREAD_CALL WINAPI
-#define THREAD_RESULT DWORD
-typedef struct {
- HANDLE mutex;
- HANDLE event[2];
-} mdbx_condpair_t;
-typedef CRITICAL_SECTION mdbx_fastmutex_t;
-
-#if !defined(_MSC_VER) && !defined(__try)
-/* *INDENT-OFF* */
-/* clang-format off */
-#define __try
-#define __except(COND) if(false)
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* stub for MSVC's __try/__except */
-
-#if MDBX_WITHOUT_MSVC_CRT
-
-#ifndef mdbx_malloc
-static inline void *mdbx_malloc(size_t bytes) {
- return HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_malloc */
-
-#ifndef mdbx_calloc
-static inline void *mdbx_calloc(size_t nelem, size_t size) {
- return HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, nelem * size);
-}
-#endif /* mdbx_calloc */
-
-#ifndef mdbx_realloc
-static inline void *mdbx_realloc(void *ptr, size_t bytes) {
- return ptr ? HeapReAlloc(GetProcessHeap(), 0, ptr, bytes)
- : HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_realloc */
-
-#ifndef mdbx_free
-static inline void mdbx_free(void *ptr) { HeapFree(GetProcessHeap(), 0, ptr); }
-#endif /* mdbx_free */
-
-#else /* MDBX_WITHOUT_MSVC_CRT */
-
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup _strdup
-
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-#ifndef snprintf
-#define snprintf _snprintf /* ntdll */
-#endif
-
-#ifndef vsnprintf
-#define vsnprintf _vsnprintf /* ntdll */
-#endif
-
-#else /*----------------------------------------------------------------------*/
-
-#include <unistd.h>
-#if !defined(_POSIX_MAPPED_FILES) || _POSIX_MAPPED_FILES < 1
-#error "libmdbx requires the _POSIX_MAPPED_FILES feature"
-#endif /* _POSIX_MAPPED_FILES */
-
-#include <pthread.h>
-#include <semaphore.h>
-#include <signal.h>
-#include <sys/file.h>
-#include <sys/ipc.h>
-#include <sys/mman.h>
-#include <sys/param.h>
-#include <sys/stat.h>
-#include <sys/statvfs.h>
-#include <sys/uio.h>
-typedef pthread_t mdbx_thread_t;
-typedef pthread_key_t mdbx_thread_key_t;
-#define INVALID_HANDLE_VALUE (-1)
-#define THREAD_CALL
-#define THREAD_RESULT void *
-typedef struct {
- pthread_mutex_t mutex;
- pthread_cond_t cond[2];
-} mdbx_condpair_t;
-typedef pthread_mutex_t mdbx_fastmutex_t;
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup strdup
-#endif /* Platform */
-
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
-/* malloc_usable_size() already provided */
-#elif defined(__APPLE__)
-#define malloc_usable_size(ptr) malloc_size(ptr)
-#elif defined(_MSC_VER) && !MDBX_WITHOUT_MSVC_CRT
-#define malloc_usable_size(ptr) _msize(ptr)
-#endif /* malloc_usable_size */
-
-#ifdef __ANDROID_API__
-#include <android/log.h>
-#if __ANDROID_API__ >= 21
-#include <sys/sendfile.h>
-#endif
-#endif /* Android */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(HAVE_SYS_STAT_H) || __has_include(<sys/stat.h>)
-#include <sys/stat.h>
-#endif
-#if defined(HAVE_SYS_TYPES_H) || __has_include(<sys/types.h>)
-#include <sys/types.h>
-#endif
-#if defined(HAVE_SYS_FILE_H) || __has_include(<sys/file.h>)
-#include <sys/file.h>
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-
-#ifndef SSIZE_MAX
-#define SSIZE_MAX INTPTR_MAX
-#endif
-
-#if !defined(MADV_DODUMP) && defined(MADV_CORE)
-#define MADV_DODUMP MADV_CORE
-#endif /* MADV_CORE -> MADV_DODUMP */
-
-#if !defined(MADV_DONTDUMP) && defined(MADV_NOCORE)
-#define MADV_DONTDUMP MADV_NOCORE
-#endif /* MADV_NOCORE -> MADV_DONTDUMP */
-
-#if defined(i386) || defined(__386) || defined(__i386) || defined(__i386__) || \
- defined(i486) || defined(__i486) || defined(__i486__) || \
- defined(i586) | defined(__i586) || defined(__i586__) || defined(i686) || \
- defined(__i686) || defined(__i686__) || defined(_M_IX86) || \
- defined(_X86_) || defined(__THW_INTEL__) || defined(__I86__) || \
- defined(__INTEL__) || defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64__) || defined(__amd64) || defined(_M_X64) || \
- defined(_M_AMD64) || defined(__IA32__) || defined(__INTEL__)
-#ifndef __ia32__
-/* LY: define neutral __ia32__ for x86 and x86-64 */
-#define __ia32__ 1
-#endif /* __ia32__ */
-#if !defined(__amd64__) && (defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64) || defined(_M_X64))
-/* LY: define trusty __amd64__ for all AMD64/x86-64 arch */
-#define __amd64__ 1
-#endif /* __amd64__ */
-#endif /* all x86 */
-
-#if (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
-#error \
- "Sanity checking failed: Two's complement, reasonably sized integer types"
-#endif
-
-#if UINTPTR_MAX > 0xffffFFFFul || ULONG_MAX > 0xffffFFFFul
-#define MDBX_WORDBITS 64
-#else
-#define MDBX_WORDBITS 32
-#endif /* MDBX_WORDBITS */
-
-#if defined(__ANDROID_API__) || defined(ANDROID)
-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "_FILE_OFFSET_BITS != MDBX_WORDBITS" (_FILE_OFFSET_BITS != MDBX_WORDBITS)
-#elif defined(__FILE_OFFSET_BITS) && __FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "__FILE_OFFSET_BITS != MDBX_WORDBITS" (__FILE_OFFSET_BITS != MDBX_WORDBITS)
-#endif
-#endif /* Android */
-
-/*----------------------------------------------------------------------------*/
-/* Compiler's includes for builtins/intrinsics */
-
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
-#include <intrin.h>
-#elif __GNUC_PREREQ(4, 4) || defined(__clang__)
-#if defined(__ia32__) || defined(__e2k__)
-#include <x86intrin.h>
-#endif /* __ia32__ */
-#if defined(__ia32__)
-#include <cpuid.h>
-#endif /* __ia32__ */
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
-#include <mbarrier.h>
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
-#include <machine/sys/inline.h>
-#elif defined(__IBMC__) && defined(__powerpc)
-#include <atomic.h>
-#elif defined(_AIX)
-#include <builtins.h>
-#include <sys/atomic_op.h>
-#elif (defined(__osf__) && defined(__DECC)) || defined(__alpha)
-#include <c_asm.h>
-#include <machine/builtins.h>
-#elif defined(__MWERKS__)
-/* CodeWarrior - troubles ? */
-#pragma gcc_extensions
-#elif defined(__SNC__)
-/* Sony PS3 - troubles ? */
-#elif defined(__hppa__) || defined(__hppa)
-#include <machine/inline.h>
-#else
-#error Unsupported C compiler, please use GNU C 4.4 or newer
-#endif /* Compiler */
-
-/*----------------------------------------------------------------------------*/
-/* Byteorder */
-
-#if !defined(__BYTE_ORDER__) || !defined(__ORDER_LITTLE_ENDIAN__) || \
- !defined(__ORDER_BIG_ENDIAN__)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(__GLIBC__) || defined(__GNU_LIBRARY__) || defined(__ANDROID_API__) || \
- defined(HAVE_ENDIAN_H) || __has_include(<endian.h>)
-#include <endian.h>
-#elif defined(__APPLE__) || defined(__MACH__) || defined(__OpenBSD__) || \
- defined(HAVE_MACHINE_ENDIAN_H) || __has_include(<machine/endian.h>)
-#include <machine/endian.h>
-#elif defined(HAVE_SYS_ISA_DEFS_H) || __has_include(<sys/isa_defs.h>)
-#include <sys/isa_defs.h>
-#elif (defined(HAVE_SYS_TYPES_H) && defined(HAVE_SYS_ENDIAN_H)) || \
- (__has_include(<sys/types.h>) && __has_include(<sys/endian.h>))
-#include <sys/endian.h>
-#include <sys/types.h>
-#elif defined(__bsdi__) || defined(__DragonFly__) || defined(__FreeBSD__) || \
- defined(__NetBSD__) || \
- defined(HAVE_SYS_PARAM_H) || __has_include(<sys/param.h>)
-#include <sys/param.h>
-#endif /* OS */
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-
-#else
-#error __BYTE_ORDER__ should be defined.
-#endif /* Arch */
-
-#endif
-#endif /* __BYTE_ORDER__ || __ORDER_LITTLE_ENDIAN__ || __ORDER_BIG_ENDIAN__ */
-
-/* Get the size of a memory page for the system.
- * This is the basic size that the platform's memory manager uses, and is
- * fundamental to the use of memory-mapped files. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline size_t
-mdbx_syspagesize(void) {
-#if defined(_WIN32) || defined(_WIN64)
- SYSTEM_INFO si;
- GetSystemInfo(&si);
- return si.dwPageSize;
-#else
- return sysconf(_SC_PAGE_SIZE);
-#endif
-}
-
-typedef struct mdbx_mmap_param {
- union {
- void *address;
- uint8_t *dxb;
- struct MDBX_lockinfo *lck;
- };
- mdbx_filehandle_t fd;
- size_t limit; /* mapping length, but NOT a size of file nor DB */
- size_t current; /* mapped region size, i.e. the size of file and DB */
- uint64_t filesize /* in-process cache of a file size */;
-#if defined(_WIN32) || defined(_WIN64)
- HANDLE section; /* memory-mapped section handle */
-#endif
-} mdbx_mmap_t;
-
-typedef union bin128 {
- __anonymous_struct_extension__ struct { uint64_t x, y; };
- __anonymous_struct_extension__ struct { uint32_t a, b, c, d; };
-} bin128_t;
-
-#if defined(_WIN32) || defined(_WIN64)
-typedef union MDBX_srwlock {
- __anonymous_struct_extension__ struct {
- long volatile readerCount;
- long volatile writerCount;
- };
- RTL_SRWLOCK native;
-} MDBX_srwlock;
-#endif /* Windows */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny);
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny);
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-#if defined(__cplusplus) && !defined(__STDC_NO_ATOMICS__) && (__has_include(<cstdatomic>) || __has_extension(cxx_atomic))
-#include <cstdatomic>
-#define MDBX_HAVE_C11ATOMICS
-#elif !defined(__cplusplus) && \
- (__STDC_VERSION__ >= 201112L || __has_extension(c_atomic)) && \
- !defined(__STDC_NO_ATOMICS__) && \
- (__GNUC_PREREQ(4, 9) || __CLANG_PREREQ(3, 8) || \
- !(defined(__GNUC__) || defined(__clang__)))
-#include <stdatomic.h>
-#define MDBX_HAVE_C11ATOMICS
-#elif defined(__GNUC__) || defined(__clang__)
-#elif defined(_MSC_VER)
-#pragma warning(disable : 4163) /* 'xyz': not available as an intrinsic */
-#pragma warning(disable : 4133) /* 'function': incompatible types - from \
- 'size_t' to 'LONGLONG' */
-#pragma warning(disable : 4244) /* 'return': conversion from 'LONGLONG' to \
- 'std::size_t', possible loss of data */
-#pragma warning(disable : 4267) /* 'function': conversion from 'size_t' to \
- 'long', possible loss of data */
-#pragma intrinsic(_InterlockedExchangeAdd, _InterlockedCompareExchange)
-#pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedCompareExchange64)
-#elif defined(__APPLE__)
-#include <libkern/OSAtomic.h>
-#else
-#error FIXME atomic-ops
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Memory/Compiler barriers, cache coherence */
-
-#if __has_include(<sys/cachectl.h>)
-#include <sys/cachectl.h>
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS should have explicit cache control */
-#include <sys/cachectl.h>
-#endif
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_compiler_barrier(void) {
-#if defined(__clang__) || defined(__GNUC__)
- __asm__ __volatile__("" ::: "memory");
-#elif defined(_MSC_VER)
- _ReadWriteBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
- __memory_barrier();
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __compiler_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_sched_fence(/* LY: no-arg meaning 'all expect ALU', e.g. 0x3D3D */);
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __fence();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_memory_barrier(void) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(memory_order_seq_cst);
-#elif defined(__ATOMIC_SEQ_CST)
-#ifdef __clang__
- __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
- __atomic_thread_fence(__ATOMIC_SEQ_CST);
-#endif
-#elif defined(__clang__) || defined(__GNUC__)
- __sync_synchronize();
-#elif defined(_WIN32) || defined(_WIN64)
- MemoryBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
-#if defined(__ia32__)
- _mm_mfence();
-#else
- __mf();
-#endif
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __machine_rw_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_mf();
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __lwsync();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* libc compatibility stuff */
-
-#if (!defined(__GLIBC__) && __GLIBC_PREREQ(2, 1)) && \
- (defined(_GNU_SOURCE) || defined(_BSD_SOURCE))
-#define mdbx_asprintf asprintf
-#define mdbx_vasprintf vasprintf
-#else
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC
- MDBX_PRINTF_ARGS(2, 3) int mdbx_asprintf(char **strp, const char *fmt, ...);
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt, va_list ap);
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* OS abstraction layer stuff */
-
-/* max bytes to write in one call */
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_WRITE UINT32_C(0x01000000)
-#else
-#define MAX_WRITE UINT32_C(0x3fff0000)
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-MDBX_INTERNAL_VAR uint32_t mdbx_linux_kernel_version;
-MDBX_INTERNAL_VAR bool mdbx_RunningOnWSL1 /* Windows Subsystem 1 for Linux */;
-#endif /* Linux */
-
-#ifndef mdbx_strdup
-LIBMDBX_API char *mdbx_strdup(const char *str);
-#endif
-
-MDBX_MAYBE_UNUSED static __inline int mdbx_get_errno(void) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD rc = GetLastError();
-#else
- int rc = errno;
-#endif
- return rc;
-}
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result);
-#endif
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr);
-#endif
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair, bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair);
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex);
-
-MDBX_INTERNAL_FUNC int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov,
- int iovcnt, uint64_t offset,
- size_t expected_written);
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t count,
- uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t count, uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t count);
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg);
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread);
-
-enum mdbx_syncmode_bits {
- MDBX_SYNC_NONE = 0,
- MDBX_SYNC_DATA = 1,
- MDBX_SYNC_SIZE = 2,
- MDBX_SYNC_IODQ = 4
-};
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- const enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length);
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos);
-MDBX_INTERNAL_FUNC int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length);
-
-enum mdbx_openfile_purpose {
- MDBX_OPEN_DXB_READ = 0,
- MDBX_OPEN_DXB_LAZY = 1,
- MDBX_OPEN_DXB_DSYNC = 2,
- MDBX_OPEN_LCK = 3,
- MDBX_OPEN_COPY = 4,
- MDBX_OPEN_DELETE = 5
-};
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait);
-
-#define MMAP_OPTION_TRUNCATE 1
-#define MMAP_OPTION_SEMAPHORE 2
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t must, const size_t limit,
- const unsigned options);
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map);
-#define MDBX_MRESIZE_MAY_MOVE 0x00000100
-#define MDBX_MRESIZE_MAY_UNMAP 0x00000200
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit);
-#if defined(_WIN32) || defined(_WIN64)
-typedef struct {
- unsigned limit, count;
- HANDLE handles[31];
-} mdbx_handle_array_t;
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array);
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array);
-#endif /* Windows */
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err);
-
-MDBX_MAYBE_UNUSED static __inline uint32_t mdbx_getpid(void) {
- STATIC_ASSERT(sizeof(mdbx_pid_t) <= sizeof(uint32_t));
-#if defined(_WIN32) || defined(_WIN64)
- return GetCurrentProcessId();
-#else
- STATIC_ASSERT(sizeof(pid_t) <= sizeof(uint32_t));
- return getpid();
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline uintptr_t mdbx_thread_self(void) {
- mdbx_tid_t thunk;
- STATIC_ASSERT(sizeof(uintptr_t) >= sizeof(thunk));
-#if defined(_WIN32) || defined(_WIN64)
- thunk = GetCurrentThreadId();
-#else
- thunk = pthread_self();
-#endif
- return (uintptr_t)thunk;
-}
-
-#if !defined(_WIN32) && !defined(_WIN64)
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void);
-#else
-static __inline int mdbx_check_tid4bionic(void) { return 0; }
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-MDBX_MAYBE_UNUSED static __inline int
-mdbx_pthread_mutex_lock(pthread_mutex_t *mutex) {
- int err = mdbx_check_tid4bionic();
- return unlikely(err) ? err : pthread_mutex_lock(mutex);
-}
-#endif /* !Windows */
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void);
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16);
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime);
-
-MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void);
-/*----------------------------------------------------------------------------*/
-/* lck stuff */
-
-/// \brief Initialization of synchronization primitives linked with MDBX_env
-/// instance both in LCK-file and within the current process.
-/// \param
-/// global_uniqueness_flag = true - denotes that there are no other processes
-/// working with DB and LCK-file. Thus the function MUST initialize
-/// shared synchronization objects in memory-mapped LCK-file.
-/// global_uniqueness_flag = false - denotes that at least one process is
-/// already working with DB and LCK-file, including the case when DB
-/// has already been opened in the current process. Thus the function
-/// MUST NOT initialize shared synchronization objects in memory-mapped
-/// LCK-file that are already in use.
-/// \return Error code or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag);
-
-/// \brief Disconnects from shared interprocess objects and destructs
-/// synchronization objects linked with MDBX_env instance
-/// within the current process.
-/// \param
-/// inprocess_neighbor = NULL - if the current process does not have other
-/// instances of MDBX_env linked with the DB being closed.
-/// Thus the function MUST check for other processes working with DB or
-/// LCK-file, and keep or destroy shared synchronization objects in
-/// memory-mapped LCK-file depending on the result.
-/// inprocess_neighbor = not-NULL - pointer to another instance of MDBX_env
-/// (anyone of there is several) working with DB or LCK-file within the
-/// current process. Thus the function MUST NOT try to acquire exclusive
-/// lock and/or try to destruct shared synchronization objects linked with
-/// DB or LCK-file. Moreover, the implementation MUST ensure correct work
-/// of other instances of MDBX_env within the current process, e.g.
-/// restore POSIX-fcntl locks after the closing of file descriptors.
-/// \return Error code (MDBX_PANIC) or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor);
-
-/// \brief Connects to shared interprocess locking objects and tries to acquire
-/// the maximum lock level (shared if exclusive is not available)
-/// Depending on implementation or/and platform (Windows) this function may
-/// acquire the non-OS super-level lock (e.g. for shared synchronization
-/// objects initialization), which will be downgraded to OS-exclusive or
-/// shared via explicit calling of mdbx_lck_downgrade().
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if an exclusive lock was acquired and thus
-/// the current process is the first and only after the last use of DB.
-/// MDBX_RESULT_FALSE (0) - if a shared lock was acquired and thus
-/// DB has already been opened and now is used by other processes.
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env);
-
-/// \brief Downgrades the level of initially acquired lock to
-/// operational level specified by argument. The reson for such downgrade:
-/// - unblocking of other processes that are waiting for access, i.e.
-/// if (env->me_flags & MDBX_EXCLUSIVE) != 0, then other processes
-/// should be made aware that access is unavailable rather than
-/// wait for it.
-/// - freeing locks that interfere file operation (especially for Windows)
-/// (env->me_flags & MDBX_EXCLUSIVE) == 0 - downgrade to shared lock.
-/// (env->me_flags & MDBX_EXCLUSIVE) != 0 - downgrade to exclusive
-/// operational lock.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env);
-
-/// \brief Locks LCK-file or/and table of readers for (de)registering.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env);
-
-/// \brief Unlocks LCK-file or/and table of readers after (de)registering.
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env);
-
-/// \brief Acquires lock for DB change (on writing transaction start)
-/// Reading transactions will not be blocked.
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-/// \return Error code or zero on success
-LIBMDBX_API int mdbx_txn_lock(MDBX_env *env, bool dont_wait);
-
-/// \brief Releases lock once DB changes is made (after writing transaction
-/// has finished).
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-LIBMDBX_API void mdbx_txn_unlock(MDBX_env *env);
-
-/// \brief Sets alive-flag of reader presence (indicative lock) for PID of
-/// the current process. The function does no more than needed for
-/// the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env);
-
-/// \brief Resets alive-flag of reader presence (indicative lock)
-/// for PID of the current process. The function does no more than needed
-/// for the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env);
-
-/// \brief Checks for reading process status with the given pid with help of
-/// alive-flag of presence (indicative lock) or using another way.
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if the reader process with the given PID is alive
-/// and working with DB (indicative lock is present).
-/// MDBX_RESULT_FALSE (0) - if the reader process with the given PID is absent
-/// or not working with DB (indicative lock is not present).
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid);
-
-#if defined(_WIN32) || defined(_WIN64)
-
-typedef void(WINAPI *MDBX_srwlock_function)(MDBX_srwlock *);
-MDBX_INTERNAL_VAR MDBX_srwlock_function mdbx_srwlock_Init,
- mdbx_srwlock_AcquireShared, mdbx_srwlock_ReleaseShared,
- mdbx_srwlock_AcquireExclusive, mdbx_srwlock_ReleaseExclusive;
-
-#if _WIN32_WINNT < 0x0600 /* prior to Windows Vista */
-typedef enum _FILE_INFO_BY_HANDLE_CLASS {
- FileBasicInfo,
- FileStandardInfo,
- FileNameInfo,
- FileRenameInfo,
- FileDispositionInfo,
- FileAllocationInfo,
- FileEndOfFileInfo,
- FileStreamInfo,
- FileCompressionInfo,
- FileAttributeTagInfo,
- FileIdBothDirectoryInfo,
- FileIdBothDirectoryRestartInfo,
- FileIoPriorityHintInfo,
- FileRemoteProtocolInfo,
- MaximumFileInfoByHandleClass
-} FILE_INFO_BY_HANDLE_CLASS,
- *PFILE_INFO_BY_HANDLE_CLASS;
-
-typedef struct _FILE_END_OF_FILE_INFO {
- LARGE_INTEGER EndOfFile;
-} FILE_END_OF_FILE_INFO, *PFILE_END_OF_FILE_INFO;
-
-#define REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK 0x00000001
-#define REMOTE_PROTOCOL_INFO_FLAG_OFFLINE 0x00000002
-
-typedef struct _FILE_REMOTE_PROTOCOL_INFO {
- USHORT StructureVersion;
- USHORT StructureSize;
- DWORD Protocol;
- USHORT ProtocolMajorVersion;
- USHORT ProtocolMinorVersion;
- USHORT ProtocolRevision;
- USHORT Reserved;
- DWORD Flags;
- struct {
- DWORD Reserved[8];
- } GenericReserved;
- struct {
- DWORD Reserved[16];
- } ProtocolSpecificReserved;
-} FILE_REMOTE_PROTOCOL_INFO, *PFILE_REMOTE_PROTOCOL_INFO;
-
-#endif /* _WIN32_WINNT < 0x0600 (prior to Windows Vista) */
-
-typedef BOOL(WINAPI *MDBX_GetFileInformationByHandleEx)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_GetFileInformationByHandleEx
- mdbx_GetFileInformationByHandleEx;
-
-typedef BOOL(WINAPI *MDBX_GetVolumeInformationByHandleW)(
- _In_ HANDLE hFile, _Out_opt_ LPWSTR lpVolumeNameBuffer,
- _In_ DWORD nVolumeNameSize, _Out_opt_ LPDWORD lpVolumeSerialNumber,
- _Out_opt_ LPDWORD lpMaximumComponentLength,
- _Out_opt_ LPDWORD lpFileSystemFlags,
- _Out_opt_ LPWSTR lpFileSystemNameBuffer, _In_ DWORD nFileSystemNameSize);
-MDBX_INTERNAL_VAR MDBX_GetVolumeInformationByHandleW
- mdbx_GetVolumeInformationByHandleW;
-
-typedef DWORD(WINAPI *MDBX_GetFinalPathNameByHandleW)(_In_ HANDLE hFile,
- _Out_ LPWSTR lpszFilePath,
- _In_ DWORD cchFilePath,
- _In_ DWORD dwFlags);
-MDBX_INTERNAL_VAR MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-
-typedef BOOL(WINAPI *MDBX_SetFileInformationByHandle)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_SetFileInformationByHandle
- mdbx_SetFileInformationByHandle;
-
-typedef NTSTATUS(NTAPI *MDBX_NtFsControlFile)(
- IN HANDLE FileHandle, IN OUT HANDLE Event,
- IN OUT PVOID /* PIO_APC_ROUTINE */ ApcRoutine, IN OUT PVOID ApcContext,
- OUT PIO_STATUS_BLOCK IoStatusBlock, IN ULONG FsControlCode,
- IN OUT PVOID InputBuffer, IN ULONG InputBufferLength,
- OUT OPTIONAL PVOID OutputBuffer, IN ULONG OutputBufferLength);
-MDBX_INTERNAL_VAR MDBX_NtFsControlFile mdbx_NtFsControlFile;
-
-typedef uint64_t(WINAPI *MDBX_GetTickCount64)(void);
-MDBX_INTERNAL_VAR MDBX_GetTickCount64 mdbx_GetTickCount64;
-
-#if !defined(_WIN32_WINNT_WIN8) || _WIN32_WINNT < _WIN32_WINNT_WIN8
-typedef struct _WIN32_MEMORY_RANGE_ENTRY {
- PVOID VirtualAddress;
- SIZE_T NumberOfBytes;
-} WIN32_MEMORY_RANGE_ENTRY, *PWIN32_MEMORY_RANGE_ENTRY;
-#endif /* Windows 8.x */
-
-typedef BOOL(WINAPI *MDBX_PrefetchVirtualMemory)(
- HANDLE hProcess, ULONG_PTR NumberOfEntries,
- PWIN32_MEMORY_RANGE_ENTRY VirtualAddresses, ULONG Flags);
-MDBX_INTERNAL_VAR MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-
-typedef enum _SECTION_INHERIT { ViewShare = 1, ViewUnmap = 2 } SECTION_INHERIT;
-
-typedef NTSTATUS(NTAPI *MDBX_NtExtendSection)(IN HANDLE SectionHandle,
- IN PLARGE_INTEGER NewSectionSize);
-MDBX_INTERNAL_VAR MDBX_NtExtendSection mdbx_NtExtendSection;
-
-static __inline bool mdbx_RunningUnderWine(void) {
- return !mdbx_NtExtendSection;
-}
-
-typedef LSTATUS(WINAPI *MDBX_RegGetValueA)(HKEY hkey, LPCSTR lpSubKey,
- LPCSTR lpValue, DWORD dwFlags,
- LPDWORD pdwType, PVOID pvData,
- LPDWORD pcbData);
-MDBX_INTERNAL_VAR MDBX_RegGetValueA mdbx_RegGetValueA;
-
-NTSYSAPI ULONG RtlRandomEx(PULONG Seed);
-
-#endif /* Windows */
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_MSC_VER) && _MSC_VER >= 1900
-/* LY: MSVC 2015/2017/2019 has buggy/inconsistent PRIuPTR/PRIxPTR macros
- * for internal format-args checker. */
-#undef PRIuPTR
-#undef PRIiPTR
-#undef PRIdPTR
-#undef PRIxPTR
-#define PRIuPTR "Iu"
-#define PRIiPTR "Ii"
-#define PRIdPTR "Id"
-#define PRIxPTR "Ix"
-#define PRIuSIZE "zu"
-#define PRIiSIZE "zi"
-#define PRIdSIZE "zd"
-#define PRIxSIZE "zx"
-#endif /* fix PRI*PTR for _MSC_VER */
-
-#ifndef PRIuSIZE
-#define PRIuSIZE PRIuPTR
-#define PRIiSIZE PRIiPTR
-#define PRIdSIZE PRIdPTR
-#define PRIxSIZE PRIxPTR
-#endif /* PRI*SIZE macros for MSVC */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-#define mdbx_sourcery_anchor XCONCAT(mdbx_sourcery_, MDBX_BUILD_SOURCERY)
-#if defined(xMDBX_TOOLS)
-extern LIBMDBX_API const char *const mdbx_sourcery_anchor;
-#endif
-
-/*******************************************************************************
- *******************************************************************************
- *******************************************************************************
- *
- *
- * #### ##### ##### # #### # # ####
- * # # # # # # # # ## # #
- * # # # # # # # # # # # ####
- * # # ##### # # # # # # # #
- * # # # # # # # # ## # #
- * #### # # # #### # # ####
- *
- *
- */
-
-/** \defgroup build_option Build options
- * The libmdbx build options.
- @{ */
-
-/** Using fcntl(F_FULLFSYNC) with 5-10 times slowdown */
-#define MDBX_OSX_WANNA_DURABILITY 0
-/** Using fsync() with chance of data lost on power failure */
-#define MDBX_OSX_WANNA_SPEED 1
-
-#ifndef MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-/** Choices \ref MDBX_OSX_WANNA_DURABILITY or \ref MDBX_OSX_WANNA_SPEED
- * for OSX & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY MDBX_OSX_WANNA_DURABILITY
-#endif /* MDBX_OSX_SPEED_INSTEADOF_DURABILITY */
-
-/** Controls checking PID against reuse DB environment after the fork() */
-#ifndef MDBX_ENV_CHECKPID
-#if defined(MADV_DONTFORK) || defined(_WIN32) || defined(_WIN64)
-/* PID check could be omitted:
- * - on Linux when madvise(MADV_DONTFORK) is available, i.e. after the fork()
- * mapped pages will not be available for child process.
- * - in Windows where fork() not available. */
-#define MDBX_ENV_CHECKPID 0
-#else
-#define MDBX_ENV_CHECKPID 1
-#endif
-#define MDBX_ENV_CHECKPID_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#else
-#define MDBX_ENV_CHECKPID_CONFIG MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#endif /* MDBX_ENV_CHECKPID */
-
-/** Controls checking transaction owner thread against misuse transactions from
- * other threads. */
-#ifndef MDBX_TXN_CHECKOWNER
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_TXN_CHECKOWNER_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#else
-#define MDBX_TXN_CHECKOWNER_CONFIG MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#endif /* MDBX_TXN_CHECKOWNER */
-
-/** Does a system have battery-backed Real-Time Clock or just a fake. */
-#ifndef MDBX_TRUST_RTC
-#if defined(__linux__) || defined(__gnu_linux__) || defined(__NetBSD__) || \
- defined(__OpenBSD__)
-#define MDBX_TRUST_RTC 0 /* a lot of embedded systems have a fake RTC */
-#else
-#define MDBX_TRUST_RTC 1
-#endif
-#define MDBX_TRUST_RTC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#else
-#define MDBX_TRUST_RTC_CONFIG MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#endif /* MDBX_TRUST_RTC */
-
-/** Controls online database auto-compactification during write-transactions. */
-#ifndef MDBX_ENABLE_REFUND
-#define MDBX_ENABLE_REFUND 1
-#elif !(MDBX_ENABLE_REFUND == 0 || MDBX_ENABLE_REFUND == 1)
-#error MDBX_ENABLE_REFUND must be defined as 0 or 1
-#endif /* MDBX_ENABLE_REFUND */
-
-/** Controls gathering statistics for page operations. */
-#ifndef MDBX_ENABLE_PGOP_STAT
-#define MDBX_ENABLE_PGOP_STAT 1
-#elif !(MDBX_ENABLE_PGOP_STAT == 0 || MDBX_ENABLE_PGOP_STAT == 1)
-#error MDBX_ENABLE_PGOP_STAT must be defined as 0 or 1
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-/** Controls use of POSIX madvise() hints and friends. */
-#ifndef MDBX_ENABLE_MADVISE
-#define MDBX_ENABLE_MADVISE 1
-#elif !(MDBX_ENABLE_MADVISE == 0 || MDBX_ENABLE_MADVISE == 1)
-#error MDBX_ENABLE_MADVISE must be defined as 0 or 1
-#endif /* MDBX_ENABLE_MADVISE */
-
-/** Disable some checks to reduce an overhead and detection probability of
- * database corruption to a values closer to the LMDB. */
-#ifndef MDBX_DISABLE_PAGECHECKS
-#define MDBX_DISABLE_PAGECHECKS 0
-#elif !(MDBX_DISABLE_PAGECHECKS == 0 || MDBX_DISABLE_PAGECHECKS == 1)
-#error MDBX_DISABLE_PAGECHECKS must be defined as 0 or 1
-#endif /* MDBX_DISABLE_PAGECHECKS */
-
-#ifndef MDBX_PNL_PREALLOC_FOR_RADIXSORT
-#define MDBX_PNL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_PNL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_PNL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_PNL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
-
-#ifndef MDBX_DPL_PREALLOC_FOR_RADIXSORT
-#define MDBX_DPL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_DPL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_DPL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_DPL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
-
-/* Basically, this build-option is for TODO. Guess it should be replaced
- * with MDBX_ENABLE_WRITEMAP_SPILLING with the three variants:
- * 0/OFF = Don't track dirty pages at all and don't spilling ones.
- * This should be by-default on Linux and may-be other systems
- * (not sure: Darwin/OSX, FreeBSD, Windows 10) where kernel provides
- * properly LRU tracking and async writing on-demand.
- * 1/ON = Lite tracking of dirty pages but with LRU labels and explicit
- * spilling with msync(MS_ASYNC). */
-#ifndef MDBX_FAKE_SPILL_WRITEMAP
-#if defined(__linux__) || defined(__gnu_linux__)
-#define MDBX_FAKE_SPILL_WRITEMAP 1 /* msync(MS_ASYNC) is no-op on Linux */
-#else
-#define MDBX_FAKE_SPILL_WRITEMAP 0
-#endif
-#elif !(MDBX_FAKE_SPILL_WRITEMAP == 0 || MDBX_FAKE_SPILL_WRITEMAP == 1)
-#error MDBX_FAKE_SPILL_WRITEMAP must be defined as 0 or 1
-#endif /* MDBX_FAKE_SPILL_WRITEMAP */
-
-/** Controls sort order of internal page number lists.
- * This mostly experimental/advanced option with not for regular MDBX users.
- * \warning The database format depend on this option and libmdbx builded with
- * different option value are incompatible. */
-#ifndef MDBX_PNL_ASCENDING
-#define MDBX_PNL_ASCENDING 0
-#elif !(MDBX_PNL_ASCENDING == 0 || MDBX_PNL_ASCENDING == 1)
-#error MDBX_PNL_ASCENDING must be defined as 0 or 1
-#endif /* MDBX_PNL_ASCENDING */
-
-/** Avoid dependence from MSVC CRT and use ntdll.dll instead. */
-#ifndef MDBX_WITHOUT_MSVC_CRT
-#define MDBX_WITHOUT_MSVC_CRT 1
-#elif !(MDBX_WITHOUT_MSVC_CRT == 0 || MDBX_WITHOUT_MSVC_CRT == 1)
-#error MDBX_WITHOUT_MSVC_CRT must be defined as 0 or 1
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-/** Size of buffer used during copying a environment/database file. */
-#ifndef MDBX_ENVCOPY_WRITEBUF
-#define MDBX_ENVCOPY_WRITEBUF 1048576u
-#elif MDBX_ENVCOPY_WRITEBUF < 65536u || MDBX_ENVCOPY_WRITEBUF > 1073741824u || \
- MDBX_ENVCOPY_WRITEBUF % 65536u
-#error MDBX_ENVCOPY_WRITEBUF must be defined in range 65536..1073741824 and be multiple of 65536
-#endif /* MDBX_ENVCOPY_WRITEBUF */
-
-/** Forces assertion checking */
-#ifndef MDBX_FORCE_ASSERTIONS
-#define MDBX_FORCE_ASSERTIONS 0
-#elif !(MDBX_FORCE_ASSERTIONS == 0 || MDBX_FORCE_ASSERTIONS == 1)
-#error MDBX_FORCE_ASSERTIONS must be defined as 0 or 1
-#endif /* MDBX_FORCE_ASSERTIONS */
-
-/** Presumed malloc size overhead for each allocation
- * to adjust allocations to be more aligned. */
-#ifndef MDBX_ASSUME_MALLOC_OVERHEAD
-#ifdef __SIZEOF_POINTER__
-#define MDBX_ASSUME_MALLOC_OVERHEAD (__SIZEOF_POINTER__ * 2u)
-#else
-#define MDBX_ASSUME_MALLOC_OVERHEAD (sizeof(void *) * 2u)
-#endif
-#elif MDBX_ASSUME_MALLOC_OVERHEAD < 0 || MDBX_ASSUME_MALLOC_OVERHEAD > 64 || \
- MDBX_ASSUME_MALLOC_OVERHEAD % 4
-#error MDBX_ASSUME_MALLOC_OVERHEAD must be defined in range 0..64 and be multiple of 4
-#endif /* MDBX_ASSUME_MALLOC_OVERHEAD */
-
-/** In case the MDBX_DEBUG is undefined set it corresponding to NDEBUG */
-#ifndef MDBX_DEBUG
-#ifdef NDEBUG
-#define MDBX_DEBUG 0
-#else
-#define MDBX_DEBUG 1
-#endif
-#endif /* MDBX_DEBUG */
-
-/** If defined then enables integration with Valgrind,
- * a memory analyzing tool. */
-#ifndef MDBX_USE_VALGRIND
-#endif /* MDBX_USE_VALGRIND */
-
-/** If defined then enables use C11 atomics,
- * otherwise detects ones availability automatically. */
-#ifndef MDBX_HAVE_C11ATOMICS
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-//------------------------------------------------------------------------------
-
-/** Win32 File Locking API for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_WIN32FILES -1
-
-/** SystemV IPC semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_SYSV 5
-
-/** POSIX-1 Shared anonymous semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX1988 1988
-
-/** POSIX-2001 Shared Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2001 2001
-
-/** POSIX-2008 Robust Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2008 2008
-
-/** BeOS Benaphores, aka Futexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_BENAPHORE 1995
-
-/** Advanced: Choices the locking implementation (autodetection by default). */
-#if defined(_WIN32) || defined(_WIN64)
-#define MDBX_LOCKING MDBX_LOCKING_WIN32FILES
-#else
-#ifndef MDBX_LOCKING
-#if defined(_POSIX_THREAD_PROCESS_SHARED) && \
- _POSIX_THREAD_PROCESS_SHARED >= 200112L && !defined(__FreeBSD__)
-
-/* Some platforms define the EOWNERDEAD error code even though they
- * don't support Robust Mutexes. If doubt compile with -MDBX_LOCKING=2001. */
-#if defined(EOWNERDEAD) && _POSIX_THREAD_PROCESS_SHARED >= 200809L && \
- ((defined(_POSIX_THREAD_ROBUST_PRIO_INHERIT) && \
- _POSIX_THREAD_ROBUST_PRIO_INHERIT > 0) || \
- (defined(_POSIX_THREAD_ROBUST_PRIO_PROTECT) && \
- _POSIX_THREAD_ROBUST_PRIO_PROTECT > 0) || \
- defined(PTHREAD_MUTEX_ROBUST) || defined(PTHREAD_MUTEX_ROBUST_NP)) && \
- (!defined(__GLIBC__) || \
- __GLIBC_PREREQ(2, 10) /* troubles with Robust mutexes before 2.10 */)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2008
-#else
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2001
-#endif
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX1988
-#else
-#define MDBX_LOCKING MDBX_LOCKING_SYSV
-#endif
-#define MDBX_LOCKING_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_LOCKING)
-#else
-#define MDBX_LOCKING_CONFIG MDBX_STRINGIFY(MDBX_LOCKING)
-#endif /* MDBX_LOCKING */
-#endif /* !Windows */
-
-/** Advanced: Using POSIX OFD-locks (autodetection by default). */
-#ifndef MDBX_USE_OFDLOCKS
-#if defined(F_OFD_SETLK) && defined(F_OFD_SETLKW) && defined(F_OFD_GETLK) && \
- !defined(MDBX_SAFE4QEMU) && \
- !defined(__sun) /* OFD-lock are broken on Solaris */
-#define MDBX_USE_OFDLOCKS 1
-#else
-#define MDBX_USE_OFDLOCKS 0
-#endif
-#define MDBX_USE_OFDLOCKS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#else
-#define MDBX_USE_OFDLOCKS_CONFIG MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#endif /* MDBX_USE_OFDLOCKS */
-
-/** Advanced: Using sendfile() syscall (autodetection by default). */
-#ifndef MDBX_USE_SENDFILE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 21)
-#define MDBX_USE_SENDFILE 1
-#else
-#define MDBX_USE_SENDFILE 0
-#endif
-#endif /* MDBX_USE_SENDFILE */
-
-/** Advanced: Using copy_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_COPYFILERANGE
-#if __GLIBC_PREREQ(2, 27) && defined(_GNU_SOURCE)
-#define MDBX_USE_COPYFILERANGE 1
-#else
-#define MDBX_USE_COPYFILERANGE 0
-#endif
-#endif /* MDBX_USE_COPYFILERANGE */
-
-/** Advanced: Using sync_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_SYNCFILERANGE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- defined(SYNC_FILE_RANGE_WRITE) && !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 26)
-#define MDBX_USE_SYNCFILERANGE 1
-#else
-#define MDBX_USE_SYNCFILERANGE 0
-#endif
-#endif /* MDBX_USE_SYNCFILERANGE */
-
-//------------------------------------------------------------------------------
-
-#ifndef MDBX_CPU_WRITEBACK_INCOHERENT
-#if defined(__ia32__) || defined(__e2k__) || defined(__hppa) || \
- defined(__hppa__) || defined(DOXYGEN)
-#define MDBX_CPU_WRITEBACK_INCOHERENT 0
-#else
-#define MDBX_CPU_WRITEBACK_INCOHERENT 1
-#endif
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-#ifndef MDBX_MMAP_INCOHERENT_FILE_WRITE
-#ifdef __OpenBSD__
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 1
-#else
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#ifndef MDBX_MMAP_INCOHERENT_CPU_CACHE
-#if defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS has cache coherency issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 1
-#else
-/* LY: assume no relevant mmap/dcache issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#ifndef MDBX_64BIT_ATOMIC
-#if MDBX_WORDBITS >= 64 || defined(DOXYGEN)
-#define MDBX_64BIT_ATOMIC 1
-#else
-#define MDBX_64BIT_ATOMIC 0
-#endif
-#define MDBX_64BIT_ATOMIC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#else
-#define MDBX_64BIT_ATOMIC_CONFIG MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#endif /* MDBX_64BIT_ATOMIC */
-
-#ifndef MDBX_64BIT_CAS
-#if defined(ATOMIC_LLONG_LOCK_FREE)
-#if ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__GCC_ATOMIC_LLONG_LOCK_FREE)
-#if __GCC_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__CLANG_ATOMIC_LLONG_LOCK_FREE)
-#if __CLANG_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(_MSC_VER) || defined(__APPLE__) || defined(DOXYGEN)
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS MDBX_64BIT_ATOMIC
-#endif
-#define MDBX_64BIT_CAS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#else
-#define MDBX_64BIT_CAS_CONFIG MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#endif /* MDBX_64BIT_CAS */
-
-#ifndef MDBX_UNALIGNED_OK
-#if defined(__ALIGNED__) || defined(__SANITIZE_UNDEFINED__)
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#elif defined(__ARM_FEATURE_UNALIGNED)
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#elif __CLANG_PREREQ(5, 0) || __GNUC_PREREQ(5, 0)
-/* expecting an optimization will well done, also this
- * hushes false-positives from UBSAN (undefined behaviour sanitizer) */
-#define MDBX_UNALIGNED_OK 0
-#elif defined(__e2k__) || defined(__elbrus__)
-#if __iset__ > 4
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#endif
-#elif defined(__ia32__)
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#endif
-#elif MDBX_UNALIGNED_OK == 1
-#undef MDBX_UNALIGNED_OK
-#define MDBX_UNALIGNED_OK 32 /* any unaligned access allowed */
-#endif /* MDBX_UNALIGNED_OK */
-
-#ifndef MDBX_CACHELINE_SIZE
-#if defined(SYSTEM_CACHE_ALIGNMENT_SIZE)
-#define MDBX_CACHELINE_SIZE SYSTEM_CACHE_ALIGNMENT_SIZE
-#elif defined(__ia64__) || defined(__ia64) || defined(_M_IA64)
-#define MDBX_CACHELINE_SIZE 128
-#else
-#define MDBX_CACHELINE_SIZE 64
-#endif
-#endif /* MDBX_CACHELINE_SIZE */
-
-/** @} end of build options */
-/*******************************************************************************
- *******************************************************************************
- ******************************************************************************/
-
-#ifdef DOXYGEN
-/* !!! Actually this is a fake definitions !!!
- * !!! for documentation generation by Doxygen !!! */
-
-/** Controls enabling of debugging features.
- *
- * - `MDBX_DEBUG = 0` (by default) Disables any debugging features at all,
- * including logging and assertion controls.
- * Logging level and corresponding debug flags changing
- * by \ref mdbx_setup_debug() will not have effect.
- * - `MDBX_DEBUG > 0` Enables code for the debugging features (logging,
- * assertions checking and internal audit).
- * Simultaneously sets the default logging level
- * to the `MDBX_DEBUG` value.
- * Also enables \ref MDBX_DBG_AUDIT if `MDBX_DEBUG >= 2`.
- *
- * \ingroup build_option */
-#define MDBX_DEBUG 0...7
-
-/** Disables using of GNU libc extensions. */
-#define MDBX_DISABLE_GNU_SOURCE 0 or 1
-
-#endif /* DOXYGEN */
-
-/* Undefine the NDEBUG if debugging is enforced by MDBX_DEBUG */
-#if MDBX_DEBUG
-#undef NDEBUG
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-enum MDBX_memory_order {
- mo_Relaxed,
- mo_AcquireRelease,
- mo_SequentialConsistency
-};
-
-typedef union {
- volatile uint32_t weak;
-#ifdef MDBX_HAVE_C11ATOMICS
- volatile _Atomic uint32_t c11a;
-#endif /* MDBX_HAVE_C11ATOMICS */
-} MDBX_atomic_uint32_t;
-
-typedef union {
- volatile uint64_t weak;
-#if defined(MDBX_HAVE_C11ATOMICS) && (MDBX_64BIT_CAS || MDBX_64BIT_ATOMIC)
- volatile _Atomic uint64_t c11a;
-#endif
-#if !defined(MDBX_HAVE_C11ATOMICS) || !MDBX_64BIT_CAS || !MDBX_64BIT_ATOMIC
- __anonymous_struct_extension__ struct {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- MDBX_atomic_uint32_t low, high;
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- MDBX_atomic_uint32_t high, low;
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- };
-#endif
-} MDBX_atomic_uint64_t;
-
-#ifdef MDBX_HAVE_C11ATOMICS
-
-/* Crutches for C11 atomic compiler's bugs */
-#if defined(__e2k__) && defined(__LCC__) && __LCC__ < /* FIXME */ 127
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->weak)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->weak)
-#elif defined(__clang__) && __clang__ < 8
-#define MDBX_c11a_ro(type, ptr) ((volatile _Atomic(type) *)&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#else
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#endif /* Crutches for C11 atomic compiler's bugs */
-
-static __always_inline memory_order mo_c11_store(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_release;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-
-static __always_inline memory_order mo_c11_load(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_acquire;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED static __always_inline void
-mdbx_memory_fence(enum MDBX_memory_order order, bool write) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(write ? mo_c11_store(order) : mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_compiler_barrier();
- if (write &&
- order > (MDBX_CPU_WRITEBACK_INCOHERENT ? mo_Relaxed : mo_AcquireRelease))
- mdbx_memory_barrier();
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_store32(MDBX_atomic_uint32_t *p, const uint32_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint32_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_load32(const MDBX_atomic_uint32_t *p, enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint32_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint32_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint32_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-/* Basic constants and types */
-
-/* A stamp that identifies a file as an MDBX file.
- * There's nothing special about this value other than that it is easily
- * recognizable, and it will reflect any byte order mismatches. */
-#define MDBX_MAGIC UINT64_C(/* 56-bit prime */ 0x59659DBDEF4C11)
-
-/* FROZEN: The version number for a database's datafile format. */
-#define MDBX_DATA_VERSION 3
-/* The version number for a database's lockfile format. */
-#define MDBX_LOCK_VERSION 4
-
-/* handle for the DB used to track free pages. */
-#define FREE_DBI 0
-/* handle for the default DB. */
-#define MAIN_DBI 1
-/* Number of DBs in metapage (free and main) - also hardcoded elsewhere */
-#define CORE_DBS 2
-
-/* Number of meta pages - also hardcoded elsewhere */
-#define NUM_METAS 3
-
-/* A page number in the database.
- *
- * MDBX uses 32 bit for page numbers. This limits database
- * size up to 2^44 bytes, in case of 4K pages. */
-typedef uint32_t pgno_t;
-typedef MDBX_atomic_uint32_t atomic_pgno_t;
-#define PRIaPGNO PRIu32
-#define MAX_PAGENO UINT32_C(0x7FFFffff)
-#define MIN_PAGENO NUM_METAS
-
-#define SAFE64_INVALID_THRESHOLD UINT64_C(0xffffFFFF00000000)
-
-/* A transaction ID. */
-typedef uint64_t txnid_t;
-typedef MDBX_atomic_uint64_t atomic_txnid_t;
-#define PRIaTXN PRIi64
-#define MIN_TXNID UINT64_C(1)
-#define MAX_TXNID (SAFE64_INVALID_THRESHOLD - 1)
-#define INITIAL_TXNID (MIN_TXNID + NUM_METAS - 1)
-#define INVALID_TXNID UINT64_MAX
-/* LY: for testing non-atomic 64-bit txnid on 32-bit arches.
- * #define xMDBX_TXNID_STEP (UINT32_MAX / 3) */
-#ifndef xMDBX_TXNID_STEP
-#if MDBX_64BIT_CAS
-#define xMDBX_TXNID_STEP 1u
-#else
-#define xMDBX_TXNID_STEP 2u
-#endif
-#endif /* xMDBX_TXNID_STEP */
-
-/* Used for offsets within a single page.
- * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
- * this is plenty. */
-typedef uint16_t indx_t;
-
-#define MEGABYTE ((size_t)1 << 20)
-
-/*----------------------------------------------------------------------------*/
-/* Core structures for database and shared memory (i.e. format definition) */
-#pragma pack(push, 4)
-
-/* Information about a single database in the environment. */
-typedef struct MDBX_db {
- uint16_t md_flags; /* see mdbx_dbi_open */
- uint16_t md_depth; /* depth of this tree */
- uint32_t md_xsize; /* key-size for MDBX_DUPFIXED (LEAF2 pages) */
- pgno_t md_root; /* the root page of this tree */
- pgno_t md_branch_pages; /* number of internal pages */
- pgno_t md_leaf_pages; /* number of leaf pages */
- pgno_t md_overflow_pages; /* number of overflow pages */
- uint64_t md_seq; /* table sequence counter */
- uint64_t md_entries; /* number of data items */
- uint64_t md_mod_txnid; /* txnid of last committed modification */
-} MDBX_db;
-
-/* database size-related parameters */
-typedef struct MDBX_geo {
- uint16_t grow_pv; /* datafile growth step as a 16-bit packed (exponential
- quantized) value */
- uint16_t shrink_pv; /* datafile shrink threshold as a 16-bit packed
- (exponential quantized) value */
- pgno_t lower; /* minimal size of datafile in pages */
- pgno_t upper; /* maximal size of datafile in pages */
- pgno_t now; /* current size of datafile in pages */
- pgno_t next; /* first unused page in the datafile,
- but actually the file may be shorter. */
-} MDBX_geo;
-
-/* Meta page content.
- * A meta page is the start point for accessing a database snapshot.
- * Pages 0-1 are meta pages. Transaction N writes meta page (N % 2). */
-typedef struct MDBX_meta {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with MDBX_DATA_VERSION. */
- uint32_t mm_magic_and_version[2];
-
- /* txnid that committed this page, the first of a two-phase-update pair */
- uint32_t mm_txnid_a[2];
-
- uint16_t mm_extra_flags; /* extra DB flags, zero (nothing) for now */
- uint8_t mm_validator_id; /* ID of checksum and page validation method,
- * zero (nothing) for now */
- uint8_t mm_extra_pagehdr; /* extra bytes in the page header,
- * zero (nothing) for now */
-
- MDBX_geo mm_geo; /* database size-related parameters */
-
- MDBX_db mm_dbs[CORE_DBS]; /* first is free space, 2nd is main db */
- /* The size of pages used in this DB */
-#define mm_psize mm_dbs[FREE_DBI].md_xsize
- MDBX_canary mm_canary;
-
-#define MDBX_DATASIGN_NONE 0u
-#define MDBX_DATASIGN_WEAK 1u
-#define SIGN_IS_STEADY(sign) ((sign) > MDBX_DATASIGN_WEAK)
-#define META_IS_STEADY(meta) \
- SIGN_IS_STEADY(unaligned_peek_u64_volatile(4, (meta)->mm_datasync_sign))
- uint32_t mm_datasync_sign[2];
-
- /* txnid that committed this page, the second of a two-phase-update pair */
- uint32_t mm_txnid_b[2];
-
- /* Number of non-meta pages which were put in GC after COW. May be 0 in case
- * DB was previously handled by libmdbx without corresponding feature.
- * This value in couple with mr_snapshot_pages_retired allows fast estimation
- * of "how much reader is restraining GC recycling". */
- uint32_t mm_pages_retired[2];
-
- /* The analogue /proc/sys/kernel/random/boot_id or similar to determine
- * whether the system was rebooted after the last use of the database files.
- * If there was no reboot, but there is no need to rollback to the last
- * steady sync point. Zeros mean that no relevant information is available
- * from the system. */
- bin128_t mm_bootid;
-
-} MDBX_meta;
-
-#pragma pack(1)
-
-/* Common header for all page types. The page type depends on mp_flags.
- *
- * P_BRANCH and P_LEAF pages have unsorted 'MDBX_node's at the end, with
- * sorted mp_ptrs[] entries referring to them. Exception: P_LEAF2 pages
- * omit mp_ptrs and pack sorted MDBX_DUPFIXED values after the page header.
- *
- * P_OVERFLOW records occupy one or more contiguous pages where only the
- * first has a page header. They hold the real data of F_BIGDATA nodes.
- *
- * P_SUBP sub-pages are small leaf "pages" with duplicate data.
- * A node with flag F_DUPDATA but not F_SUBDATA contains a sub-page.
- * (Duplicate data can also go in sub-databases, which use normal pages.)
- *
- * P_META pages contain MDBX_meta, the start point of an MDBX snapshot.
- *
- * Each non-metapage up to MDBX_meta.mm_last_pg is reachable exactly once
- * in the snapshot: Either used by a database or listed in a GC record. */
-typedef struct MDBX_page {
- union {
-#define IS_FROZEN(txn, p) ((p)->mp_txnid < (txn)->mt_txnid)
-#define IS_SPILLED(txn, p) ((p)->mp_txnid == (txn)->mt_txnid)
-#define IS_SHADOWED(txn, p) ((p)->mp_txnid > (txn)->mt_txnid)
-#define IS_VALID(txn, p) ((p)->mp_txnid <= (txn)->mt_front)
-#define IS_MODIFIABLE(txn, p) ((p)->mp_txnid == (txn)->mt_front)
- uint64_t mp_txnid;
- struct MDBX_page *mp_next; /* for in-memory list of freed pages */
- };
- uint16_t mp_leaf2_ksize; /* key size if this is a LEAF2 page */
-#define P_BRANCH 0x01 /* branch page */
-#define P_LEAF 0x02 /* leaf page */
-#define P_OVERFLOW 0x04 /* overflow page */
-#define P_META 0x08 /* meta page */
-#define P_BAD 0x10 /* explicit flag for invalid/bad page */
-#define P_LEAF2 0x20 /* for MDBX_DUPFIXED records */
-#define P_SUBP 0x40 /* for MDBX_DUPSORT sub-pages */
-#define P_SPILLED 0x2000 /* spilled in parent txn */
-#define P_LOOSE 0x4000 /* page was dirtied then freed, can be reused */
-#define P_FROZEN 0x8000 /* used for retire page with known status */
-#define P_ILL_BITS (~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_SPILLED))
- uint16_t mp_flags;
- union {
- uint32_t mp_pages; /* number of overflow pages */
- __anonymous_struct_extension__ struct {
- indx_t mp_lower; /* lower bound of free space */
- indx_t mp_upper; /* upper bound of free space */
- };
- };
- pgno_t mp_pgno; /* page number */
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- indx_t mp_ptrs[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_page;
-
-/* Size of the page header, excluding dynamic data at the end */
-#define PAGEHDRSZ ((unsigned)offsetof(MDBX_page, mp_ptrs))
-
-#pragma pack(pop)
-
-#if MDBX_ENABLE_PGOP_STAT
-/* Statistics of page operations overall of all (running, completed and aborted)
- * transactions */
-typedef struct {
- MDBX_atomic_uint64_t newly; /* Quantity of a new pages added */
- MDBX_atomic_uint64_t cow; /* Quantity of pages copied for update */
- MDBX_atomic_uint64_t clone; /* Quantity of parent's dirty pages clones
- for nested transactions */
- MDBX_atomic_uint64_t split; /* Page splits */
- MDBX_atomic_uint64_t merge; /* Page merges */
- MDBX_atomic_uint64_t spill; /* Quantity of spilled dirty pages */
- MDBX_atomic_uint64_t unspill; /* Quantity of unspilled/reloaded pages */
- MDBX_atomic_uint64_t
- wops; /* Number of explicit write operations (not a pages) to a disk */
-} MDBX_pgop_stat_t;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-#if MDBX_LOCKING == MDBX_LOCKING_WIN32FILES
-#define MDBX_CLOCK_SIGN UINT32_C(0xF10C)
-typedef void mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
-
-#define MDBX_CLOCK_SIGN UINT32_C(0xF18D)
-typedef mdbx_pid_t mdbx_ipclock_t;
-#ifndef EOWNERDEAD
-#define EOWNERDEAD MDBX_RESULT_TRUE
-#endif
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#define MDBX_CLOCK_SIGN UINT32_C(0x8017)
-typedef pthread_mutex_t mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-#define MDBX_CLOCK_SIGN UINT32_C(0xFC29)
-typedef sem_t mdbx_ipclock_t;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV && !defined(__cplusplus)
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc);
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc);
-#endif /* MDBX_LOCKING */
-
-/* Reader Lock Table
- *
- * Readers don't acquire any locks for their data access. Instead, they
- * simply record their transaction ID in the reader table. The reader
- * mutex is needed just to find an empty slot in the reader table. The
- * slot's address is saved in thread-specific data so that subsequent
- * read transactions started by the same thread need no further locking to
- * proceed.
- *
- * If MDBX_NOTLS is set, the slot address is not saved in thread-specific data.
- * No reader table is used if the database is on a read-only filesystem.
- *
- * Since the database uses multi-version concurrency control, readers don't
- * actually need any locking. This table is used to keep track of which
- * readers are using data from which old transactions, so that we'll know
- * when a particular old transaction is no longer in use. Old transactions
- * that have discarded any data pages can then have those pages reclaimed
- * for use by a later write transaction.
- *
- * The lock table is constructed such that reader slots are aligned with the
- * processor's cache line size. Any slot is only ever used by one thread.
- * This alignment guarantees that there will be no contention or cache
- * thrashing as threads update their own slot info, and also eliminates
- * any need for locking when accessing a slot.
- *
- * A writer thread will scan every slot in the table to determine the oldest
- * outstanding reader transaction. Any freed pages older than this will be
- * reclaimed by the writer. The writer doesn't use any locks when scanning
- * this table. This means that there's no guarantee that the writer will
- * see the most up-to-date reader info, but that's not required for correct
- * operation - all we need is to know the upper bound on the oldest reader,
- * we don't care at all about the newest reader. So the only consequence of
- * reading stale information here is that old pages might hang around a
- * while longer before being reclaimed. That's actually good anyway, because
- * the longer we delay reclaiming old pages, the more likely it is that a
- * string of contiguous pages can be found after coalescing old pages from
- * many old transactions together. */
-
-/* The actual reader record, with cacheline padding. */
-typedef struct MDBX_reader {
- /* Current Transaction ID when this transaction began, or (txnid_t)-1.
- * Multiple readers that start at the same time will probably have the
- * same ID here. Again, it's not important to exclude them from
- * anything; all we need to know is which version of the DB they
- * started from so we can avoid overwriting any data used in that
- * particular version. */
- MDBX_atomic_uint64_t /* txnid_t */ mr_txnid;
-
- /* The information we store in a single slot of the reader table.
- * In addition to a transaction ID, we also record the process and
- * thread ID that owns a slot, so that we can detect stale information,
- * e.g. threads or processes that went away without cleaning up.
- *
- * NOTE: We currently don't check for stale records.
- * We simply re-init the table when we know that we're the only process
- * opening the lock file. */
-
- /* The thread ID of the thread owning this txn. */
- MDBX_atomic_uint64_t mr_tid;
-
- /* The process ID of the process owning this reader txn. */
- MDBX_atomic_uint32_t mr_pid;
-
- /* The number of pages used in the reader's MVCC snapshot,
- * i.e. the value of meta->mm_geo.next and txn->mt_next_pgno */
- atomic_pgno_t mr_snapshot_pages_used;
- /* Number of retired pages at the time this reader starts transaction. So,
- * at any time the difference mm_pages_retired - mr_snapshot_pages_retired
- * will give the number of pages which this reader restraining from reuse. */
- MDBX_atomic_uint64_t mr_snapshot_pages_retired;
-} MDBX_reader;
-
-/* The header for the reader table (a memory-mapped lock file). */
-typedef struct MDBX_lockinfo {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with with MDBX_LOCK_VERSION. */
- uint64_t mti_magic_and_version;
-
- /* Format of this lock file. Must be set to MDBX_LOCK_FORMAT. */
- uint32_t mti_os_and_format;
-
- /* Flags which environment was opened. */
- MDBX_atomic_uint32_t mti_envmode;
-
- /* Threshold of un-synced-with-disk pages for auto-sync feature,
- * zero means no-threshold, i.e. auto-sync is disabled. */
- atomic_pgno_t mti_autosync_threshold;
-
- /* Low 32-bit of txnid with which meta-pages was synced,
- * i.e. for sync-polling in the MDBX_NOMETASYNC mode. */
- MDBX_atomic_uint32_t mti_meta_sync_txnid;
-
- /* Period for timed auto-sync feature, i.e. at the every steady checkpoint
- * the mti_unsynced_timeout sets to the current_time + mti_autosync_period.
- * The time value is represented in a suitable system-dependent form, for
- * example clock_gettime(CLOCK_BOOTTIME) or clock_gettime(CLOCK_MONOTONIC).
- * Zero means timed auto-sync is disabled. */
- MDBX_atomic_uint64_t mti_autosync_period;
-
- /* Marker to distinguish uniqueness of DB/CLK. */
- MDBX_atomic_uint64_t mti_bait_uniqueness;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
-#if MDBX_ENABLE_PGOP_STAT
- /* Statistics of costly ops of all (running, completed and aborted)
- * transactions */
- MDBX_pgop_stat_t mti_pgop_stat;
-#endif /* MDBX_ENABLE_PGOP_STAT*/
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Write transaction lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_wlock;
-#endif /* MDBX_LOCKING > 0 */
-
- atomic_txnid_t mti_oldest_reader;
-
- /* Timestamp of the last steady sync. Value is represented in a suitable
- * system-dependent form, for example clock_gettime(CLOCK_BOOTTIME) or
- * clock_gettime(CLOCK_MONOTONIC). */
- MDBX_atomic_uint64_t mti_sync_timestamp;
-
- /* Number un-synced-with-disk pages for auto-sync feature. */
- atomic_pgno_t mti_unsynced_pages;
-
- /* Number of page which was discarded last time by madvise(MADV_FREE). */
- atomic_pgno_t mti_discarded_tail;
-
- /* Timestamp of the last readers check. */
- MDBX_atomic_uint64_t mti_reader_check_timestamp;
-
- /* Shared anchor for tracking readahead edge and enabled/disabled status. */
- pgno_t mti_readahead_anchor;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Readeaders registration lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_rlock;
-#endif /* MDBX_LOCKING > 0 */
-
- /* The number of slots that have been used in the reader table.
- * This always records the maximum count, it is not decremented
- * when readers release their slots. */
- MDBX_atomic_uint32_t mti_numreaders;
- MDBX_atomic_uint32_t mti_readers_refresh_flag;
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
- MDBX_reader mti_readers[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_lockinfo;
-
-/* Lockfile format signature: version, features and field layout */
-#define MDBX_LOCK_FORMAT \
- (MDBX_CLOCK_SIGN * 27733 + (unsigned)sizeof(MDBX_reader) * 13 + \
- (unsigned)offsetof(MDBX_reader, mr_snapshot_pages_used) * 251 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_oldest_reader) * 83 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_numreaders) * 37 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_readers) * 29)
-
-#define MDBX_DATA_MAGIC \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + MDBX_DATA_VERSION)
-
-#define MDBX_DATA_MAGIC_LEGACY_COMPAT \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + 2)
-
-#define MDBX_DATA_MAGIC_LEGACY_DEVEL ((MDBX_MAGIC << 8) + 255)
-
-#define MDBX_LOCK_MAGIC ((MDBX_MAGIC << 8) + MDBX_LOCK_VERSION)
-
-/* The maximum size of a database page.
- *
- * It is 64K, but value-PAGEHDRSZ must fit in MDBX_page.mp_upper.
- *
- * MDBX will use database pages < OS pages if needed.
- * That causes more I/O in write transactions: The OS must
- * know (read) the whole page before writing a partial page.
- *
- * Note that we don't currently support Huge pages. On Linux,
- * regular data files cannot use Huge pages, and in general
- * Huge pages aren't actually pageable. We rely on the OS
- * demand-pager to read our data and page it out when memory
- * pressure from other processes is high. So until OSs have
- * actual paging support for Huge pages, they're not viable. */
-#define MAX_PAGESIZE MDBX_MAX_PAGESIZE
-#define MIN_PAGESIZE MDBX_MIN_PAGESIZE
-
-#define MIN_MAPSIZE (MIN_PAGESIZE * MIN_PAGENO)
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_MAPSIZE32 UINT32_C(0x38000000)
-#else
-#define MAX_MAPSIZE32 UINT32_C(0x7f000000)
-#endif
-#define MAX_MAPSIZE64 ((MAX_PAGENO + 1) * (uint64_t)MAX_PAGESIZE)
-
-#if MDBX_WORDBITS >= 64
-#define MAX_MAPSIZE MAX_MAPSIZE64
-#define MDBX_PGL_LIMIT ((size_t)MAX_PAGENO)
-#else
-#define MAX_MAPSIZE MAX_MAPSIZE32
-#define MDBX_PGL_LIMIT (MAX_MAPSIZE32 / MIN_PAGESIZE)
-#endif /* MDBX_WORDBITS */
-
-#define MDBX_READERS_LIMIT 32767
-#define MDBX_RADIXSORT_THRESHOLD 333
-
-/*----------------------------------------------------------------------------*/
-
-/* An PNL is an Page Number List, a sorted array of IDs.
- * The first element of the array is a counter for how many actual page-numbers
- * are in the list. By default PNLs are sorted in descending order, this allow
- * cut off a page with lowest pgno (at the tail) just truncating the list. The
- * sort order of PNLs is controlled by the MDBX_PNL_ASCENDING build option. */
-typedef pgno_t *MDBX_PNL;
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_ORDERED(first, last) ((first) < (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) >= (last))
-#else
-#define MDBX_PNL_ORDERED(first, last) ((first) > (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) <= (last))
-#endif
-
-/* List of txnid, only for MDBX_txn.tw.lifo_reclaimed */
-typedef txnid_t *MDBX_TXL;
-
-/* An Dirty-Page list item is an pgno/pointer pair. */
-typedef struct MDBX_dp {
- MDBX_page *ptr;
- pgno_t pgno;
- union {
- unsigned extra;
- __anonymous_struct_extension__ struct {
- unsigned multi : 1;
- unsigned lru : 31;
- };
- };
-} MDBX_dp;
-
-/* An DPL (dirty-page list) is a sorted array of MDBX_DPs. */
-typedef struct MDBX_dpl {
- unsigned sorted;
- unsigned length;
- unsigned detent; /* allocated size excluding the MDBX_DPL_RESERVE_GAP */
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_dp items[] /* dynamic size with holes at zero and after the last */;
-#endif
-} MDBX_dpl;
-
-/* PNL sizes */
-#define MDBX_PNL_GRANULATE 1024
-#define MDBX_PNL_INITIAL \
- (MDBX_PNL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(pgno_t))
-
-#define MDBX_TXL_GRANULATE 32
-#define MDBX_TXL_INITIAL \
- (MDBX_TXL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-#define MDBX_TXL_MAX \
- ((1u << 17) - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-
-#define MDBX_PNL_ALLOCLEN(pl) ((pl)[-1])
-#define MDBX_PNL_SIZE(pl) ((pl)[0])
-#define MDBX_PNL_FIRST(pl) ((pl)[1])
-#define MDBX_PNL_LAST(pl) ((pl)[MDBX_PNL_SIZE(pl)])
-#define MDBX_PNL_BEGIN(pl) (&(pl)[1])
-#define MDBX_PNL_END(pl) (&(pl)[MDBX_PNL_SIZE(pl) + 1])
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_FIRST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_LAST(pl)
-#else
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_LAST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_FIRST(pl)
-#endif
-
-#define MDBX_PNL_SIZEOF(pl) ((MDBX_PNL_SIZE(pl) + 1) * sizeof(pgno_t))
-#define MDBX_PNL_IS_EMPTY(pl) (MDBX_PNL_SIZE(pl) == 0)
-
-/*----------------------------------------------------------------------------*/
-/* Internal structures */
-
-/* Auxiliary DB info.
- * The information here is mostly static/read-only. There is
- * only a single copy of this record in the environment. */
-typedef struct MDBX_dbx {
- MDBX_val md_name; /* name of the database */
- MDBX_cmp_func *md_cmp; /* function for comparing keys */
- MDBX_cmp_func *md_dcmp; /* function for comparing data items */
- size_t md_klen_min, md_klen_max; /* min/max key length for the database */
- size_t md_vlen_min,
- md_vlen_max; /* min/max value/data length for the database */
-} MDBX_dbx;
-
-/* A database transaction.
- * Every operation requires a transaction handle. */
-struct MDBX_txn {
-#define MDBX_MT_SIGNATURE UINT32_C(0x93D53A31)
- uint32_t mt_signature;
-
- /* Transaction Flags */
- /* mdbx_txn_begin() flags */
-#define MDBX_TXN_RO_BEGIN_FLAGS (MDBX_TXN_RDONLY | MDBX_TXN_RDONLY_PREPARE)
-#define MDBX_TXN_RW_BEGIN_FLAGS \
- (MDBX_TXN_NOMETASYNC | MDBX_TXN_NOSYNC | MDBX_TXN_TRY)
- /* Additional flag for mdbx_sync_locked() */
-#define MDBX_SHRINK_ALLOWED UINT32_C(0x40000000)
-
- /* internal txn flags */
-#define MDBX_TXN_FINISHED 0x01 /* txn is finished or never began */
-#define MDBX_TXN_ERROR 0x02 /* txn is unusable after an error */
-#define MDBX_TXN_DIRTY 0x04 /* must write, even if dirty list is empty */
-#define MDBX_TXN_SPILLS 0x08 /* txn or a parent has spilled pages */
-#define MDBX_TXN_HAS_CHILD 0x10 /* txn has an MDBX_txn.mt_child */
- /* most operations on the txn are currently illegal */
-#define MDBX_TXN_BLOCKED \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_HAS_CHILD)
-
-#define TXN_FLAGS \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_DIRTY | MDBX_TXN_SPILLS | \
- MDBX_TXN_HAS_CHILD)
-
-#if (TXN_FLAGS & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS)) || \
- ((MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS | TXN_FLAGS) & \
- MDBX_SHRINK_ALLOWED)
-#error "Oops, some flags overlapped or wrong"
-#endif
- uint32_t mt_flags;
-
- MDBX_txn *mt_parent; /* parent of a nested txn */
- /* Nested txn under this txn, set together with flag MDBX_TXN_HAS_CHILD */
- MDBX_txn *mt_child;
- MDBX_geo mt_geo;
- /* next unallocated page */
-#define mt_next_pgno mt_geo.next
- /* corresponding to the current size of datafile */
-#define mt_end_pgno mt_geo.now
-
- /* The ID of this transaction. IDs are integers incrementing from 1.
- * Only committed write transactions increment the ID. If a transaction
- * aborts, the ID may be re-used by the next writer. */
- txnid_t mt_txnid;
- txnid_t mt_front;
-
- MDBX_env *mt_env; /* the DB environment */
- /* Array of records for each DB known in the environment. */
- MDBX_dbx *mt_dbxs;
- /* Array of MDBX_db records for each known DB */
- MDBX_db *mt_dbs;
- /* Array of sequence numbers for each DB handle */
- unsigned *mt_dbiseqs;
-
- /* Transaction DBI Flags */
-#define DBI_DIRTY MDBX_DBI_DIRTY /* DB was written in this txn */
-#define DBI_STALE MDBX_DBI_STALE /* Named-DB record is older than txnID */
-#define DBI_FRESH MDBX_DBI_FRESH /* Named-DB handle opened in this txn */
-#define DBI_CREAT MDBX_DBI_CREAT /* Named-DB handle created in this txn */
-#define DBI_VALID 0x10 /* DB handle is valid, see also DB_VALID */
-#define DBI_USRVALID 0x20 /* As DB_VALID, but not set for FREE_DBI */
-#define DBI_AUDITED 0x40 /* Internal flag for accounting during audit */
- /* Array of flags for each DB */
- uint8_t *mt_dbistate;
- /* Number of DB records in use, or 0 when the txn is finished.
- * This number only ever increments until the txn finishes; we
- * don't decrement it when individual DB handles are closed. */
- MDBX_dbi mt_numdbs;
- size_t mt_owner; /* thread ID that owns this transaction */
- MDBX_canary mt_canary;
- void *mt_userctx; /* User-settable context */
- MDBX_cursor **mt_cursors;
-
- union {
- struct {
- /* For read txns: This thread/txn's reader table slot, or NULL. */
- MDBX_reader *reader;
- } to;
- struct {
- /* In write txns, array of cursors for each DB */
- pgno_t *reclaimed_pglist; /* Reclaimed GC pages */
- txnid_t last_reclaimed; /* ID of last used record */
-#if MDBX_ENABLE_REFUND
- pgno_t loose_refund_wl /* FIXME: describe */;
-#endif /* MDBX_ENABLE_REFUND */
- /* dirtylist room: Dirty array size - dirty pages visible to this txn.
- * Includes ancestor txns' dirty pages not hidden by other txns'
- * dirty/spilled pages. Thus commit(nested txn) has room to merge
- * dirtylist into mt_parent after freeing hidden mt_parent pages. */
- unsigned dirtyroom;
- /* a sequence to spilling dirty page with LRU policy */
- unsigned dirtylru;
- /* For write txns: Modified pages. Sorted when not MDBX_WRITEMAP. */
- MDBX_dpl *dirtylist;
- /* The list of reclaimed txns from GC */
- MDBX_TXL lifo_reclaimed;
- /* The list of pages that became unused during this transaction. */
- MDBX_PNL retired_pages;
- /* The list of loose pages that became unused and may be reused
- * in this transaction, linked through `mp_next`. */
- MDBX_page *loose_pages;
- /* Number of loose pages (tw.loose_pages) */
- unsigned loose_count;
- /* The sorted list of dirty pages we temporarily wrote to disk
- * because the dirty list was full. page numbers in here are
- * shifted left by 1, deleted slots have the LSB set. */
- MDBX_PNL spill_pages;
- unsigned spill_least_removed;
- } tw;
- };
-};
-
-#if MDBX_WORDBITS >= 64
-#define CURSOR_STACK 32
-#else
-#define CURSOR_STACK 24
-#endif
-
-struct MDBX_xcursor;
-
-/* Cursors are used for all DB operations.
- * A cursor holds a path of (page pointer, key index) from the DB
- * root to a position in the DB, plus other state. MDBX_DUPSORT
- * cursors include an xcursor to the current data item. Write txns
- * track their cursors and keep them up to date when data moves.
- * Exception: An xcursor's pointer to a P_SUBP page can be stale.
- * (A node with F_DUPDATA but no F_SUBDATA contains a subpage). */
-struct MDBX_cursor {
-#define MDBX_MC_LIVE UINT32_C(0xFE05D5B1)
-#define MDBX_MC_READY4CLOSE UINT32_C(0x2817A047)
-#define MDBX_MC_WAIT4EOT UINT32_C(0x90E297A7)
- uint32_t mc_signature;
- /* The database handle this cursor operates on */
- MDBX_dbi mc_dbi;
- /* Next cursor on this DB in this txn */
- MDBX_cursor *mc_next;
- /* Backup of the original cursor if this cursor is a shadow */
- MDBX_cursor *mc_backup;
- /* Context used for databases with MDBX_DUPSORT, otherwise NULL */
- struct MDBX_xcursor *mc_xcursor;
- /* The transaction that owns this cursor */
- MDBX_txn *mc_txn;
- /* The database record for this cursor */
- MDBX_db *mc_db;
- /* The database auxiliary record for this cursor */
- MDBX_dbx *mc_dbx;
- /* The mt_dbistate for this database */
- uint8_t *mc_dbistate;
- unsigned mc_snum; /* number of pushed pages */
- unsigned mc_top; /* index of top page, normally mc_snum-1 */
-
- /* Cursor state flags. */
-#define C_INITIALIZED 0x01 /* cursor has been initialized and is valid */
-#define C_EOF 0x02 /* No more data */
-#define C_SUB 0x04 /* Cursor is a sub-cursor */
-#define C_DEL 0x08 /* last op was a cursor_del */
-#define C_UNTRACK 0x10 /* Un-track cursor when closing */
-#define C_RECLAIMING 0x20 /* GC lookup is prohibited */
-#define C_GCFREEZE 0x40 /* reclaimed_pglist must not be updated */
-
- /* Cursor checking flags. */
-#define C_COPYING 0x100 /* skip key-value length check (copying simplify) */
-#define C_UPDATING 0x200 /* update/rebalance pending */
-#define C_RETIRING 0x400 /* refs to child pages may be invalid */
-#define C_SKIPORD 0x800 /* don't check keys ordering */
-
- unsigned mc_flags; /* see mdbx_cursor */
- MDBX_page *mc_pg[CURSOR_STACK]; /* stack of pushed pages */
- indx_t mc_ki[CURSOR_STACK]; /* stack of page indices */
-};
-
-/* Context for sorted-dup records.
- * We could have gone to a fully recursive design, with arbitrarily
- * deep nesting of sub-databases. But for now we only handle these
- * levels - main DB, optional sub-DB, sorted-duplicate DB. */
-typedef struct MDBX_xcursor {
- /* A sub-cursor for traversing the Dup DB */
- MDBX_cursor mx_cursor;
- /* The database record for this Dup DB */
- MDBX_db mx_db;
- /* The auxiliary DB record for this Dup DB */
- MDBX_dbx mx_dbx;
-} MDBX_xcursor;
-
-typedef struct MDBX_cursor_couple {
- MDBX_cursor outer;
- void *mc_userctx; /* User-settable context */
- MDBX_xcursor inner;
-} MDBX_cursor_couple;
-
-/* The database environment. */
-struct MDBX_env {
- /* ----------------------------------------------------- mostly static part */
-#define MDBX_ME_SIGNATURE UINT32_C(0x9A899641)
- MDBX_atomic_uint32_t me_signature;
- /* Failed to update the meta page. Probably an I/O error. */
-#define MDBX_FATAL_ERROR UINT32_C(0x80000000)
- /* Some fields are initialized. */
-#define MDBX_ENV_ACTIVE UINT32_C(0x20000000)
- /* me_txkey is set */
-#define MDBX_ENV_TXKEY UINT32_C(0x10000000)
- /* Legacy MDBX_MAPASYNC (prior v0.9) */
-#define MDBX_DEPRECATED_MAPASYNC UINT32_C(0x100000)
-#define ENV_INTERNAL_FLAGS (MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE | MDBX_ENV_TXKEY)
- uint32_t me_flags;
- mdbx_mmap_t me_dxb_mmap; /* The main data file */
-#define me_map me_dxb_mmap.dxb
-#define me_lazy_fd me_dxb_mmap.fd
- mdbx_filehandle_t me_dsync_fd;
- mdbx_mmap_t me_lck_mmap; /* The lock file */
-#define me_lfd me_lck_mmap.fd
- struct MDBX_lockinfo *me_lck;
-
- unsigned me_psize; /* DB page size, initialized from me_os_psize */
- unsigned me_leaf_nodemax; /* max size of a leaf-node */
- uint8_t me_psize2log; /* log2 of DB page size */
- int8_t me_stuck_meta; /* recovery-only: target meta page or less that zero */
- uint16_t me_merge_threshold,
- me_merge_threshold_gc; /* pages emptier than this are candidates for
- merging */
- unsigned me_os_psize; /* OS page size, from mdbx_syspagesize() */
- unsigned me_maxreaders; /* size of the reader table */
- MDBX_dbi me_maxdbs; /* size of the DB table */
- uint32_t me_pid; /* process ID of this env */
- mdbx_thread_key_t me_txkey; /* thread-key for readers */
- char *me_pathname; /* path to the DB files */
- void *me_pbuf; /* scratch area for DUPSORT put() */
- MDBX_txn *me_txn0; /* preallocated write transaction */
-
- MDBX_dbx *me_dbxs; /* array of static DB info */
- uint16_t *me_dbflags; /* array of flags from MDBX_db.md_flags */
- unsigned *me_dbiseqs; /* array of dbi sequence numbers */
- unsigned
- me_maxgc_ov1page; /* Number of pgno_t fit in a single overflow page */
- uint32_t me_live_reader; /* have liveness lock in reader table */
- void *me_userctx; /* User-settable context */
- MDBX_hsr_func *me_hsr_callback; /* Callback for kicking laggard readers */
-
- struct {
- unsigned dp_reserve_limit;
- unsigned rp_augment_limit;
- unsigned dp_limit;
- unsigned dp_initial;
- uint8_t dp_loose_limit;
- uint8_t spill_max_denominator;
- uint8_t spill_min_denominator;
- uint8_t spill_parent4child_denominator;
- unsigned merge_threshold_16dot16_percent;
- union {
- unsigned all;
- /* tracks options with non-auto values but tuned by user */
- struct {
- unsigned dp_limit : 1;
- } non_auto;
- } flags;
- } me_options;
-
- /* struct me_dbgeo used for accepting db-geo params from user for the new
- * database creation, i.e. when mdbx_env_set_geometry() was called before
- * mdbx_env_open(). */
- struct {
- size_t lower; /* minimal size of datafile */
- size_t upper; /* maximal size of datafile */
- size_t now; /* current size of datafile */
- size_t grow; /* step to grow datafile */
- size_t shrink; /* threshold to shrink datafile */
- } me_dbgeo;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- union {
- key_t key;
- int semid;
- } me_sysv_ipc;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
- MDBX_env *me_lcklist_next;
-
- /* --------------------------------------------------- mostly volatile part */
-
- MDBX_txn *me_txn; /* current write transaction */
- mdbx_fastmutex_t me_dbi_lock;
- MDBX_dbi me_numdbs; /* number of DBs opened */
-
- MDBX_page *me_dp_reserve; /* list of malloc'ed blocks for re-use */
- unsigned me_dp_reserve_len;
- /* PNL of pages that became unused in a write txn */
- MDBX_PNL me_retired_pages;
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_srwlock me_remap_guard;
- /* Workaround for LockFileEx and WriteFile multithread bug */
- CRITICAL_SECTION me_windowsbug_lock;
-#else
- mdbx_fastmutex_t me_remap_guard;
-#endif
-
- /* -------------------------------------------------------------- debugging */
-
-#if MDBX_DEBUG
- MDBX_assert_func *me_assert_func; /* Callback for assertion failures */
-#endif
-#ifdef MDBX_USE_VALGRIND
- int me_valgrind_handle;
-#endif
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- pgno_t me_poison_edge;
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-#ifndef xMDBX_DEBUG_SPILLING
-#define xMDBX_DEBUG_SPILLING 0
-#endif
-#if xMDBX_DEBUG_SPILLING == 2
- unsigned debug_dirtied_est, debug_dirtied_act;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- /* ------------------------------------------------- stub for lck-less mode */
- MDBX_atomic_uint64_t
- x_lckless_stub[(sizeof(MDBX_lockinfo) + MDBX_CACHELINE_SIZE - 1) /
- sizeof(MDBX_atomic_uint64_t)];
-};
-
-#ifndef __cplusplus
-/*----------------------------------------------------------------------------*/
-/* Debug and Logging stuff */
-
-#define MDBX_RUNTIME_FLAGS_INIT \
- ((MDBX_DEBUG) > 0) * MDBX_DBG_ASSERT + ((MDBX_DEBUG) > 1) * MDBX_DBG_AUDIT
-
-extern uint8_t mdbx_runtime_flags;
-extern uint8_t mdbx_loglevel;
-extern MDBX_debug_func *mdbx_debug_logger;
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny) {
-#if MDBX_DEBUG
- if (MDBX_DBG_JITTER & mdbx_runtime_flags)
- mdbx_osal_jitter(tiny);
-#else
- (void)tiny;
-#endif
-}
-
-MDBX_INTERNAL_FUNC void MDBX_PRINTF_ARGS(4, 5)
- mdbx_debug_log(int level, const char *function, int line, const char *fmt,
- ...) MDBX_PRINTF_ARGS(4, 5);
-MDBX_INTERNAL_FUNC void mdbx_debug_log_va(int level, const char *function,
- int line, const char *fmt,
- va_list args);
-
-#if MDBX_DEBUG
-#define mdbx_log_enabled(msg) unlikely(msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() unlikely((mdbx_runtime_flags & MDBX_DBG_AUDIT))
-#else /* MDBX_DEBUG */
-#define mdbx_log_enabled(msg) (msg < MDBX_LOG_VERBOSE && msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() (0)
-#endif /* MDBX_DEBUG */
-
-#if MDBX_FORCE_ASSERTIONS
-#define mdbx_assert_enabled() (1)
-#elif MDBX_DEBUG
-#define mdbx_assert_enabled() likely((mdbx_runtime_flags & MDBX_DBG_ASSERT))
-#else
-#define mdbx_assert_enabled() (0)
-#endif /* assertions */
-
-#if !MDBX_DEBUG && defined(__ANDROID_API__)
-#define mdbx_assert_fail(env, msg, func, line) \
- __android_log_assert(msg, "mdbx", "%s:%u", func, line)
-#else
-void mdbx_assert_fail(const MDBX_env *env, const char *msg, const char *func,
- int line);
-#endif
-
-#define mdbx_debug_extra(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, __func__, __LINE__, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug_extra_print(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, NULL, 0, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_trace(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_TRACE)) \
- mdbx_debug_log(MDBX_LOG_TRACE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_DEBUG)) \
- mdbx_debug_log(MDBX_LOG_DEBUG, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_verbose(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_VERBOSE)) \
- mdbx_debug_log(MDBX_LOG_VERBOSE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_notice(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_NOTICE)) \
- mdbx_debug_log(MDBX_LOG_NOTICE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_warning(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_WARN)) \
- mdbx_debug_log(MDBX_LOG_WARN, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_error(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) \
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_fatal(fmt, ...) \
- mdbx_debug_log(MDBX_LOG_FATAL, __func__, __LINE__, fmt "\n", __VA_ARGS__);
-
-#define mdbx_ensure_msg(env, expr, msg) \
- do { \
- if (unlikely(!(expr))) \
- mdbx_assert_fail(env, msg, __func__, __LINE__); \
- } while (0)
-
-#define mdbx_ensure(env, expr) mdbx_ensure_msg(env, expr, #expr)
-
-/* assert(3) variant in environment context */
-#define mdbx_assert(env, expr) \
- do { \
- if (mdbx_assert_enabled()) \
- mdbx_ensure(env, expr); \
- } while (0)
-
-/* assert(3) variant in cursor context */
-#define mdbx_cassert(mc, expr) mdbx_assert((mc)->mc_txn->mt_env, expr)
-
-/* assert(3) variant in transaction context */
-#define mdbx_tassert(txn, expr) mdbx_assert((txn)->mt_env, expr)
-
-#ifndef xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-#undef assert
-#define assert(expr) mdbx_assert(NULL, expr)
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Cache coherence and mmap invalidation */
-
-#if MDBX_CPU_WRITEBACK_INCOHERENT
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_memory_barrier()
-#else
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_compiler_barrier()
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-MDBX_MAYBE_UNUSED static __inline void
-mdbx_flush_incoherent_mmap(void *addr, size_t nbytes, const intptr_t pagesize) {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- char *const begin = (char *)(-pagesize & (intptr_t)addr);
- char *const end =
- (char *)(-pagesize & (intptr_t)((char *)addr + nbytes + pagesize - 1));
- int err = msync(begin, end - begin, MS_SYNC | MS_INVALIDATE) ? errno : 0;
- mdbx_assert(nullptr, err == 0);
- (void)err;
-#else
- (void)pagesize;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#if MDBX_MMAP_INCOHERENT_CPU_CACHE
-#ifdef DCACHE
- /* MIPS has cache coherency issues.
- * Note: for any nbytes >= on-chip cache size, entire is flushed. */
- cacheflush(addr, nbytes, DCACHE);
-#else
-#error "Oops, cacheflush() not available"
-#endif /* DCACHE */
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#if !MDBX_MMAP_INCOHERENT_FILE_WRITE && !MDBX_MMAP_INCOHERENT_CPU_CACHE
- (void)addr;
- (void)nbytes;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* Internal prototypes */
-
-MDBX_INTERNAL_FUNC int mdbx_cleanup_dead_readers(MDBX_env *env, int rlocked,
- int *dead);
-MDBX_INTERNAL_FUNC int mdbx_rthc_alloc(mdbx_thread_key_t *key,
- MDBX_reader *begin, MDBX_reader *end);
-MDBX_INTERNAL_FUNC void mdbx_rthc_remove(const mdbx_thread_key_t key);
-
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_init(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_dtor(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_thread_dtor(void *ptr);
-
-#endif /* !__cplusplus */
-
-#define MDBX_IS_ERROR(rc) \
- ((rc) != MDBX_RESULT_TRUE && (rc) != MDBX_RESULT_FALSE)
-
-/* Internal error codes, not exposed outside libmdbx */
-#define MDBX_NO_ROOT (MDBX_LAST_ADDED_ERRCODE + 10)
-
-/* Debugging output value of a cursor DBI: Negative in a sub-cursor. */
-#define DDBI(mc) \
- (((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
-
-/* Key size which fits in a DKBUF (debug key buffer). */
-#define DKBUF_MAX 511
-#define DKBUF char _kbuf[DKBUF_MAX * 4 + 2]
-#define DKEY(x) mdbx_dump_val(x, _kbuf, DKBUF_MAX * 2 + 1)
-#define DVAL(x) mdbx_dump_val(x, _kbuf + DKBUF_MAX * 2 + 1, DKBUF_MAX * 2 + 1)
-
-#if MDBX_DEBUG
-#define DKBUF_DEBUG DKBUF
-#define DKEY_DEBUG(x) DKEY(x)
-#define DVAL_DEBUG(x) DVAL(x)
-#else
-#define DKBUF_DEBUG ((void)(0))
-#define DKEY_DEBUG(x) ("-")
-#define DVAL_DEBUG(x) ("-")
-#endif
-
-/* An invalid page number.
- * Mainly used to denote an empty tree. */
-#define P_INVALID (~(pgno_t)0)
-
-/* Test if the flags f are set in a flag word w. */
-#define F_ISSET(w, f) (((w) & (f)) == (f))
-
-/* Round n up to an even number. */
-#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
-
-/* Default size of memory map.
- * This is certainly too small for any actual applications. Apps should
- * always set the size explicitly using mdbx_env_set_geometry(). */
-#define DEFAULT_MAPSIZE MEGABYTE
-
-/* Number of slots in the reader table.
- * This value was chosen somewhat arbitrarily. The 61 is a prime number,
- * and such readers plus a couple mutexes fit into single 4KB page.
- * Applications should set the table size using mdbx_env_set_maxreaders(). */
-#define DEFAULT_READERS 61
-
-/* Test if a page is a leaf page */
-#define IS_LEAF(p) (((p)->mp_flags & P_LEAF) != 0)
-/* Test if a page is a LEAF2 page */
-#define IS_LEAF2(p) unlikely(((p)->mp_flags & P_LEAF2) != 0)
-/* Test if a page is a branch page */
-#define IS_BRANCH(p) (((p)->mp_flags & P_BRANCH) != 0)
-/* Test if a page is an overflow page */
-#define IS_OVERFLOW(p) unlikely(((p)->mp_flags & P_OVERFLOW) != 0)
-/* Test if a page is a sub page */
-#define IS_SUBP(p) (((p)->mp_flags & P_SUBP) != 0)
-
-#define PAGETYPE(p) ((p)->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW))
-
-/* Header for a single key/data pair within a page.
- * Used in pages of type P_BRANCH and P_LEAF without P_LEAF2.
- * We guarantee 2-byte alignment for 'MDBX_node's.
- *
- * Leaf node flags describe node contents. F_BIGDATA says the node's
- * data part is the page number of an overflow page with actual data.
- * F_DUPDATA and F_SUBDATA can be combined giving duplicate data in
- * a sub-page/sub-database, and named databases (just F_SUBDATA). */
-typedef struct MDBX_node {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- union {
- uint32_t mn_dsize;
- uint32_t mn_pgno32;
- };
- uint8_t mn_flags; /* see mdbx_node flags */
- uint8_t mn_extra;
- uint16_t mn_ksize; /* key size */
-#else
- uint16_t mn_ksize; /* key size */
- uint8_t mn_extra;
- uint8_t mn_flags; /* see mdbx_node flags */
- union {
- uint32_t mn_pgno32;
- uint32_t mn_dsize;
- };
-#endif /* __BYTE_ORDER__ */
-
- /* mdbx_node Flags */
-#define F_BIGDATA 0x01 /* data put on overflow page */
-#define F_SUBDATA 0x02 /* data is a sub-database */
-#define F_DUPDATA 0x04 /* data has duplicates */
-
- /* valid flags for mdbx_node_add() */
-#define NODE_ADD_FLAGS (F_DUPDATA | F_SUBDATA | MDBX_RESERVE | MDBX_APPEND)
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- uint8_t mn_data[] /* key and data are appended here */;
-#endif /* C99 */
-} MDBX_node;
-
-#define DB_PERSISTENT_FLAGS \
- (MDBX_REVERSEKEY | MDBX_DUPSORT | MDBX_INTEGERKEY | MDBX_DUPFIXED | \
- MDBX_INTEGERDUP | MDBX_REVERSEDUP)
-
-/* mdbx_dbi_open() flags */
-#define DB_USABLE_FLAGS (DB_PERSISTENT_FLAGS | MDBX_CREATE | MDBX_DB_ACCEDE)
-
-#define DB_VALID 0x8000 /* DB handle is valid, for me_dbflags */
-#define DB_INTERNAL_FLAGS DB_VALID
-
-#if DB_INTERNAL_FLAGS & DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-#if DB_PERSISTENT_FLAGS & ~DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-
-/* max number of pages to commit in one writev() call */
-#define MDBX_COMMIT_PAGES 64
-#if defined(IOV_MAX) && IOV_MAX < MDBX_COMMIT_PAGES /* sysconf(_SC_IOV_MAX) */
-#undef MDBX_COMMIT_PAGES
-#define MDBX_COMMIT_PAGES IOV_MAX
-#endif
-
-/*
- * /
- * | -1, a < b
- * CMP2INT(a,b) = < 0, a == b
- * | 1, a > b
- * \
- */
-#ifndef __e2k__
-/* LY: fast enough on most systems */
-#define CMP2INT(a, b) (((b) > (a)) ? -1 : (a) > (b))
-#else
-/* LY: more parallelable on VLIW Elbrus */
-#define CMP2INT(a, b) (((a) > (b)) - ((b) > (a)))
-#endif
-
-/* Do not spill pages to disk if txn is getting full, may fail instead */
-#define MDBX_NOSPILL 0x8000
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-int64pgno(int64_t i64) {
- if (likely(i64 >= (int64_t)MIN_PAGENO && i64 <= (int64_t)MAX_PAGENO + 1))
- return (pgno_t)i64;
- return (i64 < (int64_t)MIN_PAGENO) ? MIN_PAGENO : MAX_PAGENO;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_add(size_t base, size_t augend) {
- assert(base <= MAX_PAGENO + 1 && augend < MAX_PAGENO);
- return int64pgno(base + augend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_sub(size_t base, size_t subtrahend) {
- assert(base >= MIN_PAGENO && base <= MAX_PAGENO + 1 &&
- subtrahend < MAX_PAGENO);
- return int64pgno(base - subtrahend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline bool
-is_powerof2(size_t x) {
- return (x & (x - 1)) == 0;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-floor_powerof2(size_t value, size_t granularity) {
- assert(is_powerof2(granularity));
- return value & ~(granularity - 1);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-ceil_powerof2(size_t value, size_t granularity) {
- return floor_powerof2(value + granularity - 1, granularity);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static unsigned
-log2n_powerof2(size_t value) {
- assert(value > 0 && value < INT32_MAX && is_powerof2(value));
- assert((value & -(int32_t)value) == value);
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_ctzl)
- return __builtin_ctzl(value);
-#elif defined(_MSC_VER)
- unsigned long index;
- _BitScanForward(&index, (unsigned long)value);
- return index;
-#else
- static const uint8_t debruijn_ctz32[32] = {
- 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
- 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9};
- return debruijn_ctz32[(uint32_t)(value * 0x077CB531u) >> 27];
-#endif
-}
-
-/* Only a subset of the mdbx_env flags can be changed
- * at runtime. Changing other flags requires closing the
- * environment and re-opening it with the new flags. */
-#define ENV_CHANGEABLE_FLAGS \
- (MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC | MDBX_DEPRECATED_MAPASYNC | \
- MDBX_NOMEMINIT | MDBX_COALESCE | MDBX_PAGEPERTURB | MDBX_ACCEDE)
-#define ENV_CHANGELESS_FLAGS \
- (MDBX_NOSUBDIR | MDBX_RDONLY | MDBX_WRITEMAP | MDBX_NOTLS | MDBX_NORDAHEAD | \
- MDBX_LIFORECLAIM | MDBX_EXCLUSIVE)
-#define ENV_USABLE_FLAGS (ENV_CHANGEABLE_FLAGS | ENV_CHANGELESS_FLAGS)
-
-#if !defined(__cplusplus) || CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_MAYBE_UNUSED static void static_checks(void) {
- STATIC_ASSERT_MSG(INT16_MAX - CORE_DBS == MDBX_MAX_DBI,
- "Oops, MDBX_MAX_DBI or CORE_DBS?");
- STATIC_ASSERT_MSG((unsigned)(MDBX_DB_ACCEDE | MDBX_CREATE) ==
- ((DB_USABLE_FLAGS | DB_INTERNAL_FLAGS) &
- (ENV_USABLE_FLAGS | ENV_INTERNAL_FLAGS)),
- "Oops, some flags overlapped or wrong");
- STATIC_ASSERT_MSG((ENV_INTERNAL_FLAGS & ENV_USABLE_FLAGS) == 0,
- "Oops, some flags overlapped or wrong");
-}
-#endif /* Disabled for MSVC 19.0 (VisualStudio 2015) */
-
-#ifdef __cplusplus
-}
-#endif
-
-#define MDBX_ASAN_POISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("POISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_POISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#define MDBX_ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("UNPOISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_UNPOISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#if defined(_WIN32) || defined(_WIN64)
-/*
- * POSIX getopt for Windows
- *
- * AT&T Public License
- *
- * Code given out at the 1985 UNIFORUM conference in Dallas.
- */
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif
-#endif /* _MSC_VER (warnings) */
-
-#include <stdio.h>
-#include <string.h>
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-/*----------------------------------------------------------------------------*/
-
-#ifndef NULL
-#define NULL 0
-#endif
-
-#ifndef EOF
-#define EOF (-1)
-#endif
-
-int optind = 1;
-int optopt;
-char *optarg;
-
-int getopt(int argc, char *const argv[], const char *opts) {
- static int sp = 1;
- int c;
- const char *cp;
-
- if (sp == 1) {
- if (optind >= argc || argv[optind][0] != '-' || argv[optind][1] == '\0')
- return EOF;
- else if (strcmp(argv[optind], "--") == 0) {
- optind++;
- return EOF;
- }
- }
- optopt = c = argv[optind][sp];
- if (c == ':' || (cp = strchr(opts, c)) == NULL) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "illegal option", c);
- if (argv[optind][++sp] == '\0') {
- optind++;
- sp = 1;
- }
- return '?';
- }
- if (*++cp == ':') {
- if (argv[optind][sp + 1] != '\0')
- optarg = &argv[optind++][sp + 1];
- else if (++optind >= argc) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "option requires an argument",
- c);
- sp = 1;
- return '?';
- } else
- optarg = argv[optind++];
- sp = 1;
- } else {
- if (argv[optind][++sp] == '\0') {
- sp = 1;
- optind++;
- }
- optarg = NULL;
- }
- return c;
-}
-
-static volatile BOOL user_break;
-static BOOL WINAPI ConsoleBreakHandlerRoutine(DWORD dwCtrlType) {
- (void)dwCtrlType;
- user_break = true;
- return true;
-}
-
-#else /* WINDOWS */
-
-static volatile sig_atomic_t user_break;
-static void signal_handler(int sig) {
- (void)sig;
- user_break = 1;
-}
-
-#endif /* !WINDOWS */
-
-static void usage(const char *prog) {
- fprintf(stderr,
- "usage: %s [-V] [-q] [-c] src_path [dest_path]\n"
- " -V\t\tprint version and exit\n"
- " -q\t\tbe quiet\n"
- " -c\t\tenable compactification (skip unused pages)\n"
- " src_path\tsource database\n"
- " dest_path\tdestination (stdout if not specified)\n",
- prog);
- exit(EXIT_FAILURE);
-}
-
-int main(int argc, char *argv[]) {
- int rc;
- MDBX_env *env = NULL;
- const char *progname = argv[0], *act;
- unsigned flags = MDBX_RDONLY;
- unsigned cpflags = 0;
- bool quiet = false;
-
- for (; argc > 1 && argv[1][0] == '-'; argc--, argv++) {
- if (argv[1][1] == 'n' && argv[1][2] == '\0')
- flags |= MDBX_NOSUBDIR;
- else if (argv[1][1] == 'c' && argv[1][2] == '\0')
- cpflags |= MDBX_CP_COMPACT;
- else if (argv[1][1] == 'q' && argv[1][2] == '\0')
- quiet = true;
- else if ((argv[1][1] == 'h' && argv[1][2] == '\0') ||
- strcmp(argv[1], "--help") == 0)
- usage(progname);
- else if (argv[1][1] == 'V' && argv[1][2] == '\0') {
- printf("mdbx_copy version %d.%d.%d.%d\n"
- " - source: %s %s, commit %s, tree %s\n"
- " - anchor: %s\n"
- " - build: %s for %s by %s\n"
- " - flags: %s\n"
- " - options: %s\n",
- mdbx_version.major, mdbx_version.minor, mdbx_version.release,
- mdbx_version.revision, mdbx_version.git.describe,
- mdbx_version.git.datetime, mdbx_version.git.commit,
- mdbx_version.git.tree, mdbx_sourcery_anchor, mdbx_build.datetime,
- mdbx_build.target, mdbx_build.compiler, mdbx_build.flags,
- mdbx_build.options);
- return EXIT_SUCCESS;
- } else
- argc = 0;
- }
-
- if (argc < 2 || argc > 3)
- usage(progname);
-
-#if defined(_WIN32) || defined(_WIN64)
- SetConsoleCtrlHandler(ConsoleBreakHandlerRoutine, true);
-#else
-#ifdef SIGPIPE
- signal(SIGPIPE, signal_handler);
-#endif
-#ifdef SIGHUP
- signal(SIGHUP, signal_handler);
-#endif
- signal(SIGINT, signal_handler);
- signal(SIGTERM, signal_handler);
-#endif /* !WINDOWS */
-
- if (!quiet) {
- fprintf((argc == 2) ? stderr : stdout,
- "mdbx_copy %s (%s, T-%s)\nRunning for copy %s to %s...\n",
- mdbx_version.git.describe, mdbx_version.git.datetime,
- mdbx_version.git.tree, argv[1], (argc == 2) ? "stdout" : argv[2]);
- fflush(NULL);
- }
-
- act = "opening environment";
- rc = mdbx_env_create(&env);
- if (rc == MDBX_SUCCESS)
- rc = mdbx_env_open(env, argv[1], flags, 0);
-
- if (rc == MDBX_SUCCESS) {
- act = "copying";
- if (argc == 2) {
- mdbx_filehandle_t fd;
-#if defined(_WIN32) || defined(_WIN64)
- fd = GetStdHandle(STD_OUTPUT_HANDLE);
-#else
- fd = fileno(stdout);
-#endif
- rc = mdbx_env_copy2fd(env, fd, cpflags);
- } else
- rc = mdbx_env_copy(env, argv[2], cpflags);
- }
- if (rc)
- fprintf(stderr, "%s: %s failed, error %d (%s)\n", progname, act, rc,
- mdbx_strerror(rc));
- mdbx_env_close(env);
-
- return rc ? EXIT_FAILURE : EXIT_SUCCESS;
-}
diff --git a/3rdparty/libmdbx/mdbx_drop.c b/3rdparty/libmdbx/mdbx_drop.c
deleted file mode 100644
index 9b0487161c459e4f5382b73ce67c602d3958bc62..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx_drop.c
+++ /dev/null
@@ -1,3715 +0,0 @@
-/* mdbx_drop.c - memory-mapped database delete tool */
-
-/*
- * Copyright 2021 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- *
- * Copyright 2016-2022 Howard Chu, Symas Corp.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#ifdef _MSC_VER
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#pragma warning(disable : 4996) /* The POSIX name is deprecated... */
-#endif /* _MSC_VER (warnings) */
-
-#define xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#define MDBX_BUILD_SOURCERY c510b93d78b5848764f976ea67f06d278a7b99f141b4c98ceb095285f834f31c_v0_11_7_0_g40ec559c
-#ifdef MDBX_CONFIG_H
-#include MDBX_CONFIG_H
-#endif
-
-#define LIBMDBX_INTERNALS
-#ifdef xMDBX_TOOLS
-#define MDBX_DEPRECATED
-#endif /* xMDBX_TOOLS */
-
-#ifdef xMDBX_ALLOY
-/* Amalgamated build */
-#define MDBX_INTERNAL_FUNC static
-#define MDBX_INTERNAL_VAR static
-#else
-/* Non-amalgamated build */
-#define MDBX_INTERNAL_FUNC
-#define MDBX_INTERNAL_VAR extern
-#endif /* xMDBX_ALLOY */
-
-/** Disables using GNU/Linux libc extensions.
- * \ingroup build_option
- * \note This option couldn't be moved to the options.h since dependant
- * control macros/defined should be prepared before include the options.h */
-#ifndef MDBX_DISABLE_GNU_SOURCE
-#define MDBX_DISABLE_GNU_SOURCE 0
-#endif
-#if MDBX_DISABLE_GNU_SOURCE
-#undef _GNU_SOURCE
-#elif (defined(__linux__) || defined(__gnu_linux__)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif /* MDBX_DISABLE_GNU_SOURCE */
-
-/*----------------------------------------------------------------------------*/
-
-/* Should be defined before any includes */
-#if !defined(_FILE_OFFSET_BITS) && !defined(__ANDROID_API__) && \
- !defined(ANDROID)
-#define _FILE_OFFSET_BITS 64
-#endif
-
-#ifdef __APPLE__
-#define _DARWIN_C_SOURCE
-#endif
-
-#ifdef _MSC_VER
-#if _MSC_FULL_VER < 190024234
-/* Actually libmdbx was not tested with compilers older than 19.00.24234 (Visual
- * Studio 2015 Update 3). But you could remove this #error and try to continue
- * at your own risk. In such case please don't rise up an issues related ONLY to
- * old compilers.
- *
- * NOTE:
- * Unfortunately, there are several different builds of "Visual Studio" that
- * are called "Visual Studio 2015 Update 3".
- *
- * The 190024234 is used here because it is minimal version of Visual Studio
- * that was used for build and testing libmdbx in recent years. Soon this
- * value will be increased to 19.0.24241.7, since build and testing using
- * "Visual Studio 2015" will be performed only at https://ci.appveyor.com.
- *
- * Please ask Microsoft (but not us) for information about version differences
- * and how to and where you can obtain the latest "Visual Studio 2015" build
- * with all fixes.
- */
-#error \
- "At least \"Microsoft C/C++ Compiler\" version 19.00.24234 (Visual Studio 2015 Update 3) is required."
-#endif
-#ifndef _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#if _MSC_VER > 1913
-#pragma warning(disable : 5045) /* Compiler will insert Spectre mitigation... \
- */
-#endif
-#if _MSC_VER > 1914
-#pragma warning( \
- disable : 5105) /* winbase.h(9531): warning C5105: macro expansion \
- producing 'defined' has undefined behavior */
-#endif
-#pragma warning(disable : 4710) /* 'xyz': function not inlined */
-#pragma warning(disable : 4711) /* function 'xyz' selected for automatic \
- inline expansion */
-#pragma warning( \
- disable : 4201) /* nonstandard extension used : nameless struct / union */
-#pragma warning(disable : 4702) /* unreachable code */
-#pragma warning(disable : 4706) /* assignment within conditional expression */
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4324) /* 'xyz': structure was padded due to \
- alignment specifier */
-#pragma warning(disable : 4310) /* cast truncates constant value */
-#pragma warning( \
- disable : 4820) /* bytes padding added after data member for alignment */
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4366) /* the result of the unary '&' operator may be \
- unaligned */
-#pragma warning(disable : 4200) /* nonstandard extension used: zero-sized \
- array in struct/union */
-#pragma warning(disable : 4204) /* nonstandard extension used: non-constant \
- aggregate initializer */
-#pragma warning( \
- disable : 4505) /* unreferenced local function has been removed */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(__GNUC__) && __GNUC__ < 9
-#pragma GCC diagnostic ignored "-Wattributes"
-#endif /* GCC < 9 */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-#include "mdbx.h"
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#ifndef __GNUC_PREREQ
-# if defined(__GNUC__) && defined(__GNUC_MINOR__)
-# define __GNUC_PREREQ(maj, min) \
- ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GNUC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GNUC_PREREQ */
-
-#ifndef __CLANG_PREREQ
-# ifdef __clang__
-# define __CLANG_PREREQ(maj,min) \
- ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min))
-# else
-# define __CLANG_PREREQ(maj,min) (0)
-# endif
-#endif /* __CLANG_PREREQ */
-
-#ifndef __GLIBC_PREREQ
-# if defined(__GLIBC__) && defined(__GLIBC_MINOR__)
-# define __GLIBC_PREREQ(maj, min) \
- ((__GLIBC__ << 16) + __GLIBC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GLIBC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GLIBC_PREREQ */
-
-#ifndef __has_warning
-# define __has_warning(x) (0)
-#endif
-
-#ifndef __has_include
-# define __has_include(x) (0)
-#endif
-
-#if __has_feature(thread_sanitizer)
-# define __SANITIZE_THREAD__ 1
-#endif
-
-#if __has_feature(address_sanitizer)
-# define __SANITIZE_ADDRESS__ 1
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __extern_C
-# ifdef __cplusplus
-# define __extern_C extern "C"
-# else
-# define __extern_C
-# endif
-#endif /* __extern_C */
-
-#if !defined(nullptr) && !defined(__cplusplus) || (__cplusplus < 201103L && !defined(_MSC_VER))
-# define nullptr NULL
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __always_inline
-# if defined(__GNUC__) || __has_attribute(__always_inline__)
-# define __always_inline __inline __attribute__((__always_inline__))
-# elif defined(_MSC_VER)
-# define __always_inline __forceinline
-# else
-# define __always_inline
-# endif
-#endif /* __always_inline */
-
-#ifndef __noinline
-# if defined(__GNUC__) || __has_attribute(__noinline__)
-# define __noinline __attribute__((__noinline__))
-# elif defined(_MSC_VER)
-# define __noinline __declspec(noinline)
-# else
-# define __noinline
-# endif
-#endif /* __noinline */
-
-#ifndef __must_check_result
-# if defined(__GNUC__) || __has_attribute(__warn_unused_result__)
-# define __must_check_result __attribute__((__warn_unused_result__))
-# else
-# define __must_check_result
-# endif
-#endif /* __must_check_result */
-
-#if !defined(__noop) && !defined(_MSC_VER)
-# define __noop(...) do {} while(0)
-#endif /* __noop */
-
-#if defined(__fallthrough) && \
- (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__))
-#undef __fallthrough
-#endif /* __fallthrough workaround for MinGW */
-
-#ifndef __fallthrough
-# if defined(__cplusplus) && (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || __cplusplus >= 201703L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(8, 0) && defined(__cplusplus) && __cplusplus >= 201103L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(7, 0) && \
- (!defined(__LCC__) || (__LCC__ == 124 && __LCC_MINOR__ >= 12) || \
- (__LCC__ == 125 && __LCC_MINOR__ >= 5) || (__LCC__ >= 126))
-# define __fallthrough __attribute__((__fallthrough__))
-# elif defined(__clang__) && defined(__cplusplus) && __cplusplus >= 201103L &&\
- __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
-# define __fallthrough [[clang::fallthrough]]
-# else
-# define __fallthrough
-# endif
-#endif /* __fallthrough */
-
-#ifndef __unreachable
-# if __GNUC_PREREQ(4,5) || __has_builtin(__builtin_unreachable)
-# define __unreachable() __builtin_unreachable()
-# elif defined(_MSC_VER)
-# define __unreachable() __assume(0)
-# else
-# define __unreachable() __noop()
-# endif
-#endif /* __unreachable */
-
-#ifndef __prefetch
-# if defined(__GNUC__) || defined(__clang__) || __has_builtin(__builtin_prefetch)
-# define __prefetch(ptr) __builtin_prefetch(ptr)
-# else
-# define __prefetch(ptr) __noop(ptr)
-# endif
-#endif /* __prefetch */
-
-#ifndef __nothrow
-# if defined(__cplusplus)
-# if __cplusplus < 201703L
-# define __nothrow throw()
-# else
-# define __nothrow noexcept(true)
-# endif /* __cplusplus */
-# elif defined(__GNUC__) || __has_attribute(__nothrow__)
-# define __nothrow __attribute__((__nothrow__))
-# elif defined(_MSC_VER) && defined(__cplusplus)
-# define __nothrow __declspec(nothrow)
-# else
-# define __nothrow
-# endif
-#endif /* __nothrow */
-
-#ifndef __hidden
-# if defined(__GNUC__) || __has_attribute(__visibility__)
-# define __hidden __attribute__((__visibility__("hidden")))
-# else
-# define __hidden
-# endif
-#endif /* __hidden */
-
-#ifndef __optimize
-# if defined(__OPTIMIZE__)
-# if (defined(__GNUC__) && !defined(__clang__)) || __has_attribute(__optimize__)
-# define __optimize(ops) __attribute__((__optimize__(ops)))
-# else
-# define __optimize(ops)
-# endif
-# else
-# define __optimize(ops)
-# endif
-#endif /* __optimize */
-
-#ifndef __hot
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __hot __attribute__((__hot__)) __optimize(3)
-# elif defined(__clang__) && !__has_attribute(__hot_) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put frequently used functions in separate section */
-# define __hot __attribute__((__section__("text.hot"))) __optimize("O3")
-# elif defined(__GNUC__) || __has_attribute(__hot__)
-# define __hot __attribute__((__hot__)) __optimize("O3")
-# else
-# define __hot __optimize("O3")
-# endif
-# else
-# define __hot
-# endif
-#endif /* __hot */
-
-#ifndef __cold
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __cold __attribute__((__cold__)) __optimize(1)
-# elif defined(__clang__) && !__has_attribute(cold) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put infrequently used functions in separate section */
-# define __cold __attribute__((__section__("text.unlikely"))) __optimize("Os")
-# elif defined(__GNUC__) || __has_attribute(cold)
-# define __cold __attribute__((__cold__)) __optimize("Os")
-# else
-# define __cold __optimize("Os")
-# endif
-# else
-# define __cold
-# endif
-#endif /* __cold */
-
-#ifndef __flatten
-# if defined(__OPTIMIZE__) && (defined(__GNUC__) || __has_attribute(__flatten__))
-# define __flatten __attribute__((__flatten__))
-# else
-# define __flatten
-# endif
-#endif /* __flatten */
-
-#ifndef likely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define likely(cond) __builtin_expect(!!(cond), 1)
-# else
-# define likely(x) (!!(x))
-# endif
-#endif /* likely */
-
-#ifndef unlikely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define unlikely(cond) __builtin_expect(!!(cond), 0)
-# else
-# define unlikely(x) (!!(x))
-# endif
-#endif /* unlikely */
-
-#ifndef __anonymous_struct_extension__
-# if defined(__GNUC__)
-# define __anonymous_struct_extension__ __extension__
-# else
-# define __anonymous_struct_extension__
-# endif
-#endif /* __anonymous_struct_extension__ */
-
-#ifndef __Wpedantic_format_voidptr
- MDBX_MAYBE_UNUSED MDBX_PURE_FUNCTION static __inline const void*
- __Wpedantic_format_voidptr(const void* ptr) {return ptr;}
-# define __Wpedantic_format_voidptr(ARG) __Wpedantic_format_voidptr(ARG)
-#endif /* __Wpedantic_format_voidptr */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(MDBX_USE_VALGRIND)
-# include <valgrind/memcheck.h>
-# ifndef VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE
- /* LY: available since Valgrind 3.10 */
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# endif
-#elif !defined(RUNNING_ON_VALGRIND)
-# define VALGRIND_CREATE_MEMPOOL(h,r,z)
-# define VALGRIND_DESTROY_MEMPOOL(h)
-# define VALGRIND_MEMPOOL_TRIM(h,a,s)
-# define VALGRIND_MEMPOOL_ALLOC(h,a,s)
-# define VALGRIND_MEMPOOL_FREE(h,a)
-# define VALGRIND_MEMPOOL_CHANGE(h,a,b,s)
-# define VALGRIND_MAKE_MEM_NOACCESS(a,s)
-# define VALGRIND_MAKE_MEM_DEFINED(a,s)
-# define VALGRIND_MAKE_MEM_UNDEFINED(a,s)
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_CHECK_MEM_IS_ADDRESSABLE(a,s) (0)
-# define VALGRIND_CHECK_MEM_IS_DEFINED(a,s) (0)
-# define RUNNING_ON_VALGRIND (0)
-#endif /* MDBX_USE_VALGRIND */
-
-#ifdef __SANITIZE_ADDRESS__
-# include <sanitizer/asan_interface.h>
-#elif !defined(ASAN_POISON_MEMORY_REGION)
-# define ASAN_POISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-# define ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-#endif /* __SANITIZE_ADDRESS__ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef ARRAY_LENGTH
-# ifdef __cplusplus
- template <typename T, size_t N>
- char (&__ArraySizeHelper(T (&array)[N]))[N];
-# define ARRAY_LENGTH(array) (sizeof(::__ArraySizeHelper(array)))
-# else
-# define ARRAY_LENGTH(array) (sizeof(array) / sizeof(array[0]))
-# endif
-#endif /* ARRAY_LENGTH */
-
-#ifndef ARRAY_END
-# define ARRAY_END(array) (&array[ARRAY_LENGTH(array)])
-#endif /* ARRAY_END */
-
-#define CONCAT(a,b) a##b
-#define XCONCAT(a,b) CONCAT(a,b)
-
-#ifndef offsetof
-# define offsetof(type, member) __builtin_offsetof(type, member)
-#endif /* offsetof */
-
-#ifndef container_of
-# define container_of(ptr, type, member) \
- ((type *)((char *)(ptr) - offsetof(type, member)))
-#endif /* container_of */
-
-#define MDBX_TETRAD(a, b, c, d) \
- ((uint32_t)(a) << 24 | (uint32_t)(b) << 16 | (uint32_t)(c) << 8 | (d))
-
-#define MDBX_STRING_TETRAD(str) MDBX_TETRAD(str[0], str[1], str[2], str[3])
-
-#define FIXME "FIXME: " __FILE__ ", " MDBX_STRINGIFY(__LINE__)
-
-#ifndef STATIC_ASSERT_MSG
-# if defined(static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) static_assert(expr, msg)
-# elif defined(_STATIC_ASSERT)
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif defined(_MSC_VER)
-# include <crtdbg.h>
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) \
- || __has_feature(c_static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) _Static_assert(expr, msg)
-# else
-# define STATIC_ASSERT_MSG(expr, msg) switch (0) {case 0:case (expr):;}
-# endif
-#endif /* STATIC_ASSERT */
-
-#ifndef STATIC_ASSERT
-# define STATIC_ASSERT(expr) STATIC_ASSERT_MSG(expr, #expr)
-#endif
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__GNUC__) && !__GNUC_PREREQ(4, 2)
-/* Actually libmdbx was not tested with compilers older than GCC 4.2.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required GCC >= 4.2"
-#endif
-
-#if defined(__clang__) && !__CLANG_PREREQ(3, 8)
-/* Actually libmdbx was not tested with CLANG older than 3.8.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required CLANG >= 3.8"
-#endif
-
-#if defined(__GLIBC__) && !__GLIBC_PREREQ(2, 12)
-/* Actually libmdbx was not tested with something older than glibc 2.12.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old systems.
- */
-#warning "libmdbx was only tested with GLIBC >= 2.12."
-#endif
-
-#ifdef __SANITIZE_THREAD__
-#warning \
- "libmdbx don't compatible with ThreadSanitizer, you will get a lot of false-positive issues."
-#endif /* __SANITIZE_THREAD__ */
-
-#if __has_warning("-Wnested-anon-types")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wnested-anon-types"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wnested-anon-types"
-#else
-#pragma warning disable "nested-anon-types"
-#endif
-#endif /* -Wnested-anon-types */
-
-#if __has_warning("-Wconstant-logical-operand")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wconstant-logical-operand"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wconstant-logical-operand"
-#else
-#pragma warning disable "constant-logical-operand"
-#endif
-#endif /* -Wconstant-logical-operand */
-
-#if defined(__LCC__) && (__LCC__ <= 121)
-/* bug #2798 */
-#pragma diag_suppress alignment_reduction_ignored
-#elif defined(__ICC)
-#pragma warning(disable : 3453 1366)
-#elif __has_warning("-Walignment-reduction-ignored")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Walignment-reduction-ignored"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Walignment-reduction-ignored"
-#else
-#pragma warning disable "alignment-reduction-ignored"
-#endif
-#endif /* -Walignment-reduction-ignored */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(_WIN32) || defined(_WIN64)
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if !defined(_NO_CRT_STDIO_INLINE) && MDBX_BUILD_SHARED_LIBRARY && \
- !defined(xMDBX_TOOLS) && MDBX_WITHOUT_MSVC_CRT
-#define _NO_CRT_STDIO_INLINE
-#endif
-#elif !defined(_POSIX_C_SOURCE)
-#define _POSIX_C_SOURCE 200809L
-#endif /* Windows */
-
-/*----------------------------------------------------------------------------*/
-/* C99 includes */
-#include <inttypes.h>
-#include <stddef.h>
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <assert.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-/* C11' alignas() */
-#if __has_include(<stdalign.h>)
-#include <stdalign.h>
-#endif
-#if defined(alignas) || defined(__cplusplus)
-#define MDBX_ALIGNAS(N) alignas(N)
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
-#define MDBX_ALIGNAS(N) _Alignas(N)
-#elif defined(_MSC_VER)
-#define MDBX_ALIGNAS(N) __declspec(align(N))
-#elif __has_attribute(__aligned__) || defined(__GNUC__)
-#define MDBX_ALIGNAS(N) __attribute__((__aligned__(N)))
-#else
-#error "FIXME: Required alignas() or equivalent."
-#endif /* MDBX_ALIGNAS */
-
-/*----------------------------------------------------------------------------*/
-/* Systems includes */
-
-#ifdef __APPLE__
-#include <TargetConditionals.h>
-#endif /* Apple OSX & iOS */
-
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__)
-#include <sys/cdefs.h>
-#include <sys/mount.h>
-#include <sys/sysctl.h>
-#include <sys/types.h>
-#if defined(__FreeBSD__) || defined(__DragonFly__)
-#include <vm/vm_param.h>
-#elif defined(__OpenBSD__) || defined(__NetBSD__)
-#include <uvm/uvm_param.h>
-#else
-#define SYSCTL_LEGACY_NONCONST_MIB
-#endif
-#ifndef __MACH__
-#include <sys/vmmeter.h>
-#endif
-#else
-#include <malloc.h>
-#if !(defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(_WIN32) || defined(_WIN64))
-#include <mntent.h>
-#endif /* !Solaris */
-#endif /* !xBSD */
-
-#if defined(__FreeBSD__) || __has_include(<malloc_np.h>)
-#include <malloc_np.h>
-#endif
-
-#if defined(__APPLE__) || defined(__MACH__) || __has_include(<malloc/malloc.h>)
-#include <malloc/malloc.h>
-#endif /* MacOS */
-
-#if defined(__MACH__)
-#include <mach/host_info.h>
-#include <mach/mach_host.h>
-#include <mach/mach_port.h>
-#include <uuid/uuid.h>
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-#include <sched.h>
-#include <sys/sendfile.h>
-#include <sys/statfs.h>
-#endif /* Linux */
-
-#ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 0
-#endif
-
-#ifndef _XOPEN_SOURCE_EXTENDED
-#define _XOPEN_SOURCE_EXTENDED 0
-#else
-#include <utmpx.h>
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#include <kstat.h>
-#include <sys/mnttab.h>
-/* On Solaris, it's easier to add a missing prototype rather than find a
- * combination of #defines that break nothing. */
-__extern_C key_t ftok(const char *, int);
-#endif /* SunOS/Solaris */
-
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef _WIN32_WINNT
-#define _WIN32_WINNT 0x0601 /* Windows 7 */
-#elif _WIN32_WINNT < 0x0500
-#error At least 'Windows 2000' API is required for libmdbx.
-#endif /* _WIN32_WINNT */
-#if (defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* MinGW */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif /* WIN32_LEAN_AND_MEAN */
-#include <excpt.h>
-#include <tlhelp32.h>
-#include <windows.h>
-#include <winnt.h>
-#include <winternl.h>
-#define HAVE_SYS_STAT_H
-#define HAVE_SYS_TYPES_H
-typedef HANDLE mdbx_thread_t;
-typedef unsigned mdbx_thread_key_t;
-#define MAP_FAILED NULL
-#define HIGH_DWORD(v) ((DWORD)((sizeof(v) > 4) ? ((uint64_t)(v) >> 32) : 0))
-#define THREAD_CALL WINAPI
-#define THREAD_RESULT DWORD
-typedef struct {
- HANDLE mutex;
- HANDLE event[2];
-} mdbx_condpair_t;
-typedef CRITICAL_SECTION mdbx_fastmutex_t;
-
-#if !defined(_MSC_VER) && !defined(__try)
-/* *INDENT-OFF* */
-/* clang-format off */
-#define __try
-#define __except(COND) if(false)
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* stub for MSVC's __try/__except */
-
-#if MDBX_WITHOUT_MSVC_CRT
-
-#ifndef mdbx_malloc
-static inline void *mdbx_malloc(size_t bytes) {
- return HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_malloc */
-
-#ifndef mdbx_calloc
-static inline void *mdbx_calloc(size_t nelem, size_t size) {
- return HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, nelem * size);
-}
-#endif /* mdbx_calloc */
-
-#ifndef mdbx_realloc
-static inline void *mdbx_realloc(void *ptr, size_t bytes) {
- return ptr ? HeapReAlloc(GetProcessHeap(), 0, ptr, bytes)
- : HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_realloc */
-
-#ifndef mdbx_free
-static inline void mdbx_free(void *ptr) { HeapFree(GetProcessHeap(), 0, ptr); }
-#endif /* mdbx_free */
-
-#else /* MDBX_WITHOUT_MSVC_CRT */
-
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup _strdup
-
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-#ifndef snprintf
-#define snprintf _snprintf /* ntdll */
-#endif
-
-#ifndef vsnprintf
-#define vsnprintf _vsnprintf /* ntdll */
-#endif
-
-#else /*----------------------------------------------------------------------*/
-
-#include <unistd.h>
-#if !defined(_POSIX_MAPPED_FILES) || _POSIX_MAPPED_FILES < 1
-#error "libmdbx requires the _POSIX_MAPPED_FILES feature"
-#endif /* _POSIX_MAPPED_FILES */
-
-#include <pthread.h>
-#include <semaphore.h>
-#include <signal.h>
-#include <sys/file.h>
-#include <sys/ipc.h>
-#include <sys/mman.h>
-#include <sys/param.h>
-#include <sys/stat.h>
-#include <sys/statvfs.h>
-#include <sys/uio.h>
-typedef pthread_t mdbx_thread_t;
-typedef pthread_key_t mdbx_thread_key_t;
-#define INVALID_HANDLE_VALUE (-1)
-#define THREAD_CALL
-#define THREAD_RESULT void *
-typedef struct {
- pthread_mutex_t mutex;
- pthread_cond_t cond[2];
-} mdbx_condpair_t;
-typedef pthread_mutex_t mdbx_fastmutex_t;
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup strdup
-#endif /* Platform */
-
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
-/* malloc_usable_size() already provided */
-#elif defined(__APPLE__)
-#define malloc_usable_size(ptr) malloc_size(ptr)
-#elif defined(_MSC_VER) && !MDBX_WITHOUT_MSVC_CRT
-#define malloc_usable_size(ptr) _msize(ptr)
-#endif /* malloc_usable_size */
-
-#ifdef __ANDROID_API__
-#include <android/log.h>
-#if __ANDROID_API__ >= 21
-#include <sys/sendfile.h>
-#endif
-#endif /* Android */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(HAVE_SYS_STAT_H) || __has_include(<sys/stat.h>)
-#include <sys/stat.h>
-#endif
-#if defined(HAVE_SYS_TYPES_H) || __has_include(<sys/types.h>)
-#include <sys/types.h>
-#endif
-#if defined(HAVE_SYS_FILE_H) || __has_include(<sys/file.h>)
-#include <sys/file.h>
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-
-#ifndef SSIZE_MAX
-#define SSIZE_MAX INTPTR_MAX
-#endif
-
-#if !defined(MADV_DODUMP) && defined(MADV_CORE)
-#define MADV_DODUMP MADV_CORE
-#endif /* MADV_CORE -> MADV_DODUMP */
-
-#if !defined(MADV_DONTDUMP) && defined(MADV_NOCORE)
-#define MADV_DONTDUMP MADV_NOCORE
-#endif /* MADV_NOCORE -> MADV_DONTDUMP */
-
-#if defined(i386) || defined(__386) || defined(__i386) || defined(__i386__) || \
- defined(i486) || defined(__i486) || defined(__i486__) || \
- defined(i586) | defined(__i586) || defined(__i586__) || defined(i686) || \
- defined(__i686) || defined(__i686__) || defined(_M_IX86) || \
- defined(_X86_) || defined(__THW_INTEL__) || defined(__I86__) || \
- defined(__INTEL__) || defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64__) || defined(__amd64) || defined(_M_X64) || \
- defined(_M_AMD64) || defined(__IA32__) || defined(__INTEL__)
-#ifndef __ia32__
-/* LY: define neutral __ia32__ for x86 and x86-64 */
-#define __ia32__ 1
-#endif /* __ia32__ */
-#if !defined(__amd64__) && (defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64) || defined(_M_X64))
-/* LY: define trusty __amd64__ for all AMD64/x86-64 arch */
-#define __amd64__ 1
-#endif /* __amd64__ */
-#endif /* all x86 */
-
-#if (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
-#error \
- "Sanity checking failed: Two's complement, reasonably sized integer types"
-#endif
-
-#if UINTPTR_MAX > 0xffffFFFFul || ULONG_MAX > 0xffffFFFFul
-#define MDBX_WORDBITS 64
-#else
-#define MDBX_WORDBITS 32
-#endif /* MDBX_WORDBITS */
-
-#if defined(__ANDROID_API__) || defined(ANDROID)
-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "_FILE_OFFSET_BITS != MDBX_WORDBITS" (_FILE_OFFSET_BITS != MDBX_WORDBITS)
-#elif defined(__FILE_OFFSET_BITS) && __FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "__FILE_OFFSET_BITS != MDBX_WORDBITS" (__FILE_OFFSET_BITS != MDBX_WORDBITS)
-#endif
-#endif /* Android */
-
-/*----------------------------------------------------------------------------*/
-/* Compiler's includes for builtins/intrinsics */
-
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
-#include <intrin.h>
-#elif __GNUC_PREREQ(4, 4) || defined(__clang__)
-#if defined(__ia32__) || defined(__e2k__)
-#include <x86intrin.h>
-#endif /* __ia32__ */
-#if defined(__ia32__)
-#include <cpuid.h>
-#endif /* __ia32__ */
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
-#include <mbarrier.h>
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
-#include <machine/sys/inline.h>
-#elif defined(__IBMC__) && defined(__powerpc)
-#include <atomic.h>
-#elif defined(_AIX)
-#include <builtins.h>
-#include <sys/atomic_op.h>
-#elif (defined(__osf__) && defined(__DECC)) || defined(__alpha)
-#include <c_asm.h>
-#include <machine/builtins.h>
-#elif defined(__MWERKS__)
-/* CodeWarrior - troubles ? */
-#pragma gcc_extensions
-#elif defined(__SNC__)
-/* Sony PS3 - troubles ? */
-#elif defined(__hppa__) || defined(__hppa)
-#include <machine/inline.h>
-#else
-#error Unsupported C compiler, please use GNU C 4.4 or newer
-#endif /* Compiler */
-
-/*----------------------------------------------------------------------------*/
-/* Byteorder */
-
-#if !defined(__BYTE_ORDER__) || !defined(__ORDER_LITTLE_ENDIAN__) || \
- !defined(__ORDER_BIG_ENDIAN__)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(__GLIBC__) || defined(__GNU_LIBRARY__) || defined(__ANDROID_API__) || \
- defined(HAVE_ENDIAN_H) || __has_include(<endian.h>)
-#include <endian.h>
-#elif defined(__APPLE__) || defined(__MACH__) || defined(__OpenBSD__) || \
- defined(HAVE_MACHINE_ENDIAN_H) || __has_include(<machine/endian.h>)
-#include <machine/endian.h>
-#elif defined(HAVE_SYS_ISA_DEFS_H) || __has_include(<sys/isa_defs.h>)
-#include <sys/isa_defs.h>
-#elif (defined(HAVE_SYS_TYPES_H) && defined(HAVE_SYS_ENDIAN_H)) || \
- (__has_include(<sys/types.h>) && __has_include(<sys/endian.h>))
-#include <sys/endian.h>
-#include <sys/types.h>
-#elif defined(__bsdi__) || defined(__DragonFly__) || defined(__FreeBSD__) || \
- defined(__NetBSD__) || \
- defined(HAVE_SYS_PARAM_H) || __has_include(<sys/param.h>)
-#include <sys/param.h>
-#endif /* OS */
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-
-#else
-#error __BYTE_ORDER__ should be defined.
-#endif /* Arch */
-
-#endif
-#endif /* __BYTE_ORDER__ || __ORDER_LITTLE_ENDIAN__ || __ORDER_BIG_ENDIAN__ */
-
-/* Get the size of a memory page for the system.
- * This is the basic size that the platform's memory manager uses, and is
- * fundamental to the use of memory-mapped files. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline size_t
-mdbx_syspagesize(void) {
-#if defined(_WIN32) || defined(_WIN64)
- SYSTEM_INFO si;
- GetSystemInfo(&si);
- return si.dwPageSize;
-#else
- return sysconf(_SC_PAGE_SIZE);
-#endif
-}
-
-typedef struct mdbx_mmap_param {
- union {
- void *address;
- uint8_t *dxb;
- struct MDBX_lockinfo *lck;
- };
- mdbx_filehandle_t fd;
- size_t limit; /* mapping length, but NOT a size of file nor DB */
- size_t current; /* mapped region size, i.e. the size of file and DB */
- uint64_t filesize /* in-process cache of a file size */;
-#if defined(_WIN32) || defined(_WIN64)
- HANDLE section; /* memory-mapped section handle */
-#endif
-} mdbx_mmap_t;
-
-typedef union bin128 {
- __anonymous_struct_extension__ struct { uint64_t x, y; };
- __anonymous_struct_extension__ struct { uint32_t a, b, c, d; };
-} bin128_t;
-
-#if defined(_WIN32) || defined(_WIN64)
-typedef union MDBX_srwlock {
- __anonymous_struct_extension__ struct {
- long volatile readerCount;
- long volatile writerCount;
- };
- RTL_SRWLOCK native;
-} MDBX_srwlock;
-#endif /* Windows */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny);
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny);
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-#if defined(__cplusplus) && !defined(__STDC_NO_ATOMICS__) && (__has_include(<cstdatomic>) || __has_extension(cxx_atomic))
-#include <cstdatomic>
-#define MDBX_HAVE_C11ATOMICS
-#elif !defined(__cplusplus) && \
- (__STDC_VERSION__ >= 201112L || __has_extension(c_atomic)) && \
- !defined(__STDC_NO_ATOMICS__) && \
- (__GNUC_PREREQ(4, 9) || __CLANG_PREREQ(3, 8) || \
- !(defined(__GNUC__) || defined(__clang__)))
-#include <stdatomic.h>
-#define MDBX_HAVE_C11ATOMICS
-#elif defined(__GNUC__) || defined(__clang__)
-#elif defined(_MSC_VER)
-#pragma warning(disable : 4163) /* 'xyz': not available as an intrinsic */
-#pragma warning(disable : 4133) /* 'function': incompatible types - from \
- 'size_t' to 'LONGLONG' */
-#pragma warning(disable : 4244) /* 'return': conversion from 'LONGLONG' to \
- 'std::size_t', possible loss of data */
-#pragma warning(disable : 4267) /* 'function': conversion from 'size_t' to \
- 'long', possible loss of data */
-#pragma intrinsic(_InterlockedExchangeAdd, _InterlockedCompareExchange)
-#pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedCompareExchange64)
-#elif defined(__APPLE__)
-#include <libkern/OSAtomic.h>
-#else
-#error FIXME atomic-ops
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Memory/Compiler barriers, cache coherence */
-
-#if __has_include(<sys/cachectl.h>)
-#include <sys/cachectl.h>
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS should have explicit cache control */
-#include <sys/cachectl.h>
-#endif
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_compiler_barrier(void) {
-#if defined(__clang__) || defined(__GNUC__)
- __asm__ __volatile__("" ::: "memory");
-#elif defined(_MSC_VER)
- _ReadWriteBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
- __memory_barrier();
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __compiler_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_sched_fence(/* LY: no-arg meaning 'all expect ALU', e.g. 0x3D3D */);
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __fence();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_memory_barrier(void) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(memory_order_seq_cst);
-#elif defined(__ATOMIC_SEQ_CST)
-#ifdef __clang__
- __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
- __atomic_thread_fence(__ATOMIC_SEQ_CST);
-#endif
-#elif defined(__clang__) || defined(__GNUC__)
- __sync_synchronize();
-#elif defined(_WIN32) || defined(_WIN64)
- MemoryBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
-#if defined(__ia32__)
- _mm_mfence();
-#else
- __mf();
-#endif
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __machine_rw_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_mf();
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __lwsync();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* libc compatibility stuff */
-
-#if (!defined(__GLIBC__) && __GLIBC_PREREQ(2, 1)) && \
- (defined(_GNU_SOURCE) || defined(_BSD_SOURCE))
-#define mdbx_asprintf asprintf
-#define mdbx_vasprintf vasprintf
-#else
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC
- MDBX_PRINTF_ARGS(2, 3) int mdbx_asprintf(char **strp, const char *fmt, ...);
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt, va_list ap);
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* OS abstraction layer stuff */
-
-/* max bytes to write in one call */
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_WRITE UINT32_C(0x01000000)
-#else
-#define MAX_WRITE UINT32_C(0x3fff0000)
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-MDBX_INTERNAL_VAR uint32_t mdbx_linux_kernel_version;
-MDBX_INTERNAL_VAR bool mdbx_RunningOnWSL1 /* Windows Subsystem 1 for Linux */;
-#endif /* Linux */
-
-#ifndef mdbx_strdup
-LIBMDBX_API char *mdbx_strdup(const char *str);
-#endif
-
-MDBX_MAYBE_UNUSED static __inline int mdbx_get_errno(void) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD rc = GetLastError();
-#else
- int rc = errno;
-#endif
- return rc;
-}
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result);
-#endif
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr);
-#endif
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair, bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair);
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex);
-
-MDBX_INTERNAL_FUNC int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov,
- int iovcnt, uint64_t offset,
- size_t expected_written);
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t count,
- uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t count, uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t count);
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg);
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread);
-
-enum mdbx_syncmode_bits {
- MDBX_SYNC_NONE = 0,
- MDBX_SYNC_DATA = 1,
- MDBX_SYNC_SIZE = 2,
- MDBX_SYNC_IODQ = 4
-};
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- const enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length);
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos);
-MDBX_INTERNAL_FUNC int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length);
-
-enum mdbx_openfile_purpose {
- MDBX_OPEN_DXB_READ = 0,
- MDBX_OPEN_DXB_LAZY = 1,
- MDBX_OPEN_DXB_DSYNC = 2,
- MDBX_OPEN_LCK = 3,
- MDBX_OPEN_COPY = 4,
- MDBX_OPEN_DELETE = 5
-};
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait);
-
-#define MMAP_OPTION_TRUNCATE 1
-#define MMAP_OPTION_SEMAPHORE 2
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t must, const size_t limit,
- const unsigned options);
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map);
-#define MDBX_MRESIZE_MAY_MOVE 0x00000100
-#define MDBX_MRESIZE_MAY_UNMAP 0x00000200
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit);
-#if defined(_WIN32) || defined(_WIN64)
-typedef struct {
- unsigned limit, count;
- HANDLE handles[31];
-} mdbx_handle_array_t;
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array);
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array);
-#endif /* Windows */
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err);
-
-MDBX_MAYBE_UNUSED static __inline uint32_t mdbx_getpid(void) {
- STATIC_ASSERT(sizeof(mdbx_pid_t) <= sizeof(uint32_t));
-#if defined(_WIN32) || defined(_WIN64)
- return GetCurrentProcessId();
-#else
- STATIC_ASSERT(sizeof(pid_t) <= sizeof(uint32_t));
- return getpid();
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline uintptr_t mdbx_thread_self(void) {
- mdbx_tid_t thunk;
- STATIC_ASSERT(sizeof(uintptr_t) >= sizeof(thunk));
-#if defined(_WIN32) || defined(_WIN64)
- thunk = GetCurrentThreadId();
-#else
- thunk = pthread_self();
-#endif
- return (uintptr_t)thunk;
-}
-
-#if !defined(_WIN32) && !defined(_WIN64)
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void);
-#else
-static __inline int mdbx_check_tid4bionic(void) { return 0; }
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-MDBX_MAYBE_UNUSED static __inline int
-mdbx_pthread_mutex_lock(pthread_mutex_t *mutex) {
- int err = mdbx_check_tid4bionic();
- return unlikely(err) ? err : pthread_mutex_lock(mutex);
-}
-#endif /* !Windows */
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void);
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16);
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime);
-
-MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void);
-/*----------------------------------------------------------------------------*/
-/* lck stuff */
-
-/// \brief Initialization of synchronization primitives linked with MDBX_env
-/// instance both in LCK-file and within the current process.
-/// \param
-/// global_uniqueness_flag = true - denotes that there are no other processes
-/// working with DB and LCK-file. Thus the function MUST initialize
-/// shared synchronization objects in memory-mapped LCK-file.
-/// global_uniqueness_flag = false - denotes that at least one process is
-/// already working with DB and LCK-file, including the case when DB
-/// has already been opened in the current process. Thus the function
-/// MUST NOT initialize shared synchronization objects in memory-mapped
-/// LCK-file that are already in use.
-/// \return Error code or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag);
-
-/// \brief Disconnects from shared interprocess objects and destructs
-/// synchronization objects linked with MDBX_env instance
-/// within the current process.
-/// \param
-/// inprocess_neighbor = NULL - if the current process does not have other
-/// instances of MDBX_env linked with the DB being closed.
-/// Thus the function MUST check for other processes working with DB or
-/// LCK-file, and keep or destroy shared synchronization objects in
-/// memory-mapped LCK-file depending on the result.
-/// inprocess_neighbor = not-NULL - pointer to another instance of MDBX_env
-/// (anyone of there is several) working with DB or LCK-file within the
-/// current process. Thus the function MUST NOT try to acquire exclusive
-/// lock and/or try to destruct shared synchronization objects linked with
-/// DB or LCK-file. Moreover, the implementation MUST ensure correct work
-/// of other instances of MDBX_env within the current process, e.g.
-/// restore POSIX-fcntl locks after the closing of file descriptors.
-/// \return Error code (MDBX_PANIC) or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor);
-
-/// \brief Connects to shared interprocess locking objects and tries to acquire
-/// the maximum lock level (shared if exclusive is not available)
-/// Depending on implementation or/and platform (Windows) this function may
-/// acquire the non-OS super-level lock (e.g. for shared synchronization
-/// objects initialization), which will be downgraded to OS-exclusive or
-/// shared via explicit calling of mdbx_lck_downgrade().
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if an exclusive lock was acquired and thus
-/// the current process is the first and only after the last use of DB.
-/// MDBX_RESULT_FALSE (0) - if a shared lock was acquired and thus
-/// DB has already been opened and now is used by other processes.
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env);
-
-/// \brief Downgrades the level of initially acquired lock to
-/// operational level specified by argument. The reson for such downgrade:
-/// - unblocking of other processes that are waiting for access, i.e.
-/// if (env->me_flags & MDBX_EXCLUSIVE) != 0, then other processes
-/// should be made aware that access is unavailable rather than
-/// wait for it.
-/// - freeing locks that interfere file operation (especially for Windows)
-/// (env->me_flags & MDBX_EXCLUSIVE) == 0 - downgrade to shared lock.
-/// (env->me_flags & MDBX_EXCLUSIVE) != 0 - downgrade to exclusive
-/// operational lock.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env);
-
-/// \brief Locks LCK-file or/and table of readers for (de)registering.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env);
-
-/// \brief Unlocks LCK-file or/and table of readers after (de)registering.
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env);
-
-/// \brief Acquires lock for DB change (on writing transaction start)
-/// Reading transactions will not be blocked.
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-/// \return Error code or zero on success
-LIBMDBX_API int mdbx_txn_lock(MDBX_env *env, bool dont_wait);
-
-/// \brief Releases lock once DB changes is made (after writing transaction
-/// has finished).
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-LIBMDBX_API void mdbx_txn_unlock(MDBX_env *env);
-
-/// \brief Sets alive-flag of reader presence (indicative lock) for PID of
-/// the current process. The function does no more than needed for
-/// the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env);
-
-/// \brief Resets alive-flag of reader presence (indicative lock)
-/// for PID of the current process. The function does no more than needed
-/// for the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env);
-
-/// \brief Checks for reading process status with the given pid with help of
-/// alive-flag of presence (indicative lock) or using another way.
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if the reader process with the given PID is alive
-/// and working with DB (indicative lock is present).
-/// MDBX_RESULT_FALSE (0) - if the reader process with the given PID is absent
-/// or not working with DB (indicative lock is not present).
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid);
-
-#if defined(_WIN32) || defined(_WIN64)
-
-typedef void(WINAPI *MDBX_srwlock_function)(MDBX_srwlock *);
-MDBX_INTERNAL_VAR MDBX_srwlock_function mdbx_srwlock_Init,
- mdbx_srwlock_AcquireShared, mdbx_srwlock_ReleaseShared,
- mdbx_srwlock_AcquireExclusive, mdbx_srwlock_ReleaseExclusive;
-
-#if _WIN32_WINNT < 0x0600 /* prior to Windows Vista */
-typedef enum _FILE_INFO_BY_HANDLE_CLASS {
- FileBasicInfo,
- FileStandardInfo,
- FileNameInfo,
- FileRenameInfo,
- FileDispositionInfo,
- FileAllocationInfo,
- FileEndOfFileInfo,
- FileStreamInfo,
- FileCompressionInfo,
- FileAttributeTagInfo,
- FileIdBothDirectoryInfo,
- FileIdBothDirectoryRestartInfo,
- FileIoPriorityHintInfo,
- FileRemoteProtocolInfo,
- MaximumFileInfoByHandleClass
-} FILE_INFO_BY_HANDLE_CLASS,
- *PFILE_INFO_BY_HANDLE_CLASS;
-
-typedef struct _FILE_END_OF_FILE_INFO {
- LARGE_INTEGER EndOfFile;
-} FILE_END_OF_FILE_INFO, *PFILE_END_OF_FILE_INFO;
-
-#define REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK 0x00000001
-#define REMOTE_PROTOCOL_INFO_FLAG_OFFLINE 0x00000002
-
-typedef struct _FILE_REMOTE_PROTOCOL_INFO {
- USHORT StructureVersion;
- USHORT StructureSize;
- DWORD Protocol;
- USHORT ProtocolMajorVersion;
- USHORT ProtocolMinorVersion;
- USHORT ProtocolRevision;
- USHORT Reserved;
- DWORD Flags;
- struct {
- DWORD Reserved[8];
- } GenericReserved;
- struct {
- DWORD Reserved[16];
- } ProtocolSpecificReserved;
-} FILE_REMOTE_PROTOCOL_INFO, *PFILE_REMOTE_PROTOCOL_INFO;
-
-#endif /* _WIN32_WINNT < 0x0600 (prior to Windows Vista) */
-
-typedef BOOL(WINAPI *MDBX_GetFileInformationByHandleEx)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_GetFileInformationByHandleEx
- mdbx_GetFileInformationByHandleEx;
-
-typedef BOOL(WINAPI *MDBX_GetVolumeInformationByHandleW)(
- _In_ HANDLE hFile, _Out_opt_ LPWSTR lpVolumeNameBuffer,
- _In_ DWORD nVolumeNameSize, _Out_opt_ LPDWORD lpVolumeSerialNumber,
- _Out_opt_ LPDWORD lpMaximumComponentLength,
- _Out_opt_ LPDWORD lpFileSystemFlags,
- _Out_opt_ LPWSTR lpFileSystemNameBuffer, _In_ DWORD nFileSystemNameSize);
-MDBX_INTERNAL_VAR MDBX_GetVolumeInformationByHandleW
- mdbx_GetVolumeInformationByHandleW;
-
-typedef DWORD(WINAPI *MDBX_GetFinalPathNameByHandleW)(_In_ HANDLE hFile,
- _Out_ LPWSTR lpszFilePath,
- _In_ DWORD cchFilePath,
- _In_ DWORD dwFlags);
-MDBX_INTERNAL_VAR MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-
-typedef BOOL(WINAPI *MDBX_SetFileInformationByHandle)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_SetFileInformationByHandle
- mdbx_SetFileInformationByHandle;
-
-typedef NTSTATUS(NTAPI *MDBX_NtFsControlFile)(
- IN HANDLE FileHandle, IN OUT HANDLE Event,
- IN OUT PVOID /* PIO_APC_ROUTINE */ ApcRoutine, IN OUT PVOID ApcContext,
- OUT PIO_STATUS_BLOCK IoStatusBlock, IN ULONG FsControlCode,
- IN OUT PVOID InputBuffer, IN ULONG InputBufferLength,
- OUT OPTIONAL PVOID OutputBuffer, IN ULONG OutputBufferLength);
-MDBX_INTERNAL_VAR MDBX_NtFsControlFile mdbx_NtFsControlFile;
-
-typedef uint64_t(WINAPI *MDBX_GetTickCount64)(void);
-MDBX_INTERNAL_VAR MDBX_GetTickCount64 mdbx_GetTickCount64;
-
-#if !defined(_WIN32_WINNT_WIN8) || _WIN32_WINNT < _WIN32_WINNT_WIN8
-typedef struct _WIN32_MEMORY_RANGE_ENTRY {
- PVOID VirtualAddress;
- SIZE_T NumberOfBytes;
-} WIN32_MEMORY_RANGE_ENTRY, *PWIN32_MEMORY_RANGE_ENTRY;
-#endif /* Windows 8.x */
-
-typedef BOOL(WINAPI *MDBX_PrefetchVirtualMemory)(
- HANDLE hProcess, ULONG_PTR NumberOfEntries,
- PWIN32_MEMORY_RANGE_ENTRY VirtualAddresses, ULONG Flags);
-MDBX_INTERNAL_VAR MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-
-typedef enum _SECTION_INHERIT { ViewShare = 1, ViewUnmap = 2 } SECTION_INHERIT;
-
-typedef NTSTATUS(NTAPI *MDBX_NtExtendSection)(IN HANDLE SectionHandle,
- IN PLARGE_INTEGER NewSectionSize);
-MDBX_INTERNAL_VAR MDBX_NtExtendSection mdbx_NtExtendSection;
-
-static __inline bool mdbx_RunningUnderWine(void) {
- return !mdbx_NtExtendSection;
-}
-
-typedef LSTATUS(WINAPI *MDBX_RegGetValueA)(HKEY hkey, LPCSTR lpSubKey,
- LPCSTR lpValue, DWORD dwFlags,
- LPDWORD pdwType, PVOID pvData,
- LPDWORD pcbData);
-MDBX_INTERNAL_VAR MDBX_RegGetValueA mdbx_RegGetValueA;
-
-NTSYSAPI ULONG RtlRandomEx(PULONG Seed);
-
-#endif /* Windows */
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_MSC_VER) && _MSC_VER >= 1900
-/* LY: MSVC 2015/2017/2019 has buggy/inconsistent PRIuPTR/PRIxPTR macros
- * for internal format-args checker. */
-#undef PRIuPTR
-#undef PRIiPTR
-#undef PRIdPTR
-#undef PRIxPTR
-#define PRIuPTR "Iu"
-#define PRIiPTR "Ii"
-#define PRIdPTR "Id"
-#define PRIxPTR "Ix"
-#define PRIuSIZE "zu"
-#define PRIiSIZE "zi"
-#define PRIdSIZE "zd"
-#define PRIxSIZE "zx"
-#endif /* fix PRI*PTR for _MSC_VER */
-
-#ifndef PRIuSIZE
-#define PRIuSIZE PRIuPTR
-#define PRIiSIZE PRIiPTR
-#define PRIdSIZE PRIdPTR
-#define PRIxSIZE PRIxPTR
-#endif /* PRI*SIZE macros for MSVC */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-#define mdbx_sourcery_anchor XCONCAT(mdbx_sourcery_, MDBX_BUILD_SOURCERY)
-#if defined(xMDBX_TOOLS)
-extern LIBMDBX_API const char *const mdbx_sourcery_anchor;
-#endif
-
-/*******************************************************************************
- *******************************************************************************
- *******************************************************************************
- *
- *
- * #### ##### ##### # #### # # ####
- * # # # # # # # # ## # #
- * # # # # # # # # # # # ####
- * # # ##### # # # # # # # #
- * # # # # # # # # ## # #
- * #### # # # #### # # ####
- *
- *
- */
-
-/** \defgroup build_option Build options
- * The libmdbx build options.
- @{ */
-
-/** Using fcntl(F_FULLFSYNC) with 5-10 times slowdown */
-#define MDBX_OSX_WANNA_DURABILITY 0
-/** Using fsync() with chance of data lost on power failure */
-#define MDBX_OSX_WANNA_SPEED 1
-
-#ifndef MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-/** Choices \ref MDBX_OSX_WANNA_DURABILITY or \ref MDBX_OSX_WANNA_SPEED
- * for OSX & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY MDBX_OSX_WANNA_DURABILITY
-#endif /* MDBX_OSX_SPEED_INSTEADOF_DURABILITY */
-
-/** Controls checking PID against reuse DB environment after the fork() */
-#ifndef MDBX_ENV_CHECKPID
-#if defined(MADV_DONTFORK) || defined(_WIN32) || defined(_WIN64)
-/* PID check could be omitted:
- * - on Linux when madvise(MADV_DONTFORK) is available, i.e. after the fork()
- * mapped pages will not be available for child process.
- * - in Windows where fork() not available. */
-#define MDBX_ENV_CHECKPID 0
-#else
-#define MDBX_ENV_CHECKPID 1
-#endif
-#define MDBX_ENV_CHECKPID_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#else
-#define MDBX_ENV_CHECKPID_CONFIG MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#endif /* MDBX_ENV_CHECKPID */
-
-/** Controls checking transaction owner thread against misuse transactions from
- * other threads. */
-#ifndef MDBX_TXN_CHECKOWNER
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_TXN_CHECKOWNER_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#else
-#define MDBX_TXN_CHECKOWNER_CONFIG MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#endif /* MDBX_TXN_CHECKOWNER */
-
-/** Does a system have battery-backed Real-Time Clock or just a fake. */
-#ifndef MDBX_TRUST_RTC
-#if defined(__linux__) || defined(__gnu_linux__) || defined(__NetBSD__) || \
- defined(__OpenBSD__)
-#define MDBX_TRUST_RTC 0 /* a lot of embedded systems have a fake RTC */
-#else
-#define MDBX_TRUST_RTC 1
-#endif
-#define MDBX_TRUST_RTC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#else
-#define MDBX_TRUST_RTC_CONFIG MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#endif /* MDBX_TRUST_RTC */
-
-/** Controls online database auto-compactification during write-transactions. */
-#ifndef MDBX_ENABLE_REFUND
-#define MDBX_ENABLE_REFUND 1
-#elif !(MDBX_ENABLE_REFUND == 0 || MDBX_ENABLE_REFUND == 1)
-#error MDBX_ENABLE_REFUND must be defined as 0 or 1
-#endif /* MDBX_ENABLE_REFUND */
-
-/** Controls gathering statistics for page operations. */
-#ifndef MDBX_ENABLE_PGOP_STAT
-#define MDBX_ENABLE_PGOP_STAT 1
-#elif !(MDBX_ENABLE_PGOP_STAT == 0 || MDBX_ENABLE_PGOP_STAT == 1)
-#error MDBX_ENABLE_PGOP_STAT must be defined as 0 or 1
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-/** Controls use of POSIX madvise() hints and friends. */
-#ifndef MDBX_ENABLE_MADVISE
-#define MDBX_ENABLE_MADVISE 1
-#elif !(MDBX_ENABLE_MADVISE == 0 || MDBX_ENABLE_MADVISE == 1)
-#error MDBX_ENABLE_MADVISE must be defined as 0 or 1
-#endif /* MDBX_ENABLE_MADVISE */
-
-/** Disable some checks to reduce an overhead and detection probability of
- * database corruption to a values closer to the LMDB. */
-#ifndef MDBX_DISABLE_PAGECHECKS
-#define MDBX_DISABLE_PAGECHECKS 0
-#elif !(MDBX_DISABLE_PAGECHECKS == 0 || MDBX_DISABLE_PAGECHECKS == 1)
-#error MDBX_DISABLE_PAGECHECKS must be defined as 0 or 1
-#endif /* MDBX_DISABLE_PAGECHECKS */
-
-#ifndef MDBX_PNL_PREALLOC_FOR_RADIXSORT
-#define MDBX_PNL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_PNL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_PNL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_PNL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
-
-#ifndef MDBX_DPL_PREALLOC_FOR_RADIXSORT
-#define MDBX_DPL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_DPL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_DPL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_DPL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
-
-/* Basically, this build-option is for TODO. Guess it should be replaced
- * with MDBX_ENABLE_WRITEMAP_SPILLING with the three variants:
- * 0/OFF = Don't track dirty pages at all and don't spilling ones.
- * This should be by-default on Linux and may-be other systems
- * (not sure: Darwin/OSX, FreeBSD, Windows 10) where kernel provides
- * properly LRU tracking and async writing on-demand.
- * 1/ON = Lite tracking of dirty pages but with LRU labels and explicit
- * spilling with msync(MS_ASYNC). */
-#ifndef MDBX_FAKE_SPILL_WRITEMAP
-#if defined(__linux__) || defined(__gnu_linux__)
-#define MDBX_FAKE_SPILL_WRITEMAP 1 /* msync(MS_ASYNC) is no-op on Linux */
-#else
-#define MDBX_FAKE_SPILL_WRITEMAP 0
-#endif
-#elif !(MDBX_FAKE_SPILL_WRITEMAP == 0 || MDBX_FAKE_SPILL_WRITEMAP == 1)
-#error MDBX_FAKE_SPILL_WRITEMAP must be defined as 0 or 1
-#endif /* MDBX_FAKE_SPILL_WRITEMAP */
-
-/** Controls sort order of internal page number lists.
- * This mostly experimental/advanced option with not for regular MDBX users.
- * \warning The database format depend on this option and libmdbx builded with
- * different option value are incompatible. */
-#ifndef MDBX_PNL_ASCENDING
-#define MDBX_PNL_ASCENDING 0
-#elif !(MDBX_PNL_ASCENDING == 0 || MDBX_PNL_ASCENDING == 1)
-#error MDBX_PNL_ASCENDING must be defined as 0 or 1
-#endif /* MDBX_PNL_ASCENDING */
-
-/** Avoid dependence from MSVC CRT and use ntdll.dll instead. */
-#ifndef MDBX_WITHOUT_MSVC_CRT
-#define MDBX_WITHOUT_MSVC_CRT 1
-#elif !(MDBX_WITHOUT_MSVC_CRT == 0 || MDBX_WITHOUT_MSVC_CRT == 1)
-#error MDBX_WITHOUT_MSVC_CRT must be defined as 0 or 1
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-/** Size of buffer used during copying a environment/database file. */
-#ifndef MDBX_ENVCOPY_WRITEBUF
-#define MDBX_ENVCOPY_WRITEBUF 1048576u
-#elif MDBX_ENVCOPY_WRITEBUF < 65536u || MDBX_ENVCOPY_WRITEBUF > 1073741824u || \
- MDBX_ENVCOPY_WRITEBUF % 65536u
-#error MDBX_ENVCOPY_WRITEBUF must be defined in range 65536..1073741824 and be multiple of 65536
-#endif /* MDBX_ENVCOPY_WRITEBUF */
-
-/** Forces assertion checking */
-#ifndef MDBX_FORCE_ASSERTIONS
-#define MDBX_FORCE_ASSERTIONS 0
-#elif !(MDBX_FORCE_ASSERTIONS == 0 || MDBX_FORCE_ASSERTIONS == 1)
-#error MDBX_FORCE_ASSERTIONS must be defined as 0 or 1
-#endif /* MDBX_FORCE_ASSERTIONS */
-
-/** Presumed malloc size overhead for each allocation
- * to adjust allocations to be more aligned. */
-#ifndef MDBX_ASSUME_MALLOC_OVERHEAD
-#ifdef __SIZEOF_POINTER__
-#define MDBX_ASSUME_MALLOC_OVERHEAD (__SIZEOF_POINTER__ * 2u)
-#else
-#define MDBX_ASSUME_MALLOC_OVERHEAD (sizeof(void *) * 2u)
-#endif
-#elif MDBX_ASSUME_MALLOC_OVERHEAD < 0 || MDBX_ASSUME_MALLOC_OVERHEAD > 64 || \
- MDBX_ASSUME_MALLOC_OVERHEAD % 4
-#error MDBX_ASSUME_MALLOC_OVERHEAD must be defined in range 0..64 and be multiple of 4
-#endif /* MDBX_ASSUME_MALLOC_OVERHEAD */
-
-/** In case the MDBX_DEBUG is undefined set it corresponding to NDEBUG */
-#ifndef MDBX_DEBUG
-#ifdef NDEBUG
-#define MDBX_DEBUG 0
-#else
-#define MDBX_DEBUG 1
-#endif
-#endif /* MDBX_DEBUG */
-
-/** If defined then enables integration with Valgrind,
- * a memory analyzing tool. */
-#ifndef MDBX_USE_VALGRIND
-#endif /* MDBX_USE_VALGRIND */
-
-/** If defined then enables use C11 atomics,
- * otherwise detects ones availability automatically. */
-#ifndef MDBX_HAVE_C11ATOMICS
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-//------------------------------------------------------------------------------
-
-/** Win32 File Locking API for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_WIN32FILES -1
-
-/** SystemV IPC semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_SYSV 5
-
-/** POSIX-1 Shared anonymous semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX1988 1988
-
-/** POSIX-2001 Shared Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2001 2001
-
-/** POSIX-2008 Robust Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2008 2008
-
-/** BeOS Benaphores, aka Futexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_BENAPHORE 1995
-
-/** Advanced: Choices the locking implementation (autodetection by default). */
-#if defined(_WIN32) || defined(_WIN64)
-#define MDBX_LOCKING MDBX_LOCKING_WIN32FILES
-#else
-#ifndef MDBX_LOCKING
-#if defined(_POSIX_THREAD_PROCESS_SHARED) && \
- _POSIX_THREAD_PROCESS_SHARED >= 200112L && !defined(__FreeBSD__)
-
-/* Some platforms define the EOWNERDEAD error code even though they
- * don't support Robust Mutexes. If doubt compile with -MDBX_LOCKING=2001. */
-#if defined(EOWNERDEAD) && _POSIX_THREAD_PROCESS_SHARED >= 200809L && \
- ((defined(_POSIX_THREAD_ROBUST_PRIO_INHERIT) && \
- _POSIX_THREAD_ROBUST_PRIO_INHERIT > 0) || \
- (defined(_POSIX_THREAD_ROBUST_PRIO_PROTECT) && \
- _POSIX_THREAD_ROBUST_PRIO_PROTECT > 0) || \
- defined(PTHREAD_MUTEX_ROBUST) || defined(PTHREAD_MUTEX_ROBUST_NP)) && \
- (!defined(__GLIBC__) || \
- __GLIBC_PREREQ(2, 10) /* troubles with Robust mutexes before 2.10 */)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2008
-#else
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2001
-#endif
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX1988
-#else
-#define MDBX_LOCKING MDBX_LOCKING_SYSV
-#endif
-#define MDBX_LOCKING_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_LOCKING)
-#else
-#define MDBX_LOCKING_CONFIG MDBX_STRINGIFY(MDBX_LOCKING)
-#endif /* MDBX_LOCKING */
-#endif /* !Windows */
-
-/** Advanced: Using POSIX OFD-locks (autodetection by default). */
-#ifndef MDBX_USE_OFDLOCKS
-#if defined(F_OFD_SETLK) && defined(F_OFD_SETLKW) && defined(F_OFD_GETLK) && \
- !defined(MDBX_SAFE4QEMU) && \
- !defined(__sun) /* OFD-lock are broken on Solaris */
-#define MDBX_USE_OFDLOCKS 1
-#else
-#define MDBX_USE_OFDLOCKS 0
-#endif
-#define MDBX_USE_OFDLOCKS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#else
-#define MDBX_USE_OFDLOCKS_CONFIG MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#endif /* MDBX_USE_OFDLOCKS */
-
-/** Advanced: Using sendfile() syscall (autodetection by default). */
-#ifndef MDBX_USE_SENDFILE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 21)
-#define MDBX_USE_SENDFILE 1
-#else
-#define MDBX_USE_SENDFILE 0
-#endif
-#endif /* MDBX_USE_SENDFILE */
-
-/** Advanced: Using copy_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_COPYFILERANGE
-#if __GLIBC_PREREQ(2, 27) && defined(_GNU_SOURCE)
-#define MDBX_USE_COPYFILERANGE 1
-#else
-#define MDBX_USE_COPYFILERANGE 0
-#endif
-#endif /* MDBX_USE_COPYFILERANGE */
-
-/** Advanced: Using sync_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_SYNCFILERANGE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- defined(SYNC_FILE_RANGE_WRITE) && !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 26)
-#define MDBX_USE_SYNCFILERANGE 1
-#else
-#define MDBX_USE_SYNCFILERANGE 0
-#endif
-#endif /* MDBX_USE_SYNCFILERANGE */
-
-//------------------------------------------------------------------------------
-
-#ifndef MDBX_CPU_WRITEBACK_INCOHERENT
-#if defined(__ia32__) || defined(__e2k__) || defined(__hppa) || \
- defined(__hppa__) || defined(DOXYGEN)
-#define MDBX_CPU_WRITEBACK_INCOHERENT 0
-#else
-#define MDBX_CPU_WRITEBACK_INCOHERENT 1
-#endif
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-#ifndef MDBX_MMAP_INCOHERENT_FILE_WRITE
-#ifdef __OpenBSD__
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 1
-#else
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#ifndef MDBX_MMAP_INCOHERENT_CPU_CACHE
-#if defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS has cache coherency issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 1
-#else
-/* LY: assume no relevant mmap/dcache issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#ifndef MDBX_64BIT_ATOMIC
-#if MDBX_WORDBITS >= 64 || defined(DOXYGEN)
-#define MDBX_64BIT_ATOMIC 1
-#else
-#define MDBX_64BIT_ATOMIC 0
-#endif
-#define MDBX_64BIT_ATOMIC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#else
-#define MDBX_64BIT_ATOMIC_CONFIG MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#endif /* MDBX_64BIT_ATOMIC */
-
-#ifndef MDBX_64BIT_CAS
-#if defined(ATOMIC_LLONG_LOCK_FREE)
-#if ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__GCC_ATOMIC_LLONG_LOCK_FREE)
-#if __GCC_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__CLANG_ATOMIC_LLONG_LOCK_FREE)
-#if __CLANG_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(_MSC_VER) || defined(__APPLE__) || defined(DOXYGEN)
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS MDBX_64BIT_ATOMIC
-#endif
-#define MDBX_64BIT_CAS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#else
-#define MDBX_64BIT_CAS_CONFIG MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#endif /* MDBX_64BIT_CAS */
-
-#ifndef MDBX_UNALIGNED_OK
-#if defined(__ALIGNED__) || defined(__SANITIZE_UNDEFINED__)
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#elif defined(__ARM_FEATURE_UNALIGNED)
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#elif __CLANG_PREREQ(5, 0) || __GNUC_PREREQ(5, 0)
-/* expecting an optimization will well done, also this
- * hushes false-positives from UBSAN (undefined behaviour sanitizer) */
-#define MDBX_UNALIGNED_OK 0
-#elif defined(__e2k__) || defined(__elbrus__)
-#if __iset__ > 4
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#endif
-#elif defined(__ia32__)
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#endif
-#elif MDBX_UNALIGNED_OK == 1
-#undef MDBX_UNALIGNED_OK
-#define MDBX_UNALIGNED_OK 32 /* any unaligned access allowed */
-#endif /* MDBX_UNALIGNED_OK */
-
-#ifndef MDBX_CACHELINE_SIZE
-#if defined(SYSTEM_CACHE_ALIGNMENT_SIZE)
-#define MDBX_CACHELINE_SIZE SYSTEM_CACHE_ALIGNMENT_SIZE
-#elif defined(__ia64__) || defined(__ia64) || defined(_M_IA64)
-#define MDBX_CACHELINE_SIZE 128
-#else
-#define MDBX_CACHELINE_SIZE 64
-#endif
-#endif /* MDBX_CACHELINE_SIZE */
-
-/** @} end of build options */
-/*******************************************************************************
- *******************************************************************************
- ******************************************************************************/
-
-#ifdef DOXYGEN
-/* !!! Actually this is a fake definitions !!!
- * !!! for documentation generation by Doxygen !!! */
-
-/** Controls enabling of debugging features.
- *
- * - `MDBX_DEBUG = 0` (by default) Disables any debugging features at all,
- * including logging and assertion controls.
- * Logging level and corresponding debug flags changing
- * by \ref mdbx_setup_debug() will not have effect.
- * - `MDBX_DEBUG > 0` Enables code for the debugging features (logging,
- * assertions checking and internal audit).
- * Simultaneously sets the default logging level
- * to the `MDBX_DEBUG` value.
- * Also enables \ref MDBX_DBG_AUDIT if `MDBX_DEBUG >= 2`.
- *
- * \ingroup build_option */
-#define MDBX_DEBUG 0...7
-
-/** Disables using of GNU libc extensions. */
-#define MDBX_DISABLE_GNU_SOURCE 0 or 1
-
-#endif /* DOXYGEN */
-
-/* Undefine the NDEBUG if debugging is enforced by MDBX_DEBUG */
-#if MDBX_DEBUG
-#undef NDEBUG
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-enum MDBX_memory_order {
- mo_Relaxed,
- mo_AcquireRelease,
- mo_SequentialConsistency
-};
-
-typedef union {
- volatile uint32_t weak;
-#ifdef MDBX_HAVE_C11ATOMICS
- volatile _Atomic uint32_t c11a;
-#endif /* MDBX_HAVE_C11ATOMICS */
-} MDBX_atomic_uint32_t;
-
-typedef union {
- volatile uint64_t weak;
-#if defined(MDBX_HAVE_C11ATOMICS) && (MDBX_64BIT_CAS || MDBX_64BIT_ATOMIC)
- volatile _Atomic uint64_t c11a;
-#endif
-#if !defined(MDBX_HAVE_C11ATOMICS) || !MDBX_64BIT_CAS || !MDBX_64BIT_ATOMIC
- __anonymous_struct_extension__ struct {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- MDBX_atomic_uint32_t low, high;
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- MDBX_atomic_uint32_t high, low;
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- };
-#endif
-} MDBX_atomic_uint64_t;
-
-#ifdef MDBX_HAVE_C11ATOMICS
-
-/* Crutches for C11 atomic compiler's bugs */
-#if defined(__e2k__) && defined(__LCC__) && __LCC__ < /* FIXME */ 127
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->weak)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->weak)
-#elif defined(__clang__) && __clang__ < 8
-#define MDBX_c11a_ro(type, ptr) ((volatile _Atomic(type) *)&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#else
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#endif /* Crutches for C11 atomic compiler's bugs */
-
-static __always_inline memory_order mo_c11_store(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_release;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-
-static __always_inline memory_order mo_c11_load(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_acquire;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED static __always_inline void
-mdbx_memory_fence(enum MDBX_memory_order order, bool write) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(write ? mo_c11_store(order) : mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_compiler_barrier();
- if (write &&
- order > (MDBX_CPU_WRITEBACK_INCOHERENT ? mo_Relaxed : mo_AcquireRelease))
- mdbx_memory_barrier();
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_store32(MDBX_atomic_uint32_t *p, const uint32_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint32_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_load32(const MDBX_atomic_uint32_t *p, enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint32_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint32_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint32_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-/* Basic constants and types */
-
-/* A stamp that identifies a file as an MDBX file.
- * There's nothing special about this value other than that it is easily
- * recognizable, and it will reflect any byte order mismatches. */
-#define MDBX_MAGIC UINT64_C(/* 56-bit prime */ 0x59659DBDEF4C11)
-
-/* FROZEN: The version number for a database's datafile format. */
-#define MDBX_DATA_VERSION 3
-/* The version number for a database's lockfile format. */
-#define MDBX_LOCK_VERSION 4
-
-/* handle for the DB used to track free pages. */
-#define FREE_DBI 0
-/* handle for the default DB. */
-#define MAIN_DBI 1
-/* Number of DBs in metapage (free and main) - also hardcoded elsewhere */
-#define CORE_DBS 2
-
-/* Number of meta pages - also hardcoded elsewhere */
-#define NUM_METAS 3
-
-/* A page number in the database.
- *
- * MDBX uses 32 bit for page numbers. This limits database
- * size up to 2^44 bytes, in case of 4K pages. */
-typedef uint32_t pgno_t;
-typedef MDBX_atomic_uint32_t atomic_pgno_t;
-#define PRIaPGNO PRIu32
-#define MAX_PAGENO UINT32_C(0x7FFFffff)
-#define MIN_PAGENO NUM_METAS
-
-#define SAFE64_INVALID_THRESHOLD UINT64_C(0xffffFFFF00000000)
-
-/* A transaction ID. */
-typedef uint64_t txnid_t;
-typedef MDBX_atomic_uint64_t atomic_txnid_t;
-#define PRIaTXN PRIi64
-#define MIN_TXNID UINT64_C(1)
-#define MAX_TXNID (SAFE64_INVALID_THRESHOLD - 1)
-#define INITIAL_TXNID (MIN_TXNID + NUM_METAS - 1)
-#define INVALID_TXNID UINT64_MAX
-/* LY: for testing non-atomic 64-bit txnid on 32-bit arches.
- * #define xMDBX_TXNID_STEP (UINT32_MAX / 3) */
-#ifndef xMDBX_TXNID_STEP
-#if MDBX_64BIT_CAS
-#define xMDBX_TXNID_STEP 1u
-#else
-#define xMDBX_TXNID_STEP 2u
-#endif
-#endif /* xMDBX_TXNID_STEP */
-
-/* Used for offsets within a single page.
- * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
- * this is plenty. */
-typedef uint16_t indx_t;
-
-#define MEGABYTE ((size_t)1 << 20)
-
-/*----------------------------------------------------------------------------*/
-/* Core structures for database and shared memory (i.e. format definition) */
-#pragma pack(push, 4)
-
-/* Information about a single database in the environment. */
-typedef struct MDBX_db {
- uint16_t md_flags; /* see mdbx_dbi_open */
- uint16_t md_depth; /* depth of this tree */
- uint32_t md_xsize; /* key-size for MDBX_DUPFIXED (LEAF2 pages) */
- pgno_t md_root; /* the root page of this tree */
- pgno_t md_branch_pages; /* number of internal pages */
- pgno_t md_leaf_pages; /* number of leaf pages */
- pgno_t md_overflow_pages; /* number of overflow pages */
- uint64_t md_seq; /* table sequence counter */
- uint64_t md_entries; /* number of data items */
- uint64_t md_mod_txnid; /* txnid of last committed modification */
-} MDBX_db;
-
-/* database size-related parameters */
-typedef struct MDBX_geo {
- uint16_t grow_pv; /* datafile growth step as a 16-bit packed (exponential
- quantized) value */
- uint16_t shrink_pv; /* datafile shrink threshold as a 16-bit packed
- (exponential quantized) value */
- pgno_t lower; /* minimal size of datafile in pages */
- pgno_t upper; /* maximal size of datafile in pages */
- pgno_t now; /* current size of datafile in pages */
- pgno_t next; /* first unused page in the datafile,
- but actually the file may be shorter. */
-} MDBX_geo;
-
-/* Meta page content.
- * A meta page is the start point for accessing a database snapshot.
- * Pages 0-1 are meta pages. Transaction N writes meta page (N % 2). */
-typedef struct MDBX_meta {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with MDBX_DATA_VERSION. */
- uint32_t mm_magic_and_version[2];
-
- /* txnid that committed this page, the first of a two-phase-update pair */
- uint32_t mm_txnid_a[2];
-
- uint16_t mm_extra_flags; /* extra DB flags, zero (nothing) for now */
- uint8_t mm_validator_id; /* ID of checksum and page validation method,
- * zero (nothing) for now */
- uint8_t mm_extra_pagehdr; /* extra bytes in the page header,
- * zero (nothing) for now */
-
- MDBX_geo mm_geo; /* database size-related parameters */
-
- MDBX_db mm_dbs[CORE_DBS]; /* first is free space, 2nd is main db */
- /* The size of pages used in this DB */
-#define mm_psize mm_dbs[FREE_DBI].md_xsize
- MDBX_canary mm_canary;
-
-#define MDBX_DATASIGN_NONE 0u
-#define MDBX_DATASIGN_WEAK 1u
-#define SIGN_IS_STEADY(sign) ((sign) > MDBX_DATASIGN_WEAK)
-#define META_IS_STEADY(meta) \
- SIGN_IS_STEADY(unaligned_peek_u64_volatile(4, (meta)->mm_datasync_sign))
- uint32_t mm_datasync_sign[2];
-
- /* txnid that committed this page, the second of a two-phase-update pair */
- uint32_t mm_txnid_b[2];
-
- /* Number of non-meta pages which were put in GC after COW. May be 0 in case
- * DB was previously handled by libmdbx without corresponding feature.
- * This value in couple with mr_snapshot_pages_retired allows fast estimation
- * of "how much reader is restraining GC recycling". */
- uint32_t mm_pages_retired[2];
-
- /* The analogue /proc/sys/kernel/random/boot_id or similar to determine
- * whether the system was rebooted after the last use of the database files.
- * If there was no reboot, but there is no need to rollback to the last
- * steady sync point. Zeros mean that no relevant information is available
- * from the system. */
- bin128_t mm_bootid;
-
-} MDBX_meta;
-
-#pragma pack(1)
-
-/* Common header for all page types. The page type depends on mp_flags.
- *
- * P_BRANCH and P_LEAF pages have unsorted 'MDBX_node's at the end, with
- * sorted mp_ptrs[] entries referring to them. Exception: P_LEAF2 pages
- * omit mp_ptrs and pack sorted MDBX_DUPFIXED values after the page header.
- *
- * P_OVERFLOW records occupy one or more contiguous pages where only the
- * first has a page header. They hold the real data of F_BIGDATA nodes.
- *
- * P_SUBP sub-pages are small leaf "pages" with duplicate data.
- * A node with flag F_DUPDATA but not F_SUBDATA contains a sub-page.
- * (Duplicate data can also go in sub-databases, which use normal pages.)
- *
- * P_META pages contain MDBX_meta, the start point of an MDBX snapshot.
- *
- * Each non-metapage up to MDBX_meta.mm_last_pg is reachable exactly once
- * in the snapshot: Either used by a database or listed in a GC record. */
-typedef struct MDBX_page {
- union {
-#define IS_FROZEN(txn, p) ((p)->mp_txnid < (txn)->mt_txnid)
-#define IS_SPILLED(txn, p) ((p)->mp_txnid == (txn)->mt_txnid)
-#define IS_SHADOWED(txn, p) ((p)->mp_txnid > (txn)->mt_txnid)
-#define IS_VALID(txn, p) ((p)->mp_txnid <= (txn)->mt_front)
-#define IS_MODIFIABLE(txn, p) ((p)->mp_txnid == (txn)->mt_front)
- uint64_t mp_txnid;
- struct MDBX_page *mp_next; /* for in-memory list of freed pages */
- };
- uint16_t mp_leaf2_ksize; /* key size if this is a LEAF2 page */
-#define P_BRANCH 0x01 /* branch page */
-#define P_LEAF 0x02 /* leaf page */
-#define P_OVERFLOW 0x04 /* overflow page */
-#define P_META 0x08 /* meta page */
-#define P_BAD 0x10 /* explicit flag for invalid/bad page */
-#define P_LEAF2 0x20 /* for MDBX_DUPFIXED records */
-#define P_SUBP 0x40 /* for MDBX_DUPSORT sub-pages */
-#define P_SPILLED 0x2000 /* spilled in parent txn */
-#define P_LOOSE 0x4000 /* page was dirtied then freed, can be reused */
-#define P_FROZEN 0x8000 /* used for retire page with known status */
-#define P_ILL_BITS (~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_SPILLED))
- uint16_t mp_flags;
- union {
- uint32_t mp_pages; /* number of overflow pages */
- __anonymous_struct_extension__ struct {
- indx_t mp_lower; /* lower bound of free space */
- indx_t mp_upper; /* upper bound of free space */
- };
- };
- pgno_t mp_pgno; /* page number */
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- indx_t mp_ptrs[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_page;
-
-/* Size of the page header, excluding dynamic data at the end */
-#define PAGEHDRSZ ((unsigned)offsetof(MDBX_page, mp_ptrs))
-
-#pragma pack(pop)
-
-#if MDBX_ENABLE_PGOP_STAT
-/* Statistics of page operations overall of all (running, completed and aborted)
- * transactions */
-typedef struct {
- MDBX_atomic_uint64_t newly; /* Quantity of a new pages added */
- MDBX_atomic_uint64_t cow; /* Quantity of pages copied for update */
- MDBX_atomic_uint64_t clone; /* Quantity of parent's dirty pages clones
- for nested transactions */
- MDBX_atomic_uint64_t split; /* Page splits */
- MDBX_atomic_uint64_t merge; /* Page merges */
- MDBX_atomic_uint64_t spill; /* Quantity of spilled dirty pages */
- MDBX_atomic_uint64_t unspill; /* Quantity of unspilled/reloaded pages */
- MDBX_atomic_uint64_t
- wops; /* Number of explicit write operations (not a pages) to a disk */
-} MDBX_pgop_stat_t;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-#if MDBX_LOCKING == MDBX_LOCKING_WIN32FILES
-#define MDBX_CLOCK_SIGN UINT32_C(0xF10C)
-typedef void mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
-
-#define MDBX_CLOCK_SIGN UINT32_C(0xF18D)
-typedef mdbx_pid_t mdbx_ipclock_t;
-#ifndef EOWNERDEAD
-#define EOWNERDEAD MDBX_RESULT_TRUE
-#endif
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#define MDBX_CLOCK_SIGN UINT32_C(0x8017)
-typedef pthread_mutex_t mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-#define MDBX_CLOCK_SIGN UINT32_C(0xFC29)
-typedef sem_t mdbx_ipclock_t;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV && !defined(__cplusplus)
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc);
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc);
-#endif /* MDBX_LOCKING */
-
-/* Reader Lock Table
- *
- * Readers don't acquire any locks for their data access. Instead, they
- * simply record their transaction ID in the reader table. The reader
- * mutex is needed just to find an empty slot in the reader table. The
- * slot's address is saved in thread-specific data so that subsequent
- * read transactions started by the same thread need no further locking to
- * proceed.
- *
- * If MDBX_NOTLS is set, the slot address is not saved in thread-specific data.
- * No reader table is used if the database is on a read-only filesystem.
- *
- * Since the database uses multi-version concurrency control, readers don't
- * actually need any locking. This table is used to keep track of which
- * readers are using data from which old transactions, so that we'll know
- * when a particular old transaction is no longer in use. Old transactions
- * that have discarded any data pages can then have those pages reclaimed
- * for use by a later write transaction.
- *
- * The lock table is constructed such that reader slots are aligned with the
- * processor's cache line size. Any slot is only ever used by one thread.
- * This alignment guarantees that there will be no contention or cache
- * thrashing as threads update their own slot info, and also eliminates
- * any need for locking when accessing a slot.
- *
- * A writer thread will scan every slot in the table to determine the oldest
- * outstanding reader transaction. Any freed pages older than this will be
- * reclaimed by the writer. The writer doesn't use any locks when scanning
- * this table. This means that there's no guarantee that the writer will
- * see the most up-to-date reader info, but that's not required for correct
- * operation - all we need is to know the upper bound on the oldest reader,
- * we don't care at all about the newest reader. So the only consequence of
- * reading stale information here is that old pages might hang around a
- * while longer before being reclaimed. That's actually good anyway, because
- * the longer we delay reclaiming old pages, the more likely it is that a
- * string of contiguous pages can be found after coalescing old pages from
- * many old transactions together. */
-
-/* The actual reader record, with cacheline padding. */
-typedef struct MDBX_reader {
- /* Current Transaction ID when this transaction began, or (txnid_t)-1.
- * Multiple readers that start at the same time will probably have the
- * same ID here. Again, it's not important to exclude them from
- * anything; all we need to know is which version of the DB they
- * started from so we can avoid overwriting any data used in that
- * particular version. */
- MDBX_atomic_uint64_t /* txnid_t */ mr_txnid;
-
- /* The information we store in a single slot of the reader table.
- * In addition to a transaction ID, we also record the process and
- * thread ID that owns a slot, so that we can detect stale information,
- * e.g. threads or processes that went away without cleaning up.
- *
- * NOTE: We currently don't check for stale records.
- * We simply re-init the table when we know that we're the only process
- * opening the lock file. */
-
- /* The thread ID of the thread owning this txn. */
- MDBX_atomic_uint64_t mr_tid;
-
- /* The process ID of the process owning this reader txn. */
- MDBX_atomic_uint32_t mr_pid;
-
- /* The number of pages used in the reader's MVCC snapshot,
- * i.e. the value of meta->mm_geo.next and txn->mt_next_pgno */
- atomic_pgno_t mr_snapshot_pages_used;
- /* Number of retired pages at the time this reader starts transaction. So,
- * at any time the difference mm_pages_retired - mr_snapshot_pages_retired
- * will give the number of pages which this reader restraining from reuse. */
- MDBX_atomic_uint64_t mr_snapshot_pages_retired;
-} MDBX_reader;
-
-/* The header for the reader table (a memory-mapped lock file). */
-typedef struct MDBX_lockinfo {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with with MDBX_LOCK_VERSION. */
- uint64_t mti_magic_and_version;
-
- /* Format of this lock file. Must be set to MDBX_LOCK_FORMAT. */
- uint32_t mti_os_and_format;
-
- /* Flags which environment was opened. */
- MDBX_atomic_uint32_t mti_envmode;
-
- /* Threshold of un-synced-with-disk pages for auto-sync feature,
- * zero means no-threshold, i.e. auto-sync is disabled. */
- atomic_pgno_t mti_autosync_threshold;
-
- /* Low 32-bit of txnid with which meta-pages was synced,
- * i.e. for sync-polling in the MDBX_NOMETASYNC mode. */
- MDBX_atomic_uint32_t mti_meta_sync_txnid;
-
- /* Period for timed auto-sync feature, i.e. at the every steady checkpoint
- * the mti_unsynced_timeout sets to the current_time + mti_autosync_period.
- * The time value is represented in a suitable system-dependent form, for
- * example clock_gettime(CLOCK_BOOTTIME) or clock_gettime(CLOCK_MONOTONIC).
- * Zero means timed auto-sync is disabled. */
- MDBX_atomic_uint64_t mti_autosync_period;
-
- /* Marker to distinguish uniqueness of DB/CLK. */
- MDBX_atomic_uint64_t mti_bait_uniqueness;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
-#if MDBX_ENABLE_PGOP_STAT
- /* Statistics of costly ops of all (running, completed and aborted)
- * transactions */
- MDBX_pgop_stat_t mti_pgop_stat;
-#endif /* MDBX_ENABLE_PGOP_STAT*/
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Write transaction lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_wlock;
-#endif /* MDBX_LOCKING > 0 */
-
- atomic_txnid_t mti_oldest_reader;
-
- /* Timestamp of the last steady sync. Value is represented in a suitable
- * system-dependent form, for example clock_gettime(CLOCK_BOOTTIME) or
- * clock_gettime(CLOCK_MONOTONIC). */
- MDBX_atomic_uint64_t mti_sync_timestamp;
-
- /* Number un-synced-with-disk pages for auto-sync feature. */
- atomic_pgno_t mti_unsynced_pages;
-
- /* Number of page which was discarded last time by madvise(MADV_FREE). */
- atomic_pgno_t mti_discarded_tail;
-
- /* Timestamp of the last readers check. */
- MDBX_atomic_uint64_t mti_reader_check_timestamp;
-
- /* Shared anchor for tracking readahead edge and enabled/disabled status. */
- pgno_t mti_readahead_anchor;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Readeaders registration lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_rlock;
-#endif /* MDBX_LOCKING > 0 */
-
- /* The number of slots that have been used in the reader table.
- * This always records the maximum count, it is not decremented
- * when readers release their slots. */
- MDBX_atomic_uint32_t mti_numreaders;
- MDBX_atomic_uint32_t mti_readers_refresh_flag;
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
- MDBX_reader mti_readers[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_lockinfo;
-
-/* Lockfile format signature: version, features and field layout */
-#define MDBX_LOCK_FORMAT \
- (MDBX_CLOCK_SIGN * 27733 + (unsigned)sizeof(MDBX_reader) * 13 + \
- (unsigned)offsetof(MDBX_reader, mr_snapshot_pages_used) * 251 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_oldest_reader) * 83 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_numreaders) * 37 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_readers) * 29)
-
-#define MDBX_DATA_MAGIC \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + MDBX_DATA_VERSION)
-
-#define MDBX_DATA_MAGIC_LEGACY_COMPAT \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + 2)
-
-#define MDBX_DATA_MAGIC_LEGACY_DEVEL ((MDBX_MAGIC << 8) + 255)
-
-#define MDBX_LOCK_MAGIC ((MDBX_MAGIC << 8) + MDBX_LOCK_VERSION)
-
-/* The maximum size of a database page.
- *
- * It is 64K, but value-PAGEHDRSZ must fit in MDBX_page.mp_upper.
- *
- * MDBX will use database pages < OS pages if needed.
- * That causes more I/O in write transactions: The OS must
- * know (read) the whole page before writing a partial page.
- *
- * Note that we don't currently support Huge pages. On Linux,
- * regular data files cannot use Huge pages, and in general
- * Huge pages aren't actually pageable. We rely on the OS
- * demand-pager to read our data and page it out when memory
- * pressure from other processes is high. So until OSs have
- * actual paging support for Huge pages, they're not viable. */
-#define MAX_PAGESIZE MDBX_MAX_PAGESIZE
-#define MIN_PAGESIZE MDBX_MIN_PAGESIZE
-
-#define MIN_MAPSIZE (MIN_PAGESIZE * MIN_PAGENO)
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_MAPSIZE32 UINT32_C(0x38000000)
-#else
-#define MAX_MAPSIZE32 UINT32_C(0x7f000000)
-#endif
-#define MAX_MAPSIZE64 ((MAX_PAGENO + 1) * (uint64_t)MAX_PAGESIZE)
-
-#if MDBX_WORDBITS >= 64
-#define MAX_MAPSIZE MAX_MAPSIZE64
-#define MDBX_PGL_LIMIT ((size_t)MAX_PAGENO)
-#else
-#define MAX_MAPSIZE MAX_MAPSIZE32
-#define MDBX_PGL_LIMIT (MAX_MAPSIZE32 / MIN_PAGESIZE)
-#endif /* MDBX_WORDBITS */
-
-#define MDBX_READERS_LIMIT 32767
-#define MDBX_RADIXSORT_THRESHOLD 333
-
-/*----------------------------------------------------------------------------*/
-
-/* An PNL is an Page Number List, a sorted array of IDs.
- * The first element of the array is a counter for how many actual page-numbers
- * are in the list. By default PNLs are sorted in descending order, this allow
- * cut off a page with lowest pgno (at the tail) just truncating the list. The
- * sort order of PNLs is controlled by the MDBX_PNL_ASCENDING build option. */
-typedef pgno_t *MDBX_PNL;
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_ORDERED(first, last) ((first) < (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) >= (last))
-#else
-#define MDBX_PNL_ORDERED(first, last) ((first) > (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) <= (last))
-#endif
-
-/* List of txnid, only for MDBX_txn.tw.lifo_reclaimed */
-typedef txnid_t *MDBX_TXL;
-
-/* An Dirty-Page list item is an pgno/pointer pair. */
-typedef struct MDBX_dp {
- MDBX_page *ptr;
- pgno_t pgno;
- union {
- unsigned extra;
- __anonymous_struct_extension__ struct {
- unsigned multi : 1;
- unsigned lru : 31;
- };
- };
-} MDBX_dp;
-
-/* An DPL (dirty-page list) is a sorted array of MDBX_DPs. */
-typedef struct MDBX_dpl {
- unsigned sorted;
- unsigned length;
- unsigned detent; /* allocated size excluding the MDBX_DPL_RESERVE_GAP */
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_dp items[] /* dynamic size with holes at zero and after the last */;
-#endif
-} MDBX_dpl;
-
-/* PNL sizes */
-#define MDBX_PNL_GRANULATE 1024
-#define MDBX_PNL_INITIAL \
- (MDBX_PNL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(pgno_t))
-
-#define MDBX_TXL_GRANULATE 32
-#define MDBX_TXL_INITIAL \
- (MDBX_TXL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-#define MDBX_TXL_MAX \
- ((1u << 17) - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-
-#define MDBX_PNL_ALLOCLEN(pl) ((pl)[-1])
-#define MDBX_PNL_SIZE(pl) ((pl)[0])
-#define MDBX_PNL_FIRST(pl) ((pl)[1])
-#define MDBX_PNL_LAST(pl) ((pl)[MDBX_PNL_SIZE(pl)])
-#define MDBX_PNL_BEGIN(pl) (&(pl)[1])
-#define MDBX_PNL_END(pl) (&(pl)[MDBX_PNL_SIZE(pl) + 1])
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_FIRST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_LAST(pl)
-#else
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_LAST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_FIRST(pl)
-#endif
-
-#define MDBX_PNL_SIZEOF(pl) ((MDBX_PNL_SIZE(pl) + 1) * sizeof(pgno_t))
-#define MDBX_PNL_IS_EMPTY(pl) (MDBX_PNL_SIZE(pl) == 0)
-
-/*----------------------------------------------------------------------------*/
-/* Internal structures */
-
-/* Auxiliary DB info.
- * The information here is mostly static/read-only. There is
- * only a single copy of this record in the environment. */
-typedef struct MDBX_dbx {
- MDBX_val md_name; /* name of the database */
- MDBX_cmp_func *md_cmp; /* function for comparing keys */
- MDBX_cmp_func *md_dcmp; /* function for comparing data items */
- size_t md_klen_min, md_klen_max; /* min/max key length for the database */
- size_t md_vlen_min,
- md_vlen_max; /* min/max value/data length for the database */
-} MDBX_dbx;
-
-/* A database transaction.
- * Every operation requires a transaction handle. */
-struct MDBX_txn {
-#define MDBX_MT_SIGNATURE UINT32_C(0x93D53A31)
- uint32_t mt_signature;
-
- /* Transaction Flags */
- /* mdbx_txn_begin() flags */
-#define MDBX_TXN_RO_BEGIN_FLAGS (MDBX_TXN_RDONLY | MDBX_TXN_RDONLY_PREPARE)
-#define MDBX_TXN_RW_BEGIN_FLAGS \
- (MDBX_TXN_NOMETASYNC | MDBX_TXN_NOSYNC | MDBX_TXN_TRY)
- /* Additional flag for mdbx_sync_locked() */
-#define MDBX_SHRINK_ALLOWED UINT32_C(0x40000000)
-
- /* internal txn flags */
-#define MDBX_TXN_FINISHED 0x01 /* txn is finished or never began */
-#define MDBX_TXN_ERROR 0x02 /* txn is unusable after an error */
-#define MDBX_TXN_DIRTY 0x04 /* must write, even if dirty list is empty */
-#define MDBX_TXN_SPILLS 0x08 /* txn or a parent has spilled pages */
-#define MDBX_TXN_HAS_CHILD 0x10 /* txn has an MDBX_txn.mt_child */
- /* most operations on the txn are currently illegal */
-#define MDBX_TXN_BLOCKED \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_HAS_CHILD)
-
-#define TXN_FLAGS \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_DIRTY | MDBX_TXN_SPILLS | \
- MDBX_TXN_HAS_CHILD)
-
-#if (TXN_FLAGS & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS)) || \
- ((MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS | TXN_FLAGS) & \
- MDBX_SHRINK_ALLOWED)
-#error "Oops, some flags overlapped or wrong"
-#endif
- uint32_t mt_flags;
-
- MDBX_txn *mt_parent; /* parent of a nested txn */
- /* Nested txn under this txn, set together with flag MDBX_TXN_HAS_CHILD */
- MDBX_txn *mt_child;
- MDBX_geo mt_geo;
- /* next unallocated page */
-#define mt_next_pgno mt_geo.next
- /* corresponding to the current size of datafile */
-#define mt_end_pgno mt_geo.now
-
- /* The ID of this transaction. IDs are integers incrementing from 1.
- * Only committed write transactions increment the ID. If a transaction
- * aborts, the ID may be re-used by the next writer. */
- txnid_t mt_txnid;
- txnid_t mt_front;
-
- MDBX_env *mt_env; /* the DB environment */
- /* Array of records for each DB known in the environment. */
- MDBX_dbx *mt_dbxs;
- /* Array of MDBX_db records for each known DB */
- MDBX_db *mt_dbs;
- /* Array of sequence numbers for each DB handle */
- unsigned *mt_dbiseqs;
-
- /* Transaction DBI Flags */
-#define DBI_DIRTY MDBX_DBI_DIRTY /* DB was written in this txn */
-#define DBI_STALE MDBX_DBI_STALE /* Named-DB record is older than txnID */
-#define DBI_FRESH MDBX_DBI_FRESH /* Named-DB handle opened in this txn */
-#define DBI_CREAT MDBX_DBI_CREAT /* Named-DB handle created in this txn */
-#define DBI_VALID 0x10 /* DB handle is valid, see also DB_VALID */
-#define DBI_USRVALID 0x20 /* As DB_VALID, but not set for FREE_DBI */
-#define DBI_AUDITED 0x40 /* Internal flag for accounting during audit */
- /* Array of flags for each DB */
- uint8_t *mt_dbistate;
- /* Number of DB records in use, or 0 when the txn is finished.
- * This number only ever increments until the txn finishes; we
- * don't decrement it when individual DB handles are closed. */
- MDBX_dbi mt_numdbs;
- size_t mt_owner; /* thread ID that owns this transaction */
- MDBX_canary mt_canary;
- void *mt_userctx; /* User-settable context */
- MDBX_cursor **mt_cursors;
-
- union {
- struct {
- /* For read txns: This thread/txn's reader table slot, or NULL. */
- MDBX_reader *reader;
- } to;
- struct {
- /* In write txns, array of cursors for each DB */
- pgno_t *reclaimed_pglist; /* Reclaimed GC pages */
- txnid_t last_reclaimed; /* ID of last used record */
-#if MDBX_ENABLE_REFUND
- pgno_t loose_refund_wl /* FIXME: describe */;
-#endif /* MDBX_ENABLE_REFUND */
- /* dirtylist room: Dirty array size - dirty pages visible to this txn.
- * Includes ancestor txns' dirty pages not hidden by other txns'
- * dirty/spilled pages. Thus commit(nested txn) has room to merge
- * dirtylist into mt_parent after freeing hidden mt_parent pages. */
- unsigned dirtyroom;
- /* a sequence to spilling dirty page with LRU policy */
- unsigned dirtylru;
- /* For write txns: Modified pages. Sorted when not MDBX_WRITEMAP. */
- MDBX_dpl *dirtylist;
- /* The list of reclaimed txns from GC */
- MDBX_TXL lifo_reclaimed;
- /* The list of pages that became unused during this transaction. */
- MDBX_PNL retired_pages;
- /* The list of loose pages that became unused and may be reused
- * in this transaction, linked through `mp_next`. */
- MDBX_page *loose_pages;
- /* Number of loose pages (tw.loose_pages) */
- unsigned loose_count;
- /* The sorted list of dirty pages we temporarily wrote to disk
- * because the dirty list was full. page numbers in here are
- * shifted left by 1, deleted slots have the LSB set. */
- MDBX_PNL spill_pages;
- unsigned spill_least_removed;
- } tw;
- };
-};
-
-#if MDBX_WORDBITS >= 64
-#define CURSOR_STACK 32
-#else
-#define CURSOR_STACK 24
-#endif
-
-struct MDBX_xcursor;
-
-/* Cursors are used for all DB operations.
- * A cursor holds a path of (page pointer, key index) from the DB
- * root to a position in the DB, plus other state. MDBX_DUPSORT
- * cursors include an xcursor to the current data item. Write txns
- * track their cursors and keep them up to date when data moves.
- * Exception: An xcursor's pointer to a P_SUBP page can be stale.
- * (A node with F_DUPDATA but no F_SUBDATA contains a subpage). */
-struct MDBX_cursor {
-#define MDBX_MC_LIVE UINT32_C(0xFE05D5B1)
-#define MDBX_MC_READY4CLOSE UINT32_C(0x2817A047)
-#define MDBX_MC_WAIT4EOT UINT32_C(0x90E297A7)
- uint32_t mc_signature;
- /* The database handle this cursor operates on */
- MDBX_dbi mc_dbi;
- /* Next cursor on this DB in this txn */
- MDBX_cursor *mc_next;
- /* Backup of the original cursor if this cursor is a shadow */
- MDBX_cursor *mc_backup;
- /* Context used for databases with MDBX_DUPSORT, otherwise NULL */
- struct MDBX_xcursor *mc_xcursor;
- /* The transaction that owns this cursor */
- MDBX_txn *mc_txn;
- /* The database record for this cursor */
- MDBX_db *mc_db;
- /* The database auxiliary record for this cursor */
- MDBX_dbx *mc_dbx;
- /* The mt_dbistate for this database */
- uint8_t *mc_dbistate;
- unsigned mc_snum; /* number of pushed pages */
- unsigned mc_top; /* index of top page, normally mc_snum-1 */
-
- /* Cursor state flags. */
-#define C_INITIALIZED 0x01 /* cursor has been initialized and is valid */
-#define C_EOF 0x02 /* No more data */
-#define C_SUB 0x04 /* Cursor is a sub-cursor */
-#define C_DEL 0x08 /* last op was a cursor_del */
-#define C_UNTRACK 0x10 /* Un-track cursor when closing */
-#define C_RECLAIMING 0x20 /* GC lookup is prohibited */
-#define C_GCFREEZE 0x40 /* reclaimed_pglist must not be updated */
-
- /* Cursor checking flags. */
-#define C_COPYING 0x100 /* skip key-value length check (copying simplify) */
-#define C_UPDATING 0x200 /* update/rebalance pending */
-#define C_RETIRING 0x400 /* refs to child pages may be invalid */
-#define C_SKIPORD 0x800 /* don't check keys ordering */
-
- unsigned mc_flags; /* see mdbx_cursor */
- MDBX_page *mc_pg[CURSOR_STACK]; /* stack of pushed pages */
- indx_t mc_ki[CURSOR_STACK]; /* stack of page indices */
-};
-
-/* Context for sorted-dup records.
- * We could have gone to a fully recursive design, with arbitrarily
- * deep nesting of sub-databases. But for now we only handle these
- * levels - main DB, optional sub-DB, sorted-duplicate DB. */
-typedef struct MDBX_xcursor {
- /* A sub-cursor for traversing the Dup DB */
- MDBX_cursor mx_cursor;
- /* The database record for this Dup DB */
- MDBX_db mx_db;
- /* The auxiliary DB record for this Dup DB */
- MDBX_dbx mx_dbx;
-} MDBX_xcursor;
-
-typedef struct MDBX_cursor_couple {
- MDBX_cursor outer;
- void *mc_userctx; /* User-settable context */
- MDBX_xcursor inner;
-} MDBX_cursor_couple;
-
-/* The database environment. */
-struct MDBX_env {
- /* ----------------------------------------------------- mostly static part */
-#define MDBX_ME_SIGNATURE UINT32_C(0x9A899641)
- MDBX_atomic_uint32_t me_signature;
- /* Failed to update the meta page. Probably an I/O error. */
-#define MDBX_FATAL_ERROR UINT32_C(0x80000000)
- /* Some fields are initialized. */
-#define MDBX_ENV_ACTIVE UINT32_C(0x20000000)
- /* me_txkey is set */
-#define MDBX_ENV_TXKEY UINT32_C(0x10000000)
- /* Legacy MDBX_MAPASYNC (prior v0.9) */
-#define MDBX_DEPRECATED_MAPASYNC UINT32_C(0x100000)
-#define ENV_INTERNAL_FLAGS (MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE | MDBX_ENV_TXKEY)
- uint32_t me_flags;
- mdbx_mmap_t me_dxb_mmap; /* The main data file */
-#define me_map me_dxb_mmap.dxb
-#define me_lazy_fd me_dxb_mmap.fd
- mdbx_filehandle_t me_dsync_fd;
- mdbx_mmap_t me_lck_mmap; /* The lock file */
-#define me_lfd me_lck_mmap.fd
- struct MDBX_lockinfo *me_lck;
-
- unsigned me_psize; /* DB page size, initialized from me_os_psize */
- unsigned me_leaf_nodemax; /* max size of a leaf-node */
- uint8_t me_psize2log; /* log2 of DB page size */
- int8_t me_stuck_meta; /* recovery-only: target meta page or less that zero */
- uint16_t me_merge_threshold,
- me_merge_threshold_gc; /* pages emptier than this are candidates for
- merging */
- unsigned me_os_psize; /* OS page size, from mdbx_syspagesize() */
- unsigned me_maxreaders; /* size of the reader table */
- MDBX_dbi me_maxdbs; /* size of the DB table */
- uint32_t me_pid; /* process ID of this env */
- mdbx_thread_key_t me_txkey; /* thread-key for readers */
- char *me_pathname; /* path to the DB files */
- void *me_pbuf; /* scratch area for DUPSORT put() */
- MDBX_txn *me_txn0; /* preallocated write transaction */
-
- MDBX_dbx *me_dbxs; /* array of static DB info */
- uint16_t *me_dbflags; /* array of flags from MDBX_db.md_flags */
- unsigned *me_dbiseqs; /* array of dbi sequence numbers */
- unsigned
- me_maxgc_ov1page; /* Number of pgno_t fit in a single overflow page */
- uint32_t me_live_reader; /* have liveness lock in reader table */
- void *me_userctx; /* User-settable context */
- MDBX_hsr_func *me_hsr_callback; /* Callback for kicking laggard readers */
-
- struct {
- unsigned dp_reserve_limit;
- unsigned rp_augment_limit;
- unsigned dp_limit;
- unsigned dp_initial;
- uint8_t dp_loose_limit;
- uint8_t spill_max_denominator;
- uint8_t spill_min_denominator;
- uint8_t spill_parent4child_denominator;
- unsigned merge_threshold_16dot16_percent;
- union {
- unsigned all;
- /* tracks options with non-auto values but tuned by user */
- struct {
- unsigned dp_limit : 1;
- } non_auto;
- } flags;
- } me_options;
-
- /* struct me_dbgeo used for accepting db-geo params from user for the new
- * database creation, i.e. when mdbx_env_set_geometry() was called before
- * mdbx_env_open(). */
- struct {
- size_t lower; /* minimal size of datafile */
- size_t upper; /* maximal size of datafile */
- size_t now; /* current size of datafile */
- size_t grow; /* step to grow datafile */
- size_t shrink; /* threshold to shrink datafile */
- } me_dbgeo;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- union {
- key_t key;
- int semid;
- } me_sysv_ipc;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
- MDBX_env *me_lcklist_next;
-
- /* --------------------------------------------------- mostly volatile part */
-
- MDBX_txn *me_txn; /* current write transaction */
- mdbx_fastmutex_t me_dbi_lock;
- MDBX_dbi me_numdbs; /* number of DBs opened */
-
- MDBX_page *me_dp_reserve; /* list of malloc'ed blocks for re-use */
- unsigned me_dp_reserve_len;
- /* PNL of pages that became unused in a write txn */
- MDBX_PNL me_retired_pages;
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_srwlock me_remap_guard;
- /* Workaround for LockFileEx and WriteFile multithread bug */
- CRITICAL_SECTION me_windowsbug_lock;
-#else
- mdbx_fastmutex_t me_remap_guard;
-#endif
-
- /* -------------------------------------------------------------- debugging */
-
-#if MDBX_DEBUG
- MDBX_assert_func *me_assert_func; /* Callback for assertion failures */
-#endif
-#ifdef MDBX_USE_VALGRIND
- int me_valgrind_handle;
-#endif
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- pgno_t me_poison_edge;
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-#ifndef xMDBX_DEBUG_SPILLING
-#define xMDBX_DEBUG_SPILLING 0
-#endif
-#if xMDBX_DEBUG_SPILLING == 2
- unsigned debug_dirtied_est, debug_dirtied_act;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- /* ------------------------------------------------- stub for lck-less mode */
- MDBX_atomic_uint64_t
- x_lckless_stub[(sizeof(MDBX_lockinfo) + MDBX_CACHELINE_SIZE - 1) /
- sizeof(MDBX_atomic_uint64_t)];
-};
-
-#ifndef __cplusplus
-/*----------------------------------------------------------------------------*/
-/* Debug and Logging stuff */
-
-#define MDBX_RUNTIME_FLAGS_INIT \
- ((MDBX_DEBUG) > 0) * MDBX_DBG_ASSERT + ((MDBX_DEBUG) > 1) * MDBX_DBG_AUDIT
-
-extern uint8_t mdbx_runtime_flags;
-extern uint8_t mdbx_loglevel;
-extern MDBX_debug_func *mdbx_debug_logger;
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny) {
-#if MDBX_DEBUG
- if (MDBX_DBG_JITTER & mdbx_runtime_flags)
- mdbx_osal_jitter(tiny);
-#else
- (void)tiny;
-#endif
-}
-
-MDBX_INTERNAL_FUNC void MDBX_PRINTF_ARGS(4, 5)
- mdbx_debug_log(int level, const char *function, int line, const char *fmt,
- ...) MDBX_PRINTF_ARGS(4, 5);
-MDBX_INTERNAL_FUNC void mdbx_debug_log_va(int level, const char *function,
- int line, const char *fmt,
- va_list args);
-
-#if MDBX_DEBUG
-#define mdbx_log_enabled(msg) unlikely(msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() unlikely((mdbx_runtime_flags & MDBX_DBG_AUDIT))
-#else /* MDBX_DEBUG */
-#define mdbx_log_enabled(msg) (msg < MDBX_LOG_VERBOSE && msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() (0)
-#endif /* MDBX_DEBUG */
-
-#if MDBX_FORCE_ASSERTIONS
-#define mdbx_assert_enabled() (1)
-#elif MDBX_DEBUG
-#define mdbx_assert_enabled() likely((mdbx_runtime_flags & MDBX_DBG_ASSERT))
-#else
-#define mdbx_assert_enabled() (0)
-#endif /* assertions */
-
-#if !MDBX_DEBUG && defined(__ANDROID_API__)
-#define mdbx_assert_fail(env, msg, func, line) \
- __android_log_assert(msg, "mdbx", "%s:%u", func, line)
-#else
-void mdbx_assert_fail(const MDBX_env *env, const char *msg, const char *func,
- int line);
-#endif
-
-#define mdbx_debug_extra(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, __func__, __LINE__, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug_extra_print(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, NULL, 0, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_trace(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_TRACE)) \
- mdbx_debug_log(MDBX_LOG_TRACE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_DEBUG)) \
- mdbx_debug_log(MDBX_LOG_DEBUG, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_verbose(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_VERBOSE)) \
- mdbx_debug_log(MDBX_LOG_VERBOSE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_notice(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_NOTICE)) \
- mdbx_debug_log(MDBX_LOG_NOTICE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_warning(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_WARN)) \
- mdbx_debug_log(MDBX_LOG_WARN, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_error(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) \
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_fatal(fmt, ...) \
- mdbx_debug_log(MDBX_LOG_FATAL, __func__, __LINE__, fmt "\n", __VA_ARGS__);
-
-#define mdbx_ensure_msg(env, expr, msg) \
- do { \
- if (unlikely(!(expr))) \
- mdbx_assert_fail(env, msg, __func__, __LINE__); \
- } while (0)
-
-#define mdbx_ensure(env, expr) mdbx_ensure_msg(env, expr, #expr)
-
-/* assert(3) variant in environment context */
-#define mdbx_assert(env, expr) \
- do { \
- if (mdbx_assert_enabled()) \
- mdbx_ensure(env, expr); \
- } while (0)
-
-/* assert(3) variant in cursor context */
-#define mdbx_cassert(mc, expr) mdbx_assert((mc)->mc_txn->mt_env, expr)
-
-/* assert(3) variant in transaction context */
-#define mdbx_tassert(txn, expr) mdbx_assert((txn)->mt_env, expr)
-
-#ifndef xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-#undef assert
-#define assert(expr) mdbx_assert(NULL, expr)
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Cache coherence and mmap invalidation */
-
-#if MDBX_CPU_WRITEBACK_INCOHERENT
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_memory_barrier()
-#else
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_compiler_barrier()
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-MDBX_MAYBE_UNUSED static __inline void
-mdbx_flush_incoherent_mmap(void *addr, size_t nbytes, const intptr_t pagesize) {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- char *const begin = (char *)(-pagesize & (intptr_t)addr);
- char *const end =
- (char *)(-pagesize & (intptr_t)((char *)addr + nbytes + pagesize - 1));
- int err = msync(begin, end - begin, MS_SYNC | MS_INVALIDATE) ? errno : 0;
- mdbx_assert(nullptr, err == 0);
- (void)err;
-#else
- (void)pagesize;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#if MDBX_MMAP_INCOHERENT_CPU_CACHE
-#ifdef DCACHE
- /* MIPS has cache coherency issues.
- * Note: for any nbytes >= on-chip cache size, entire is flushed. */
- cacheflush(addr, nbytes, DCACHE);
-#else
-#error "Oops, cacheflush() not available"
-#endif /* DCACHE */
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#if !MDBX_MMAP_INCOHERENT_FILE_WRITE && !MDBX_MMAP_INCOHERENT_CPU_CACHE
- (void)addr;
- (void)nbytes;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* Internal prototypes */
-
-MDBX_INTERNAL_FUNC int mdbx_cleanup_dead_readers(MDBX_env *env, int rlocked,
- int *dead);
-MDBX_INTERNAL_FUNC int mdbx_rthc_alloc(mdbx_thread_key_t *key,
- MDBX_reader *begin, MDBX_reader *end);
-MDBX_INTERNAL_FUNC void mdbx_rthc_remove(const mdbx_thread_key_t key);
-
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_init(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_dtor(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_thread_dtor(void *ptr);
-
-#endif /* !__cplusplus */
-
-#define MDBX_IS_ERROR(rc) \
- ((rc) != MDBX_RESULT_TRUE && (rc) != MDBX_RESULT_FALSE)
-
-/* Internal error codes, not exposed outside libmdbx */
-#define MDBX_NO_ROOT (MDBX_LAST_ADDED_ERRCODE + 10)
-
-/* Debugging output value of a cursor DBI: Negative in a sub-cursor. */
-#define DDBI(mc) \
- (((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
-
-/* Key size which fits in a DKBUF (debug key buffer). */
-#define DKBUF_MAX 511
-#define DKBUF char _kbuf[DKBUF_MAX * 4 + 2]
-#define DKEY(x) mdbx_dump_val(x, _kbuf, DKBUF_MAX * 2 + 1)
-#define DVAL(x) mdbx_dump_val(x, _kbuf + DKBUF_MAX * 2 + 1, DKBUF_MAX * 2 + 1)
-
-#if MDBX_DEBUG
-#define DKBUF_DEBUG DKBUF
-#define DKEY_DEBUG(x) DKEY(x)
-#define DVAL_DEBUG(x) DVAL(x)
-#else
-#define DKBUF_DEBUG ((void)(0))
-#define DKEY_DEBUG(x) ("-")
-#define DVAL_DEBUG(x) ("-")
-#endif
-
-/* An invalid page number.
- * Mainly used to denote an empty tree. */
-#define P_INVALID (~(pgno_t)0)
-
-/* Test if the flags f are set in a flag word w. */
-#define F_ISSET(w, f) (((w) & (f)) == (f))
-
-/* Round n up to an even number. */
-#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
-
-/* Default size of memory map.
- * This is certainly too small for any actual applications. Apps should
- * always set the size explicitly using mdbx_env_set_geometry(). */
-#define DEFAULT_MAPSIZE MEGABYTE
-
-/* Number of slots in the reader table.
- * This value was chosen somewhat arbitrarily. The 61 is a prime number,
- * and such readers plus a couple mutexes fit into single 4KB page.
- * Applications should set the table size using mdbx_env_set_maxreaders(). */
-#define DEFAULT_READERS 61
-
-/* Test if a page is a leaf page */
-#define IS_LEAF(p) (((p)->mp_flags & P_LEAF) != 0)
-/* Test if a page is a LEAF2 page */
-#define IS_LEAF2(p) unlikely(((p)->mp_flags & P_LEAF2) != 0)
-/* Test if a page is a branch page */
-#define IS_BRANCH(p) (((p)->mp_flags & P_BRANCH) != 0)
-/* Test if a page is an overflow page */
-#define IS_OVERFLOW(p) unlikely(((p)->mp_flags & P_OVERFLOW) != 0)
-/* Test if a page is a sub page */
-#define IS_SUBP(p) (((p)->mp_flags & P_SUBP) != 0)
-
-#define PAGETYPE(p) ((p)->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW))
-
-/* Header for a single key/data pair within a page.
- * Used in pages of type P_BRANCH and P_LEAF without P_LEAF2.
- * We guarantee 2-byte alignment for 'MDBX_node's.
- *
- * Leaf node flags describe node contents. F_BIGDATA says the node's
- * data part is the page number of an overflow page with actual data.
- * F_DUPDATA and F_SUBDATA can be combined giving duplicate data in
- * a sub-page/sub-database, and named databases (just F_SUBDATA). */
-typedef struct MDBX_node {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- union {
- uint32_t mn_dsize;
- uint32_t mn_pgno32;
- };
- uint8_t mn_flags; /* see mdbx_node flags */
- uint8_t mn_extra;
- uint16_t mn_ksize; /* key size */
-#else
- uint16_t mn_ksize; /* key size */
- uint8_t mn_extra;
- uint8_t mn_flags; /* see mdbx_node flags */
- union {
- uint32_t mn_pgno32;
- uint32_t mn_dsize;
- };
-#endif /* __BYTE_ORDER__ */
-
- /* mdbx_node Flags */
-#define F_BIGDATA 0x01 /* data put on overflow page */
-#define F_SUBDATA 0x02 /* data is a sub-database */
-#define F_DUPDATA 0x04 /* data has duplicates */
-
- /* valid flags for mdbx_node_add() */
-#define NODE_ADD_FLAGS (F_DUPDATA | F_SUBDATA | MDBX_RESERVE | MDBX_APPEND)
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- uint8_t mn_data[] /* key and data are appended here */;
-#endif /* C99 */
-} MDBX_node;
-
-#define DB_PERSISTENT_FLAGS \
- (MDBX_REVERSEKEY | MDBX_DUPSORT | MDBX_INTEGERKEY | MDBX_DUPFIXED | \
- MDBX_INTEGERDUP | MDBX_REVERSEDUP)
-
-/* mdbx_dbi_open() flags */
-#define DB_USABLE_FLAGS (DB_PERSISTENT_FLAGS | MDBX_CREATE | MDBX_DB_ACCEDE)
-
-#define DB_VALID 0x8000 /* DB handle is valid, for me_dbflags */
-#define DB_INTERNAL_FLAGS DB_VALID
-
-#if DB_INTERNAL_FLAGS & DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-#if DB_PERSISTENT_FLAGS & ~DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-
-/* max number of pages to commit in one writev() call */
-#define MDBX_COMMIT_PAGES 64
-#if defined(IOV_MAX) && IOV_MAX < MDBX_COMMIT_PAGES /* sysconf(_SC_IOV_MAX) */
-#undef MDBX_COMMIT_PAGES
-#define MDBX_COMMIT_PAGES IOV_MAX
-#endif
-
-/*
- * /
- * | -1, a < b
- * CMP2INT(a,b) = < 0, a == b
- * | 1, a > b
- * \
- */
-#ifndef __e2k__
-/* LY: fast enough on most systems */
-#define CMP2INT(a, b) (((b) > (a)) ? -1 : (a) > (b))
-#else
-/* LY: more parallelable on VLIW Elbrus */
-#define CMP2INT(a, b) (((a) > (b)) - ((b) > (a)))
-#endif
-
-/* Do not spill pages to disk if txn is getting full, may fail instead */
-#define MDBX_NOSPILL 0x8000
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-int64pgno(int64_t i64) {
- if (likely(i64 >= (int64_t)MIN_PAGENO && i64 <= (int64_t)MAX_PAGENO + 1))
- return (pgno_t)i64;
- return (i64 < (int64_t)MIN_PAGENO) ? MIN_PAGENO : MAX_PAGENO;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_add(size_t base, size_t augend) {
- assert(base <= MAX_PAGENO + 1 && augend < MAX_PAGENO);
- return int64pgno(base + augend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_sub(size_t base, size_t subtrahend) {
- assert(base >= MIN_PAGENO && base <= MAX_PAGENO + 1 &&
- subtrahend < MAX_PAGENO);
- return int64pgno(base - subtrahend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline bool
-is_powerof2(size_t x) {
- return (x & (x - 1)) == 0;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-floor_powerof2(size_t value, size_t granularity) {
- assert(is_powerof2(granularity));
- return value & ~(granularity - 1);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-ceil_powerof2(size_t value, size_t granularity) {
- return floor_powerof2(value + granularity - 1, granularity);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static unsigned
-log2n_powerof2(size_t value) {
- assert(value > 0 && value < INT32_MAX && is_powerof2(value));
- assert((value & -(int32_t)value) == value);
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_ctzl)
- return __builtin_ctzl(value);
-#elif defined(_MSC_VER)
- unsigned long index;
- _BitScanForward(&index, (unsigned long)value);
- return index;
-#else
- static const uint8_t debruijn_ctz32[32] = {
- 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
- 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9};
- return debruijn_ctz32[(uint32_t)(value * 0x077CB531u) >> 27];
-#endif
-}
-
-/* Only a subset of the mdbx_env flags can be changed
- * at runtime. Changing other flags requires closing the
- * environment and re-opening it with the new flags. */
-#define ENV_CHANGEABLE_FLAGS \
- (MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC | MDBX_DEPRECATED_MAPASYNC | \
- MDBX_NOMEMINIT | MDBX_COALESCE | MDBX_PAGEPERTURB | MDBX_ACCEDE)
-#define ENV_CHANGELESS_FLAGS \
- (MDBX_NOSUBDIR | MDBX_RDONLY | MDBX_WRITEMAP | MDBX_NOTLS | MDBX_NORDAHEAD | \
- MDBX_LIFORECLAIM | MDBX_EXCLUSIVE)
-#define ENV_USABLE_FLAGS (ENV_CHANGEABLE_FLAGS | ENV_CHANGELESS_FLAGS)
-
-#if !defined(__cplusplus) || CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_MAYBE_UNUSED static void static_checks(void) {
- STATIC_ASSERT_MSG(INT16_MAX - CORE_DBS == MDBX_MAX_DBI,
- "Oops, MDBX_MAX_DBI or CORE_DBS?");
- STATIC_ASSERT_MSG((unsigned)(MDBX_DB_ACCEDE | MDBX_CREATE) ==
- ((DB_USABLE_FLAGS | DB_INTERNAL_FLAGS) &
- (ENV_USABLE_FLAGS | ENV_INTERNAL_FLAGS)),
- "Oops, some flags overlapped or wrong");
- STATIC_ASSERT_MSG((ENV_INTERNAL_FLAGS & ENV_USABLE_FLAGS) == 0,
- "Oops, some flags overlapped or wrong");
-}
-#endif /* Disabled for MSVC 19.0 (VisualStudio 2015) */
-
-#ifdef __cplusplus
-}
-#endif
-
-#define MDBX_ASAN_POISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("POISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_POISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#define MDBX_ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("UNPOISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_UNPOISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#include <ctype.h>
-
-#if defined(_WIN32) || defined(_WIN64)
-/*
- * POSIX getopt for Windows
- *
- * AT&T Public License
- *
- * Code given out at the 1985 UNIFORUM conference in Dallas.
- */
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif
-#endif /* _MSC_VER (warnings) */
-
-#include <stdio.h>
-#include <string.h>
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-/*----------------------------------------------------------------------------*/
-
-#ifndef NULL
-#define NULL 0
-#endif
-
-#ifndef EOF
-#define EOF (-1)
-#endif
-
-int optind = 1;
-int optopt;
-char *optarg;
-
-int getopt(int argc, char *const argv[], const char *opts) {
- static int sp = 1;
- int c;
- const char *cp;
-
- if (sp == 1) {
- if (optind >= argc || argv[optind][0] != '-' || argv[optind][1] == '\0')
- return EOF;
- else if (strcmp(argv[optind], "--") == 0) {
- optind++;
- return EOF;
- }
- }
- optopt = c = argv[optind][sp];
- if (c == ':' || (cp = strchr(opts, c)) == NULL) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "illegal option", c);
- if (argv[optind][++sp] == '\0') {
- optind++;
- sp = 1;
- }
- return '?';
- }
- if (*++cp == ':') {
- if (argv[optind][sp + 1] != '\0')
- optarg = &argv[optind++][sp + 1];
- else if (++optind >= argc) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "option requires an argument",
- c);
- sp = 1;
- return '?';
- } else
- optarg = argv[optind++];
- sp = 1;
- } else {
- if (argv[optind][++sp] == '\0') {
- sp = 1;
- optind++;
- }
- optarg = NULL;
- }
- return c;
-}
-
-static volatile BOOL user_break;
-static BOOL WINAPI ConsoleBreakHandlerRoutine(DWORD dwCtrlType) {
- (void)dwCtrlType;
- user_break = true;
- return true;
-}
-
-#else /* WINDOWS */
-
-static volatile sig_atomic_t user_break;
-static void signal_handler(int sig) {
- (void)sig;
- user_break = 1;
-}
-
-#endif /* !WINDOWS */
-
-static char *prog;
-bool quiet = false;
-static void usage(void) {
- fprintf(stderr,
- "usage: %s [-V] [-q] [-d] [-s name] dbpath\n"
- " -V\t\tprint version and exit\n"
- " -q\t\tbe quiet\n"
- " -d\t\tdelete the specified database, don't just empty it\n"
- " -s name\tdrop the specified named subDB\n"
- " \t\tby default empty the main DB\n",
- prog);
- exit(EXIT_FAILURE);
-}
-
-static void error(const char *func, int rc) {
- if (!quiet)
- fprintf(stderr, "%s: %s() error %d %s\n", prog, func, rc,
- mdbx_strerror(rc));
-}
-
-int main(int argc, char *argv[]) {
- int i, rc;
- MDBX_env *env;
- MDBX_txn *txn;
- MDBX_dbi dbi;
- char *envname = nullptr;
- char *subname = nullptr;
- int envflags = MDBX_ACCEDE;
- bool delete = false;
-
- prog = argv[0];
- if (argc < 2)
- usage();
-
- while ((i = getopt(argc, argv,
- "d"
- "s:"
- "n"
- "q"
- "V")) != EOF) {
- switch (i) {
- case 'V':
- printf("mdbx_drop version %d.%d.%d.%d\n"
- " - source: %s %s, commit %s, tree %s\n"
- " - anchor: %s\n"
- " - build: %s for %s by %s\n"
- " - flags: %s\n"
- " - options: %s\n",
- mdbx_version.major, mdbx_version.minor, mdbx_version.release,
- mdbx_version.revision, mdbx_version.git.describe,
- mdbx_version.git.datetime, mdbx_version.git.commit,
- mdbx_version.git.tree, mdbx_sourcery_anchor, mdbx_build.datetime,
- mdbx_build.target, mdbx_build.compiler, mdbx_build.flags,
- mdbx_build.options);
- return EXIT_SUCCESS;
- case 'q':
- quiet = true;
- break;
- case 'd':
- delete = true;
- break;
- case 'n':
- break;
- case 's':
- subname = optarg;
- break;
- default:
- usage();
- }
- }
-
- if (optind != argc - 1)
- usage();
-
-#if defined(_WIN32) || defined(_WIN64)
- SetConsoleCtrlHandler(ConsoleBreakHandlerRoutine, true);
-#else
-#ifdef SIGPIPE
- signal(SIGPIPE, signal_handler);
-#endif
-#ifdef SIGHUP
- signal(SIGHUP, signal_handler);
-#endif
- signal(SIGINT, signal_handler);
- signal(SIGTERM, signal_handler);
-#endif /* !WINDOWS */
-
- envname = argv[optind];
- if (!quiet) {
- printf("mdbx_drop %s (%s, T-%s)\nRunning for %s/%s...\n",
- mdbx_version.git.describe, mdbx_version.git.datetime,
- mdbx_version.git.tree, envname, subname ? subname : "@MAIN");
- fflush(nullptr);
- }
-
- rc = mdbx_env_create(&env);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_create", rc);
- return EXIT_FAILURE;
- }
-
- if (subname) {
- rc = mdbx_env_set_maxdbs(env, 2);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_set_maxdbs", rc);
- goto env_close;
- }
- }
-
- rc = mdbx_env_open(env, envname, envflags, 0);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_open", rc);
- goto env_close;
- }
-
- rc = mdbx_txn_begin(env, NULL, 0, &txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_begin", rc);
- goto env_close;
- }
-
- rc = mdbx_dbi_open(txn, subname, MDBX_DB_ACCEDE, &dbi);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_open", rc);
- goto txn_abort;
- }
-
- rc = mdbx_drop(txn, dbi, delete);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_drop", rc);
- goto txn_abort;
- }
-
- rc = mdbx_txn_commit(txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_commit", rc);
- goto txn_abort;
- }
- txn = nullptr;
-
-txn_abort:
- if (txn)
- mdbx_txn_abort(txn);
-env_close:
- mdbx_env_close(env);
-
- return rc ? EXIT_FAILURE : EXIT_SUCCESS;
-}
diff --git a/3rdparty/libmdbx/mdbx_dump.c b/3rdparty/libmdbx/mdbx_dump.c
deleted file mode 100644
index e806fa8c99ab4922fc7949138e6c921ed7b79cdf..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx_dump.c
+++ /dev/null
@@ -1,4020 +0,0 @@
-/* mdbx_dump.c - memory-mapped database dump tool */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#ifdef _MSC_VER
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#pragma warning(disable : 4996) /* The POSIX name is deprecated... */
-#endif /* _MSC_VER (warnings) */
-
-#define xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#define MDBX_BUILD_SOURCERY c510b93d78b5848764f976ea67f06d278a7b99f141b4c98ceb095285f834f31c_v0_11_7_0_g40ec559c
-#ifdef MDBX_CONFIG_H
-#include MDBX_CONFIG_H
-#endif
-
-#define LIBMDBX_INTERNALS
-#ifdef xMDBX_TOOLS
-#define MDBX_DEPRECATED
-#endif /* xMDBX_TOOLS */
-
-#ifdef xMDBX_ALLOY
-/* Amalgamated build */
-#define MDBX_INTERNAL_FUNC static
-#define MDBX_INTERNAL_VAR static
-#else
-/* Non-amalgamated build */
-#define MDBX_INTERNAL_FUNC
-#define MDBX_INTERNAL_VAR extern
-#endif /* xMDBX_ALLOY */
-
-/** Disables using GNU/Linux libc extensions.
- * \ingroup build_option
- * \note This option couldn't be moved to the options.h since dependant
- * control macros/defined should be prepared before include the options.h */
-#ifndef MDBX_DISABLE_GNU_SOURCE
-#define MDBX_DISABLE_GNU_SOURCE 0
-#endif
-#if MDBX_DISABLE_GNU_SOURCE
-#undef _GNU_SOURCE
-#elif (defined(__linux__) || defined(__gnu_linux__)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif /* MDBX_DISABLE_GNU_SOURCE */
-
-/*----------------------------------------------------------------------------*/
-
-/* Should be defined before any includes */
-#if !defined(_FILE_OFFSET_BITS) && !defined(__ANDROID_API__) && \
- !defined(ANDROID)
-#define _FILE_OFFSET_BITS 64
-#endif
-
-#ifdef __APPLE__
-#define _DARWIN_C_SOURCE
-#endif
-
-#ifdef _MSC_VER
-#if _MSC_FULL_VER < 190024234
-/* Actually libmdbx was not tested with compilers older than 19.00.24234 (Visual
- * Studio 2015 Update 3). But you could remove this #error and try to continue
- * at your own risk. In such case please don't rise up an issues related ONLY to
- * old compilers.
- *
- * NOTE:
- * Unfortunately, there are several different builds of "Visual Studio" that
- * are called "Visual Studio 2015 Update 3".
- *
- * The 190024234 is used here because it is minimal version of Visual Studio
- * that was used for build and testing libmdbx in recent years. Soon this
- * value will be increased to 19.0.24241.7, since build and testing using
- * "Visual Studio 2015" will be performed only at https://ci.appveyor.com.
- *
- * Please ask Microsoft (but not us) for information about version differences
- * and how to and where you can obtain the latest "Visual Studio 2015" build
- * with all fixes.
- */
-#error \
- "At least \"Microsoft C/C++ Compiler\" version 19.00.24234 (Visual Studio 2015 Update 3) is required."
-#endif
-#ifndef _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#if _MSC_VER > 1913
-#pragma warning(disable : 5045) /* Compiler will insert Spectre mitigation... \
- */
-#endif
-#if _MSC_VER > 1914
-#pragma warning( \
- disable : 5105) /* winbase.h(9531): warning C5105: macro expansion \
- producing 'defined' has undefined behavior */
-#endif
-#pragma warning(disable : 4710) /* 'xyz': function not inlined */
-#pragma warning(disable : 4711) /* function 'xyz' selected for automatic \
- inline expansion */
-#pragma warning( \
- disable : 4201) /* nonstandard extension used : nameless struct / union */
-#pragma warning(disable : 4702) /* unreachable code */
-#pragma warning(disable : 4706) /* assignment within conditional expression */
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4324) /* 'xyz': structure was padded due to \
- alignment specifier */
-#pragma warning(disable : 4310) /* cast truncates constant value */
-#pragma warning( \
- disable : 4820) /* bytes padding added after data member for alignment */
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4366) /* the result of the unary '&' operator may be \
- unaligned */
-#pragma warning(disable : 4200) /* nonstandard extension used: zero-sized \
- array in struct/union */
-#pragma warning(disable : 4204) /* nonstandard extension used: non-constant \
- aggregate initializer */
-#pragma warning( \
- disable : 4505) /* unreferenced local function has been removed */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(__GNUC__) && __GNUC__ < 9
-#pragma GCC diagnostic ignored "-Wattributes"
-#endif /* GCC < 9 */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-#include "mdbx.h"
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#ifndef __GNUC_PREREQ
-# if defined(__GNUC__) && defined(__GNUC_MINOR__)
-# define __GNUC_PREREQ(maj, min) \
- ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GNUC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GNUC_PREREQ */
-
-#ifndef __CLANG_PREREQ
-# ifdef __clang__
-# define __CLANG_PREREQ(maj,min) \
- ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min))
-# else
-# define __CLANG_PREREQ(maj,min) (0)
-# endif
-#endif /* __CLANG_PREREQ */
-
-#ifndef __GLIBC_PREREQ
-# if defined(__GLIBC__) && defined(__GLIBC_MINOR__)
-# define __GLIBC_PREREQ(maj, min) \
- ((__GLIBC__ << 16) + __GLIBC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GLIBC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GLIBC_PREREQ */
-
-#ifndef __has_warning
-# define __has_warning(x) (0)
-#endif
-
-#ifndef __has_include
-# define __has_include(x) (0)
-#endif
-
-#if __has_feature(thread_sanitizer)
-# define __SANITIZE_THREAD__ 1
-#endif
-
-#if __has_feature(address_sanitizer)
-# define __SANITIZE_ADDRESS__ 1
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __extern_C
-# ifdef __cplusplus
-# define __extern_C extern "C"
-# else
-# define __extern_C
-# endif
-#endif /* __extern_C */
-
-#if !defined(nullptr) && !defined(__cplusplus) || (__cplusplus < 201103L && !defined(_MSC_VER))
-# define nullptr NULL
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __always_inline
-# if defined(__GNUC__) || __has_attribute(__always_inline__)
-# define __always_inline __inline __attribute__((__always_inline__))
-# elif defined(_MSC_VER)
-# define __always_inline __forceinline
-# else
-# define __always_inline
-# endif
-#endif /* __always_inline */
-
-#ifndef __noinline
-# if defined(__GNUC__) || __has_attribute(__noinline__)
-# define __noinline __attribute__((__noinline__))
-# elif defined(_MSC_VER)
-# define __noinline __declspec(noinline)
-# else
-# define __noinline
-# endif
-#endif /* __noinline */
-
-#ifndef __must_check_result
-# if defined(__GNUC__) || __has_attribute(__warn_unused_result__)
-# define __must_check_result __attribute__((__warn_unused_result__))
-# else
-# define __must_check_result
-# endif
-#endif /* __must_check_result */
-
-#if !defined(__noop) && !defined(_MSC_VER)
-# define __noop(...) do {} while(0)
-#endif /* __noop */
-
-#if defined(__fallthrough) && \
- (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__))
-#undef __fallthrough
-#endif /* __fallthrough workaround for MinGW */
-
-#ifndef __fallthrough
-# if defined(__cplusplus) && (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || __cplusplus >= 201703L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(8, 0) && defined(__cplusplus) && __cplusplus >= 201103L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(7, 0) && \
- (!defined(__LCC__) || (__LCC__ == 124 && __LCC_MINOR__ >= 12) || \
- (__LCC__ == 125 && __LCC_MINOR__ >= 5) || (__LCC__ >= 126))
-# define __fallthrough __attribute__((__fallthrough__))
-# elif defined(__clang__) && defined(__cplusplus) && __cplusplus >= 201103L &&\
- __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
-# define __fallthrough [[clang::fallthrough]]
-# else
-# define __fallthrough
-# endif
-#endif /* __fallthrough */
-
-#ifndef __unreachable
-# if __GNUC_PREREQ(4,5) || __has_builtin(__builtin_unreachable)
-# define __unreachable() __builtin_unreachable()
-# elif defined(_MSC_VER)
-# define __unreachable() __assume(0)
-# else
-# define __unreachable() __noop()
-# endif
-#endif /* __unreachable */
-
-#ifndef __prefetch
-# if defined(__GNUC__) || defined(__clang__) || __has_builtin(__builtin_prefetch)
-# define __prefetch(ptr) __builtin_prefetch(ptr)
-# else
-# define __prefetch(ptr) __noop(ptr)
-# endif
-#endif /* __prefetch */
-
-#ifndef __nothrow
-# if defined(__cplusplus)
-# if __cplusplus < 201703L
-# define __nothrow throw()
-# else
-# define __nothrow noexcept(true)
-# endif /* __cplusplus */
-# elif defined(__GNUC__) || __has_attribute(__nothrow__)
-# define __nothrow __attribute__((__nothrow__))
-# elif defined(_MSC_VER) && defined(__cplusplus)
-# define __nothrow __declspec(nothrow)
-# else
-# define __nothrow
-# endif
-#endif /* __nothrow */
-
-#ifndef __hidden
-# if defined(__GNUC__) || __has_attribute(__visibility__)
-# define __hidden __attribute__((__visibility__("hidden")))
-# else
-# define __hidden
-# endif
-#endif /* __hidden */
-
-#ifndef __optimize
-# if defined(__OPTIMIZE__)
-# if (defined(__GNUC__) && !defined(__clang__)) || __has_attribute(__optimize__)
-# define __optimize(ops) __attribute__((__optimize__(ops)))
-# else
-# define __optimize(ops)
-# endif
-# else
-# define __optimize(ops)
-# endif
-#endif /* __optimize */
-
-#ifndef __hot
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __hot __attribute__((__hot__)) __optimize(3)
-# elif defined(__clang__) && !__has_attribute(__hot_) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put frequently used functions in separate section */
-# define __hot __attribute__((__section__("text.hot"))) __optimize("O3")
-# elif defined(__GNUC__) || __has_attribute(__hot__)
-# define __hot __attribute__((__hot__)) __optimize("O3")
-# else
-# define __hot __optimize("O3")
-# endif
-# else
-# define __hot
-# endif
-#endif /* __hot */
-
-#ifndef __cold
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __cold __attribute__((__cold__)) __optimize(1)
-# elif defined(__clang__) && !__has_attribute(cold) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put infrequently used functions in separate section */
-# define __cold __attribute__((__section__("text.unlikely"))) __optimize("Os")
-# elif defined(__GNUC__) || __has_attribute(cold)
-# define __cold __attribute__((__cold__)) __optimize("Os")
-# else
-# define __cold __optimize("Os")
-# endif
-# else
-# define __cold
-# endif
-#endif /* __cold */
-
-#ifndef __flatten
-# if defined(__OPTIMIZE__) && (defined(__GNUC__) || __has_attribute(__flatten__))
-# define __flatten __attribute__((__flatten__))
-# else
-# define __flatten
-# endif
-#endif /* __flatten */
-
-#ifndef likely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define likely(cond) __builtin_expect(!!(cond), 1)
-# else
-# define likely(x) (!!(x))
-# endif
-#endif /* likely */
-
-#ifndef unlikely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define unlikely(cond) __builtin_expect(!!(cond), 0)
-# else
-# define unlikely(x) (!!(x))
-# endif
-#endif /* unlikely */
-
-#ifndef __anonymous_struct_extension__
-# if defined(__GNUC__)
-# define __anonymous_struct_extension__ __extension__
-# else
-# define __anonymous_struct_extension__
-# endif
-#endif /* __anonymous_struct_extension__ */
-
-#ifndef __Wpedantic_format_voidptr
- MDBX_MAYBE_UNUSED MDBX_PURE_FUNCTION static __inline const void*
- __Wpedantic_format_voidptr(const void* ptr) {return ptr;}
-# define __Wpedantic_format_voidptr(ARG) __Wpedantic_format_voidptr(ARG)
-#endif /* __Wpedantic_format_voidptr */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(MDBX_USE_VALGRIND)
-# include <valgrind/memcheck.h>
-# ifndef VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE
- /* LY: available since Valgrind 3.10 */
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# endif
-#elif !defined(RUNNING_ON_VALGRIND)
-# define VALGRIND_CREATE_MEMPOOL(h,r,z)
-# define VALGRIND_DESTROY_MEMPOOL(h)
-# define VALGRIND_MEMPOOL_TRIM(h,a,s)
-# define VALGRIND_MEMPOOL_ALLOC(h,a,s)
-# define VALGRIND_MEMPOOL_FREE(h,a)
-# define VALGRIND_MEMPOOL_CHANGE(h,a,b,s)
-# define VALGRIND_MAKE_MEM_NOACCESS(a,s)
-# define VALGRIND_MAKE_MEM_DEFINED(a,s)
-# define VALGRIND_MAKE_MEM_UNDEFINED(a,s)
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_CHECK_MEM_IS_ADDRESSABLE(a,s) (0)
-# define VALGRIND_CHECK_MEM_IS_DEFINED(a,s) (0)
-# define RUNNING_ON_VALGRIND (0)
-#endif /* MDBX_USE_VALGRIND */
-
-#ifdef __SANITIZE_ADDRESS__
-# include <sanitizer/asan_interface.h>
-#elif !defined(ASAN_POISON_MEMORY_REGION)
-# define ASAN_POISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-# define ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-#endif /* __SANITIZE_ADDRESS__ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef ARRAY_LENGTH
-# ifdef __cplusplus
- template <typename T, size_t N>
- char (&__ArraySizeHelper(T (&array)[N]))[N];
-# define ARRAY_LENGTH(array) (sizeof(::__ArraySizeHelper(array)))
-# else
-# define ARRAY_LENGTH(array) (sizeof(array) / sizeof(array[0]))
-# endif
-#endif /* ARRAY_LENGTH */
-
-#ifndef ARRAY_END
-# define ARRAY_END(array) (&array[ARRAY_LENGTH(array)])
-#endif /* ARRAY_END */
-
-#define CONCAT(a,b) a##b
-#define XCONCAT(a,b) CONCAT(a,b)
-
-#ifndef offsetof
-# define offsetof(type, member) __builtin_offsetof(type, member)
-#endif /* offsetof */
-
-#ifndef container_of
-# define container_of(ptr, type, member) \
- ((type *)((char *)(ptr) - offsetof(type, member)))
-#endif /* container_of */
-
-#define MDBX_TETRAD(a, b, c, d) \
- ((uint32_t)(a) << 24 | (uint32_t)(b) << 16 | (uint32_t)(c) << 8 | (d))
-
-#define MDBX_STRING_TETRAD(str) MDBX_TETRAD(str[0], str[1], str[2], str[3])
-
-#define FIXME "FIXME: " __FILE__ ", " MDBX_STRINGIFY(__LINE__)
-
-#ifndef STATIC_ASSERT_MSG
-# if defined(static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) static_assert(expr, msg)
-# elif defined(_STATIC_ASSERT)
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif defined(_MSC_VER)
-# include <crtdbg.h>
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) \
- || __has_feature(c_static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) _Static_assert(expr, msg)
-# else
-# define STATIC_ASSERT_MSG(expr, msg) switch (0) {case 0:case (expr):;}
-# endif
-#endif /* STATIC_ASSERT */
-
-#ifndef STATIC_ASSERT
-# define STATIC_ASSERT(expr) STATIC_ASSERT_MSG(expr, #expr)
-#endif
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__GNUC__) && !__GNUC_PREREQ(4, 2)
-/* Actually libmdbx was not tested with compilers older than GCC 4.2.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required GCC >= 4.2"
-#endif
-
-#if defined(__clang__) && !__CLANG_PREREQ(3, 8)
-/* Actually libmdbx was not tested with CLANG older than 3.8.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required CLANG >= 3.8"
-#endif
-
-#if defined(__GLIBC__) && !__GLIBC_PREREQ(2, 12)
-/* Actually libmdbx was not tested with something older than glibc 2.12.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old systems.
- */
-#warning "libmdbx was only tested with GLIBC >= 2.12."
-#endif
-
-#ifdef __SANITIZE_THREAD__
-#warning \
- "libmdbx don't compatible with ThreadSanitizer, you will get a lot of false-positive issues."
-#endif /* __SANITIZE_THREAD__ */
-
-#if __has_warning("-Wnested-anon-types")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wnested-anon-types"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wnested-anon-types"
-#else
-#pragma warning disable "nested-anon-types"
-#endif
-#endif /* -Wnested-anon-types */
-
-#if __has_warning("-Wconstant-logical-operand")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wconstant-logical-operand"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wconstant-logical-operand"
-#else
-#pragma warning disable "constant-logical-operand"
-#endif
-#endif /* -Wconstant-logical-operand */
-
-#if defined(__LCC__) && (__LCC__ <= 121)
-/* bug #2798 */
-#pragma diag_suppress alignment_reduction_ignored
-#elif defined(__ICC)
-#pragma warning(disable : 3453 1366)
-#elif __has_warning("-Walignment-reduction-ignored")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Walignment-reduction-ignored"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Walignment-reduction-ignored"
-#else
-#pragma warning disable "alignment-reduction-ignored"
-#endif
-#endif /* -Walignment-reduction-ignored */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(_WIN32) || defined(_WIN64)
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if !defined(_NO_CRT_STDIO_INLINE) && MDBX_BUILD_SHARED_LIBRARY && \
- !defined(xMDBX_TOOLS) && MDBX_WITHOUT_MSVC_CRT
-#define _NO_CRT_STDIO_INLINE
-#endif
-#elif !defined(_POSIX_C_SOURCE)
-#define _POSIX_C_SOURCE 200809L
-#endif /* Windows */
-
-/*----------------------------------------------------------------------------*/
-/* C99 includes */
-#include <inttypes.h>
-#include <stddef.h>
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <assert.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-/* C11' alignas() */
-#if __has_include(<stdalign.h>)
-#include <stdalign.h>
-#endif
-#if defined(alignas) || defined(__cplusplus)
-#define MDBX_ALIGNAS(N) alignas(N)
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
-#define MDBX_ALIGNAS(N) _Alignas(N)
-#elif defined(_MSC_VER)
-#define MDBX_ALIGNAS(N) __declspec(align(N))
-#elif __has_attribute(__aligned__) || defined(__GNUC__)
-#define MDBX_ALIGNAS(N) __attribute__((__aligned__(N)))
-#else
-#error "FIXME: Required alignas() or equivalent."
-#endif /* MDBX_ALIGNAS */
-
-/*----------------------------------------------------------------------------*/
-/* Systems includes */
-
-#ifdef __APPLE__
-#include <TargetConditionals.h>
-#endif /* Apple OSX & iOS */
-
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__)
-#include <sys/cdefs.h>
-#include <sys/mount.h>
-#include <sys/sysctl.h>
-#include <sys/types.h>
-#if defined(__FreeBSD__) || defined(__DragonFly__)
-#include <vm/vm_param.h>
-#elif defined(__OpenBSD__) || defined(__NetBSD__)
-#include <uvm/uvm_param.h>
-#else
-#define SYSCTL_LEGACY_NONCONST_MIB
-#endif
-#ifndef __MACH__
-#include <sys/vmmeter.h>
-#endif
-#else
-#include <malloc.h>
-#if !(defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(_WIN32) || defined(_WIN64))
-#include <mntent.h>
-#endif /* !Solaris */
-#endif /* !xBSD */
-
-#if defined(__FreeBSD__) || __has_include(<malloc_np.h>)
-#include <malloc_np.h>
-#endif
-
-#if defined(__APPLE__) || defined(__MACH__) || __has_include(<malloc/malloc.h>)
-#include <malloc/malloc.h>
-#endif /* MacOS */
-
-#if defined(__MACH__)
-#include <mach/host_info.h>
-#include <mach/mach_host.h>
-#include <mach/mach_port.h>
-#include <uuid/uuid.h>
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-#include <sched.h>
-#include <sys/sendfile.h>
-#include <sys/statfs.h>
-#endif /* Linux */
-
-#ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 0
-#endif
-
-#ifndef _XOPEN_SOURCE_EXTENDED
-#define _XOPEN_SOURCE_EXTENDED 0
-#else
-#include <utmpx.h>
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#include <kstat.h>
-#include <sys/mnttab.h>
-/* On Solaris, it's easier to add a missing prototype rather than find a
- * combination of #defines that break nothing. */
-__extern_C key_t ftok(const char *, int);
-#endif /* SunOS/Solaris */
-
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef _WIN32_WINNT
-#define _WIN32_WINNT 0x0601 /* Windows 7 */
-#elif _WIN32_WINNT < 0x0500
-#error At least 'Windows 2000' API is required for libmdbx.
-#endif /* _WIN32_WINNT */
-#if (defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* MinGW */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif /* WIN32_LEAN_AND_MEAN */
-#include <excpt.h>
-#include <tlhelp32.h>
-#include <windows.h>
-#include <winnt.h>
-#include <winternl.h>
-#define HAVE_SYS_STAT_H
-#define HAVE_SYS_TYPES_H
-typedef HANDLE mdbx_thread_t;
-typedef unsigned mdbx_thread_key_t;
-#define MAP_FAILED NULL
-#define HIGH_DWORD(v) ((DWORD)((sizeof(v) > 4) ? ((uint64_t)(v) >> 32) : 0))
-#define THREAD_CALL WINAPI
-#define THREAD_RESULT DWORD
-typedef struct {
- HANDLE mutex;
- HANDLE event[2];
-} mdbx_condpair_t;
-typedef CRITICAL_SECTION mdbx_fastmutex_t;
-
-#if !defined(_MSC_VER) && !defined(__try)
-/* *INDENT-OFF* */
-/* clang-format off */
-#define __try
-#define __except(COND) if(false)
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* stub for MSVC's __try/__except */
-
-#if MDBX_WITHOUT_MSVC_CRT
-
-#ifndef mdbx_malloc
-static inline void *mdbx_malloc(size_t bytes) {
- return HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_malloc */
-
-#ifndef mdbx_calloc
-static inline void *mdbx_calloc(size_t nelem, size_t size) {
- return HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, nelem * size);
-}
-#endif /* mdbx_calloc */
-
-#ifndef mdbx_realloc
-static inline void *mdbx_realloc(void *ptr, size_t bytes) {
- return ptr ? HeapReAlloc(GetProcessHeap(), 0, ptr, bytes)
- : HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_realloc */
-
-#ifndef mdbx_free
-static inline void mdbx_free(void *ptr) { HeapFree(GetProcessHeap(), 0, ptr); }
-#endif /* mdbx_free */
-
-#else /* MDBX_WITHOUT_MSVC_CRT */
-
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup _strdup
-
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-#ifndef snprintf
-#define snprintf _snprintf /* ntdll */
-#endif
-
-#ifndef vsnprintf
-#define vsnprintf _vsnprintf /* ntdll */
-#endif
-
-#else /*----------------------------------------------------------------------*/
-
-#include <unistd.h>
-#if !defined(_POSIX_MAPPED_FILES) || _POSIX_MAPPED_FILES < 1
-#error "libmdbx requires the _POSIX_MAPPED_FILES feature"
-#endif /* _POSIX_MAPPED_FILES */
-
-#include <pthread.h>
-#include <semaphore.h>
-#include <signal.h>
-#include <sys/file.h>
-#include <sys/ipc.h>
-#include <sys/mman.h>
-#include <sys/param.h>
-#include <sys/stat.h>
-#include <sys/statvfs.h>
-#include <sys/uio.h>
-typedef pthread_t mdbx_thread_t;
-typedef pthread_key_t mdbx_thread_key_t;
-#define INVALID_HANDLE_VALUE (-1)
-#define THREAD_CALL
-#define THREAD_RESULT void *
-typedef struct {
- pthread_mutex_t mutex;
- pthread_cond_t cond[2];
-} mdbx_condpair_t;
-typedef pthread_mutex_t mdbx_fastmutex_t;
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup strdup
-#endif /* Platform */
-
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
-/* malloc_usable_size() already provided */
-#elif defined(__APPLE__)
-#define malloc_usable_size(ptr) malloc_size(ptr)
-#elif defined(_MSC_VER) && !MDBX_WITHOUT_MSVC_CRT
-#define malloc_usable_size(ptr) _msize(ptr)
-#endif /* malloc_usable_size */
-
-#ifdef __ANDROID_API__
-#include <android/log.h>
-#if __ANDROID_API__ >= 21
-#include <sys/sendfile.h>
-#endif
-#endif /* Android */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(HAVE_SYS_STAT_H) || __has_include(<sys/stat.h>)
-#include <sys/stat.h>
-#endif
-#if defined(HAVE_SYS_TYPES_H) || __has_include(<sys/types.h>)
-#include <sys/types.h>
-#endif
-#if defined(HAVE_SYS_FILE_H) || __has_include(<sys/file.h>)
-#include <sys/file.h>
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-
-#ifndef SSIZE_MAX
-#define SSIZE_MAX INTPTR_MAX
-#endif
-
-#if !defined(MADV_DODUMP) && defined(MADV_CORE)
-#define MADV_DODUMP MADV_CORE
-#endif /* MADV_CORE -> MADV_DODUMP */
-
-#if !defined(MADV_DONTDUMP) && defined(MADV_NOCORE)
-#define MADV_DONTDUMP MADV_NOCORE
-#endif /* MADV_NOCORE -> MADV_DONTDUMP */
-
-#if defined(i386) || defined(__386) || defined(__i386) || defined(__i386__) || \
- defined(i486) || defined(__i486) || defined(__i486__) || \
- defined(i586) | defined(__i586) || defined(__i586__) || defined(i686) || \
- defined(__i686) || defined(__i686__) || defined(_M_IX86) || \
- defined(_X86_) || defined(__THW_INTEL__) || defined(__I86__) || \
- defined(__INTEL__) || defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64__) || defined(__amd64) || defined(_M_X64) || \
- defined(_M_AMD64) || defined(__IA32__) || defined(__INTEL__)
-#ifndef __ia32__
-/* LY: define neutral __ia32__ for x86 and x86-64 */
-#define __ia32__ 1
-#endif /* __ia32__ */
-#if !defined(__amd64__) && (defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64) || defined(_M_X64))
-/* LY: define trusty __amd64__ for all AMD64/x86-64 arch */
-#define __amd64__ 1
-#endif /* __amd64__ */
-#endif /* all x86 */
-
-#if (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
-#error \
- "Sanity checking failed: Two's complement, reasonably sized integer types"
-#endif
-
-#if UINTPTR_MAX > 0xffffFFFFul || ULONG_MAX > 0xffffFFFFul
-#define MDBX_WORDBITS 64
-#else
-#define MDBX_WORDBITS 32
-#endif /* MDBX_WORDBITS */
-
-#if defined(__ANDROID_API__) || defined(ANDROID)
-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "_FILE_OFFSET_BITS != MDBX_WORDBITS" (_FILE_OFFSET_BITS != MDBX_WORDBITS)
-#elif defined(__FILE_OFFSET_BITS) && __FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "__FILE_OFFSET_BITS != MDBX_WORDBITS" (__FILE_OFFSET_BITS != MDBX_WORDBITS)
-#endif
-#endif /* Android */
-
-/*----------------------------------------------------------------------------*/
-/* Compiler's includes for builtins/intrinsics */
-
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
-#include <intrin.h>
-#elif __GNUC_PREREQ(4, 4) || defined(__clang__)
-#if defined(__ia32__) || defined(__e2k__)
-#include <x86intrin.h>
-#endif /* __ia32__ */
-#if defined(__ia32__)
-#include <cpuid.h>
-#endif /* __ia32__ */
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
-#include <mbarrier.h>
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
-#include <machine/sys/inline.h>
-#elif defined(__IBMC__) && defined(__powerpc)
-#include <atomic.h>
-#elif defined(_AIX)
-#include <builtins.h>
-#include <sys/atomic_op.h>
-#elif (defined(__osf__) && defined(__DECC)) || defined(__alpha)
-#include <c_asm.h>
-#include <machine/builtins.h>
-#elif defined(__MWERKS__)
-/* CodeWarrior - troubles ? */
-#pragma gcc_extensions
-#elif defined(__SNC__)
-/* Sony PS3 - troubles ? */
-#elif defined(__hppa__) || defined(__hppa)
-#include <machine/inline.h>
-#else
-#error Unsupported C compiler, please use GNU C 4.4 or newer
-#endif /* Compiler */
-
-/*----------------------------------------------------------------------------*/
-/* Byteorder */
-
-#if !defined(__BYTE_ORDER__) || !defined(__ORDER_LITTLE_ENDIAN__) || \
- !defined(__ORDER_BIG_ENDIAN__)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(__GLIBC__) || defined(__GNU_LIBRARY__) || defined(__ANDROID_API__) || \
- defined(HAVE_ENDIAN_H) || __has_include(<endian.h>)
-#include <endian.h>
-#elif defined(__APPLE__) || defined(__MACH__) || defined(__OpenBSD__) || \
- defined(HAVE_MACHINE_ENDIAN_H) || __has_include(<machine/endian.h>)
-#include <machine/endian.h>
-#elif defined(HAVE_SYS_ISA_DEFS_H) || __has_include(<sys/isa_defs.h>)
-#include <sys/isa_defs.h>
-#elif (defined(HAVE_SYS_TYPES_H) && defined(HAVE_SYS_ENDIAN_H)) || \
- (__has_include(<sys/types.h>) && __has_include(<sys/endian.h>))
-#include <sys/endian.h>
-#include <sys/types.h>
-#elif defined(__bsdi__) || defined(__DragonFly__) || defined(__FreeBSD__) || \
- defined(__NetBSD__) || \
- defined(HAVE_SYS_PARAM_H) || __has_include(<sys/param.h>)
-#include <sys/param.h>
-#endif /* OS */
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-
-#else
-#error __BYTE_ORDER__ should be defined.
-#endif /* Arch */
-
-#endif
-#endif /* __BYTE_ORDER__ || __ORDER_LITTLE_ENDIAN__ || __ORDER_BIG_ENDIAN__ */
-
-/* Get the size of a memory page for the system.
- * This is the basic size that the platform's memory manager uses, and is
- * fundamental to the use of memory-mapped files. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline size_t
-mdbx_syspagesize(void) {
-#if defined(_WIN32) || defined(_WIN64)
- SYSTEM_INFO si;
- GetSystemInfo(&si);
- return si.dwPageSize;
-#else
- return sysconf(_SC_PAGE_SIZE);
-#endif
-}
-
-typedef struct mdbx_mmap_param {
- union {
- void *address;
- uint8_t *dxb;
- struct MDBX_lockinfo *lck;
- };
- mdbx_filehandle_t fd;
- size_t limit; /* mapping length, but NOT a size of file nor DB */
- size_t current; /* mapped region size, i.e. the size of file and DB */
- uint64_t filesize /* in-process cache of a file size */;
-#if defined(_WIN32) || defined(_WIN64)
- HANDLE section; /* memory-mapped section handle */
-#endif
-} mdbx_mmap_t;
-
-typedef union bin128 {
- __anonymous_struct_extension__ struct { uint64_t x, y; };
- __anonymous_struct_extension__ struct { uint32_t a, b, c, d; };
-} bin128_t;
-
-#if defined(_WIN32) || defined(_WIN64)
-typedef union MDBX_srwlock {
- __anonymous_struct_extension__ struct {
- long volatile readerCount;
- long volatile writerCount;
- };
- RTL_SRWLOCK native;
-} MDBX_srwlock;
-#endif /* Windows */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny);
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny);
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-#if defined(__cplusplus) && !defined(__STDC_NO_ATOMICS__) && (__has_include(<cstdatomic>) || __has_extension(cxx_atomic))
-#include <cstdatomic>
-#define MDBX_HAVE_C11ATOMICS
-#elif !defined(__cplusplus) && \
- (__STDC_VERSION__ >= 201112L || __has_extension(c_atomic)) && \
- !defined(__STDC_NO_ATOMICS__) && \
- (__GNUC_PREREQ(4, 9) || __CLANG_PREREQ(3, 8) || \
- !(defined(__GNUC__) || defined(__clang__)))
-#include <stdatomic.h>
-#define MDBX_HAVE_C11ATOMICS
-#elif defined(__GNUC__) || defined(__clang__)
-#elif defined(_MSC_VER)
-#pragma warning(disable : 4163) /* 'xyz': not available as an intrinsic */
-#pragma warning(disable : 4133) /* 'function': incompatible types - from \
- 'size_t' to 'LONGLONG' */
-#pragma warning(disable : 4244) /* 'return': conversion from 'LONGLONG' to \
- 'std::size_t', possible loss of data */
-#pragma warning(disable : 4267) /* 'function': conversion from 'size_t' to \
- 'long', possible loss of data */
-#pragma intrinsic(_InterlockedExchangeAdd, _InterlockedCompareExchange)
-#pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedCompareExchange64)
-#elif defined(__APPLE__)
-#include <libkern/OSAtomic.h>
-#else
-#error FIXME atomic-ops
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Memory/Compiler barriers, cache coherence */
-
-#if __has_include(<sys/cachectl.h>)
-#include <sys/cachectl.h>
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS should have explicit cache control */
-#include <sys/cachectl.h>
-#endif
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_compiler_barrier(void) {
-#if defined(__clang__) || defined(__GNUC__)
- __asm__ __volatile__("" ::: "memory");
-#elif defined(_MSC_VER)
- _ReadWriteBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
- __memory_barrier();
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __compiler_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_sched_fence(/* LY: no-arg meaning 'all expect ALU', e.g. 0x3D3D */);
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __fence();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_memory_barrier(void) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(memory_order_seq_cst);
-#elif defined(__ATOMIC_SEQ_CST)
-#ifdef __clang__
- __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
- __atomic_thread_fence(__ATOMIC_SEQ_CST);
-#endif
-#elif defined(__clang__) || defined(__GNUC__)
- __sync_synchronize();
-#elif defined(_WIN32) || defined(_WIN64)
- MemoryBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
-#if defined(__ia32__)
- _mm_mfence();
-#else
- __mf();
-#endif
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __machine_rw_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_mf();
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __lwsync();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* libc compatibility stuff */
-
-#if (!defined(__GLIBC__) && __GLIBC_PREREQ(2, 1)) && \
- (defined(_GNU_SOURCE) || defined(_BSD_SOURCE))
-#define mdbx_asprintf asprintf
-#define mdbx_vasprintf vasprintf
-#else
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC
- MDBX_PRINTF_ARGS(2, 3) int mdbx_asprintf(char **strp, const char *fmt, ...);
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt, va_list ap);
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* OS abstraction layer stuff */
-
-/* max bytes to write in one call */
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_WRITE UINT32_C(0x01000000)
-#else
-#define MAX_WRITE UINT32_C(0x3fff0000)
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-MDBX_INTERNAL_VAR uint32_t mdbx_linux_kernel_version;
-MDBX_INTERNAL_VAR bool mdbx_RunningOnWSL1 /* Windows Subsystem 1 for Linux */;
-#endif /* Linux */
-
-#ifndef mdbx_strdup
-LIBMDBX_API char *mdbx_strdup(const char *str);
-#endif
-
-MDBX_MAYBE_UNUSED static __inline int mdbx_get_errno(void) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD rc = GetLastError();
-#else
- int rc = errno;
-#endif
- return rc;
-}
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result);
-#endif
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr);
-#endif
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair, bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair);
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex);
-
-MDBX_INTERNAL_FUNC int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov,
- int iovcnt, uint64_t offset,
- size_t expected_written);
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t count,
- uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t count, uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t count);
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg);
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread);
-
-enum mdbx_syncmode_bits {
- MDBX_SYNC_NONE = 0,
- MDBX_SYNC_DATA = 1,
- MDBX_SYNC_SIZE = 2,
- MDBX_SYNC_IODQ = 4
-};
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- const enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length);
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos);
-MDBX_INTERNAL_FUNC int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length);
-
-enum mdbx_openfile_purpose {
- MDBX_OPEN_DXB_READ = 0,
- MDBX_OPEN_DXB_LAZY = 1,
- MDBX_OPEN_DXB_DSYNC = 2,
- MDBX_OPEN_LCK = 3,
- MDBX_OPEN_COPY = 4,
- MDBX_OPEN_DELETE = 5
-};
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait);
-
-#define MMAP_OPTION_TRUNCATE 1
-#define MMAP_OPTION_SEMAPHORE 2
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t must, const size_t limit,
- const unsigned options);
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map);
-#define MDBX_MRESIZE_MAY_MOVE 0x00000100
-#define MDBX_MRESIZE_MAY_UNMAP 0x00000200
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit);
-#if defined(_WIN32) || defined(_WIN64)
-typedef struct {
- unsigned limit, count;
- HANDLE handles[31];
-} mdbx_handle_array_t;
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array);
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array);
-#endif /* Windows */
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err);
-
-MDBX_MAYBE_UNUSED static __inline uint32_t mdbx_getpid(void) {
- STATIC_ASSERT(sizeof(mdbx_pid_t) <= sizeof(uint32_t));
-#if defined(_WIN32) || defined(_WIN64)
- return GetCurrentProcessId();
-#else
- STATIC_ASSERT(sizeof(pid_t) <= sizeof(uint32_t));
- return getpid();
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline uintptr_t mdbx_thread_self(void) {
- mdbx_tid_t thunk;
- STATIC_ASSERT(sizeof(uintptr_t) >= sizeof(thunk));
-#if defined(_WIN32) || defined(_WIN64)
- thunk = GetCurrentThreadId();
-#else
- thunk = pthread_self();
-#endif
- return (uintptr_t)thunk;
-}
-
-#if !defined(_WIN32) && !defined(_WIN64)
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void);
-#else
-static __inline int mdbx_check_tid4bionic(void) { return 0; }
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-MDBX_MAYBE_UNUSED static __inline int
-mdbx_pthread_mutex_lock(pthread_mutex_t *mutex) {
- int err = mdbx_check_tid4bionic();
- return unlikely(err) ? err : pthread_mutex_lock(mutex);
-}
-#endif /* !Windows */
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void);
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16);
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime);
-
-MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void);
-/*----------------------------------------------------------------------------*/
-/* lck stuff */
-
-/// \brief Initialization of synchronization primitives linked with MDBX_env
-/// instance both in LCK-file and within the current process.
-/// \param
-/// global_uniqueness_flag = true - denotes that there are no other processes
-/// working with DB and LCK-file. Thus the function MUST initialize
-/// shared synchronization objects in memory-mapped LCK-file.
-/// global_uniqueness_flag = false - denotes that at least one process is
-/// already working with DB and LCK-file, including the case when DB
-/// has already been opened in the current process. Thus the function
-/// MUST NOT initialize shared synchronization objects in memory-mapped
-/// LCK-file that are already in use.
-/// \return Error code or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag);
-
-/// \brief Disconnects from shared interprocess objects and destructs
-/// synchronization objects linked with MDBX_env instance
-/// within the current process.
-/// \param
-/// inprocess_neighbor = NULL - if the current process does not have other
-/// instances of MDBX_env linked with the DB being closed.
-/// Thus the function MUST check for other processes working with DB or
-/// LCK-file, and keep or destroy shared synchronization objects in
-/// memory-mapped LCK-file depending on the result.
-/// inprocess_neighbor = not-NULL - pointer to another instance of MDBX_env
-/// (anyone of there is several) working with DB or LCK-file within the
-/// current process. Thus the function MUST NOT try to acquire exclusive
-/// lock and/or try to destruct shared synchronization objects linked with
-/// DB or LCK-file. Moreover, the implementation MUST ensure correct work
-/// of other instances of MDBX_env within the current process, e.g.
-/// restore POSIX-fcntl locks after the closing of file descriptors.
-/// \return Error code (MDBX_PANIC) or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor);
-
-/// \brief Connects to shared interprocess locking objects and tries to acquire
-/// the maximum lock level (shared if exclusive is not available)
-/// Depending on implementation or/and platform (Windows) this function may
-/// acquire the non-OS super-level lock (e.g. for shared synchronization
-/// objects initialization), which will be downgraded to OS-exclusive or
-/// shared via explicit calling of mdbx_lck_downgrade().
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if an exclusive lock was acquired and thus
-/// the current process is the first and only after the last use of DB.
-/// MDBX_RESULT_FALSE (0) - if a shared lock was acquired and thus
-/// DB has already been opened and now is used by other processes.
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env);
-
-/// \brief Downgrades the level of initially acquired lock to
-/// operational level specified by argument. The reson for such downgrade:
-/// - unblocking of other processes that are waiting for access, i.e.
-/// if (env->me_flags & MDBX_EXCLUSIVE) != 0, then other processes
-/// should be made aware that access is unavailable rather than
-/// wait for it.
-/// - freeing locks that interfere file operation (especially for Windows)
-/// (env->me_flags & MDBX_EXCLUSIVE) == 0 - downgrade to shared lock.
-/// (env->me_flags & MDBX_EXCLUSIVE) != 0 - downgrade to exclusive
-/// operational lock.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env);
-
-/// \brief Locks LCK-file or/and table of readers for (de)registering.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env);
-
-/// \brief Unlocks LCK-file or/and table of readers after (de)registering.
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env);
-
-/// \brief Acquires lock for DB change (on writing transaction start)
-/// Reading transactions will not be blocked.
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-/// \return Error code or zero on success
-LIBMDBX_API int mdbx_txn_lock(MDBX_env *env, bool dont_wait);
-
-/// \brief Releases lock once DB changes is made (after writing transaction
-/// has finished).
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-LIBMDBX_API void mdbx_txn_unlock(MDBX_env *env);
-
-/// \brief Sets alive-flag of reader presence (indicative lock) for PID of
-/// the current process. The function does no more than needed for
-/// the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env);
-
-/// \brief Resets alive-flag of reader presence (indicative lock)
-/// for PID of the current process. The function does no more than needed
-/// for the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env);
-
-/// \brief Checks for reading process status with the given pid with help of
-/// alive-flag of presence (indicative lock) or using another way.
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if the reader process with the given PID is alive
-/// and working with DB (indicative lock is present).
-/// MDBX_RESULT_FALSE (0) - if the reader process with the given PID is absent
-/// or not working with DB (indicative lock is not present).
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid);
-
-#if defined(_WIN32) || defined(_WIN64)
-
-typedef void(WINAPI *MDBX_srwlock_function)(MDBX_srwlock *);
-MDBX_INTERNAL_VAR MDBX_srwlock_function mdbx_srwlock_Init,
- mdbx_srwlock_AcquireShared, mdbx_srwlock_ReleaseShared,
- mdbx_srwlock_AcquireExclusive, mdbx_srwlock_ReleaseExclusive;
-
-#if _WIN32_WINNT < 0x0600 /* prior to Windows Vista */
-typedef enum _FILE_INFO_BY_HANDLE_CLASS {
- FileBasicInfo,
- FileStandardInfo,
- FileNameInfo,
- FileRenameInfo,
- FileDispositionInfo,
- FileAllocationInfo,
- FileEndOfFileInfo,
- FileStreamInfo,
- FileCompressionInfo,
- FileAttributeTagInfo,
- FileIdBothDirectoryInfo,
- FileIdBothDirectoryRestartInfo,
- FileIoPriorityHintInfo,
- FileRemoteProtocolInfo,
- MaximumFileInfoByHandleClass
-} FILE_INFO_BY_HANDLE_CLASS,
- *PFILE_INFO_BY_HANDLE_CLASS;
-
-typedef struct _FILE_END_OF_FILE_INFO {
- LARGE_INTEGER EndOfFile;
-} FILE_END_OF_FILE_INFO, *PFILE_END_OF_FILE_INFO;
-
-#define REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK 0x00000001
-#define REMOTE_PROTOCOL_INFO_FLAG_OFFLINE 0x00000002
-
-typedef struct _FILE_REMOTE_PROTOCOL_INFO {
- USHORT StructureVersion;
- USHORT StructureSize;
- DWORD Protocol;
- USHORT ProtocolMajorVersion;
- USHORT ProtocolMinorVersion;
- USHORT ProtocolRevision;
- USHORT Reserved;
- DWORD Flags;
- struct {
- DWORD Reserved[8];
- } GenericReserved;
- struct {
- DWORD Reserved[16];
- } ProtocolSpecificReserved;
-} FILE_REMOTE_PROTOCOL_INFO, *PFILE_REMOTE_PROTOCOL_INFO;
-
-#endif /* _WIN32_WINNT < 0x0600 (prior to Windows Vista) */
-
-typedef BOOL(WINAPI *MDBX_GetFileInformationByHandleEx)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_GetFileInformationByHandleEx
- mdbx_GetFileInformationByHandleEx;
-
-typedef BOOL(WINAPI *MDBX_GetVolumeInformationByHandleW)(
- _In_ HANDLE hFile, _Out_opt_ LPWSTR lpVolumeNameBuffer,
- _In_ DWORD nVolumeNameSize, _Out_opt_ LPDWORD lpVolumeSerialNumber,
- _Out_opt_ LPDWORD lpMaximumComponentLength,
- _Out_opt_ LPDWORD lpFileSystemFlags,
- _Out_opt_ LPWSTR lpFileSystemNameBuffer, _In_ DWORD nFileSystemNameSize);
-MDBX_INTERNAL_VAR MDBX_GetVolumeInformationByHandleW
- mdbx_GetVolumeInformationByHandleW;
-
-typedef DWORD(WINAPI *MDBX_GetFinalPathNameByHandleW)(_In_ HANDLE hFile,
- _Out_ LPWSTR lpszFilePath,
- _In_ DWORD cchFilePath,
- _In_ DWORD dwFlags);
-MDBX_INTERNAL_VAR MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-
-typedef BOOL(WINAPI *MDBX_SetFileInformationByHandle)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_SetFileInformationByHandle
- mdbx_SetFileInformationByHandle;
-
-typedef NTSTATUS(NTAPI *MDBX_NtFsControlFile)(
- IN HANDLE FileHandle, IN OUT HANDLE Event,
- IN OUT PVOID /* PIO_APC_ROUTINE */ ApcRoutine, IN OUT PVOID ApcContext,
- OUT PIO_STATUS_BLOCK IoStatusBlock, IN ULONG FsControlCode,
- IN OUT PVOID InputBuffer, IN ULONG InputBufferLength,
- OUT OPTIONAL PVOID OutputBuffer, IN ULONG OutputBufferLength);
-MDBX_INTERNAL_VAR MDBX_NtFsControlFile mdbx_NtFsControlFile;
-
-typedef uint64_t(WINAPI *MDBX_GetTickCount64)(void);
-MDBX_INTERNAL_VAR MDBX_GetTickCount64 mdbx_GetTickCount64;
-
-#if !defined(_WIN32_WINNT_WIN8) || _WIN32_WINNT < _WIN32_WINNT_WIN8
-typedef struct _WIN32_MEMORY_RANGE_ENTRY {
- PVOID VirtualAddress;
- SIZE_T NumberOfBytes;
-} WIN32_MEMORY_RANGE_ENTRY, *PWIN32_MEMORY_RANGE_ENTRY;
-#endif /* Windows 8.x */
-
-typedef BOOL(WINAPI *MDBX_PrefetchVirtualMemory)(
- HANDLE hProcess, ULONG_PTR NumberOfEntries,
- PWIN32_MEMORY_RANGE_ENTRY VirtualAddresses, ULONG Flags);
-MDBX_INTERNAL_VAR MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-
-typedef enum _SECTION_INHERIT { ViewShare = 1, ViewUnmap = 2 } SECTION_INHERIT;
-
-typedef NTSTATUS(NTAPI *MDBX_NtExtendSection)(IN HANDLE SectionHandle,
- IN PLARGE_INTEGER NewSectionSize);
-MDBX_INTERNAL_VAR MDBX_NtExtendSection mdbx_NtExtendSection;
-
-static __inline bool mdbx_RunningUnderWine(void) {
- return !mdbx_NtExtendSection;
-}
-
-typedef LSTATUS(WINAPI *MDBX_RegGetValueA)(HKEY hkey, LPCSTR lpSubKey,
- LPCSTR lpValue, DWORD dwFlags,
- LPDWORD pdwType, PVOID pvData,
- LPDWORD pcbData);
-MDBX_INTERNAL_VAR MDBX_RegGetValueA mdbx_RegGetValueA;
-
-NTSYSAPI ULONG RtlRandomEx(PULONG Seed);
-
-#endif /* Windows */
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_MSC_VER) && _MSC_VER >= 1900
-/* LY: MSVC 2015/2017/2019 has buggy/inconsistent PRIuPTR/PRIxPTR macros
- * for internal format-args checker. */
-#undef PRIuPTR
-#undef PRIiPTR
-#undef PRIdPTR
-#undef PRIxPTR
-#define PRIuPTR "Iu"
-#define PRIiPTR "Ii"
-#define PRIdPTR "Id"
-#define PRIxPTR "Ix"
-#define PRIuSIZE "zu"
-#define PRIiSIZE "zi"
-#define PRIdSIZE "zd"
-#define PRIxSIZE "zx"
-#endif /* fix PRI*PTR for _MSC_VER */
-
-#ifndef PRIuSIZE
-#define PRIuSIZE PRIuPTR
-#define PRIiSIZE PRIiPTR
-#define PRIdSIZE PRIdPTR
-#define PRIxSIZE PRIxPTR
-#endif /* PRI*SIZE macros for MSVC */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-#define mdbx_sourcery_anchor XCONCAT(mdbx_sourcery_, MDBX_BUILD_SOURCERY)
-#if defined(xMDBX_TOOLS)
-extern LIBMDBX_API const char *const mdbx_sourcery_anchor;
-#endif
-
-/*******************************************************************************
- *******************************************************************************
- *******************************************************************************
- *
- *
- * #### ##### ##### # #### # # ####
- * # # # # # # # # ## # #
- * # # # # # # # # # # # ####
- * # # ##### # # # # # # # #
- * # # # # # # # # ## # #
- * #### # # # #### # # ####
- *
- *
- */
-
-/** \defgroup build_option Build options
- * The libmdbx build options.
- @{ */
-
-/** Using fcntl(F_FULLFSYNC) with 5-10 times slowdown */
-#define MDBX_OSX_WANNA_DURABILITY 0
-/** Using fsync() with chance of data lost on power failure */
-#define MDBX_OSX_WANNA_SPEED 1
-
-#ifndef MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-/** Choices \ref MDBX_OSX_WANNA_DURABILITY or \ref MDBX_OSX_WANNA_SPEED
- * for OSX & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY MDBX_OSX_WANNA_DURABILITY
-#endif /* MDBX_OSX_SPEED_INSTEADOF_DURABILITY */
-
-/** Controls checking PID against reuse DB environment after the fork() */
-#ifndef MDBX_ENV_CHECKPID
-#if defined(MADV_DONTFORK) || defined(_WIN32) || defined(_WIN64)
-/* PID check could be omitted:
- * - on Linux when madvise(MADV_DONTFORK) is available, i.e. after the fork()
- * mapped pages will not be available for child process.
- * - in Windows where fork() not available. */
-#define MDBX_ENV_CHECKPID 0
-#else
-#define MDBX_ENV_CHECKPID 1
-#endif
-#define MDBX_ENV_CHECKPID_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#else
-#define MDBX_ENV_CHECKPID_CONFIG MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#endif /* MDBX_ENV_CHECKPID */
-
-/** Controls checking transaction owner thread against misuse transactions from
- * other threads. */
-#ifndef MDBX_TXN_CHECKOWNER
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_TXN_CHECKOWNER_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#else
-#define MDBX_TXN_CHECKOWNER_CONFIG MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#endif /* MDBX_TXN_CHECKOWNER */
-
-/** Does a system have battery-backed Real-Time Clock or just a fake. */
-#ifndef MDBX_TRUST_RTC
-#if defined(__linux__) || defined(__gnu_linux__) || defined(__NetBSD__) || \
- defined(__OpenBSD__)
-#define MDBX_TRUST_RTC 0 /* a lot of embedded systems have a fake RTC */
-#else
-#define MDBX_TRUST_RTC 1
-#endif
-#define MDBX_TRUST_RTC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#else
-#define MDBX_TRUST_RTC_CONFIG MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#endif /* MDBX_TRUST_RTC */
-
-/** Controls online database auto-compactification during write-transactions. */
-#ifndef MDBX_ENABLE_REFUND
-#define MDBX_ENABLE_REFUND 1
-#elif !(MDBX_ENABLE_REFUND == 0 || MDBX_ENABLE_REFUND == 1)
-#error MDBX_ENABLE_REFUND must be defined as 0 or 1
-#endif /* MDBX_ENABLE_REFUND */
-
-/** Controls gathering statistics for page operations. */
-#ifndef MDBX_ENABLE_PGOP_STAT
-#define MDBX_ENABLE_PGOP_STAT 1
-#elif !(MDBX_ENABLE_PGOP_STAT == 0 || MDBX_ENABLE_PGOP_STAT == 1)
-#error MDBX_ENABLE_PGOP_STAT must be defined as 0 or 1
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-/** Controls use of POSIX madvise() hints and friends. */
-#ifndef MDBX_ENABLE_MADVISE
-#define MDBX_ENABLE_MADVISE 1
-#elif !(MDBX_ENABLE_MADVISE == 0 || MDBX_ENABLE_MADVISE == 1)
-#error MDBX_ENABLE_MADVISE must be defined as 0 or 1
-#endif /* MDBX_ENABLE_MADVISE */
-
-/** Disable some checks to reduce an overhead and detection probability of
- * database corruption to a values closer to the LMDB. */
-#ifndef MDBX_DISABLE_PAGECHECKS
-#define MDBX_DISABLE_PAGECHECKS 0
-#elif !(MDBX_DISABLE_PAGECHECKS == 0 || MDBX_DISABLE_PAGECHECKS == 1)
-#error MDBX_DISABLE_PAGECHECKS must be defined as 0 or 1
-#endif /* MDBX_DISABLE_PAGECHECKS */
-
-#ifndef MDBX_PNL_PREALLOC_FOR_RADIXSORT
-#define MDBX_PNL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_PNL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_PNL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_PNL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
-
-#ifndef MDBX_DPL_PREALLOC_FOR_RADIXSORT
-#define MDBX_DPL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_DPL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_DPL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_DPL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
-
-/* Basically, this build-option is for TODO. Guess it should be replaced
- * with MDBX_ENABLE_WRITEMAP_SPILLING with the three variants:
- * 0/OFF = Don't track dirty pages at all and don't spilling ones.
- * This should be by-default on Linux and may-be other systems
- * (not sure: Darwin/OSX, FreeBSD, Windows 10) where kernel provides
- * properly LRU tracking and async writing on-demand.
- * 1/ON = Lite tracking of dirty pages but with LRU labels and explicit
- * spilling with msync(MS_ASYNC). */
-#ifndef MDBX_FAKE_SPILL_WRITEMAP
-#if defined(__linux__) || defined(__gnu_linux__)
-#define MDBX_FAKE_SPILL_WRITEMAP 1 /* msync(MS_ASYNC) is no-op on Linux */
-#else
-#define MDBX_FAKE_SPILL_WRITEMAP 0
-#endif
-#elif !(MDBX_FAKE_SPILL_WRITEMAP == 0 || MDBX_FAKE_SPILL_WRITEMAP == 1)
-#error MDBX_FAKE_SPILL_WRITEMAP must be defined as 0 or 1
-#endif /* MDBX_FAKE_SPILL_WRITEMAP */
-
-/** Controls sort order of internal page number lists.
- * This mostly experimental/advanced option with not for regular MDBX users.
- * \warning The database format depend on this option and libmdbx builded with
- * different option value are incompatible. */
-#ifndef MDBX_PNL_ASCENDING
-#define MDBX_PNL_ASCENDING 0
-#elif !(MDBX_PNL_ASCENDING == 0 || MDBX_PNL_ASCENDING == 1)
-#error MDBX_PNL_ASCENDING must be defined as 0 or 1
-#endif /* MDBX_PNL_ASCENDING */
-
-/** Avoid dependence from MSVC CRT and use ntdll.dll instead. */
-#ifndef MDBX_WITHOUT_MSVC_CRT
-#define MDBX_WITHOUT_MSVC_CRT 1
-#elif !(MDBX_WITHOUT_MSVC_CRT == 0 || MDBX_WITHOUT_MSVC_CRT == 1)
-#error MDBX_WITHOUT_MSVC_CRT must be defined as 0 or 1
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-/** Size of buffer used during copying a environment/database file. */
-#ifndef MDBX_ENVCOPY_WRITEBUF
-#define MDBX_ENVCOPY_WRITEBUF 1048576u
-#elif MDBX_ENVCOPY_WRITEBUF < 65536u || MDBX_ENVCOPY_WRITEBUF > 1073741824u || \
- MDBX_ENVCOPY_WRITEBUF % 65536u
-#error MDBX_ENVCOPY_WRITEBUF must be defined in range 65536..1073741824 and be multiple of 65536
-#endif /* MDBX_ENVCOPY_WRITEBUF */
-
-/** Forces assertion checking */
-#ifndef MDBX_FORCE_ASSERTIONS
-#define MDBX_FORCE_ASSERTIONS 0
-#elif !(MDBX_FORCE_ASSERTIONS == 0 || MDBX_FORCE_ASSERTIONS == 1)
-#error MDBX_FORCE_ASSERTIONS must be defined as 0 or 1
-#endif /* MDBX_FORCE_ASSERTIONS */
-
-/** Presumed malloc size overhead for each allocation
- * to adjust allocations to be more aligned. */
-#ifndef MDBX_ASSUME_MALLOC_OVERHEAD
-#ifdef __SIZEOF_POINTER__
-#define MDBX_ASSUME_MALLOC_OVERHEAD (__SIZEOF_POINTER__ * 2u)
-#else
-#define MDBX_ASSUME_MALLOC_OVERHEAD (sizeof(void *) * 2u)
-#endif
-#elif MDBX_ASSUME_MALLOC_OVERHEAD < 0 || MDBX_ASSUME_MALLOC_OVERHEAD > 64 || \
- MDBX_ASSUME_MALLOC_OVERHEAD % 4
-#error MDBX_ASSUME_MALLOC_OVERHEAD must be defined in range 0..64 and be multiple of 4
-#endif /* MDBX_ASSUME_MALLOC_OVERHEAD */
-
-/** In case the MDBX_DEBUG is undefined set it corresponding to NDEBUG */
-#ifndef MDBX_DEBUG
-#ifdef NDEBUG
-#define MDBX_DEBUG 0
-#else
-#define MDBX_DEBUG 1
-#endif
-#endif /* MDBX_DEBUG */
-
-/** If defined then enables integration with Valgrind,
- * a memory analyzing tool. */
-#ifndef MDBX_USE_VALGRIND
-#endif /* MDBX_USE_VALGRIND */
-
-/** If defined then enables use C11 atomics,
- * otherwise detects ones availability automatically. */
-#ifndef MDBX_HAVE_C11ATOMICS
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-//------------------------------------------------------------------------------
-
-/** Win32 File Locking API for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_WIN32FILES -1
-
-/** SystemV IPC semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_SYSV 5
-
-/** POSIX-1 Shared anonymous semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX1988 1988
-
-/** POSIX-2001 Shared Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2001 2001
-
-/** POSIX-2008 Robust Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2008 2008
-
-/** BeOS Benaphores, aka Futexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_BENAPHORE 1995
-
-/** Advanced: Choices the locking implementation (autodetection by default). */
-#if defined(_WIN32) || defined(_WIN64)
-#define MDBX_LOCKING MDBX_LOCKING_WIN32FILES
-#else
-#ifndef MDBX_LOCKING
-#if defined(_POSIX_THREAD_PROCESS_SHARED) && \
- _POSIX_THREAD_PROCESS_SHARED >= 200112L && !defined(__FreeBSD__)
-
-/* Some platforms define the EOWNERDEAD error code even though they
- * don't support Robust Mutexes. If doubt compile with -MDBX_LOCKING=2001. */
-#if defined(EOWNERDEAD) && _POSIX_THREAD_PROCESS_SHARED >= 200809L && \
- ((defined(_POSIX_THREAD_ROBUST_PRIO_INHERIT) && \
- _POSIX_THREAD_ROBUST_PRIO_INHERIT > 0) || \
- (defined(_POSIX_THREAD_ROBUST_PRIO_PROTECT) && \
- _POSIX_THREAD_ROBUST_PRIO_PROTECT > 0) || \
- defined(PTHREAD_MUTEX_ROBUST) || defined(PTHREAD_MUTEX_ROBUST_NP)) && \
- (!defined(__GLIBC__) || \
- __GLIBC_PREREQ(2, 10) /* troubles with Robust mutexes before 2.10 */)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2008
-#else
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2001
-#endif
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX1988
-#else
-#define MDBX_LOCKING MDBX_LOCKING_SYSV
-#endif
-#define MDBX_LOCKING_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_LOCKING)
-#else
-#define MDBX_LOCKING_CONFIG MDBX_STRINGIFY(MDBX_LOCKING)
-#endif /* MDBX_LOCKING */
-#endif /* !Windows */
-
-/** Advanced: Using POSIX OFD-locks (autodetection by default). */
-#ifndef MDBX_USE_OFDLOCKS
-#if defined(F_OFD_SETLK) && defined(F_OFD_SETLKW) && defined(F_OFD_GETLK) && \
- !defined(MDBX_SAFE4QEMU) && \
- !defined(__sun) /* OFD-lock are broken on Solaris */
-#define MDBX_USE_OFDLOCKS 1
-#else
-#define MDBX_USE_OFDLOCKS 0
-#endif
-#define MDBX_USE_OFDLOCKS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#else
-#define MDBX_USE_OFDLOCKS_CONFIG MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#endif /* MDBX_USE_OFDLOCKS */
-
-/** Advanced: Using sendfile() syscall (autodetection by default). */
-#ifndef MDBX_USE_SENDFILE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 21)
-#define MDBX_USE_SENDFILE 1
-#else
-#define MDBX_USE_SENDFILE 0
-#endif
-#endif /* MDBX_USE_SENDFILE */
-
-/** Advanced: Using copy_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_COPYFILERANGE
-#if __GLIBC_PREREQ(2, 27) && defined(_GNU_SOURCE)
-#define MDBX_USE_COPYFILERANGE 1
-#else
-#define MDBX_USE_COPYFILERANGE 0
-#endif
-#endif /* MDBX_USE_COPYFILERANGE */
-
-/** Advanced: Using sync_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_SYNCFILERANGE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- defined(SYNC_FILE_RANGE_WRITE) && !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 26)
-#define MDBX_USE_SYNCFILERANGE 1
-#else
-#define MDBX_USE_SYNCFILERANGE 0
-#endif
-#endif /* MDBX_USE_SYNCFILERANGE */
-
-//------------------------------------------------------------------------------
-
-#ifndef MDBX_CPU_WRITEBACK_INCOHERENT
-#if defined(__ia32__) || defined(__e2k__) || defined(__hppa) || \
- defined(__hppa__) || defined(DOXYGEN)
-#define MDBX_CPU_WRITEBACK_INCOHERENT 0
-#else
-#define MDBX_CPU_WRITEBACK_INCOHERENT 1
-#endif
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-#ifndef MDBX_MMAP_INCOHERENT_FILE_WRITE
-#ifdef __OpenBSD__
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 1
-#else
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#ifndef MDBX_MMAP_INCOHERENT_CPU_CACHE
-#if defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS has cache coherency issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 1
-#else
-/* LY: assume no relevant mmap/dcache issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#ifndef MDBX_64BIT_ATOMIC
-#if MDBX_WORDBITS >= 64 || defined(DOXYGEN)
-#define MDBX_64BIT_ATOMIC 1
-#else
-#define MDBX_64BIT_ATOMIC 0
-#endif
-#define MDBX_64BIT_ATOMIC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#else
-#define MDBX_64BIT_ATOMIC_CONFIG MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#endif /* MDBX_64BIT_ATOMIC */
-
-#ifndef MDBX_64BIT_CAS
-#if defined(ATOMIC_LLONG_LOCK_FREE)
-#if ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__GCC_ATOMIC_LLONG_LOCK_FREE)
-#if __GCC_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__CLANG_ATOMIC_LLONG_LOCK_FREE)
-#if __CLANG_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(_MSC_VER) || defined(__APPLE__) || defined(DOXYGEN)
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS MDBX_64BIT_ATOMIC
-#endif
-#define MDBX_64BIT_CAS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#else
-#define MDBX_64BIT_CAS_CONFIG MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#endif /* MDBX_64BIT_CAS */
-
-#ifndef MDBX_UNALIGNED_OK
-#if defined(__ALIGNED__) || defined(__SANITIZE_UNDEFINED__)
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#elif defined(__ARM_FEATURE_UNALIGNED)
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#elif __CLANG_PREREQ(5, 0) || __GNUC_PREREQ(5, 0)
-/* expecting an optimization will well done, also this
- * hushes false-positives from UBSAN (undefined behaviour sanitizer) */
-#define MDBX_UNALIGNED_OK 0
-#elif defined(__e2k__) || defined(__elbrus__)
-#if __iset__ > 4
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#endif
-#elif defined(__ia32__)
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#endif
-#elif MDBX_UNALIGNED_OK == 1
-#undef MDBX_UNALIGNED_OK
-#define MDBX_UNALIGNED_OK 32 /* any unaligned access allowed */
-#endif /* MDBX_UNALIGNED_OK */
-
-#ifndef MDBX_CACHELINE_SIZE
-#if defined(SYSTEM_CACHE_ALIGNMENT_SIZE)
-#define MDBX_CACHELINE_SIZE SYSTEM_CACHE_ALIGNMENT_SIZE
-#elif defined(__ia64__) || defined(__ia64) || defined(_M_IA64)
-#define MDBX_CACHELINE_SIZE 128
-#else
-#define MDBX_CACHELINE_SIZE 64
-#endif
-#endif /* MDBX_CACHELINE_SIZE */
-
-/** @} end of build options */
-/*******************************************************************************
- *******************************************************************************
- ******************************************************************************/
-
-#ifdef DOXYGEN
-/* !!! Actually this is a fake definitions !!!
- * !!! for documentation generation by Doxygen !!! */
-
-/** Controls enabling of debugging features.
- *
- * - `MDBX_DEBUG = 0` (by default) Disables any debugging features at all,
- * including logging and assertion controls.
- * Logging level and corresponding debug flags changing
- * by \ref mdbx_setup_debug() will not have effect.
- * - `MDBX_DEBUG > 0` Enables code for the debugging features (logging,
- * assertions checking and internal audit).
- * Simultaneously sets the default logging level
- * to the `MDBX_DEBUG` value.
- * Also enables \ref MDBX_DBG_AUDIT if `MDBX_DEBUG >= 2`.
- *
- * \ingroup build_option */
-#define MDBX_DEBUG 0...7
-
-/** Disables using of GNU libc extensions. */
-#define MDBX_DISABLE_GNU_SOURCE 0 or 1
-
-#endif /* DOXYGEN */
-
-/* Undefine the NDEBUG if debugging is enforced by MDBX_DEBUG */
-#if MDBX_DEBUG
-#undef NDEBUG
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-enum MDBX_memory_order {
- mo_Relaxed,
- mo_AcquireRelease,
- mo_SequentialConsistency
-};
-
-typedef union {
- volatile uint32_t weak;
-#ifdef MDBX_HAVE_C11ATOMICS
- volatile _Atomic uint32_t c11a;
-#endif /* MDBX_HAVE_C11ATOMICS */
-} MDBX_atomic_uint32_t;
-
-typedef union {
- volatile uint64_t weak;
-#if defined(MDBX_HAVE_C11ATOMICS) && (MDBX_64BIT_CAS || MDBX_64BIT_ATOMIC)
- volatile _Atomic uint64_t c11a;
-#endif
-#if !defined(MDBX_HAVE_C11ATOMICS) || !MDBX_64BIT_CAS || !MDBX_64BIT_ATOMIC
- __anonymous_struct_extension__ struct {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- MDBX_atomic_uint32_t low, high;
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- MDBX_atomic_uint32_t high, low;
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- };
-#endif
-} MDBX_atomic_uint64_t;
-
-#ifdef MDBX_HAVE_C11ATOMICS
-
-/* Crutches for C11 atomic compiler's bugs */
-#if defined(__e2k__) && defined(__LCC__) && __LCC__ < /* FIXME */ 127
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->weak)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->weak)
-#elif defined(__clang__) && __clang__ < 8
-#define MDBX_c11a_ro(type, ptr) ((volatile _Atomic(type) *)&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#else
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#endif /* Crutches for C11 atomic compiler's bugs */
-
-static __always_inline memory_order mo_c11_store(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_release;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-
-static __always_inline memory_order mo_c11_load(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_acquire;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED static __always_inline void
-mdbx_memory_fence(enum MDBX_memory_order order, bool write) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(write ? mo_c11_store(order) : mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_compiler_barrier();
- if (write &&
- order > (MDBX_CPU_WRITEBACK_INCOHERENT ? mo_Relaxed : mo_AcquireRelease))
- mdbx_memory_barrier();
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_store32(MDBX_atomic_uint32_t *p, const uint32_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint32_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_load32(const MDBX_atomic_uint32_t *p, enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint32_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint32_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint32_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-/* Basic constants and types */
-
-/* A stamp that identifies a file as an MDBX file.
- * There's nothing special about this value other than that it is easily
- * recognizable, and it will reflect any byte order mismatches. */
-#define MDBX_MAGIC UINT64_C(/* 56-bit prime */ 0x59659DBDEF4C11)
-
-/* FROZEN: The version number for a database's datafile format. */
-#define MDBX_DATA_VERSION 3
-/* The version number for a database's lockfile format. */
-#define MDBX_LOCK_VERSION 4
-
-/* handle for the DB used to track free pages. */
-#define FREE_DBI 0
-/* handle for the default DB. */
-#define MAIN_DBI 1
-/* Number of DBs in metapage (free and main) - also hardcoded elsewhere */
-#define CORE_DBS 2
-
-/* Number of meta pages - also hardcoded elsewhere */
-#define NUM_METAS 3
-
-/* A page number in the database.
- *
- * MDBX uses 32 bit for page numbers. This limits database
- * size up to 2^44 bytes, in case of 4K pages. */
-typedef uint32_t pgno_t;
-typedef MDBX_atomic_uint32_t atomic_pgno_t;
-#define PRIaPGNO PRIu32
-#define MAX_PAGENO UINT32_C(0x7FFFffff)
-#define MIN_PAGENO NUM_METAS
-
-#define SAFE64_INVALID_THRESHOLD UINT64_C(0xffffFFFF00000000)
-
-/* A transaction ID. */
-typedef uint64_t txnid_t;
-typedef MDBX_atomic_uint64_t atomic_txnid_t;
-#define PRIaTXN PRIi64
-#define MIN_TXNID UINT64_C(1)
-#define MAX_TXNID (SAFE64_INVALID_THRESHOLD - 1)
-#define INITIAL_TXNID (MIN_TXNID + NUM_METAS - 1)
-#define INVALID_TXNID UINT64_MAX
-/* LY: for testing non-atomic 64-bit txnid on 32-bit arches.
- * #define xMDBX_TXNID_STEP (UINT32_MAX / 3) */
-#ifndef xMDBX_TXNID_STEP
-#if MDBX_64BIT_CAS
-#define xMDBX_TXNID_STEP 1u
-#else
-#define xMDBX_TXNID_STEP 2u
-#endif
-#endif /* xMDBX_TXNID_STEP */
-
-/* Used for offsets within a single page.
- * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
- * this is plenty. */
-typedef uint16_t indx_t;
-
-#define MEGABYTE ((size_t)1 << 20)
-
-/*----------------------------------------------------------------------------*/
-/* Core structures for database and shared memory (i.e. format definition) */
-#pragma pack(push, 4)
-
-/* Information about a single database in the environment. */
-typedef struct MDBX_db {
- uint16_t md_flags; /* see mdbx_dbi_open */
- uint16_t md_depth; /* depth of this tree */
- uint32_t md_xsize; /* key-size for MDBX_DUPFIXED (LEAF2 pages) */
- pgno_t md_root; /* the root page of this tree */
- pgno_t md_branch_pages; /* number of internal pages */
- pgno_t md_leaf_pages; /* number of leaf pages */
- pgno_t md_overflow_pages; /* number of overflow pages */
- uint64_t md_seq; /* table sequence counter */
- uint64_t md_entries; /* number of data items */
- uint64_t md_mod_txnid; /* txnid of last committed modification */
-} MDBX_db;
-
-/* database size-related parameters */
-typedef struct MDBX_geo {
- uint16_t grow_pv; /* datafile growth step as a 16-bit packed (exponential
- quantized) value */
- uint16_t shrink_pv; /* datafile shrink threshold as a 16-bit packed
- (exponential quantized) value */
- pgno_t lower; /* minimal size of datafile in pages */
- pgno_t upper; /* maximal size of datafile in pages */
- pgno_t now; /* current size of datafile in pages */
- pgno_t next; /* first unused page in the datafile,
- but actually the file may be shorter. */
-} MDBX_geo;
-
-/* Meta page content.
- * A meta page is the start point for accessing a database snapshot.
- * Pages 0-1 are meta pages. Transaction N writes meta page (N % 2). */
-typedef struct MDBX_meta {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with MDBX_DATA_VERSION. */
- uint32_t mm_magic_and_version[2];
-
- /* txnid that committed this page, the first of a two-phase-update pair */
- uint32_t mm_txnid_a[2];
-
- uint16_t mm_extra_flags; /* extra DB flags, zero (nothing) for now */
- uint8_t mm_validator_id; /* ID of checksum and page validation method,
- * zero (nothing) for now */
- uint8_t mm_extra_pagehdr; /* extra bytes in the page header,
- * zero (nothing) for now */
-
- MDBX_geo mm_geo; /* database size-related parameters */
-
- MDBX_db mm_dbs[CORE_DBS]; /* first is free space, 2nd is main db */
- /* The size of pages used in this DB */
-#define mm_psize mm_dbs[FREE_DBI].md_xsize
- MDBX_canary mm_canary;
-
-#define MDBX_DATASIGN_NONE 0u
-#define MDBX_DATASIGN_WEAK 1u
-#define SIGN_IS_STEADY(sign) ((sign) > MDBX_DATASIGN_WEAK)
-#define META_IS_STEADY(meta) \
- SIGN_IS_STEADY(unaligned_peek_u64_volatile(4, (meta)->mm_datasync_sign))
- uint32_t mm_datasync_sign[2];
-
- /* txnid that committed this page, the second of a two-phase-update pair */
- uint32_t mm_txnid_b[2];
-
- /* Number of non-meta pages which were put in GC after COW. May be 0 in case
- * DB was previously handled by libmdbx without corresponding feature.
- * This value in couple with mr_snapshot_pages_retired allows fast estimation
- * of "how much reader is restraining GC recycling". */
- uint32_t mm_pages_retired[2];
-
- /* The analogue /proc/sys/kernel/random/boot_id or similar to determine
- * whether the system was rebooted after the last use of the database files.
- * If there was no reboot, but there is no need to rollback to the last
- * steady sync point. Zeros mean that no relevant information is available
- * from the system. */
- bin128_t mm_bootid;
-
-} MDBX_meta;
-
-#pragma pack(1)
-
-/* Common header for all page types. The page type depends on mp_flags.
- *
- * P_BRANCH and P_LEAF pages have unsorted 'MDBX_node's at the end, with
- * sorted mp_ptrs[] entries referring to them. Exception: P_LEAF2 pages
- * omit mp_ptrs and pack sorted MDBX_DUPFIXED values after the page header.
- *
- * P_OVERFLOW records occupy one or more contiguous pages where only the
- * first has a page header. They hold the real data of F_BIGDATA nodes.
- *
- * P_SUBP sub-pages are small leaf "pages" with duplicate data.
- * A node with flag F_DUPDATA but not F_SUBDATA contains a sub-page.
- * (Duplicate data can also go in sub-databases, which use normal pages.)
- *
- * P_META pages contain MDBX_meta, the start point of an MDBX snapshot.
- *
- * Each non-metapage up to MDBX_meta.mm_last_pg is reachable exactly once
- * in the snapshot: Either used by a database or listed in a GC record. */
-typedef struct MDBX_page {
- union {
-#define IS_FROZEN(txn, p) ((p)->mp_txnid < (txn)->mt_txnid)
-#define IS_SPILLED(txn, p) ((p)->mp_txnid == (txn)->mt_txnid)
-#define IS_SHADOWED(txn, p) ((p)->mp_txnid > (txn)->mt_txnid)
-#define IS_VALID(txn, p) ((p)->mp_txnid <= (txn)->mt_front)
-#define IS_MODIFIABLE(txn, p) ((p)->mp_txnid == (txn)->mt_front)
- uint64_t mp_txnid;
- struct MDBX_page *mp_next; /* for in-memory list of freed pages */
- };
- uint16_t mp_leaf2_ksize; /* key size if this is a LEAF2 page */
-#define P_BRANCH 0x01 /* branch page */
-#define P_LEAF 0x02 /* leaf page */
-#define P_OVERFLOW 0x04 /* overflow page */
-#define P_META 0x08 /* meta page */
-#define P_BAD 0x10 /* explicit flag for invalid/bad page */
-#define P_LEAF2 0x20 /* for MDBX_DUPFIXED records */
-#define P_SUBP 0x40 /* for MDBX_DUPSORT sub-pages */
-#define P_SPILLED 0x2000 /* spilled in parent txn */
-#define P_LOOSE 0x4000 /* page was dirtied then freed, can be reused */
-#define P_FROZEN 0x8000 /* used for retire page with known status */
-#define P_ILL_BITS (~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_SPILLED))
- uint16_t mp_flags;
- union {
- uint32_t mp_pages; /* number of overflow pages */
- __anonymous_struct_extension__ struct {
- indx_t mp_lower; /* lower bound of free space */
- indx_t mp_upper; /* upper bound of free space */
- };
- };
- pgno_t mp_pgno; /* page number */
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- indx_t mp_ptrs[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_page;
-
-/* Size of the page header, excluding dynamic data at the end */
-#define PAGEHDRSZ ((unsigned)offsetof(MDBX_page, mp_ptrs))
-
-#pragma pack(pop)
-
-#if MDBX_ENABLE_PGOP_STAT
-/* Statistics of page operations overall of all (running, completed and aborted)
- * transactions */
-typedef struct {
- MDBX_atomic_uint64_t newly; /* Quantity of a new pages added */
- MDBX_atomic_uint64_t cow; /* Quantity of pages copied for update */
- MDBX_atomic_uint64_t clone; /* Quantity of parent's dirty pages clones
- for nested transactions */
- MDBX_atomic_uint64_t split; /* Page splits */
- MDBX_atomic_uint64_t merge; /* Page merges */
- MDBX_atomic_uint64_t spill; /* Quantity of spilled dirty pages */
- MDBX_atomic_uint64_t unspill; /* Quantity of unspilled/reloaded pages */
- MDBX_atomic_uint64_t
- wops; /* Number of explicit write operations (not a pages) to a disk */
-} MDBX_pgop_stat_t;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-#if MDBX_LOCKING == MDBX_LOCKING_WIN32FILES
-#define MDBX_CLOCK_SIGN UINT32_C(0xF10C)
-typedef void mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
-
-#define MDBX_CLOCK_SIGN UINT32_C(0xF18D)
-typedef mdbx_pid_t mdbx_ipclock_t;
-#ifndef EOWNERDEAD
-#define EOWNERDEAD MDBX_RESULT_TRUE
-#endif
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#define MDBX_CLOCK_SIGN UINT32_C(0x8017)
-typedef pthread_mutex_t mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-#define MDBX_CLOCK_SIGN UINT32_C(0xFC29)
-typedef sem_t mdbx_ipclock_t;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV && !defined(__cplusplus)
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc);
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc);
-#endif /* MDBX_LOCKING */
-
-/* Reader Lock Table
- *
- * Readers don't acquire any locks for their data access. Instead, they
- * simply record their transaction ID in the reader table. The reader
- * mutex is needed just to find an empty slot in the reader table. The
- * slot's address is saved in thread-specific data so that subsequent
- * read transactions started by the same thread need no further locking to
- * proceed.
- *
- * If MDBX_NOTLS is set, the slot address is not saved in thread-specific data.
- * No reader table is used if the database is on a read-only filesystem.
- *
- * Since the database uses multi-version concurrency control, readers don't
- * actually need any locking. This table is used to keep track of which
- * readers are using data from which old transactions, so that we'll know
- * when a particular old transaction is no longer in use. Old transactions
- * that have discarded any data pages can then have those pages reclaimed
- * for use by a later write transaction.
- *
- * The lock table is constructed such that reader slots are aligned with the
- * processor's cache line size. Any slot is only ever used by one thread.
- * This alignment guarantees that there will be no contention or cache
- * thrashing as threads update their own slot info, and also eliminates
- * any need for locking when accessing a slot.
- *
- * A writer thread will scan every slot in the table to determine the oldest
- * outstanding reader transaction. Any freed pages older than this will be
- * reclaimed by the writer. The writer doesn't use any locks when scanning
- * this table. This means that there's no guarantee that the writer will
- * see the most up-to-date reader info, but that's not required for correct
- * operation - all we need is to know the upper bound on the oldest reader,
- * we don't care at all about the newest reader. So the only consequence of
- * reading stale information here is that old pages might hang around a
- * while longer before being reclaimed. That's actually good anyway, because
- * the longer we delay reclaiming old pages, the more likely it is that a
- * string of contiguous pages can be found after coalescing old pages from
- * many old transactions together. */
-
-/* The actual reader record, with cacheline padding. */
-typedef struct MDBX_reader {
- /* Current Transaction ID when this transaction began, or (txnid_t)-1.
- * Multiple readers that start at the same time will probably have the
- * same ID here. Again, it's not important to exclude them from
- * anything; all we need to know is which version of the DB they
- * started from so we can avoid overwriting any data used in that
- * particular version. */
- MDBX_atomic_uint64_t /* txnid_t */ mr_txnid;
-
- /* The information we store in a single slot of the reader table.
- * In addition to a transaction ID, we also record the process and
- * thread ID that owns a slot, so that we can detect stale information,
- * e.g. threads or processes that went away without cleaning up.
- *
- * NOTE: We currently don't check for stale records.
- * We simply re-init the table when we know that we're the only process
- * opening the lock file. */
-
- /* The thread ID of the thread owning this txn. */
- MDBX_atomic_uint64_t mr_tid;
-
- /* The process ID of the process owning this reader txn. */
- MDBX_atomic_uint32_t mr_pid;
-
- /* The number of pages used in the reader's MVCC snapshot,
- * i.e. the value of meta->mm_geo.next and txn->mt_next_pgno */
- atomic_pgno_t mr_snapshot_pages_used;
- /* Number of retired pages at the time this reader starts transaction. So,
- * at any time the difference mm_pages_retired - mr_snapshot_pages_retired
- * will give the number of pages which this reader restraining from reuse. */
- MDBX_atomic_uint64_t mr_snapshot_pages_retired;
-} MDBX_reader;
-
-/* The header for the reader table (a memory-mapped lock file). */
-typedef struct MDBX_lockinfo {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with with MDBX_LOCK_VERSION. */
- uint64_t mti_magic_and_version;
-
- /* Format of this lock file. Must be set to MDBX_LOCK_FORMAT. */
- uint32_t mti_os_and_format;
-
- /* Flags which environment was opened. */
- MDBX_atomic_uint32_t mti_envmode;
-
- /* Threshold of un-synced-with-disk pages for auto-sync feature,
- * zero means no-threshold, i.e. auto-sync is disabled. */
- atomic_pgno_t mti_autosync_threshold;
-
- /* Low 32-bit of txnid with which meta-pages was synced,
- * i.e. for sync-polling in the MDBX_NOMETASYNC mode. */
- MDBX_atomic_uint32_t mti_meta_sync_txnid;
-
- /* Period for timed auto-sync feature, i.e. at the every steady checkpoint
- * the mti_unsynced_timeout sets to the current_time + mti_autosync_period.
- * The time value is represented in a suitable system-dependent form, for
- * example clock_gettime(CLOCK_BOOTTIME) or clock_gettime(CLOCK_MONOTONIC).
- * Zero means timed auto-sync is disabled. */
- MDBX_atomic_uint64_t mti_autosync_period;
-
- /* Marker to distinguish uniqueness of DB/CLK. */
- MDBX_atomic_uint64_t mti_bait_uniqueness;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
-#if MDBX_ENABLE_PGOP_STAT
- /* Statistics of costly ops of all (running, completed and aborted)
- * transactions */
- MDBX_pgop_stat_t mti_pgop_stat;
-#endif /* MDBX_ENABLE_PGOP_STAT*/
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Write transaction lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_wlock;
-#endif /* MDBX_LOCKING > 0 */
-
- atomic_txnid_t mti_oldest_reader;
-
- /* Timestamp of the last steady sync. Value is represented in a suitable
- * system-dependent form, for example clock_gettime(CLOCK_BOOTTIME) or
- * clock_gettime(CLOCK_MONOTONIC). */
- MDBX_atomic_uint64_t mti_sync_timestamp;
-
- /* Number un-synced-with-disk pages for auto-sync feature. */
- atomic_pgno_t mti_unsynced_pages;
-
- /* Number of page which was discarded last time by madvise(MADV_FREE). */
- atomic_pgno_t mti_discarded_tail;
-
- /* Timestamp of the last readers check. */
- MDBX_atomic_uint64_t mti_reader_check_timestamp;
-
- /* Shared anchor for tracking readahead edge and enabled/disabled status. */
- pgno_t mti_readahead_anchor;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Readeaders registration lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_rlock;
-#endif /* MDBX_LOCKING > 0 */
-
- /* The number of slots that have been used in the reader table.
- * This always records the maximum count, it is not decremented
- * when readers release their slots. */
- MDBX_atomic_uint32_t mti_numreaders;
- MDBX_atomic_uint32_t mti_readers_refresh_flag;
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
- MDBX_reader mti_readers[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_lockinfo;
-
-/* Lockfile format signature: version, features and field layout */
-#define MDBX_LOCK_FORMAT \
- (MDBX_CLOCK_SIGN * 27733 + (unsigned)sizeof(MDBX_reader) * 13 + \
- (unsigned)offsetof(MDBX_reader, mr_snapshot_pages_used) * 251 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_oldest_reader) * 83 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_numreaders) * 37 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_readers) * 29)
-
-#define MDBX_DATA_MAGIC \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + MDBX_DATA_VERSION)
-
-#define MDBX_DATA_MAGIC_LEGACY_COMPAT \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + 2)
-
-#define MDBX_DATA_MAGIC_LEGACY_DEVEL ((MDBX_MAGIC << 8) + 255)
-
-#define MDBX_LOCK_MAGIC ((MDBX_MAGIC << 8) + MDBX_LOCK_VERSION)
-
-/* The maximum size of a database page.
- *
- * It is 64K, but value-PAGEHDRSZ must fit in MDBX_page.mp_upper.
- *
- * MDBX will use database pages < OS pages if needed.
- * That causes more I/O in write transactions: The OS must
- * know (read) the whole page before writing a partial page.
- *
- * Note that we don't currently support Huge pages. On Linux,
- * regular data files cannot use Huge pages, and in general
- * Huge pages aren't actually pageable. We rely on the OS
- * demand-pager to read our data and page it out when memory
- * pressure from other processes is high. So until OSs have
- * actual paging support for Huge pages, they're not viable. */
-#define MAX_PAGESIZE MDBX_MAX_PAGESIZE
-#define MIN_PAGESIZE MDBX_MIN_PAGESIZE
-
-#define MIN_MAPSIZE (MIN_PAGESIZE * MIN_PAGENO)
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_MAPSIZE32 UINT32_C(0x38000000)
-#else
-#define MAX_MAPSIZE32 UINT32_C(0x7f000000)
-#endif
-#define MAX_MAPSIZE64 ((MAX_PAGENO + 1) * (uint64_t)MAX_PAGESIZE)
-
-#if MDBX_WORDBITS >= 64
-#define MAX_MAPSIZE MAX_MAPSIZE64
-#define MDBX_PGL_LIMIT ((size_t)MAX_PAGENO)
-#else
-#define MAX_MAPSIZE MAX_MAPSIZE32
-#define MDBX_PGL_LIMIT (MAX_MAPSIZE32 / MIN_PAGESIZE)
-#endif /* MDBX_WORDBITS */
-
-#define MDBX_READERS_LIMIT 32767
-#define MDBX_RADIXSORT_THRESHOLD 333
-
-/*----------------------------------------------------------------------------*/
-
-/* An PNL is an Page Number List, a sorted array of IDs.
- * The first element of the array is a counter for how many actual page-numbers
- * are in the list. By default PNLs are sorted in descending order, this allow
- * cut off a page with lowest pgno (at the tail) just truncating the list. The
- * sort order of PNLs is controlled by the MDBX_PNL_ASCENDING build option. */
-typedef pgno_t *MDBX_PNL;
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_ORDERED(first, last) ((first) < (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) >= (last))
-#else
-#define MDBX_PNL_ORDERED(first, last) ((first) > (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) <= (last))
-#endif
-
-/* List of txnid, only for MDBX_txn.tw.lifo_reclaimed */
-typedef txnid_t *MDBX_TXL;
-
-/* An Dirty-Page list item is an pgno/pointer pair. */
-typedef struct MDBX_dp {
- MDBX_page *ptr;
- pgno_t pgno;
- union {
- unsigned extra;
- __anonymous_struct_extension__ struct {
- unsigned multi : 1;
- unsigned lru : 31;
- };
- };
-} MDBX_dp;
-
-/* An DPL (dirty-page list) is a sorted array of MDBX_DPs. */
-typedef struct MDBX_dpl {
- unsigned sorted;
- unsigned length;
- unsigned detent; /* allocated size excluding the MDBX_DPL_RESERVE_GAP */
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_dp items[] /* dynamic size with holes at zero and after the last */;
-#endif
-} MDBX_dpl;
-
-/* PNL sizes */
-#define MDBX_PNL_GRANULATE 1024
-#define MDBX_PNL_INITIAL \
- (MDBX_PNL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(pgno_t))
-
-#define MDBX_TXL_GRANULATE 32
-#define MDBX_TXL_INITIAL \
- (MDBX_TXL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-#define MDBX_TXL_MAX \
- ((1u << 17) - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-
-#define MDBX_PNL_ALLOCLEN(pl) ((pl)[-1])
-#define MDBX_PNL_SIZE(pl) ((pl)[0])
-#define MDBX_PNL_FIRST(pl) ((pl)[1])
-#define MDBX_PNL_LAST(pl) ((pl)[MDBX_PNL_SIZE(pl)])
-#define MDBX_PNL_BEGIN(pl) (&(pl)[1])
-#define MDBX_PNL_END(pl) (&(pl)[MDBX_PNL_SIZE(pl) + 1])
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_FIRST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_LAST(pl)
-#else
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_LAST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_FIRST(pl)
-#endif
-
-#define MDBX_PNL_SIZEOF(pl) ((MDBX_PNL_SIZE(pl) + 1) * sizeof(pgno_t))
-#define MDBX_PNL_IS_EMPTY(pl) (MDBX_PNL_SIZE(pl) == 0)
-
-/*----------------------------------------------------------------------------*/
-/* Internal structures */
-
-/* Auxiliary DB info.
- * The information here is mostly static/read-only. There is
- * only a single copy of this record in the environment. */
-typedef struct MDBX_dbx {
- MDBX_val md_name; /* name of the database */
- MDBX_cmp_func *md_cmp; /* function for comparing keys */
- MDBX_cmp_func *md_dcmp; /* function for comparing data items */
- size_t md_klen_min, md_klen_max; /* min/max key length for the database */
- size_t md_vlen_min,
- md_vlen_max; /* min/max value/data length for the database */
-} MDBX_dbx;
-
-/* A database transaction.
- * Every operation requires a transaction handle. */
-struct MDBX_txn {
-#define MDBX_MT_SIGNATURE UINT32_C(0x93D53A31)
- uint32_t mt_signature;
-
- /* Transaction Flags */
- /* mdbx_txn_begin() flags */
-#define MDBX_TXN_RO_BEGIN_FLAGS (MDBX_TXN_RDONLY | MDBX_TXN_RDONLY_PREPARE)
-#define MDBX_TXN_RW_BEGIN_FLAGS \
- (MDBX_TXN_NOMETASYNC | MDBX_TXN_NOSYNC | MDBX_TXN_TRY)
- /* Additional flag for mdbx_sync_locked() */
-#define MDBX_SHRINK_ALLOWED UINT32_C(0x40000000)
-
- /* internal txn flags */
-#define MDBX_TXN_FINISHED 0x01 /* txn is finished or never began */
-#define MDBX_TXN_ERROR 0x02 /* txn is unusable after an error */
-#define MDBX_TXN_DIRTY 0x04 /* must write, even if dirty list is empty */
-#define MDBX_TXN_SPILLS 0x08 /* txn or a parent has spilled pages */
-#define MDBX_TXN_HAS_CHILD 0x10 /* txn has an MDBX_txn.mt_child */
- /* most operations on the txn are currently illegal */
-#define MDBX_TXN_BLOCKED \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_HAS_CHILD)
-
-#define TXN_FLAGS \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_DIRTY | MDBX_TXN_SPILLS | \
- MDBX_TXN_HAS_CHILD)
-
-#if (TXN_FLAGS & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS)) || \
- ((MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS | TXN_FLAGS) & \
- MDBX_SHRINK_ALLOWED)
-#error "Oops, some flags overlapped or wrong"
-#endif
- uint32_t mt_flags;
-
- MDBX_txn *mt_parent; /* parent of a nested txn */
- /* Nested txn under this txn, set together with flag MDBX_TXN_HAS_CHILD */
- MDBX_txn *mt_child;
- MDBX_geo mt_geo;
- /* next unallocated page */
-#define mt_next_pgno mt_geo.next
- /* corresponding to the current size of datafile */
-#define mt_end_pgno mt_geo.now
-
- /* The ID of this transaction. IDs are integers incrementing from 1.
- * Only committed write transactions increment the ID. If a transaction
- * aborts, the ID may be re-used by the next writer. */
- txnid_t mt_txnid;
- txnid_t mt_front;
-
- MDBX_env *mt_env; /* the DB environment */
- /* Array of records for each DB known in the environment. */
- MDBX_dbx *mt_dbxs;
- /* Array of MDBX_db records for each known DB */
- MDBX_db *mt_dbs;
- /* Array of sequence numbers for each DB handle */
- unsigned *mt_dbiseqs;
-
- /* Transaction DBI Flags */
-#define DBI_DIRTY MDBX_DBI_DIRTY /* DB was written in this txn */
-#define DBI_STALE MDBX_DBI_STALE /* Named-DB record is older than txnID */
-#define DBI_FRESH MDBX_DBI_FRESH /* Named-DB handle opened in this txn */
-#define DBI_CREAT MDBX_DBI_CREAT /* Named-DB handle created in this txn */
-#define DBI_VALID 0x10 /* DB handle is valid, see also DB_VALID */
-#define DBI_USRVALID 0x20 /* As DB_VALID, but not set for FREE_DBI */
-#define DBI_AUDITED 0x40 /* Internal flag for accounting during audit */
- /* Array of flags for each DB */
- uint8_t *mt_dbistate;
- /* Number of DB records in use, or 0 when the txn is finished.
- * This number only ever increments until the txn finishes; we
- * don't decrement it when individual DB handles are closed. */
- MDBX_dbi mt_numdbs;
- size_t mt_owner; /* thread ID that owns this transaction */
- MDBX_canary mt_canary;
- void *mt_userctx; /* User-settable context */
- MDBX_cursor **mt_cursors;
-
- union {
- struct {
- /* For read txns: This thread/txn's reader table slot, or NULL. */
- MDBX_reader *reader;
- } to;
- struct {
- /* In write txns, array of cursors for each DB */
- pgno_t *reclaimed_pglist; /* Reclaimed GC pages */
- txnid_t last_reclaimed; /* ID of last used record */
-#if MDBX_ENABLE_REFUND
- pgno_t loose_refund_wl /* FIXME: describe */;
-#endif /* MDBX_ENABLE_REFUND */
- /* dirtylist room: Dirty array size - dirty pages visible to this txn.
- * Includes ancestor txns' dirty pages not hidden by other txns'
- * dirty/spilled pages. Thus commit(nested txn) has room to merge
- * dirtylist into mt_parent after freeing hidden mt_parent pages. */
- unsigned dirtyroom;
- /* a sequence to spilling dirty page with LRU policy */
- unsigned dirtylru;
- /* For write txns: Modified pages. Sorted when not MDBX_WRITEMAP. */
- MDBX_dpl *dirtylist;
- /* The list of reclaimed txns from GC */
- MDBX_TXL lifo_reclaimed;
- /* The list of pages that became unused during this transaction. */
- MDBX_PNL retired_pages;
- /* The list of loose pages that became unused and may be reused
- * in this transaction, linked through `mp_next`. */
- MDBX_page *loose_pages;
- /* Number of loose pages (tw.loose_pages) */
- unsigned loose_count;
- /* The sorted list of dirty pages we temporarily wrote to disk
- * because the dirty list was full. page numbers in here are
- * shifted left by 1, deleted slots have the LSB set. */
- MDBX_PNL spill_pages;
- unsigned spill_least_removed;
- } tw;
- };
-};
-
-#if MDBX_WORDBITS >= 64
-#define CURSOR_STACK 32
-#else
-#define CURSOR_STACK 24
-#endif
-
-struct MDBX_xcursor;
-
-/* Cursors are used for all DB operations.
- * A cursor holds a path of (page pointer, key index) from the DB
- * root to a position in the DB, plus other state. MDBX_DUPSORT
- * cursors include an xcursor to the current data item. Write txns
- * track their cursors and keep them up to date when data moves.
- * Exception: An xcursor's pointer to a P_SUBP page can be stale.
- * (A node with F_DUPDATA but no F_SUBDATA contains a subpage). */
-struct MDBX_cursor {
-#define MDBX_MC_LIVE UINT32_C(0xFE05D5B1)
-#define MDBX_MC_READY4CLOSE UINT32_C(0x2817A047)
-#define MDBX_MC_WAIT4EOT UINT32_C(0x90E297A7)
- uint32_t mc_signature;
- /* The database handle this cursor operates on */
- MDBX_dbi mc_dbi;
- /* Next cursor on this DB in this txn */
- MDBX_cursor *mc_next;
- /* Backup of the original cursor if this cursor is a shadow */
- MDBX_cursor *mc_backup;
- /* Context used for databases with MDBX_DUPSORT, otherwise NULL */
- struct MDBX_xcursor *mc_xcursor;
- /* The transaction that owns this cursor */
- MDBX_txn *mc_txn;
- /* The database record for this cursor */
- MDBX_db *mc_db;
- /* The database auxiliary record for this cursor */
- MDBX_dbx *mc_dbx;
- /* The mt_dbistate for this database */
- uint8_t *mc_dbistate;
- unsigned mc_snum; /* number of pushed pages */
- unsigned mc_top; /* index of top page, normally mc_snum-1 */
-
- /* Cursor state flags. */
-#define C_INITIALIZED 0x01 /* cursor has been initialized and is valid */
-#define C_EOF 0x02 /* No more data */
-#define C_SUB 0x04 /* Cursor is a sub-cursor */
-#define C_DEL 0x08 /* last op was a cursor_del */
-#define C_UNTRACK 0x10 /* Un-track cursor when closing */
-#define C_RECLAIMING 0x20 /* GC lookup is prohibited */
-#define C_GCFREEZE 0x40 /* reclaimed_pglist must not be updated */
-
- /* Cursor checking flags. */
-#define C_COPYING 0x100 /* skip key-value length check (copying simplify) */
-#define C_UPDATING 0x200 /* update/rebalance pending */
-#define C_RETIRING 0x400 /* refs to child pages may be invalid */
-#define C_SKIPORD 0x800 /* don't check keys ordering */
-
- unsigned mc_flags; /* see mdbx_cursor */
- MDBX_page *mc_pg[CURSOR_STACK]; /* stack of pushed pages */
- indx_t mc_ki[CURSOR_STACK]; /* stack of page indices */
-};
-
-/* Context for sorted-dup records.
- * We could have gone to a fully recursive design, with arbitrarily
- * deep nesting of sub-databases. But for now we only handle these
- * levels - main DB, optional sub-DB, sorted-duplicate DB. */
-typedef struct MDBX_xcursor {
- /* A sub-cursor for traversing the Dup DB */
- MDBX_cursor mx_cursor;
- /* The database record for this Dup DB */
- MDBX_db mx_db;
- /* The auxiliary DB record for this Dup DB */
- MDBX_dbx mx_dbx;
-} MDBX_xcursor;
-
-typedef struct MDBX_cursor_couple {
- MDBX_cursor outer;
- void *mc_userctx; /* User-settable context */
- MDBX_xcursor inner;
-} MDBX_cursor_couple;
-
-/* The database environment. */
-struct MDBX_env {
- /* ----------------------------------------------------- mostly static part */
-#define MDBX_ME_SIGNATURE UINT32_C(0x9A899641)
- MDBX_atomic_uint32_t me_signature;
- /* Failed to update the meta page. Probably an I/O error. */
-#define MDBX_FATAL_ERROR UINT32_C(0x80000000)
- /* Some fields are initialized. */
-#define MDBX_ENV_ACTIVE UINT32_C(0x20000000)
- /* me_txkey is set */
-#define MDBX_ENV_TXKEY UINT32_C(0x10000000)
- /* Legacy MDBX_MAPASYNC (prior v0.9) */
-#define MDBX_DEPRECATED_MAPASYNC UINT32_C(0x100000)
-#define ENV_INTERNAL_FLAGS (MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE | MDBX_ENV_TXKEY)
- uint32_t me_flags;
- mdbx_mmap_t me_dxb_mmap; /* The main data file */
-#define me_map me_dxb_mmap.dxb
-#define me_lazy_fd me_dxb_mmap.fd
- mdbx_filehandle_t me_dsync_fd;
- mdbx_mmap_t me_lck_mmap; /* The lock file */
-#define me_lfd me_lck_mmap.fd
- struct MDBX_lockinfo *me_lck;
-
- unsigned me_psize; /* DB page size, initialized from me_os_psize */
- unsigned me_leaf_nodemax; /* max size of a leaf-node */
- uint8_t me_psize2log; /* log2 of DB page size */
- int8_t me_stuck_meta; /* recovery-only: target meta page or less that zero */
- uint16_t me_merge_threshold,
- me_merge_threshold_gc; /* pages emptier than this are candidates for
- merging */
- unsigned me_os_psize; /* OS page size, from mdbx_syspagesize() */
- unsigned me_maxreaders; /* size of the reader table */
- MDBX_dbi me_maxdbs; /* size of the DB table */
- uint32_t me_pid; /* process ID of this env */
- mdbx_thread_key_t me_txkey; /* thread-key for readers */
- char *me_pathname; /* path to the DB files */
- void *me_pbuf; /* scratch area for DUPSORT put() */
- MDBX_txn *me_txn0; /* preallocated write transaction */
-
- MDBX_dbx *me_dbxs; /* array of static DB info */
- uint16_t *me_dbflags; /* array of flags from MDBX_db.md_flags */
- unsigned *me_dbiseqs; /* array of dbi sequence numbers */
- unsigned
- me_maxgc_ov1page; /* Number of pgno_t fit in a single overflow page */
- uint32_t me_live_reader; /* have liveness lock in reader table */
- void *me_userctx; /* User-settable context */
- MDBX_hsr_func *me_hsr_callback; /* Callback for kicking laggard readers */
-
- struct {
- unsigned dp_reserve_limit;
- unsigned rp_augment_limit;
- unsigned dp_limit;
- unsigned dp_initial;
- uint8_t dp_loose_limit;
- uint8_t spill_max_denominator;
- uint8_t spill_min_denominator;
- uint8_t spill_parent4child_denominator;
- unsigned merge_threshold_16dot16_percent;
- union {
- unsigned all;
- /* tracks options with non-auto values but tuned by user */
- struct {
- unsigned dp_limit : 1;
- } non_auto;
- } flags;
- } me_options;
-
- /* struct me_dbgeo used for accepting db-geo params from user for the new
- * database creation, i.e. when mdbx_env_set_geometry() was called before
- * mdbx_env_open(). */
- struct {
- size_t lower; /* minimal size of datafile */
- size_t upper; /* maximal size of datafile */
- size_t now; /* current size of datafile */
- size_t grow; /* step to grow datafile */
- size_t shrink; /* threshold to shrink datafile */
- } me_dbgeo;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- union {
- key_t key;
- int semid;
- } me_sysv_ipc;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
- MDBX_env *me_lcklist_next;
-
- /* --------------------------------------------------- mostly volatile part */
-
- MDBX_txn *me_txn; /* current write transaction */
- mdbx_fastmutex_t me_dbi_lock;
- MDBX_dbi me_numdbs; /* number of DBs opened */
-
- MDBX_page *me_dp_reserve; /* list of malloc'ed blocks for re-use */
- unsigned me_dp_reserve_len;
- /* PNL of pages that became unused in a write txn */
- MDBX_PNL me_retired_pages;
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_srwlock me_remap_guard;
- /* Workaround for LockFileEx and WriteFile multithread bug */
- CRITICAL_SECTION me_windowsbug_lock;
-#else
- mdbx_fastmutex_t me_remap_guard;
-#endif
-
- /* -------------------------------------------------------------- debugging */
-
-#if MDBX_DEBUG
- MDBX_assert_func *me_assert_func; /* Callback for assertion failures */
-#endif
-#ifdef MDBX_USE_VALGRIND
- int me_valgrind_handle;
-#endif
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- pgno_t me_poison_edge;
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-#ifndef xMDBX_DEBUG_SPILLING
-#define xMDBX_DEBUG_SPILLING 0
-#endif
-#if xMDBX_DEBUG_SPILLING == 2
- unsigned debug_dirtied_est, debug_dirtied_act;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- /* ------------------------------------------------- stub for lck-less mode */
- MDBX_atomic_uint64_t
- x_lckless_stub[(sizeof(MDBX_lockinfo) + MDBX_CACHELINE_SIZE - 1) /
- sizeof(MDBX_atomic_uint64_t)];
-};
-
-#ifndef __cplusplus
-/*----------------------------------------------------------------------------*/
-/* Debug and Logging stuff */
-
-#define MDBX_RUNTIME_FLAGS_INIT \
- ((MDBX_DEBUG) > 0) * MDBX_DBG_ASSERT + ((MDBX_DEBUG) > 1) * MDBX_DBG_AUDIT
-
-extern uint8_t mdbx_runtime_flags;
-extern uint8_t mdbx_loglevel;
-extern MDBX_debug_func *mdbx_debug_logger;
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny) {
-#if MDBX_DEBUG
- if (MDBX_DBG_JITTER & mdbx_runtime_flags)
- mdbx_osal_jitter(tiny);
-#else
- (void)tiny;
-#endif
-}
-
-MDBX_INTERNAL_FUNC void MDBX_PRINTF_ARGS(4, 5)
- mdbx_debug_log(int level, const char *function, int line, const char *fmt,
- ...) MDBX_PRINTF_ARGS(4, 5);
-MDBX_INTERNAL_FUNC void mdbx_debug_log_va(int level, const char *function,
- int line, const char *fmt,
- va_list args);
-
-#if MDBX_DEBUG
-#define mdbx_log_enabled(msg) unlikely(msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() unlikely((mdbx_runtime_flags & MDBX_DBG_AUDIT))
-#else /* MDBX_DEBUG */
-#define mdbx_log_enabled(msg) (msg < MDBX_LOG_VERBOSE && msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() (0)
-#endif /* MDBX_DEBUG */
-
-#if MDBX_FORCE_ASSERTIONS
-#define mdbx_assert_enabled() (1)
-#elif MDBX_DEBUG
-#define mdbx_assert_enabled() likely((mdbx_runtime_flags & MDBX_DBG_ASSERT))
-#else
-#define mdbx_assert_enabled() (0)
-#endif /* assertions */
-
-#if !MDBX_DEBUG && defined(__ANDROID_API__)
-#define mdbx_assert_fail(env, msg, func, line) \
- __android_log_assert(msg, "mdbx", "%s:%u", func, line)
-#else
-void mdbx_assert_fail(const MDBX_env *env, const char *msg, const char *func,
- int line);
-#endif
-
-#define mdbx_debug_extra(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, __func__, __LINE__, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug_extra_print(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, NULL, 0, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_trace(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_TRACE)) \
- mdbx_debug_log(MDBX_LOG_TRACE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_DEBUG)) \
- mdbx_debug_log(MDBX_LOG_DEBUG, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_verbose(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_VERBOSE)) \
- mdbx_debug_log(MDBX_LOG_VERBOSE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_notice(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_NOTICE)) \
- mdbx_debug_log(MDBX_LOG_NOTICE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_warning(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_WARN)) \
- mdbx_debug_log(MDBX_LOG_WARN, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_error(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) \
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_fatal(fmt, ...) \
- mdbx_debug_log(MDBX_LOG_FATAL, __func__, __LINE__, fmt "\n", __VA_ARGS__);
-
-#define mdbx_ensure_msg(env, expr, msg) \
- do { \
- if (unlikely(!(expr))) \
- mdbx_assert_fail(env, msg, __func__, __LINE__); \
- } while (0)
-
-#define mdbx_ensure(env, expr) mdbx_ensure_msg(env, expr, #expr)
-
-/* assert(3) variant in environment context */
-#define mdbx_assert(env, expr) \
- do { \
- if (mdbx_assert_enabled()) \
- mdbx_ensure(env, expr); \
- } while (0)
-
-/* assert(3) variant in cursor context */
-#define mdbx_cassert(mc, expr) mdbx_assert((mc)->mc_txn->mt_env, expr)
-
-/* assert(3) variant in transaction context */
-#define mdbx_tassert(txn, expr) mdbx_assert((txn)->mt_env, expr)
-
-#ifndef xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-#undef assert
-#define assert(expr) mdbx_assert(NULL, expr)
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Cache coherence and mmap invalidation */
-
-#if MDBX_CPU_WRITEBACK_INCOHERENT
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_memory_barrier()
-#else
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_compiler_barrier()
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-MDBX_MAYBE_UNUSED static __inline void
-mdbx_flush_incoherent_mmap(void *addr, size_t nbytes, const intptr_t pagesize) {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- char *const begin = (char *)(-pagesize & (intptr_t)addr);
- char *const end =
- (char *)(-pagesize & (intptr_t)((char *)addr + nbytes + pagesize - 1));
- int err = msync(begin, end - begin, MS_SYNC | MS_INVALIDATE) ? errno : 0;
- mdbx_assert(nullptr, err == 0);
- (void)err;
-#else
- (void)pagesize;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#if MDBX_MMAP_INCOHERENT_CPU_CACHE
-#ifdef DCACHE
- /* MIPS has cache coherency issues.
- * Note: for any nbytes >= on-chip cache size, entire is flushed. */
- cacheflush(addr, nbytes, DCACHE);
-#else
-#error "Oops, cacheflush() not available"
-#endif /* DCACHE */
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#if !MDBX_MMAP_INCOHERENT_FILE_WRITE && !MDBX_MMAP_INCOHERENT_CPU_CACHE
- (void)addr;
- (void)nbytes;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* Internal prototypes */
-
-MDBX_INTERNAL_FUNC int mdbx_cleanup_dead_readers(MDBX_env *env, int rlocked,
- int *dead);
-MDBX_INTERNAL_FUNC int mdbx_rthc_alloc(mdbx_thread_key_t *key,
- MDBX_reader *begin, MDBX_reader *end);
-MDBX_INTERNAL_FUNC void mdbx_rthc_remove(const mdbx_thread_key_t key);
-
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_init(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_dtor(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_thread_dtor(void *ptr);
-
-#endif /* !__cplusplus */
-
-#define MDBX_IS_ERROR(rc) \
- ((rc) != MDBX_RESULT_TRUE && (rc) != MDBX_RESULT_FALSE)
-
-/* Internal error codes, not exposed outside libmdbx */
-#define MDBX_NO_ROOT (MDBX_LAST_ADDED_ERRCODE + 10)
-
-/* Debugging output value of a cursor DBI: Negative in a sub-cursor. */
-#define DDBI(mc) \
- (((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
-
-/* Key size which fits in a DKBUF (debug key buffer). */
-#define DKBUF_MAX 511
-#define DKBUF char _kbuf[DKBUF_MAX * 4 + 2]
-#define DKEY(x) mdbx_dump_val(x, _kbuf, DKBUF_MAX * 2 + 1)
-#define DVAL(x) mdbx_dump_val(x, _kbuf + DKBUF_MAX * 2 + 1, DKBUF_MAX * 2 + 1)
-
-#if MDBX_DEBUG
-#define DKBUF_DEBUG DKBUF
-#define DKEY_DEBUG(x) DKEY(x)
-#define DVAL_DEBUG(x) DVAL(x)
-#else
-#define DKBUF_DEBUG ((void)(0))
-#define DKEY_DEBUG(x) ("-")
-#define DVAL_DEBUG(x) ("-")
-#endif
-
-/* An invalid page number.
- * Mainly used to denote an empty tree. */
-#define P_INVALID (~(pgno_t)0)
-
-/* Test if the flags f are set in a flag word w. */
-#define F_ISSET(w, f) (((w) & (f)) == (f))
-
-/* Round n up to an even number. */
-#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
-
-/* Default size of memory map.
- * This is certainly too small for any actual applications. Apps should
- * always set the size explicitly using mdbx_env_set_geometry(). */
-#define DEFAULT_MAPSIZE MEGABYTE
-
-/* Number of slots in the reader table.
- * This value was chosen somewhat arbitrarily. The 61 is a prime number,
- * and such readers plus a couple mutexes fit into single 4KB page.
- * Applications should set the table size using mdbx_env_set_maxreaders(). */
-#define DEFAULT_READERS 61
-
-/* Test if a page is a leaf page */
-#define IS_LEAF(p) (((p)->mp_flags & P_LEAF) != 0)
-/* Test if a page is a LEAF2 page */
-#define IS_LEAF2(p) unlikely(((p)->mp_flags & P_LEAF2) != 0)
-/* Test if a page is a branch page */
-#define IS_BRANCH(p) (((p)->mp_flags & P_BRANCH) != 0)
-/* Test if a page is an overflow page */
-#define IS_OVERFLOW(p) unlikely(((p)->mp_flags & P_OVERFLOW) != 0)
-/* Test if a page is a sub page */
-#define IS_SUBP(p) (((p)->mp_flags & P_SUBP) != 0)
-
-#define PAGETYPE(p) ((p)->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW))
-
-/* Header for a single key/data pair within a page.
- * Used in pages of type P_BRANCH and P_LEAF without P_LEAF2.
- * We guarantee 2-byte alignment for 'MDBX_node's.
- *
- * Leaf node flags describe node contents. F_BIGDATA says the node's
- * data part is the page number of an overflow page with actual data.
- * F_DUPDATA and F_SUBDATA can be combined giving duplicate data in
- * a sub-page/sub-database, and named databases (just F_SUBDATA). */
-typedef struct MDBX_node {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- union {
- uint32_t mn_dsize;
- uint32_t mn_pgno32;
- };
- uint8_t mn_flags; /* see mdbx_node flags */
- uint8_t mn_extra;
- uint16_t mn_ksize; /* key size */
-#else
- uint16_t mn_ksize; /* key size */
- uint8_t mn_extra;
- uint8_t mn_flags; /* see mdbx_node flags */
- union {
- uint32_t mn_pgno32;
- uint32_t mn_dsize;
- };
-#endif /* __BYTE_ORDER__ */
-
- /* mdbx_node Flags */
-#define F_BIGDATA 0x01 /* data put on overflow page */
-#define F_SUBDATA 0x02 /* data is a sub-database */
-#define F_DUPDATA 0x04 /* data has duplicates */
-
- /* valid flags for mdbx_node_add() */
-#define NODE_ADD_FLAGS (F_DUPDATA | F_SUBDATA | MDBX_RESERVE | MDBX_APPEND)
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- uint8_t mn_data[] /* key and data are appended here */;
-#endif /* C99 */
-} MDBX_node;
-
-#define DB_PERSISTENT_FLAGS \
- (MDBX_REVERSEKEY | MDBX_DUPSORT | MDBX_INTEGERKEY | MDBX_DUPFIXED | \
- MDBX_INTEGERDUP | MDBX_REVERSEDUP)
-
-/* mdbx_dbi_open() flags */
-#define DB_USABLE_FLAGS (DB_PERSISTENT_FLAGS | MDBX_CREATE | MDBX_DB_ACCEDE)
-
-#define DB_VALID 0x8000 /* DB handle is valid, for me_dbflags */
-#define DB_INTERNAL_FLAGS DB_VALID
-
-#if DB_INTERNAL_FLAGS & DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-#if DB_PERSISTENT_FLAGS & ~DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-
-/* max number of pages to commit in one writev() call */
-#define MDBX_COMMIT_PAGES 64
-#if defined(IOV_MAX) && IOV_MAX < MDBX_COMMIT_PAGES /* sysconf(_SC_IOV_MAX) */
-#undef MDBX_COMMIT_PAGES
-#define MDBX_COMMIT_PAGES IOV_MAX
-#endif
-
-/*
- * /
- * | -1, a < b
- * CMP2INT(a,b) = < 0, a == b
- * | 1, a > b
- * \
- */
-#ifndef __e2k__
-/* LY: fast enough on most systems */
-#define CMP2INT(a, b) (((b) > (a)) ? -1 : (a) > (b))
-#else
-/* LY: more parallelable on VLIW Elbrus */
-#define CMP2INT(a, b) (((a) > (b)) - ((b) > (a)))
-#endif
-
-/* Do not spill pages to disk if txn is getting full, may fail instead */
-#define MDBX_NOSPILL 0x8000
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-int64pgno(int64_t i64) {
- if (likely(i64 >= (int64_t)MIN_PAGENO && i64 <= (int64_t)MAX_PAGENO + 1))
- return (pgno_t)i64;
- return (i64 < (int64_t)MIN_PAGENO) ? MIN_PAGENO : MAX_PAGENO;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_add(size_t base, size_t augend) {
- assert(base <= MAX_PAGENO + 1 && augend < MAX_PAGENO);
- return int64pgno(base + augend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_sub(size_t base, size_t subtrahend) {
- assert(base >= MIN_PAGENO && base <= MAX_PAGENO + 1 &&
- subtrahend < MAX_PAGENO);
- return int64pgno(base - subtrahend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline bool
-is_powerof2(size_t x) {
- return (x & (x - 1)) == 0;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-floor_powerof2(size_t value, size_t granularity) {
- assert(is_powerof2(granularity));
- return value & ~(granularity - 1);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-ceil_powerof2(size_t value, size_t granularity) {
- return floor_powerof2(value + granularity - 1, granularity);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static unsigned
-log2n_powerof2(size_t value) {
- assert(value > 0 && value < INT32_MAX && is_powerof2(value));
- assert((value & -(int32_t)value) == value);
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_ctzl)
- return __builtin_ctzl(value);
-#elif defined(_MSC_VER)
- unsigned long index;
- _BitScanForward(&index, (unsigned long)value);
- return index;
-#else
- static const uint8_t debruijn_ctz32[32] = {
- 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
- 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9};
- return debruijn_ctz32[(uint32_t)(value * 0x077CB531u) >> 27];
-#endif
-}
-
-/* Only a subset of the mdbx_env flags can be changed
- * at runtime. Changing other flags requires closing the
- * environment and re-opening it with the new flags. */
-#define ENV_CHANGEABLE_FLAGS \
- (MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC | MDBX_DEPRECATED_MAPASYNC | \
- MDBX_NOMEMINIT | MDBX_COALESCE | MDBX_PAGEPERTURB | MDBX_ACCEDE)
-#define ENV_CHANGELESS_FLAGS \
- (MDBX_NOSUBDIR | MDBX_RDONLY | MDBX_WRITEMAP | MDBX_NOTLS | MDBX_NORDAHEAD | \
- MDBX_LIFORECLAIM | MDBX_EXCLUSIVE)
-#define ENV_USABLE_FLAGS (ENV_CHANGEABLE_FLAGS | ENV_CHANGELESS_FLAGS)
-
-#if !defined(__cplusplus) || CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_MAYBE_UNUSED static void static_checks(void) {
- STATIC_ASSERT_MSG(INT16_MAX - CORE_DBS == MDBX_MAX_DBI,
- "Oops, MDBX_MAX_DBI or CORE_DBS?");
- STATIC_ASSERT_MSG((unsigned)(MDBX_DB_ACCEDE | MDBX_CREATE) ==
- ((DB_USABLE_FLAGS | DB_INTERNAL_FLAGS) &
- (ENV_USABLE_FLAGS | ENV_INTERNAL_FLAGS)),
- "Oops, some flags overlapped or wrong");
- STATIC_ASSERT_MSG((ENV_INTERNAL_FLAGS & ENV_USABLE_FLAGS) == 0,
- "Oops, some flags overlapped or wrong");
-}
-#endif /* Disabled for MSVC 19.0 (VisualStudio 2015) */
-
-#ifdef __cplusplus
-}
-#endif
-
-#define MDBX_ASAN_POISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("POISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_POISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#define MDBX_ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("UNPOISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_UNPOISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#include <ctype.h>
-
-#define PRINT 1
-#define GLOBAL 2
-static int mode = GLOBAL;
-
-typedef struct flagbit {
- int bit;
- char *name;
-} flagbit;
-
-flagbit dbflags[] = {{MDBX_REVERSEKEY, "reversekey"},
- {MDBX_DUPSORT, "dupsort"},
- {MDBX_INTEGERKEY, "integerkey"},
- {MDBX_DUPFIXED, "dupfixed"},
- {MDBX_INTEGERDUP, "integerdup"},
- {MDBX_REVERSEDUP, "reversedup"},
- {0, nullptr}};
-
-#if defined(_WIN32) || defined(_WIN64)
-/*
- * POSIX getopt for Windows
- *
- * AT&T Public License
- *
- * Code given out at the 1985 UNIFORUM conference in Dallas.
- */
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif
-#endif /* _MSC_VER (warnings) */
-
-#include <stdio.h>
-#include <string.h>
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-/*----------------------------------------------------------------------------*/
-
-#ifndef NULL
-#define NULL 0
-#endif
-
-#ifndef EOF
-#define EOF (-1)
-#endif
-
-int optind = 1;
-int optopt;
-char *optarg;
-
-int getopt(int argc, char *const argv[], const char *opts) {
- static int sp = 1;
- int c;
- const char *cp;
-
- if (sp == 1) {
- if (optind >= argc || argv[optind][0] != '-' || argv[optind][1] == '\0')
- return EOF;
- else if (strcmp(argv[optind], "--") == 0) {
- optind++;
- return EOF;
- }
- }
- optopt = c = argv[optind][sp];
- if (c == ':' || (cp = strchr(opts, c)) == NULL) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "illegal option", c);
- if (argv[optind][++sp] == '\0') {
- optind++;
- sp = 1;
- }
- return '?';
- }
- if (*++cp == ':') {
- if (argv[optind][sp + 1] != '\0')
- optarg = &argv[optind++][sp + 1];
- else if (++optind >= argc) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "option requires an argument",
- c);
- sp = 1;
- return '?';
- } else
- optarg = argv[optind++];
- sp = 1;
- } else {
- if (argv[optind][++sp] == '\0') {
- sp = 1;
- optind++;
- }
- optarg = NULL;
- }
- return c;
-}
-
-static volatile BOOL user_break;
-static BOOL WINAPI ConsoleBreakHandlerRoutine(DWORD dwCtrlType) {
- (void)dwCtrlType;
- user_break = true;
- return true;
-}
-
-#else /* WINDOWS */
-
-static volatile sig_atomic_t user_break;
-static void signal_handler(int sig) {
- (void)sig;
- user_break = 1;
-}
-
-#endif /* !WINDOWS */
-
-static const char hexc[] = "0123456789abcdef";
-
-static void dumpbyte(unsigned char c) {
- putchar(hexc[c >> 4]);
- putchar(hexc[c & 0xf]);
-}
-
-static void text(MDBX_val *v) {
- unsigned char *c, *end;
-
- putchar(' ');
- c = v->iov_base;
- end = c + v->iov_len;
- while (c < end) {
- if (isprint(*c) && *c != '\\') {
- putchar(*c);
- } else {
- putchar('\\');
- dumpbyte(*c);
- }
- c++;
- }
- putchar('\n');
-}
-
-static void dumpval(MDBX_val *v) {
- unsigned char *c, *end;
-
- putchar(' ');
- c = v->iov_base;
- end = c + v->iov_len;
- while (c < end)
- dumpbyte(*c++);
- putchar('\n');
-}
-
-bool quiet = false, rescue = false;
-const char *prog;
-static void error(const char *func, int rc) {
- if (!quiet)
- fprintf(stderr, "%s: %s() error %d %s\n", prog, func, rc,
- mdbx_strerror(rc));
-}
-
-/* Dump in BDB-compatible format */
-static int dump_sdb(MDBX_txn *txn, MDBX_dbi dbi, char *name) {
- unsigned int flags;
- int rc = mdbx_dbi_flags(txn, dbi, &flags);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_flags", rc);
- return rc;
- }
-
- MDBX_stat ms;
- rc = mdbx_dbi_stat(txn, dbi, &ms, sizeof(ms));
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_stat", rc);
- return rc;
- }
-
- MDBX_envinfo info;
- rc = mdbx_env_info_ex(mdbx_txn_env(txn), txn, &info, sizeof(info));
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_info_ex", rc);
- return rc;
- }
-
- printf("VERSION=3\n");
- if (mode & GLOBAL) {
- mode -= GLOBAL;
- if (info.mi_geo.upper != info.mi_geo.lower)
- printf("geometry=l%" PRIu64 ",c%" PRIu64 ",u%" PRIu64 ",s%" PRIu64
- ",g%" PRIu64 "\n",
- info.mi_geo.lower, info.mi_geo.current, info.mi_geo.upper,
- info.mi_geo.shrink, info.mi_geo.grow);
- printf("mapsize=%" PRIu64 "\n", info.mi_geo.upper);
- printf("maxreaders=%u\n", info.mi_maxreaders);
-
- MDBX_canary canary;
- rc = mdbx_canary_get(txn, &canary);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_canary_get", rc);
- return rc;
- }
- if (canary.v)
- printf("canary=v%" PRIu64 ",x%" PRIu64 ",y%" PRIu64 ",z%" PRIu64 "\n",
- canary.v, canary.x, canary.y, canary.z);
- }
- printf("format=%s\n", mode & PRINT ? "print" : "bytevalue");
- if (name)
- printf("database=%s\n", name);
- printf("type=btree\n");
- printf("db_pagesize=%u\n", ms.ms_psize);
- /* if (ms.ms_mod_txnid)
- printf("txnid=%" PRIaTXN "\n", ms.ms_mod_txnid);
- else if (!name)
- printf("txnid=%" PRIaTXN "\n", mdbx_txn_id(txn)); */
-
- printf("duplicates=%d\n", (flags & (MDBX_DUPSORT | MDBX_DUPFIXED |
- MDBX_INTEGERDUP | MDBX_REVERSEDUP))
- ? 1
- : 0);
- for (int i = 0; dbflags[i].bit; i++)
- if (flags & dbflags[i].bit)
- printf("%s=1\n", dbflags[i].name);
-
- uint64_t sequence;
- rc = mdbx_dbi_sequence(txn, dbi, &sequence, 0);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_sequence", rc);
- return rc;
- }
- if (sequence)
- printf("sequence=%" PRIu64 "\n", sequence);
-
- printf("HEADER=END\n"); /*-------------------------------------------------*/
-
- MDBX_cursor *cursor;
- MDBX_val key, data;
- rc = mdbx_cursor_open(txn, dbi, &cursor);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_cursor_open", rc);
- return rc;
- }
- if (MDBX_DEBUG > 0 && rescue) {
- cursor->mc_flags |= C_SKIPORD;
- if (cursor->mc_xcursor)
- cursor->mc_xcursor->mx_cursor.mc_flags |= C_SKIPORD;
- }
-
- while ((rc = mdbx_cursor_get(cursor, &key, &data, MDBX_NEXT)) ==
- MDBX_SUCCESS) {
- if (user_break) {
- rc = MDBX_EINTR;
- break;
- }
- if (mode & PRINT) {
- text(&key);
- text(&data);
- } else {
- dumpval(&key);
- dumpval(&data);
- }
- }
- printf("DATA=END\n");
- if (rc == MDBX_NOTFOUND)
- rc = MDBX_SUCCESS;
- if (unlikely(rc != MDBX_SUCCESS))
- error("mdbx_cursor_get", rc);
-
- mdbx_cursor_close(cursor);
- return rc;
-}
-
-static void usage(void) {
- fprintf(stderr,
- "usage: %s [-V] [-q] [-f file] [-l] [-p] [-r] [-a|-s subdb] "
- "dbpath\n"
- " -V\t\tprint version and exit\n"
- " -q\t\tbe quiet\n"
- " -f\t\twrite to file instead of stdout\n"
- " -l\t\tlist subDBs and exit\n"
- " -p\t\tuse printable characters\n"
- " -r\t\trescue mode (ignore errors to dump corrupted DB)\n"
- " -a\t\tdump main DB and all subDBs\n"
- " -s name\tdump only the specified named subDB\n"
- " \t\tby default dump only the main DB\n",
- prog);
- exit(EXIT_FAILURE);
-}
-
-static int equal_or_greater(const MDBX_val *a, const MDBX_val *b) {
- return (a->iov_len == b->iov_len &&
- memcmp(a->iov_base, b->iov_base, a->iov_len) == 0)
- ? 0
- : 1;
-}
-
-int main(int argc, char *argv[]) {
- int i, rc;
- MDBX_env *env;
- MDBX_txn *txn;
- MDBX_dbi dbi;
- prog = argv[0];
- char *envname;
- char *subname = nullptr, *buf4free = nullptr;
- unsigned envflags = 0;
- bool alldbs = false, list = false;
-
- if (argc < 2)
- usage();
-
- while ((i = getopt(argc, argv,
- "a"
- "f:"
- "l"
- "n"
- "p"
- "s:"
- "V"
- "r"
- "q")) != EOF) {
- switch (i) {
- case 'V':
- printf("mdbx_dump version %d.%d.%d.%d\n"
- " - source: %s %s, commit %s, tree %s\n"
- " - anchor: %s\n"
- " - build: %s for %s by %s\n"
- " - flags: %s\n"
- " - options: %s\n",
- mdbx_version.major, mdbx_version.minor, mdbx_version.release,
- mdbx_version.revision, mdbx_version.git.describe,
- mdbx_version.git.datetime, mdbx_version.git.commit,
- mdbx_version.git.tree, mdbx_sourcery_anchor, mdbx_build.datetime,
- mdbx_build.target, mdbx_build.compiler, mdbx_build.flags,
- mdbx_build.options);
- return EXIT_SUCCESS;
- case 'l':
- list = true;
- /*FALLTHROUGH*/;
- __fallthrough;
- case 'a':
- if (subname)
- usage();
- alldbs = true;
- break;
- case 'f':
- if (freopen(optarg, "w", stdout) == nullptr) {
- fprintf(stderr, "%s: %s: reopen: %s\n", prog, optarg,
- mdbx_strerror(errno));
- exit(EXIT_FAILURE);
- }
- break;
- case 'n':
- break;
- case 'p':
- mode |= PRINT;
- break;
- case 's':
- if (alldbs)
- usage();
- subname = optarg;
- break;
- case 'q':
- quiet = true;
- break;
- case 'r':
- rescue = true;
- break;
- default:
- usage();
- }
- }
-
- if (optind != argc - 1)
- usage();
-
-#if defined(_WIN32) || defined(_WIN64)
- SetConsoleCtrlHandler(ConsoleBreakHandlerRoutine, true);
-#else
-#ifdef SIGPIPE
- signal(SIGPIPE, signal_handler);
-#endif
-#ifdef SIGHUP
- signal(SIGHUP, signal_handler);
-#endif
- signal(SIGINT, signal_handler);
- signal(SIGTERM, signal_handler);
-#endif /* !WINDOWS */
-
- envname = argv[optind];
- if (!quiet) {
- fprintf(stderr, "mdbx_dump %s (%s, T-%s)\nRunning for %s...\n",
- mdbx_version.git.describe, mdbx_version.git.datetime,
- mdbx_version.git.tree, envname);
- fflush(nullptr);
- }
-
- rc = mdbx_env_create(&env);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_create", rc);
- return EXIT_FAILURE;
- }
-
- if (alldbs || subname) {
- rc = mdbx_env_set_maxdbs(env, 2);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_set_maxdbs", rc);
- goto env_close;
- }
- }
-
- rc = mdbx_env_open(
- env, envname,
- envflags | (rescue ? MDBX_RDONLY | MDBX_EXCLUSIVE : MDBX_RDONLY), 0);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_open", rc);
- goto env_close;
- }
-
- rc = mdbx_txn_begin(env, nullptr, MDBX_TXN_RDONLY, &txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_begin", rc);
- goto env_close;
- }
-
- rc = mdbx_dbi_open(txn, subname, MDBX_DB_ACCEDE, &dbi);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_open", rc);
- goto txn_abort;
- }
-
- if (alldbs) {
- assert(dbi == MAIN_DBI);
-
- MDBX_cursor *cursor;
- rc = mdbx_cursor_open(txn, MAIN_DBI, &cursor);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_cursor_open", rc);
- goto txn_abort;
- }
- if (MDBX_DEBUG > 0 && rescue) {
- cursor->mc_flags |= C_SKIPORD;
- if (cursor->mc_xcursor)
- cursor->mc_xcursor->mx_cursor.mc_flags |= C_SKIPORD;
- }
-
- bool have_raw = false;
- int count = 0;
- MDBX_val key;
- while (MDBX_SUCCESS ==
- (rc = mdbx_cursor_get(cursor, &key, nullptr, MDBX_NEXT_NODUP))) {
- if (user_break) {
- rc = MDBX_EINTR;
- break;
- }
-
- if (memchr(key.iov_base, '\0', key.iov_len))
- continue;
- subname = mdbx_realloc(buf4free, key.iov_len + 1);
- if (!subname) {
- rc = MDBX_ENOMEM;
- break;
- }
-
- buf4free = subname;
- memcpy(subname, key.iov_base, key.iov_len);
- subname[key.iov_len] = '\0';
-
- MDBX_dbi sub_dbi;
- rc = mdbx_dbi_open_ex(txn, subname, MDBX_DB_ACCEDE, &sub_dbi,
- rescue ? equal_or_greater : nullptr,
- rescue ? equal_or_greater : nullptr);
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc == MDBX_INCOMPATIBLE) {
- have_raw = true;
- continue;
- }
- error("mdbx_dbi_open", rc);
- if (!rescue)
- break;
- } else {
- count++;
- if (list) {
- printf("%s\n", subname);
- } else {
- rc = dump_sdb(txn, sub_dbi, subname);
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (!rescue)
- break;
- if (!quiet)
- fprintf(stderr, "%s: %s: ignore %s for `%s` and continue\n", prog,
- envname, mdbx_strerror(rc), subname);
- /* Here is a hack for rescue mode, don't do that:
- * - we should restart transaction in case error due
- * database corruption;
- * - but we won't close cursor, reopen and re-positioning it
- * for new a transaction;
- * - this is possible since DB is opened in read-only exclusive
- * mode and transaction is the same, i.e. has the same address
- * and so on. */
- rc = mdbx_txn_reset(txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_reset", rc);
- goto env_close;
- }
- rc = mdbx_txn_renew(txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_renew", rc);
- goto env_close;
- }
- }
- }
- rc = mdbx_dbi_close(env, sub_dbi);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_close", rc);
- break;
- }
- }
- }
- mdbx_cursor_close(cursor);
- cursor = nullptr;
-
- if (have_raw && (!count /* || rescue */))
- rc = dump_sdb(txn, MAIN_DBI, nullptr);
- else if (!count) {
- if (!quiet)
- fprintf(stderr, "%s: %s does not contain multiple databases\n", prog,
- envname);
- rc = MDBX_NOTFOUND;
- }
- } else {
- rc = dump_sdb(txn, dbi, subname);
- }
-
- switch (rc) {
- case MDBX_NOTFOUND:
- rc = MDBX_SUCCESS;
- case MDBX_SUCCESS:
- break;
- case MDBX_EINTR:
- if (!quiet)
- fprintf(stderr, "Interrupted by signal/user\n");
- break;
- default:
- if (unlikely(rc != MDBX_SUCCESS))
- error("mdbx_cursor_get", rc);
- }
-
- mdbx_dbi_close(env, dbi);
-txn_abort:
- mdbx_txn_abort(txn);
-env_close:
- mdbx_env_close(env);
- free(buf4free);
-
- return rc ? EXIT_FAILURE : EXIT_SUCCESS;
-}
diff --git a/3rdparty/libmdbx/mdbx_load.c b/3rdparty/libmdbx/mdbx_load.c
deleted file mode 100644
index 098ccd536b976d5f1464d49c0cb5d36f22c6f2af..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx_load.c
+++ /dev/null
@@ -1,4391 +0,0 @@
-/* mdbx_load.c - memory-mapped database load tool */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#ifdef _MSC_VER
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#pragma warning(disable : 4996) /* The POSIX name is deprecated... */
-#endif /* _MSC_VER (warnings) */
-
-#define xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#define MDBX_BUILD_SOURCERY c510b93d78b5848764f976ea67f06d278a7b99f141b4c98ceb095285f834f31c_v0_11_7_0_g40ec559c
-#ifdef MDBX_CONFIG_H
-#include MDBX_CONFIG_H
-#endif
-
-#define LIBMDBX_INTERNALS
-#ifdef xMDBX_TOOLS
-#define MDBX_DEPRECATED
-#endif /* xMDBX_TOOLS */
-
-#ifdef xMDBX_ALLOY
-/* Amalgamated build */
-#define MDBX_INTERNAL_FUNC static
-#define MDBX_INTERNAL_VAR static
-#else
-/* Non-amalgamated build */
-#define MDBX_INTERNAL_FUNC
-#define MDBX_INTERNAL_VAR extern
-#endif /* xMDBX_ALLOY */
-
-/** Disables using GNU/Linux libc extensions.
- * \ingroup build_option
- * \note This option couldn't be moved to the options.h since dependant
- * control macros/defined should be prepared before include the options.h */
-#ifndef MDBX_DISABLE_GNU_SOURCE
-#define MDBX_DISABLE_GNU_SOURCE 0
-#endif
-#if MDBX_DISABLE_GNU_SOURCE
-#undef _GNU_SOURCE
-#elif (defined(__linux__) || defined(__gnu_linux__)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif /* MDBX_DISABLE_GNU_SOURCE */
-
-/*----------------------------------------------------------------------------*/
-
-/* Should be defined before any includes */
-#if !defined(_FILE_OFFSET_BITS) && !defined(__ANDROID_API__) && \
- !defined(ANDROID)
-#define _FILE_OFFSET_BITS 64
-#endif
-
-#ifdef __APPLE__
-#define _DARWIN_C_SOURCE
-#endif
-
-#ifdef _MSC_VER
-#if _MSC_FULL_VER < 190024234
-/* Actually libmdbx was not tested with compilers older than 19.00.24234 (Visual
- * Studio 2015 Update 3). But you could remove this #error and try to continue
- * at your own risk. In such case please don't rise up an issues related ONLY to
- * old compilers.
- *
- * NOTE:
- * Unfortunately, there are several different builds of "Visual Studio" that
- * are called "Visual Studio 2015 Update 3".
- *
- * The 190024234 is used here because it is minimal version of Visual Studio
- * that was used for build and testing libmdbx in recent years. Soon this
- * value will be increased to 19.0.24241.7, since build and testing using
- * "Visual Studio 2015" will be performed only at https://ci.appveyor.com.
- *
- * Please ask Microsoft (but not us) for information about version differences
- * and how to and where you can obtain the latest "Visual Studio 2015" build
- * with all fixes.
- */
-#error \
- "At least \"Microsoft C/C++ Compiler\" version 19.00.24234 (Visual Studio 2015 Update 3) is required."
-#endif
-#ifndef _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#if _MSC_VER > 1913
-#pragma warning(disable : 5045) /* Compiler will insert Spectre mitigation... \
- */
-#endif
-#if _MSC_VER > 1914
-#pragma warning( \
- disable : 5105) /* winbase.h(9531): warning C5105: macro expansion \
- producing 'defined' has undefined behavior */
-#endif
-#pragma warning(disable : 4710) /* 'xyz': function not inlined */
-#pragma warning(disable : 4711) /* function 'xyz' selected for automatic \
- inline expansion */
-#pragma warning( \
- disable : 4201) /* nonstandard extension used : nameless struct / union */
-#pragma warning(disable : 4702) /* unreachable code */
-#pragma warning(disable : 4706) /* assignment within conditional expression */
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4324) /* 'xyz': structure was padded due to \
- alignment specifier */
-#pragma warning(disable : 4310) /* cast truncates constant value */
-#pragma warning( \
- disable : 4820) /* bytes padding added after data member for alignment */
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4366) /* the result of the unary '&' operator may be \
- unaligned */
-#pragma warning(disable : 4200) /* nonstandard extension used: zero-sized \
- array in struct/union */
-#pragma warning(disable : 4204) /* nonstandard extension used: non-constant \
- aggregate initializer */
-#pragma warning( \
- disable : 4505) /* unreferenced local function has been removed */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(__GNUC__) && __GNUC__ < 9
-#pragma GCC diagnostic ignored "-Wattributes"
-#endif /* GCC < 9 */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-#include "mdbx.h"
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#ifndef __GNUC_PREREQ
-# if defined(__GNUC__) && defined(__GNUC_MINOR__)
-# define __GNUC_PREREQ(maj, min) \
- ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GNUC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GNUC_PREREQ */
-
-#ifndef __CLANG_PREREQ
-# ifdef __clang__
-# define __CLANG_PREREQ(maj,min) \
- ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min))
-# else
-# define __CLANG_PREREQ(maj,min) (0)
-# endif
-#endif /* __CLANG_PREREQ */
-
-#ifndef __GLIBC_PREREQ
-# if defined(__GLIBC__) && defined(__GLIBC_MINOR__)
-# define __GLIBC_PREREQ(maj, min) \
- ((__GLIBC__ << 16) + __GLIBC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GLIBC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GLIBC_PREREQ */
-
-#ifndef __has_warning
-# define __has_warning(x) (0)
-#endif
-
-#ifndef __has_include
-# define __has_include(x) (0)
-#endif
-
-#if __has_feature(thread_sanitizer)
-# define __SANITIZE_THREAD__ 1
-#endif
-
-#if __has_feature(address_sanitizer)
-# define __SANITIZE_ADDRESS__ 1
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __extern_C
-# ifdef __cplusplus
-# define __extern_C extern "C"
-# else
-# define __extern_C
-# endif
-#endif /* __extern_C */
-
-#if !defined(nullptr) && !defined(__cplusplus) || (__cplusplus < 201103L && !defined(_MSC_VER))
-# define nullptr NULL
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __always_inline
-# if defined(__GNUC__) || __has_attribute(__always_inline__)
-# define __always_inline __inline __attribute__((__always_inline__))
-# elif defined(_MSC_VER)
-# define __always_inline __forceinline
-# else
-# define __always_inline
-# endif
-#endif /* __always_inline */
-
-#ifndef __noinline
-# if defined(__GNUC__) || __has_attribute(__noinline__)
-# define __noinline __attribute__((__noinline__))
-# elif defined(_MSC_VER)
-# define __noinline __declspec(noinline)
-# else
-# define __noinline
-# endif
-#endif /* __noinline */
-
-#ifndef __must_check_result
-# if defined(__GNUC__) || __has_attribute(__warn_unused_result__)
-# define __must_check_result __attribute__((__warn_unused_result__))
-# else
-# define __must_check_result
-# endif
-#endif /* __must_check_result */
-
-#if !defined(__noop) && !defined(_MSC_VER)
-# define __noop(...) do {} while(0)
-#endif /* __noop */
-
-#if defined(__fallthrough) && \
- (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__))
-#undef __fallthrough
-#endif /* __fallthrough workaround for MinGW */
-
-#ifndef __fallthrough
-# if defined(__cplusplus) && (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || __cplusplus >= 201703L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(8, 0) && defined(__cplusplus) && __cplusplus >= 201103L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(7, 0) && \
- (!defined(__LCC__) || (__LCC__ == 124 && __LCC_MINOR__ >= 12) || \
- (__LCC__ == 125 && __LCC_MINOR__ >= 5) || (__LCC__ >= 126))
-# define __fallthrough __attribute__((__fallthrough__))
-# elif defined(__clang__) && defined(__cplusplus) && __cplusplus >= 201103L &&\
- __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
-# define __fallthrough [[clang::fallthrough]]
-# else
-# define __fallthrough
-# endif
-#endif /* __fallthrough */
-
-#ifndef __unreachable
-# if __GNUC_PREREQ(4,5) || __has_builtin(__builtin_unreachable)
-# define __unreachable() __builtin_unreachable()
-# elif defined(_MSC_VER)
-# define __unreachable() __assume(0)
-# else
-# define __unreachable() __noop()
-# endif
-#endif /* __unreachable */
-
-#ifndef __prefetch
-# if defined(__GNUC__) || defined(__clang__) || __has_builtin(__builtin_prefetch)
-# define __prefetch(ptr) __builtin_prefetch(ptr)
-# else
-# define __prefetch(ptr) __noop(ptr)
-# endif
-#endif /* __prefetch */
-
-#ifndef __nothrow
-# if defined(__cplusplus)
-# if __cplusplus < 201703L
-# define __nothrow throw()
-# else
-# define __nothrow noexcept(true)
-# endif /* __cplusplus */
-# elif defined(__GNUC__) || __has_attribute(__nothrow__)
-# define __nothrow __attribute__((__nothrow__))
-# elif defined(_MSC_VER) && defined(__cplusplus)
-# define __nothrow __declspec(nothrow)
-# else
-# define __nothrow
-# endif
-#endif /* __nothrow */
-
-#ifndef __hidden
-# if defined(__GNUC__) || __has_attribute(__visibility__)
-# define __hidden __attribute__((__visibility__("hidden")))
-# else
-# define __hidden
-# endif
-#endif /* __hidden */
-
-#ifndef __optimize
-# if defined(__OPTIMIZE__)
-# if (defined(__GNUC__) && !defined(__clang__)) || __has_attribute(__optimize__)
-# define __optimize(ops) __attribute__((__optimize__(ops)))
-# else
-# define __optimize(ops)
-# endif
-# else
-# define __optimize(ops)
-# endif
-#endif /* __optimize */
-
-#ifndef __hot
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __hot __attribute__((__hot__)) __optimize(3)
-# elif defined(__clang__) && !__has_attribute(__hot_) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put frequently used functions in separate section */
-# define __hot __attribute__((__section__("text.hot"))) __optimize("O3")
-# elif defined(__GNUC__) || __has_attribute(__hot__)
-# define __hot __attribute__((__hot__)) __optimize("O3")
-# else
-# define __hot __optimize("O3")
-# endif
-# else
-# define __hot
-# endif
-#endif /* __hot */
-
-#ifndef __cold
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __cold __attribute__((__cold__)) __optimize(1)
-# elif defined(__clang__) && !__has_attribute(cold) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put infrequently used functions in separate section */
-# define __cold __attribute__((__section__("text.unlikely"))) __optimize("Os")
-# elif defined(__GNUC__) || __has_attribute(cold)
-# define __cold __attribute__((__cold__)) __optimize("Os")
-# else
-# define __cold __optimize("Os")
-# endif
-# else
-# define __cold
-# endif
-#endif /* __cold */
-
-#ifndef __flatten
-# if defined(__OPTIMIZE__) && (defined(__GNUC__) || __has_attribute(__flatten__))
-# define __flatten __attribute__((__flatten__))
-# else
-# define __flatten
-# endif
-#endif /* __flatten */
-
-#ifndef likely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define likely(cond) __builtin_expect(!!(cond), 1)
-# else
-# define likely(x) (!!(x))
-# endif
-#endif /* likely */
-
-#ifndef unlikely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define unlikely(cond) __builtin_expect(!!(cond), 0)
-# else
-# define unlikely(x) (!!(x))
-# endif
-#endif /* unlikely */
-
-#ifndef __anonymous_struct_extension__
-# if defined(__GNUC__)
-# define __anonymous_struct_extension__ __extension__
-# else
-# define __anonymous_struct_extension__
-# endif
-#endif /* __anonymous_struct_extension__ */
-
-#ifndef __Wpedantic_format_voidptr
- MDBX_MAYBE_UNUSED MDBX_PURE_FUNCTION static __inline const void*
- __Wpedantic_format_voidptr(const void* ptr) {return ptr;}
-# define __Wpedantic_format_voidptr(ARG) __Wpedantic_format_voidptr(ARG)
-#endif /* __Wpedantic_format_voidptr */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(MDBX_USE_VALGRIND)
-# include <valgrind/memcheck.h>
-# ifndef VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE
- /* LY: available since Valgrind 3.10 */
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# endif
-#elif !defined(RUNNING_ON_VALGRIND)
-# define VALGRIND_CREATE_MEMPOOL(h,r,z)
-# define VALGRIND_DESTROY_MEMPOOL(h)
-# define VALGRIND_MEMPOOL_TRIM(h,a,s)
-# define VALGRIND_MEMPOOL_ALLOC(h,a,s)
-# define VALGRIND_MEMPOOL_FREE(h,a)
-# define VALGRIND_MEMPOOL_CHANGE(h,a,b,s)
-# define VALGRIND_MAKE_MEM_NOACCESS(a,s)
-# define VALGRIND_MAKE_MEM_DEFINED(a,s)
-# define VALGRIND_MAKE_MEM_UNDEFINED(a,s)
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_CHECK_MEM_IS_ADDRESSABLE(a,s) (0)
-# define VALGRIND_CHECK_MEM_IS_DEFINED(a,s) (0)
-# define RUNNING_ON_VALGRIND (0)
-#endif /* MDBX_USE_VALGRIND */
-
-#ifdef __SANITIZE_ADDRESS__
-# include <sanitizer/asan_interface.h>
-#elif !defined(ASAN_POISON_MEMORY_REGION)
-# define ASAN_POISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-# define ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-#endif /* __SANITIZE_ADDRESS__ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef ARRAY_LENGTH
-# ifdef __cplusplus
- template <typename T, size_t N>
- char (&__ArraySizeHelper(T (&array)[N]))[N];
-# define ARRAY_LENGTH(array) (sizeof(::__ArraySizeHelper(array)))
-# else
-# define ARRAY_LENGTH(array) (sizeof(array) / sizeof(array[0]))
-# endif
-#endif /* ARRAY_LENGTH */
-
-#ifndef ARRAY_END
-# define ARRAY_END(array) (&array[ARRAY_LENGTH(array)])
-#endif /* ARRAY_END */
-
-#define CONCAT(a,b) a##b
-#define XCONCAT(a,b) CONCAT(a,b)
-
-#ifndef offsetof
-# define offsetof(type, member) __builtin_offsetof(type, member)
-#endif /* offsetof */
-
-#ifndef container_of
-# define container_of(ptr, type, member) \
- ((type *)((char *)(ptr) - offsetof(type, member)))
-#endif /* container_of */
-
-#define MDBX_TETRAD(a, b, c, d) \
- ((uint32_t)(a) << 24 | (uint32_t)(b) << 16 | (uint32_t)(c) << 8 | (d))
-
-#define MDBX_STRING_TETRAD(str) MDBX_TETRAD(str[0], str[1], str[2], str[3])
-
-#define FIXME "FIXME: " __FILE__ ", " MDBX_STRINGIFY(__LINE__)
-
-#ifndef STATIC_ASSERT_MSG
-# if defined(static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) static_assert(expr, msg)
-# elif defined(_STATIC_ASSERT)
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif defined(_MSC_VER)
-# include <crtdbg.h>
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) \
- || __has_feature(c_static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) _Static_assert(expr, msg)
-# else
-# define STATIC_ASSERT_MSG(expr, msg) switch (0) {case 0:case (expr):;}
-# endif
-#endif /* STATIC_ASSERT */
-
-#ifndef STATIC_ASSERT
-# define STATIC_ASSERT(expr) STATIC_ASSERT_MSG(expr, #expr)
-#endif
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__GNUC__) && !__GNUC_PREREQ(4, 2)
-/* Actually libmdbx was not tested with compilers older than GCC 4.2.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required GCC >= 4.2"
-#endif
-
-#if defined(__clang__) && !__CLANG_PREREQ(3, 8)
-/* Actually libmdbx was not tested with CLANG older than 3.8.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required CLANG >= 3.8"
-#endif
-
-#if defined(__GLIBC__) && !__GLIBC_PREREQ(2, 12)
-/* Actually libmdbx was not tested with something older than glibc 2.12.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old systems.
- */
-#warning "libmdbx was only tested with GLIBC >= 2.12."
-#endif
-
-#ifdef __SANITIZE_THREAD__
-#warning \
- "libmdbx don't compatible with ThreadSanitizer, you will get a lot of false-positive issues."
-#endif /* __SANITIZE_THREAD__ */
-
-#if __has_warning("-Wnested-anon-types")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wnested-anon-types"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wnested-anon-types"
-#else
-#pragma warning disable "nested-anon-types"
-#endif
-#endif /* -Wnested-anon-types */
-
-#if __has_warning("-Wconstant-logical-operand")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wconstant-logical-operand"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wconstant-logical-operand"
-#else
-#pragma warning disable "constant-logical-operand"
-#endif
-#endif /* -Wconstant-logical-operand */
-
-#if defined(__LCC__) && (__LCC__ <= 121)
-/* bug #2798 */
-#pragma diag_suppress alignment_reduction_ignored
-#elif defined(__ICC)
-#pragma warning(disable : 3453 1366)
-#elif __has_warning("-Walignment-reduction-ignored")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Walignment-reduction-ignored"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Walignment-reduction-ignored"
-#else
-#pragma warning disable "alignment-reduction-ignored"
-#endif
-#endif /* -Walignment-reduction-ignored */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(_WIN32) || defined(_WIN64)
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if !defined(_NO_CRT_STDIO_INLINE) && MDBX_BUILD_SHARED_LIBRARY && \
- !defined(xMDBX_TOOLS) && MDBX_WITHOUT_MSVC_CRT
-#define _NO_CRT_STDIO_INLINE
-#endif
-#elif !defined(_POSIX_C_SOURCE)
-#define _POSIX_C_SOURCE 200809L
-#endif /* Windows */
-
-/*----------------------------------------------------------------------------*/
-/* C99 includes */
-#include <inttypes.h>
-#include <stddef.h>
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <assert.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-/* C11' alignas() */
-#if __has_include(<stdalign.h>)
-#include <stdalign.h>
-#endif
-#if defined(alignas) || defined(__cplusplus)
-#define MDBX_ALIGNAS(N) alignas(N)
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
-#define MDBX_ALIGNAS(N) _Alignas(N)
-#elif defined(_MSC_VER)
-#define MDBX_ALIGNAS(N) __declspec(align(N))
-#elif __has_attribute(__aligned__) || defined(__GNUC__)
-#define MDBX_ALIGNAS(N) __attribute__((__aligned__(N)))
-#else
-#error "FIXME: Required alignas() or equivalent."
-#endif /* MDBX_ALIGNAS */
-
-/*----------------------------------------------------------------------------*/
-/* Systems includes */
-
-#ifdef __APPLE__
-#include <TargetConditionals.h>
-#endif /* Apple OSX & iOS */
-
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__)
-#include <sys/cdefs.h>
-#include <sys/mount.h>
-#include <sys/sysctl.h>
-#include <sys/types.h>
-#if defined(__FreeBSD__) || defined(__DragonFly__)
-#include <vm/vm_param.h>
-#elif defined(__OpenBSD__) || defined(__NetBSD__)
-#include <uvm/uvm_param.h>
-#else
-#define SYSCTL_LEGACY_NONCONST_MIB
-#endif
-#ifndef __MACH__
-#include <sys/vmmeter.h>
-#endif
-#else
-#include <malloc.h>
-#if !(defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(_WIN32) || defined(_WIN64))
-#include <mntent.h>
-#endif /* !Solaris */
-#endif /* !xBSD */
-
-#if defined(__FreeBSD__) || __has_include(<malloc_np.h>)
-#include <malloc_np.h>
-#endif
-
-#if defined(__APPLE__) || defined(__MACH__) || __has_include(<malloc/malloc.h>)
-#include <malloc/malloc.h>
-#endif /* MacOS */
-
-#if defined(__MACH__)
-#include <mach/host_info.h>
-#include <mach/mach_host.h>
-#include <mach/mach_port.h>
-#include <uuid/uuid.h>
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-#include <sched.h>
-#include <sys/sendfile.h>
-#include <sys/statfs.h>
-#endif /* Linux */
-
-#ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 0
-#endif
-
-#ifndef _XOPEN_SOURCE_EXTENDED
-#define _XOPEN_SOURCE_EXTENDED 0
-#else
-#include <utmpx.h>
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#include <kstat.h>
-#include <sys/mnttab.h>
-/* On Solaris, it's easier to add a missing prototype rather than find a
- * combination of #defines that break nothing. */
-__extern_C key_t ftok(const char *, int);
-#endif /* SunOS/Solaris */
-
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef _WIN32_WINNT
-#define _WIN32_WINNT 0x0601 /* Windows 7 */
-#elif _WIN32_WINNT < 0x0500
-#error At least 'Windows 2000' API is required for libmdbx.
-#endif /* _WIN32_WINNT */
-#if (defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* MinGW */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif /* WIN32_LEAN_AND_MEAN */
-#include <excpt.h>
-#include <tlhelp32.h>
-#include <windows.h>
-#include <winnt.h>
-#include <winternl.h>
-#define HAVE_SYS_STAT_H
-#define HAVE_SYS_TYPES_H
-typedef HANDLE mdbx_thread_t;
-typedef unsigned mdbx_thread_key_t;
-#define MAP_FAILED NULL
-#define HIGH_DWORD(v) ((DWORD)((sizeof(v) > 4) ? ((uint64_t)(v) >> 32) : 0))
-#define THREAD_CALL WINAPI
-#define THREAD_RESULT DWORD
-typedef struct {
- HANDLE mutex;
- HANDLE event[2];
-} mdbx_condpair_t;
-typedef CRITICAL_SECTION mdbx_fastmutex_t;
-
-#if !defined(_MSC_VER) && !defined(__try)
-/* *INDENT-OFF* */
-/* clang-format off */
-#define __try
-#define __except(COND) if(false)
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* stub for MSVC's __try/__except */
-
-#if MDBX_WITHOUT_MSVC_CRT
-
-#ifndef mdbx_malloc
-static inline void *mdbx_malloc(size_t bytes) {
- return HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_malloc */
-
-#ifndef mdbx_calloc
-static inline void *mdbx_calloc(size_t nelem, size_t size) {
- return HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, nelem * size);
-}
-#endif /* mdbx_calloc */
-
-#ifndef mdbx_realloc
-static inline void *mdbx_realloc(void *ptr, size_t bytes) {
- return ptr ? HeapReAlloc(GetProcessHeap(), 0, ptr, bytes)
- : HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_realloc */
-
-#ifndef mdbx_free
-static inline void mdbx_free(void *ptr) { HeapFree(GetProcessHeap(), 0, ptr); }
-#endif /* mdbx_free */
-
-#else /* MDBX_WITHOUT_MSVC_CRT */
-
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup _strdup
-
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-#ifndef snprintf
-#define snprintf _snprintf /* ntdll */
-#endif
-
-#ifndef vsnprintf
-#define vsnprintf _vsnprintf /* ntdll */
-#endif
-
-#else /*----------------------------------------------------------------------*/
-
-#include <unistd.h>
-#if !defined(_POSIX_MAPPED_FILES) || _POSIX_MAPPED_FILES < 1
-#error "libmdbx requires the _POSIX_MAPPED_FILES feature"
-#endif /* _POSIX_MAPPED_FILES */
-
-#include <pthread.h>
-#include <semaphore.h>
-#include <signal.h>
-#include <sys/file.h>
-#include <sys/ipc.h>
-#include <sys/mman.h>
-#include <sys/param.h>
-#include <sys/stat.h>
-#include <sys/statvfs.h>
-#include <sys/uio.h>
-typedef pthread_t mdbx_thread_t;
-typedef pthread_key_t mdbx_thread_key_t;
-#define INVALID_HANDLE_VALUE (-1)
-#define THREAD_CALL
-#define THREAD_RESULT void *
-typedef struct {
- pthread_mutex_t mutex;
- pthread_cond_t cond[2];
-} mdbx_condpair_t;
-typedef pthread_mutex_t mdbx_fastmutex_t;
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup strdup
-#endif /* Platform */
-
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
-/* malloc_usable_size() already provided */
-#elif defined(__APPLE__)
-#define malloc_usable_size(ptr) malloc_size(ptr)
-#elif defined(_MSC_VER) && !MDBX_WITHOUT_MSVC_CRT
-#define malloc_usable_size(ptr) _msize(ptr)
-#endif /* malloc_usable_size */
-
-#ifdef __ANDROID_API__
-#include <android/log.h>
-#if __ANDROID_API__ >= 21
-#include <sys/sendfile.h>
-#endif
-#endif /* Android */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(HAVE_SYS_STAT_H) || __has_include(<sys/stat.h>)
-#include <sys/stat.h>
-#endif
-#if defined(HAVE_SYS_TYPES_H) || __has_include(<sys/types.h>)
-#include <sys/types.h>
-#endif
-#if defined(HAVE_SYS_FILE_H) || __has_include(<sys/file.h>)
-#include <sys/file.h>
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-
-#ifndef SSIZE_MAX
-#define SSIZE_MAX INTPTR_MAX
-#endif
-
-#if !defined(MADV_DODUMP) && defined(MADV_CORE)
-#define MADV_DODUMP MADV_CORE
-#endif /* MADV_CORE -> MADV_DODUMP */
-
-#if !defined(MADV_DONTDUMP) && defined(MADV_NOCORE)
-#define MADV_DONTDUMP MADV_NOCORE
-#endif /* MADV_NOCORE -> MADV_DONTDUMP */
-
-#if defined(i386) || defined(__386) || defined(__i386) || defined(__i386__) || \
- defined(i486) || defined(__i486) || defined(__i486__) || \
- defined(i586) | defined(__i586) || defined(__i586__) || defined(i686) || \
- defined(__i686) || defined(__i686__) || defined(_M_IX86) || \
- defined(_X86_) || defined(__THW_INTEL__) || defined(__I86__) || \
- defined(__INTEL__) || defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64__) || defined(__amd64) || defined(_M_X64) || \
- defined(_M_AMD64) || defined(__IA32__) || defined(__INTEL__)
-#ifndef __ia32__
-/* LY: define neutral __ia32__ for x86 and x86-64 */
-#define __ia32__ 1
-#endif /* __ia32__ */
-#if !defined(__amd64__) && (defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64) || defined(_M_X64))
-/* LY: define trusty __amd64__ for all AMD64/x86-64 arch */
-#define __amd64__ 1
-#endif /* __amd64__ */
-#endif /* all x86 */
-
-#if (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
-#error \
- "Sanity checking failed: Two's complement, reasonably sized integer types"
-#endif
-
-#if UINTPTR_MAX > 0xffffFFFFul || ULONG_MAX > 0xffffFFFFul
-#define MDBX_WORDBITS 64
-#else
-#define MDBX_WORDBITS 32
-#endif /* MDBX_WORDBITS */
-
-#if defined(__ANDROID_API__) || defined(ANDROID)
-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "_FILE_OFFSET_BITS != MDBX_WORDBITS" (_FILE_OFFSET_BITS != MDBX_WORDBITS)
-#elif defined(__FILE_OFFSET_BITS) && __FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "__FILE_OFFSET_BITS != MDBX_WORDBITS" (__FILE_OFFSET_BITS != MDBX_WORDBITS)
-#endif
-#endif /* Android */
-
-/*----------------------------------------------------------------------------*/
-/* Compiler's includes for builtins/intrinsics */
-
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
-#include <intrin.h>
-#elif __GNUC_PREREQ(4, 4) || defined(__clang__)
-#if defined(__ia32__) || defined(__e2k__)
-#include <x86intrin.h>
-#endif /* __ia32__ */
-#if defined(__ia32__)
-#include <cpuid.h>
-#endif /* __ia32__ */
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
-#include <mbarrier.h>
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
-#include <machine/sys/inline.h>
-#elif defined(__IBMC__) && defined(__powerpc)
-#include <atomic.h>
-#elif defined(_AIX)
-#include <builtins.h>
-#include <sys/atomic_op.h>
-#elif (defined(__osf__) && defined(__DECC)) || defined(__alpha)
-#include <c_asm.h>
-#include <machine/builtins.h>
-#elif defined(__MWERKS__)
-/* CodeWarrior - troubles ? */
-#pragma gcc_extensions
-#elif defined(__SNC__)
-/* Sony PS3 - troubles ? */
-#elif defined(__hppa__) || defined(__hppa)
-#include <machine/inline.h>
-#else
-#error Unsupported C compiler, please use GNU C 4.4 or newer
-#endif /* Compiler */
-
-/*----------------------------------------------------------------------------*/
-/* Byteorder */
-
-#if !defined(__BYTE_ORDER__) || !defined(__ORDER_LITTLE_ENDIAN__) || \
- !defined(__ORDER_BIG_ENDIAN__)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(__GLIBC__) || defined(__GNU_LIBRARY__) || defined(__ANDROID_API__) || \
- defined(HAVE_ENDIAN_H) || __has_include(<endian.h>)
-#include <endian.h>
-#elif defined(__APPLE__) || defined(__MACH__) || defined(__OpenBSD__) || \
- defined(HAVE_MACHINE_ENDIAN_H) || __has_include(<machine/endian.h>)
-#include <machine/endian.h>
-#elif defined(HAVE_SYS_ISA_DEFS_H) || __has_include(<sys/isa_defs.h>)
-#include <sys/isa_defs.h>
-#elif (defined(HAVE_SYS_TYPES_H) && defined(HAVE_SYS_ENDIAN_H)) || \
- (__has_include(<sys/types.h>) && __has_include(<sys/endian.h>))
-#include <sys/endian.h>
-#include <sys/types.h>
-#elif defined(__bsdi__) || defined(__DragonFly__) || defined(__FreeBSD__) || \
- defined(__NetBSD__) || \
- defined(HAVE_SYS_PARAM_H) || __has_include(<sys/param.h>)
-#include <sys/param.h>
-#endif /* OS */
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-
-#else
-#error __BYTE_ORDER__ should be defined.
-#endif /* Arch */
-
-#endif
-#endif /* __BYTE_ORDER__ || __ORDER_LITTLE_ENDIAN__ || __ORDER_BIG_ENDIAN__ */
-
-/* Get the size of a memory page for the system.
- * This is the basic size that the platform's memory manager uses, and is
- * fundamental to the use of memory-mapped files. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline size_t
-mdbx_syspagesize(void) {
-#if defined(_WIN32) || defined(_WIN64)
- SYSTEM_INFO si;
- GetSystemInfo(&si);
- return si.dwPageSize;
-#else
- return sysconf(_SC_PAGE_SIZE);
-#endif
-}
-
-typedef struct mdbx_mmap_param {
- union {
- void *address;
- uint8_t *dxb;
- struct MDBX_lockinfo *lck;
- };
- mdbx_filehandle_t fd;
- size_t limit; /* mapping length, but NOT a size of file nor DB */
- size_t current; /* mapped region size, i.e. the size of file and DB */
- uint64_t filesize /* in-process cache of a file size */;
-#if defined(_WIN32) || defined(_WIN64)
- HANDLE section; /* memory-mapped section handle */
-#endif
-} mdbx_mmap_t;
-
-typedef union bin128 {
- __anonymous_struct_extension__ struct { uint64_t x, y; };
- __anonymous_struct_extension__ struct { uint32_t a, b, c, d; };
-} bin128_t;
-
-#if defined(_WIN32) || defined(_WIN64)
-typedef union MDBX_srwlock {
- __anonymous_struct_extension__ struct {
- long volatile readerCount;
- long volatile writerCount;
- };
- RTL_SRWLOCK native;
-} MDBX_srwlock;
-#endif /* Windows */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny);
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny);
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-#if defined(__cplusplus) && !defined(__STDC_NO_ATOMICS__) && (__has_include(<cstdatomic>) || __has_extension(cxx_atomic))
-#include <cstdatomic>
-#define MDBX_HAVE_C11ATOMICS
-#elif !defined(__cplusplus) && \
- (__STDC_VERSION__ >= 201112L || __has_extension(c_atomic)) && \
- !defined(__STDC_NO_ATOMICS__) && \
- (__GNUC_PREREQ(4, 9) || __CLANG_PREREQ(3, 8) || \
- !(defined(__GNUC__) || defined(__clang__)))
-#include <stdatomic.h>
-#define MDBX_HAVE_C11ATOMICS
-#elif defined(__GNUC__) || defined(__clang__)
-#elif defined(_MSC_VER)
-#pragma warning(disable : 4163) /* 'xyz': not available as an intrinsic */
-#pragma warning(disable : 4133) /* 'function': incompatible types - from \
- 'size_t' to 'LONGLONG' */
-#pragma warning(disable : 4244) /* 'return': conversion from 'LONGLONG' to \
- 'std::size_t', possible loss of data */
-#pragma warning(disable : 4267) /* 'function': conversion from 'size_t' to \
- 'long', possible loss of data */
-#pragma intrinsic(_InterlockedExchangeAdd, _InterlockedCompareExchange)
-#pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedCompareExchange64)
-#elif defined(__APPLE__)
-#include <libkern/OSAtomic.h>
-#else
-#error FIXME atomic-ops
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Memory/Compiler barriers, cache coherence */
-
-#if __has_include(<sys/cachectl.h>)
-#include <sys/cachectl.h>
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS should have explicit cache control */
-#include <sys/cachectl.h>
-#endif
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_compiler_barrier(void) {
-#if defined(__clang__) || defined(__GNUC__)
- __asm__ __volatile__("" ::: "memory");
-#elif defined(_MSC_VER)
- _ReadWriteBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
- __memory_barrier();
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __compiler_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_sched_fence(/* LY: no-arg meaning 'all expect ALU', e.g. 0x3D3D */);
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __fence();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_memory_barrier(void) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(memory_order_seq_cst);
-#elif defined(__ATOMIC_SEQ_CST)
-#ifdef __clang__
- __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
- __atomic_thread_fence(__ATOMIC_SEQ_CST);
-#endif
-#elif defined(__clang__) || defined(__GNUC__)
- __sync_synchronize();
-#elif defined(_WIN32) || defined(_WIN64)
- MemoryBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
-#if defined(__ia32__)
- _mm_mfence();
-#else
- __mf();
-#endif
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __machine_rw_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_mf();
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __lwsync();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* libc compatibility stuff */
-
-#if (!defined(__GLIBC__) && __GLIBC_PREREQ(2, 1)) && \
- (defined(_GNU_SOURCE) || defined(_BSD_SOURCE))
-#define mdbx_asprintf asprintf
-#define mdbx_vasprintf vasprintf
-#else
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC
- MDBX_PRINTF_ARGS(2, 3) int mdbx_asprintf(char **strp, const char *fmt, ...);
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt, va_list ap);
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* OS abstraction layer stuff */
-
-/* max bytes to write in one call */
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_WRITE UINT32_C(0x01000000)
-#else
-#define MAX_WRITE UINT32_C(0x3fff0000)
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-MDBX_INTERNAL_VAR uint32_t mdbx_linux_kernel_version;
-MDBX_INTERNAL_VAR bool mdbx_RunningOnWSL1 /* Windows Subsystem 1 for Linux */;
-#endif /* Linux */
-
-#ifndef mdbx_strdup
-LIBMDBX_API char *mdbx_strdup(const char *str);
-#endif
-
-MDBX_MAYBE_UNUSED static __inline int mdbx_get_errno(void) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD rc = GetLastError();
-#else
- int rc = errno;
-#endif
- return rc;
-}
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result);
-#endif
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr);
-#endif
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair, bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair);
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex);
-
-MDBX_INTERNAL_FUNC int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov,
- int iovcnt, uint64_t offset,
- size_t expected_written);
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t count,
- uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t count, uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t count);
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg);
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread);
-
-enum mdbx_syncmode_bits {
- MDBX_SYNC_NONE = 0,
- MDBX_SYNC_DATA = 1,
- MDBX_SYNC_SIZE = 2,
- MDBX_SYNC_IODQ = 4
-};
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- const enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length);
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos);
-MDBX_INTERNAL_FUNC int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length);
-
-enum mdbx_openfile_purpose {
- MDBX_OPEN_DXB_READ = 0,
- MDBX_OPEN_DXB_LAZY = 1,
- MDBX_OPEN_DXB_DSYNC = 2,
- MDBX_OPEN_LCK = 3,
- MDBX_OPEN_COPY = 4,
- MDBX_OPEN_DELETE = 5
-};
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait);
-
-#define MMAP_OPTION_TRUNCATE 1
-#define MMAP_OPTION_SEMAPHORE 2
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t must, const size_t limit,
- const unsigned options);
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map);
-#define MDBX_MRESIZE_MAY_MOVE 0x00000100
-#define MDBX_MRESIZE_MAY_UNMAP 0x00000200
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit);
-#if defined(_WIN32) || defined(_WIN64)
-typedef struct {
- unsigned limit, count;
- HANDLE handles[31];
-} mdbx_handle_array_t;
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array);
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array);
-#endif /* Windows */
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err);
-
-MDBX_MAYBE_UNUSED static __inline uint32_t mdbx_getpid(void) {
- STATIC_ASSERT(sizeof(mdbx_pid_t) <= sizeof(uint32_t));
-#if defined(_WIN32) || defined(_WIN64)
- return GetCurrentProcessId();
-#else
- STATIC_ASSERT(sizeof(pid_t) <= sizeof(uint32_t));
- return getpid();
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline uintptr_t mdbx_thread_self(void) {
- mdbx_tid_t thunk;
- STATIC_ASSERT(sizeof(uintptr_t) >= sizeof(thunk));
-#if defined(_WIN32) || defined(_WIN64)
- thunk = GetCurrentThreadId();
-#else
- thunk = pthread_self();
-#endif
- return (uintptr_t)thunk;
-}
-
-#if !defined(_WIN32) && !defined(_WIN64)
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void);
-#else
-static __inline int mdbx_check_tid4bionic(void) { return 0; }
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-MDBX_MAYBE_UNUSED static __inline int
-mdbx_pthread_mutex_lock(pthread_mutex_t *mutex) {
- int err = mdbx_check_tid4bionic();
- return unlikely(err) ? err : pthread_mutex_lock(mutex);
-}
-#endif /* !Windows */
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void);
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16);
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime);
-
-MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void);
-/*----------------------------------------------------------------------------*/
-/* lck stuff */
-
-/// \brief Initialization of synchronization primitives linked with MDBX_env
-/// instance both in LCK-file and within the current process.
-/// \param
-/// global_uniqueness_flag = true - denotes that there are no other processes
-/// working with DB and LCK-file. Thus the function MUST initialize
-/// shared synchronization objects in memory-mapped LCK-file.
-/// global_uniqueness_flag = false - denotes that at least one process is
-/// already working with DB and LCK-file, including the case when DB
-/// has already been opened in the current process. Thus the function
-/// MUST NOT initialize shared synchronization objects in memory-mapped
-/// LCK-file that are already in use.
-/// \return Error code or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag);
-
-/// \brief Disconnects from shared interprocess objects and destructs
-/// synchronization objects linked with MDBX_env instance
-/// within the current process.
-/// \param
-/// inprocess_neighbor = NULL - if the current process does not have other
-/// instances of MDBX_env linked with the DB being closed.
-/// Thus the function MUST check for other processes working with DB or
-/// LCK-file, and keep or destroy shared synchronization objects in
-/// memory-mapped LCK-file depending on the result.
-/// inprocess_neighbor = not-NULL - pointer to another instance of MDBX_env
-/// (anyone of there is several) working with DB or LCK-file within the
-/// current process. Thus the function MUST NOT try to acquire exclusive
-/// lock and/or try to destruct shared synchronization objects linked with
-/// DB or LCK-file. Moreover, the implementation MUST ensure correct work
-/// of other instances of MDBX_env within the current process, e.g.
-/// restore POSIX-fcntl locks after the closing of file descriptors.
-/// \return Error code (MDBX_PANIC) or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor);
-
-/// \brief Connects to shared interprocess locking objects and tries to acquire
-/// the maximum lock level (shared if exclusive is not available)
-/// Depending on implementation or/and platform (Windows) this function may
-/// acquire the non-OS super-level lock (e.g. for shared synchronization
-/// objects initialization), which will be downgraded to OS-exclusive or
-/// shared via explicit calling of mdbx_lck_downgrade().
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if an exclusive lock was acquired and thus
-/// the current process is the first and only after the last use of DB.
-/// MDBX_RESULT_FALSE (0) - if a shared lock was acquired and thus
-/// DB has already been opened and now is used by other processes.
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env);
-
-/// \brief Downgrades the level of initially acquired lock to
-/// operational level specified by argument. The reson for such downgrade:
-/// - unblocking of other processes that are waiting for access, i.e.
-/// if (env->me_flags & MDBX_EXCLUSIVE) != 0, then other processes
-/// should be made aware that access is unavailable rather than
-/// wait for it.
-/// - freeing locks that interfere file operation (especially for Windows)
-/// (env->me_flags & MDBX_EXCLUSIVE) == 0 - downgrade to shared lock.
-/// (env->me_flags & MDBX_EXCLUSIVE) != 0 - downgrade to exclusive
-/// operational lock.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env);
-
-/// \brief Locks LCK-file or/and table of readers for (de)registering.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env);
-
-/// \brief Unlocks LCK-file or/and table of readers after (de)registering.
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env);
-
-/// \brief Acquires lock for DB change (on writing transaction start)
-/// Reading transactions will not be blocked.
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-/// \return Error code or zero on success
-LIBMDBX_API int mdbx_txn_lock(MDBX_env *env, bool dont_wait);
-
-/// \brief Releases lock once DB changes is made (after writing transaction
-/// has finished).
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-LIBMDBX_API void mdbx_txn_unlock(MDBX_env *env);
-
-/// \brief Sets alive-flag of reader presence (indicative lock) for PID of
-/// the current process. The function does no more than needed for
-/// the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env);
-
-/// \brief Resets alive-flag of reader presence (indicative lock)
-/// for PID of the current process. The function does no more than needed
-/// for the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env);
-
-/// \brief Checks for reading process status with the given pid with help of
-/// alive-flag of presence (indicative lock) or using another way.
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if the reader process with the given PID is alive
-/// and working with DB (indicative lock is present).
-/// MDBX_RESULT_FALSE (0) - if the reader process with the given PID is absent
-/// or not working with DB (indicative lock is not present).
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid);
-
-#if defined(_WIN32) || defined(_WIN64)
-
-typedef void(WINAPI *MDBX_srwlock_function)(MDBX_srwlock *);
-MDBX_INTERNAL_VAR MDBX_srwlock_function mdbx_srwlock_Init,
- mdbx_srwlock_AcquireShared, mdbx_srwlock_ReleaseShared,
- mdbx_srwlock_AcquireExclusive, mdbx_srwlock_ReleaseExclusive;
-
-#if _WIN32_WINNT < 0x0600 /* prior to Windows Vista */
-typedef enum _FILE_INFO_BY_HANDLE_CLASS {
- FileBasicInfo,
- FileStandardInfo,
- FileNameInfo,
- FileRenameInfo,
- FileDispositionInfo,
- FileAllocationInfo,
- FileEndOfFileInfo,
- FileStreamInfo,
- FileCompressionInfo,
- FileAttributeTagInfo,
- FileIdBothDirectoryInfo,
- FileIdBothDirectoryRestartInfo,
- FileIoPriorityHintInfo,
- FileRemoteProtocolInfo,
- MaximumFileInfoByHandleClass
-} FILE_INFO_BY_HANDLE_CLASS,
- *PFILE_INFO_BY_HANDLE_CLASS;
-
-typedef struct _FILE_END_OF_FILE_INFO {
- LARGE_INTEGER EndOfFile;
-} FILE_END_OF_FILE_INFO, *PFILE_END_OF_FILE_INFO;
-
-#define REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK 0x00000001
-#define REMOTE_PROTOCOL_INFO_FLAG_OFFLINE 0x00000002
-
-typedef struct _FILE_REMOTE_PROTOCOL_INFO {
- USHORT StructureVersion;
- USHORT StructureSize;
- DWORD Protocol;
- USHORT ProtocolMajorVersion;
- USHORT ProtocolMinorVersion;
- USHORT ProtocolRevision;
- USHORT Reserved;
- DWORD Flags;
- struct {
- DWORD Reserved[8];
- } GenericReserved;
- struct {
- DWORD Reserved[16];
- } ProtocolSpecificReserved;
-} FILE_REMOTE_PROTOCOL_INFO, *PFILE_REMOTE_PROTOCOL_INFO;
-
-#endif /* _WIN32_WINNT < 0x0600 (prior to Windows Vista) */
-
-typedef BOOL(WINAPI *MDBX_GetFileInformationByHandleEx)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_GetFileInformationByHandleEx
- mdbx_GetFileInformationByHandleEx;
-
-typedef BOOL(WINAPI *MDBX_GetVolumeInformationByHandleW)(
- _In_ HANDLE hFile, _Out_opt_ LPWSTR lpVolumeNameBuffer,
- _In_ DWORD nVolumeNameSize, _Out_opt_ LPDWORD lpVolumeSerialNumber,
- _Out_opt_ LPDWORD lpMaximumComponentLength,
- _Out_opt_ LPDWORD lpFileSystemFlags,
- _Out_opt_ LPWSTR lpFileSystemNameBuffer, _In_ DWORD nFileSystemNameSize);
-MDBX_INTERNAL_VAR MDBX_GetVolumeInformationByHandleW
- mdbx_GetVolumeInformationByHandleW;
-
-typedef DWORD(WINAPI *MDBX_GetFinalPathNameByHandleW)(_In_ HANDLE hFile,
- _Out_ LPWSTR lpszFilePath,
- _In_ DWORD cchFilePath,
- _In_ DWORD dwFlags);
-MDBX_INTERNAL_VAR MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-
-typedef BOOL(WINAPI *MDBX_SetFileInformationByHandle)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_SetFileInformationByHandle
- mdbx_SetFileInformationByHandle;
-
-typedef NTSTATUS(NTAPI *MDBX_NtFsControlFile)(
- IN HANDLE FileHandle, IN OUT HANDLE Event,
- IN OUT PVOID /* PIO_APC_ROUTINE */ ApcRoutine, IN OUT PVOID ApcContext,
- OUT PIO_STATUS_BLOCK IoStatusBlock, IN ULONG FsControlCode,
- IN OUT PVOID InputBuffer, IN ULONG InputBufferLength,
- OUT OPTIONAL PVOID OutputBuffer, IN ULONG OutputBufferLength);
-MDBX_INTERNAL_VAR MDBX_NtFsControlFile mdbx_NtFsControlFile;
-
-typedef uint64_t(WINAPI *MDBX_GetTickCount64)(void);
-MDBX_INTERNAL_VAR MDBX_GetTickCount64 mdbx_GetTickCount64;
-
-#if !defined(_WIN32_WINNT_WIN8) || _WIN32_WINNT < _WIN32_WINNT_WIN8
-typedef struct _WIN32_MEMORY_RANGE_ENTRY {
- PVOID VirtualAddress;
- SIZE_T NumberOfBytes;
-} WIN32_MEMORY_RANGE_ENTRY, *PWIN32_MEMORY_RANGE_ENTRY;
-#endif /* Windows 8.x */
-
-typedef BOOL(WINAPI *MDBX_PrefetchVirtualMemory)(
- HANDLE hProcess, ULONG_PTR NumberOfEntries,
- PWIN32_MEMORY_RANGE_ENTRY VirtualAddresses, ULONG Flags);
-MDBX_INTERNAL_VAR MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-
-typedef enum _SECTION_INHERIT { ViewShare = 1, ViewUnmap = 2 } SECTION_INHERIT;
-
-typedef NTSTATUS(NTAPI *MDBX_NtExtendSection)(IN HANDLE SectionHandle,
- IN PLARGE_INTEGER NewSectionSize);
-MDBX_INTERNAL_VAR MDBX_NtExtendSection mdbx_NtExtendSection;
-
-static __inline bool mdbx_RunningUnderWine(void) {
- return !mdbx_NtExtendSection;
-}
-
-typedef LSTATUS(WINAPI *MDBX_RegGetValueA)(HKEY hkey, LPCSTR lpSubKey,
- LPCSTR lpValue, DWORD dwFlags,
- LPDWORD pdwType, PVOID pvData,
- LPDWORD pcbData);
-MDBX_INTERNAL_VAR MDBX_RegGetValueA mdbx_RegGetValueA;
-
-NTSYSAPI ULONG RtlRandomEx(PULONG Seed);
-
-#endif /* Windows */
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_MSC_VER) && _MSC_VER >= 1900
-/* LY: MSVC 2015/2017/2019 has buggy/inconsistent PRIuPTR/PRIxPTR macros
- * for internal format-args checker. */
-#undef PRIuPTR
-#undef PRIiPTR
-#undef PRIdPTR
-#undef PRIxPTR
-#define PRIuPTR "Iu"
-#define PRIiPTR "Ii"
-#define PRIdPTR "Id"
-#define PRIxPTR "Ix"
-#define PRIuSIZE "zu"
-#define PRIiSIZE "zi"
-#define PRIdSIZE "zd"
-#define PRIxSIZE "zx"
-#endif /* fix PRI*PTR for _MSC_VER */
-
-#ifndef PRIuSIZE
-#define PRIuSIZE PRIuPTR
-#define PRIiSIZE PRIiPTR
-#define PRIdSIZE PRIdPTR
-#define PRIxSIZE PRIxPTR
-#endif /* PRI*SIZE macros for MSVC */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-#define mdbx_sourcery_anchor XCONCAT(mdbx_sourcery_, MDBX_BUILD_SOURCERY)
-#if defined(xMDBX_TOOLS)
-extern LIBMDBX_API const char *const mdbx_sourcery_anchor;
-#endif
-
-/*******************************************************************************
- *******************************************************************************
- *******************************************************************************
- *
- *
- * #### ##### ##### # #### # # ####
- * # # # # # # # # ## # #
- * # # # # # # # # # # # ####
- * # # ##### # # # # # # # #
- * # # # # # # # # ## # #
- * #### # # # #### # # ####
- *
- *
- */
-
-/** \defgroup build_option Build options
- * The libmdbx build options.
- @{ */
-
-/** Using fcntl(F_FULLFSYNC) with 5-10 times slowdown */
-#define MDBX_OSX_WANNA_DURABILITY 0
-/** Using fsync() with chance of data lost on power failure */
-#define MDBX_OSX_WANNA_SPEED 1
-
-#ifndef MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-/** Choices \ref MDBX_OSX_WANNA_DURABILITY or \ref MDBX_OSX_WANNA_SPEED
- * for OSX & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY MDBX_OSX_WANNA_DURABILITY
-#endif /* MDBX_OSX_SPEED_INSTEADOF_DURABILITY */
-
-/** Controls checking PID against reuse DB environment after the fork() */
-#ifndef MDBX_ENV_CHECKPID
-#if defined(MADV_DONTFORK) || defined(_WIN32) || defined(_WIN64)
-/* PID check could be omitted:
- * - on Linux when madvise(MADV_DONTFORK) is available, i.e. after the fork()
- * mapped pages will not be available for child process.
- * - in Windows where fork() not available. */
-#define MDBX_ENV_CHECKPID 0
-#else
-#define MDBX_ENV_CHECKPID 1
-#endif
-#define MDBX_ENV_CHECKPID_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#else
-#define MDBX_ENV_CHECKPID_CONFIG MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#endif /* MDBX_ENV_CHECKPID */
-
-/** Controls checking transaction owner thread against misuse transactions from
- * other threads. */
-#ifndef MDBX_TXN_CHECKOWNER
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_TXN_CHECKOWNER_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#else
-#define MDBX_TXN_CHECKOWNER_CONFIG MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#endif /* MDBX_TXN_CHECKOWNER */
-
-/** Does a system have battery-backed Real-Time Clock or just a fake. */
-#ifndef MDBX_TRUST_RTC
-#if defined(__linux__) || defined(__gnu_linux__) || defined(__NetBSD__) || \
- defined(__OpenBSD__)
-#define MDBX_TRUST_RTC 0 /* a lot of embedded systems have a fake RTC */
-#else
-#define MDBX_TRUST_RTC 1
-#endif
-#define MDBX_TRUST_RTC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#else
-#define MDBX_TRUST_RTC_CONFIG MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#endif /* MDBX_TRUST_RTC */
-
-/** Controls online database auto-compactification during write-transactions. */
-#ifndef MDBX_ENABLE_REFUND
-#define MDBX_ENABLE_REFUND 1
-#elif !(MDBX_ENABLE_REFUND == 0 || MDBX_ENABLE_REFUND == 1)
-#error MDBX_ENABLE_REFUND must be defined as 0 or 1
-#endif /* MDBX_ENABLE_REFUND */
-
-/** Controls gathering statistics for page operations. */
-#ifndef MDBX_ENABLE_PGOP_STAT
-#define MDBX_ENABLE_PGOP_STAT 1
-#elif !(MDBX_ENABLE_PGOP_STAT == 0 || MDBX_ENABLE_PGOP_STAT == 1)
-#error MDBX_ENABLE_PGOP_STAT must be defined as 0 or 1
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-/** Controls use of POSIX madvise() hints and friends. */
-#ifndef MDBX_ENABLE_MADVISE
-#define MDBX_ENABLE_MADVISE 1
-#elif !(MDBX_ENABLE_MADVISE == 0 || MDBX_ENABLE_MADVISE == 1)
-#error MDBX_ENABLE_MADVISE must be defined as 0 or 1
-#endif /* MDBX_ENABLE_MADVISE */
-
-/** Disable some checks to reduce an overhead and detection probability of
- * database corruption to a values closer to the LMDB. */
-#ifndef MDBX_DISABLE_PAGECHECKS
-#define MDBX_DISABLE_PAGECHECKS 0
-#elif !(MDBX_DISABLE_PAGECHECKS == 0 || MDBX_DISABLE_PAGECHECKS == 1)
-#error MDBX_DISABLE_PAGECHECKS must be defined as 0 or 1
-#endif /* MDBX_DISABLE_PAGECHECKS */
-
-#ifndef MDBX_PNL_PREALLOC_FOR_RADIXSORT
-#define MDBX_PNL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_PNL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_PNL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_PNL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
-
-#ifndef MDBX_DPL_PREALLOC_FOR_RADIXSORT
-#define MDBX_DPL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_DPL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_DPL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_DPL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
-
-/* Basically, this build-option is for TODO. Guess it should be replaced
- * with MDBX_ENABLE_WRITEMAP_SPILLING with the three variants:
- * 0/OFF = Don't track dirty pages at all and don't spilling ones.
- * This should be by-default on Linux and may-be other systems
- * (not sure: Darwin/OSX, FreeBSD, Windows 10) where kernel provides
- * properly LRU tracking and async writing on-demand.
- * 1/ON = Lite tracking of dirty pages but with LRU labels and explicit
- * spilling with msync(MS_ASYNC). */
-#ifndef MDBX_FAKE_SPILL_WRITEMAP
-#if defined(__linux__) || defined(__gnu_linux__)
-#define MDBX_FAKE_SPILL_WRITEMAP 1 /* msync(MS_ASYNC) is no-op on Linux */
-#else
-#define MDBX_FAKE_SPILL_WRITEMAP 0
-#endif
-#elif !(MDBX_FAKE_SPILL_WRITEMAP == 0 || MDBX_FAKE_SPILL_WRITEMAP == 1)
-#error MDBX_FAKE_SPILL_WRITEMAP must be defined as 0 or 1
-#endif /* MDBX_FAKE_SPILL_WRITEMAP */
-
-/** Controls sort order of internal page number lists.
- * This mostly experimental/advanced option with not for regular MDBX users.
- * \warning The database format depend on this option and libmdbx builded with
- * different option value are incompatible. */
-#ifndef MDBX_PNL_ASCENDING
-#define MDBX_PNL_ASCENDING 0
-#elif !(MDBX_PNL_ASCENDING == 0 || MDBX_PNL_ASCENDING == 1)
-#error MDBX_PNL_ASCENDING must be defined as 0 or 1
-#endif /* MDBX_PNL_ASCENDING */
-
-/** Avoid dependence from MSVC CRT and use ntdll.dll instead. */
-#ifndef MDBX_WITHOUT_MSVC_CRT
-#define MDBX_WITHOUT_MSVC_CRT 1
-#elif !(MDBX_WITHOUT_MSVC_CRT == 0 || MDBX_WITHOUT_MSVC_CRT == 1)
-#error MDBX_WITHOUT_MSVC_CRT must be defined as 0 or 1
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-/** Size of buffer used during copying a environment/database file. */
-#ifndef MDBX_ENVCOPY_WRITEBUF
-#define MDBX_ENVCOPY_WRITEBUF 1048576u
-#elif MDBX_ENVCOPY_WRITEBUF < 65536u || MDBX_ENVCOPY_WRITEBUF > 1073741824u || \
- MDBX_ENVCOPY_WRITEBUF % 65536u
-#error MDBX_ENVCOPY_WRITEBUF must be defined in range 65536..1073741824 and be multiple of 65536
-#endif /* MDBX_ENVCOPY_WRITEBUF */
-
-/** Forces assertion checking */
-#ifndef MDBX_FORCE_ASSERTIONS
-#define MDBX_FORCE_ASSERTIONS 0
-#elif !(MDBX_FORCE_ASSERTIONS == 0 || MDBX_FORCE_ASSERTIONS == 1)
-#error MDBX_FORCE_ASSERTIONS must be defined as 0 or 1
-#endif /* MDBX_FORCE_ASSERTIONS */
-
-/** Presumed malloc size overhead for each allocation
- * to adjust allocations to be more aligned. */
-#ifndef MDBX_ASSUME_MALLOC_OVERHEAD
-#ifdef __SIZEOF_POINTER__
-#define MDBX_ASSUME_MALLOC_OVERHEAD (__SIZEOF_POINTER__ * 2u)
-#else
-#define MDBX_ASSUME_MALLOC_OVERHEAD (sizeof(void *) * 2u)
-#endif
-#elif MDBX_ASSUME_MALLOC_OVERHEAD < 0 || MDBX_ASSUME_MALLOC_OVERHEAD > 64 || \
- MDBX_ASSUME_MALLOC_OVERHEAD % 4
-#error MDBX_ASSUME_MALLOC_OVERHEAD must be defined in range 0..64 and be multiple of 4
-#endif /* MDBX_ASSUME_MALLOC_OVERHEAD */
-
-/** In case the MDBX_DEBUG is undefined set it corresponding to NDEBUG */
-#ifndef MDBX_DEBUG
-#ifdef NDEBUG
-#define MDBX_DEBUG 0
-#else
-#define MDBX_DEBUG 1
-#endif
-#endif /* MDBX_DEBUG */
-
-/** If defined then enables integration with Valgrind,
- * a memory analyzing tool. */
-#ifndef MDBX_USE_VALGRIND
-#endif /* MDBX_USE_VALGRIND */
-
-/** If defined then enables use C11 atomics,
- * otherwise detects ones availability automatically. */
-#ifndef MDBX_HAVE_C11ATOMICS
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-//------------------------------------------------------------------------------
-
-/** Win32 File Locking API for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_WIN32FILES -1
-
-/** SystemV IPC semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_SYSV 5
-
-/** POSIX-1 Shared anonymous semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX1988 1988
-
-/** POSIX-2001 Shared Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2001 2001
-
-/** POSIX-2008 Robust Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2008 2008
-
-/** BeOS Benaphores, aka Futexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_BENAPHORE 1995
-
-/** Advanced: Choices the locking implementation (autodetection by default). */
-#if defined(_WIN32) || defined(_WIN64)
-#define MDBX_LOCKING MDBX_LOCKING_WIN32FILES
-#else
-#ifndef MDBX_LOCKING
-#if defined(_POSIX_THREAD_PROCESS_SHARED) && \
- _POSIX_THREAD_PROCESS_SHARED >= 200112L && !defined(__FreeBSD__)
-
-/* Some platforms define the EOWNERDEAD error code even though they
- * don't support Robust Mutexes. If doubt compile with -MDBX_LOCKING=2001. */
-#if defined(EOWNERDEAD) && _POSIX_THREAD_PROCESS_SHARED >= 200809L && \
- ((defined(_POSIX_THREAD_ROBUST_PRIO_INHERIT) && \
- _POSIX_THREAD_ROBUST_PRIO_INHERIT > 0) || \
- (defined(_POSIX_THREAD_ROBUST_PRIO_PROTECT) && \
- _POSIX_THREAD_ROBUST_PRIO_PROTECT > 0) || \
- defined(PTHREAD_MUTEX_ROBUST) || defined(PTHREAD_MUTEX_ROBUST_NP)) && \
- (!defined(__GLIBC__) || \
- __GLIBC_PREREQ(2, 10) /* troubles with Robust mutexes before 2.10 */)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2008
-#else
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2001
-#endif
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX1988
-#else
-#define MDBX_LOCKING MDBX_LOCKING_SYSV
-#endif
-#define MDBX_LOCKING_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_LOCKING)
-#else
-#define MDBX_LOCKING_CONFIG MDBX_STRINGIFY(MDBX_LOCKING)
-#endif /* MDBX_LOCKING */
-#endif /* !Windows */
-
-/** Advanced: Using POSIX OFD-locks (autodetection by default). */
-#ifndef MDBX_USE_OFDLOCKS
-#if defined(F_OFD_SETLK) && defined(F_OFD_SETLKW) && defined(F_OFD_GETLK) && \
- !defined(MDBX_SAFE4QEMU) && \
- !defined(__sun) /* OFD-lock are broken on Solaris */
-#define MDBX_USE_OFDLOCKS 1
-#else
-#define MDBX_USE_OFDLOCKS 0
-#endif
-#define MDBX_USE_OFDLOCKS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#else
-#define MDBX_USE_OFDLOCKS_CONFIG MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#endif /* MDBX_USE_OFDLOCKS */
-
-/** Advanced: Using sendfile() syscall (autodetection by default). */
-#ifndef MDBX_USE_SENDFILE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 21)
-#define MDBX_USE_SENDFILE 1
-#else
-#define MDBX_USE_SENDFILE 0
-#endif
-#endif /* MDBX_USE_SENDFILE */
-
-/** Advanced: Using copy_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_COPYFILERANGE
-#if __GLIBC_PREREQ(2, 27) && defined(_GNU_SOURCE)
-#define MDBX_USE_COPYFILERANGE 1
-#else
-#define MDBX_USE_COPYFILERANGE 0
-#endif
-#endif /* MDBX_USE_COPYFILERANGE */
-
-/** Advanced: Using sync_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_SYNCFILERANGE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- defined(SYNC_FILE_RANGE_WRITE) && !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 26)
-#define MDBX_USE_SYNCFILERANGE 1
-#else
-#define MDBX_USE_SYNCFILERANGE 0
-#endif
-#endif /* MDBX_USE_SYNCFILERANGE */
-
-//------------------------------------------------------------------------------
-
-#ifndef MDBX_CPU_WRITEBACK_INCOHERENT
-#if defined(__ia32__) || defined(__e2k__) || defined(__hppa) || \
- defined(__hppa__) || defined(DOXYGEN)
-#define MDBX_CPU_WRITEBACK_INCOHERENT 0
-#else
-#define MDBX_CPU_WRITEBACK_INCOHERENT 1
-#endif
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-#ifndef MDBX_MMAP_INCOHERENT_FILE_WRITE
-#ifdef __OpenBSD__
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 1
-#else
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#ifndef MDBX_MMAP_INCOHERENT_CPU_CACHE
-#if defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS has cache coherency issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 1
-#else
-/* LY: assume no relevant mmap/dcache issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#ifndef MDBX_64BIT_ATOMIC
-#if MDBX_WORDBITS >= 64 || defined(DOXYGEN)
-#define MDBX_64BIT_ATOMIC 1
-#else
-#define MDBX_64BIT_ATOMIC 0
-#endif
-#define MDBX_64BIT_ATOMIC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#else
-#define MDBX_64BIT_ATOMIC_CONFIG MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#endif /* MDBX_64BIT_ATOMIC */
-
-#ifndef MDBX_64BIT_CAS
-#if defined(ATOMIC_LLONG_LOCK_FREE)
-#if ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__GCC_ATOMIC_LLONG_LOCK_FREE)
-#if __GCC_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__CLANG_ATOMIC_LLONG_LOCK_FREE)
-#if __CLANG_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(_MSC_VER) || defined(__APPLE__) || defined(DOXYGEN)
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS MDBX_64BIT_ATOMIC
-#endif
-#define MDBX_64BIT_CAS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#else
-#define MDBX_64BIT_CAS_CONFIG MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#endif /* MDBX_64BIT_CAS */
-
-#ifndef MDBX_UNALIGNED_OK
-#if defined(__ALIGNED__) || defined(__SANITIZE_UNDEFINED__)
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#elif defined(__ARM_FEATURE_UNALIGNED)
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#elif __CLANG_PREREQ(5, 0) || __GNUC_PREREQ(5, 0)
-/* expecting an optimization will well done, also this
- * hushes false-positives from UBSAN (undefined behaviour sanitizer) */
-#define MDBX_UNALIGNED_OK 0
-#elif defined(__e2k__) || defined(__elbrus__)
-#if __iset__ > 4
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#endif
-#elif defined(__ia32__)
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#endif
-#elif MDBX_UNALIGNED_OK == 1
-#undef MDBX_UNALIGNED_OK
-#define MDBX_UNALIGNED_OK 32 /* any unaligned access allowed */
-#endif /* MDBX_UNALIGNED_OK */
-
-#ifndef MDBX_CACHELINE_SIZE
-#if defined(SYSTEM_CACHE_ALIGNMENT_SIZE)
-#define MDBX_CACHELINE_SIZE SYSTEM_CACHE_ALIGNMENT_SIZE
-#elif defined(__ia64__) || defined(__ia64) || defined(_M_IA64)
-#define MDBX_CACHELINE_SIZE 128
-#else
-#define MDBX_CACHELINE_SIZE 64
-#endif
-#endif /* MDBX_CACHELINE_SIZE */
-
-/** @} end of build options */
-/*******************************************************************************
- *******************************************************************************
- ******************************************************************************/
-
-#ifdef DOXYGEN
-/* !!! Actually this is a fake definitions !!!
- * !!! for documentation generation by Doxygen !!! */
-
-/** Controls enabling of debugging features.
- *
- * - `MDBX_DEBUG = 0` (by default) Disables any debugging features at all,
- * including logging and assertion controls.
- * Logging level and corresponding debug flags changing
- * by \ref mdbx_setup_debug() will not have effect.
- * - `MDBX_DEBUG > 0` Enables code for the debugging features (logging,
- * assertions checking and internal audit).
- * Simultaneously sets the default logging level
- * to the `MDBX_DEBUG` value.
- * Also enables \ref MDBX_DBG_AUDIT if `MDBX_DEBUG >= 2`.
- *
- * \ingroup build_option */
-#define MDBX_DEBUG 0...7
-
-/** Disables using of GNU libc extensions. */
-#define MDBX_DISABLE_GNU_SOURCE 0 or 1
-
-#endif /* DOXYGEN */
-
-/* Undefine the NDEBUG if debugging is enforced by MDBX_DEBUG */
-#if MDBX_DEBUG
-#undef NDEBUG
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-enum MDBX_memory_order {
- mo_Relaxed,
- mo_AcquireRelease,
- mo_SequentialConsistency
-};
-
-typedef union {
- volatile uint32_t weak;
-#ifdef MDBX_HAVE_C11ATOMICS
- volatile _Atomic uint32_t c11a;
-#endif /* MDBX_HAVE_C11ATOMICS */
-} MDBX_atomic_uint32_t;
-
-typedef union {
- volatile uint64_t weak;
-#if defined(MDBX_HAVE_C11ATOMICS) && (MDBX_64BIT_CAS || MDBX_64BIT_ATOMIC)
- volatile _Atomic uint64_t c11a;
-#endif
-#if !defined(MDBX_HAVE_C11ATOMICS) || !MDBX_64BIT_CAS || !MDBX_64BIT_ATOMIC
- __anonymous_struct_extension__ struct {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- MDBX_atomic_uint32_t low, high;
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- MDBX_atomic_uint32_t high, low;
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- };
-#endif
-} MDBX_atomic_uint64_t;
-
-#ifdef MDBX_HAVE_C11ATOMICS
-
-/* Crutches for C11 atomic compiler's bugs */
-#if defined(__e2k__) && defined(__LCC__) && __LCC__ < /* FIXME */ 127
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->weak)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->weak)
-#elif defined(__clang__) && __clang__ < 8
-#define MDBX_c11a_ro(type, ptr) ((volatile _Atomic(type) *)&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#else
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#endif /* Crutches for C11 atomic compiler's bugs */
-
-static __always_inline memory_order mo_c11_store(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_release;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-
-static __always_inline memory_order mo_c11_load(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_acquire;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED static __always_inline void
-mdbx_memory_fence(enum MDBX_memory_order order, bool write) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(write ? mo_c11_store(order) : mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_compiler_barrier();
- if (write &&
- order > (MDBX_CPU_WRITEBACK_INCOHERENT ? mo_Relaxed : mo_AcquireRelease))
- mdbx_memory_barrier();
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_store32(MDBX_atomic_uint32_t *p, const uint32_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint32_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_load32(const MDBX_atomic_uint32_t *p, enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint32_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint32_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint32_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-/* Basic constants and types */
-
-/* A stamp that identifies a file as an MDBX file.
- * There's nothing special about this value other than that it is easily
- * recognizable, and it will reflect any byte order mismatches. */
-#define MDBX_MAGIC UINT64_C(/* 56-bit prime */ 0x59659DBDEF4C11)
-
-/* FROZEN: The version number for a database's datafile format. */
-#define MDBX_DATA_VERSION 3
-/* The version number for a database's lockfile format. */
-#define MDBX_LOCK_VERSION 4
-
-/* handle for the DB used to track free pages. */
-#define FREE_DBI 0
-/* handle for the default DB. */
-#define MAIN_DBI 1
-/* Number of DBs in metapage (free and main) - also hardcoded elsewhere */
-#define CORE_DBS 2
-
-/* Number of meta pages - also hardcoded elsewhere */
-#define NUM_METAS 3
-
-/* A page number in the database.
- *
- * MDBX uses 32 bit for page numbers. This limits database
- * size up to 2^44 bytes, in case of 4K pages. */
-typedef uint32_t pgno_t;
-typedef MDBX_atomic_uint32_t atomic_pgno_t;
-#define PRIaPGNO PRIu32
-#define MAX_PAGENO UINT32_C(0x7FFFffff)
-#define MIN_PAGENO NUM_METAS
-
-#define SAFE64_INVALID_THRESHOLD UINT64_C(0xffffFFFF00000000)
-
-/* A transaction ID. */
-typedef uint64_t txnid_t;
-typedef MDBX_atomic_uint64_t atomic_txnid_t;
-#define PRIaTXN PRIi64
-#define MIN_TXNID UINT64_C(1)
-#define MAX_TXNID (SAFE64_INVALID_THRESHOLD - 1)
-#define INITIAL_TXNID (MIN_TXNID + NUM_METAS - 1)
-#define INVALID_TXNID UINT64_MAX
-/* LY: for testing non-atomic 64-bit txnid on 32-bit arches.
- * #define xMDBX_TXNID_STEP (UINT32_MAX / 3) */
-#ifndef xMDBX_TXNID_STEP
-#if MDBX_64BIT_CAS
-#define xMDBX_TXNID_STEP 1u
-#else
-#define xMDBX_TXNID_STEP 2u
-#endif
-#endif /* xMDBX_TXNID_STEP */
-
-/* Used for offsets within a single page.
- * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
- * this is plenty. */
-typedef uint16_t indx_t;
-
-#define MEGABYTE ((size_t)1 << 20)
-
-/*----------------------------------------------------------------------------*/
-/* Core structures for database and shared memory (i.e. format definition) */
-#pragma pack(push, 4)
-
-/* Information about a single database in the environment. */
-typedef struct MDBX_db {
- uint16_t md_flags; /* see mdbx_dbi_open */
- uint16_t md_depth; /* depth of this tree */
- uint32_t md_xsize; /* key-size for MDBX_DUPFIXED (LEAF2 pages) */
- pgno_t md_root; /* the root page of this tree */
- pgno_t md_branch_pages; /* number of internal pages */
- pgno_t md_leaf_pages; /* number of leaf pages */
- pgno_t md_overflow_pages; /* number of overflow pages */
- uint64_t md_seq; /* table sequence counter */
- uint64_t md_entries; /* number of data items */
- uint64_t md_mod_txnid; /* txnid of last committed modification */
-} MDBX_db;
-
-/* database size-related parameters */
-typedef struct MDBX_geo {
- uint16_t grow_pv; /* datafile growth step as a 16-bit packed (exponential
- quantized) value */
- uint16_t shrink_pv; /* datafile shrink threshold as a 16-bit packed
- (exponential quantized) value */
- pgno_t lower; /* minimal size of datafile in pages */
- pgno_t upper; /* maximal size of datafile in pages */
- pgno_t now; /* current size of datafile in pages */
- pgno_t next; /* first unused page in the datafile,
- but actually the file may be shorter. */
-} MDBX_geo;
-
-/* Meta page content.
- * A meta page is the start point for accessing a database snapshot.
- * Pages 0-1 are meta pages. Transaction N writes meta page (N % 2). */
-typedef struct MDBX_meta {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with MDBX_DATA_VERSION. */
- uint32_t mm_magic_and_version[2];
-
- /* txnid that committed this page, the first of a two-phase-update pair */
- uint32_t mm_txnid_a[2];
-
- uint16_t mm_extra_flags; /* extra DB flags, zero (nothing) for now */
- uint8_t mm_validator_id; /* ID of checksum and page validation method,
- * zero (nothing) for now */
- uint8_t mm_extra_pagehdr; /* extra bytes in the page header,
- * zero (nothing) for now */
-
- MDBX_geo mm_geo; /* database size-related parameters */
-
- MDBX_db mm_dbs[CORE_DBS]; /* first is free space, 2nd is main db */
- /* The size of pages used in this DB */
-#define mm_psize mm_dbs[FREE_DBI].md_xsize
- MDBX_canary mm_canary;
-
-#define MDBX_DATASIGN_NONE 0u
-#define MDBX_DATASIGN_WEAK 1u
-#define SIGN_IS_STEADY(sign) ((sign) > MDBX_DATASIGN_WEAK)
-#define META_IS_STEADY(meta) \
- SIGN_IS_STEADY(unaligned_peek_u64_volatile(4, (meta)->mm_datasync_sign))
- uint32_t mm_datasync_sign[2];
-
- /* txnid that committed this page, the second of a two-phase-update pair */
- uint32_t mm_txnid_b[2];
-
- /* Number of non-meta pages which were put in GC after COW. May be 0 in case
- * DB was previously handled by libmdbx without corresponding feature.
- * This value in couple with mr_snapshot_pages_retired allows fast estimation
- * of "how much reader is restraining GC recycling". */
- uint32_t mm_pages_retired[2];
-
- /* The analogue /proc/sys/kernel/random/boot_id or similar to determine
- * whether the system was rebooted after the last use of the database files.
- * If there was no reboot, but there is no need to rollback to the last
- * steady sync point. Zeros mean that no relevant information is available
- * from the system. */
- bin128_t mm_bootid;
-
-} MDBX_meta;
-
-#pragma pack(1)
-
-/* Common header for all page types. The page type depends on mp_flags.
- *
- * P_BRANCH and P_LEAF pages have unsorted 'MDBX_node's at the end, with
- * sorted mp_ptrs[] entries referring to them. Exception: P_LEAF2 pages
- * omit mp_ptrs and pack sorted MDBX_DUPFIXED values after the page header.
- *
- * P_OVERFLOW records occupy one or more contiguous pages where only the
- * first has a page header. They hold the real data of F_BIGDATA nodes.
- *
- * P_SUBP sub-pages are small leaf "pages" with duplicate data.
- * A node with flag F_DUPDATA but not F_SUBDATA contains a sub-page.
- * (Duplicate data can also go in sub-databases, which use normal pages.)
- *
- * P_META pages contain MDBX_meta, the start point of an MDBX snapshot.
- *
- * Each non-metapage up to MDBX_meta.mm_last_pg is reachable exactly once
- * in the snapshot: Either used by a database or listed in a GC record. */
-typedef struct MDBX_page {
- union {
-#define IS_FROZEN(txn, p) ((p)->mp_txnid < (txn)->mt_txnid)
-#define IS_SPILLED(txn, p) ((p)->mp_txnid == (txn)->mt_txnid)
-#define IS_SHADOWED(txn, p) ((p)->mp_txnid > (txn)->mt_txnid)
-#define IS_VALID(txn, p) ((p)->mp_txnid <= (txn)->mt_front)
-#define IS_MODIFIABLE(txn, p) ((p)->mp_txnid == (txn)->mt_front)
- uint64_t mp_txnid;
- struct MDBX_page *mp_next; /* for in-memory list of freed pages */
- };
- uint16_t mp_leaf2_ksize; /* key size if this is a LEAF2 page */
-#define P_BRANCH 0x01 /* branch page */
-#define P_LEAF 0x02 /* leaf page */
-#define P_OVERFLOW 0x04 /* overflow page */
-#define P_META 0x08 /* meta page */
-#define P_BAD 0x10 /* explicit flag for invalid/bad page */
-#define P_LEAF2 0x20 /* for MDBX_DUPFIXED records */
-#define P_SUBP 0x40 /* for MDBX_DUPSORT sub-pages */
-#define P_SPILLED 0x2000 /* spilled in parent txn */
-#define P_LOOSE 0x4000 /* page was dirtied then freed, can be reused */
-#define P_FROZEN 0x8000 /* used for retire page with known status */
-#define P_ILL_BITS (~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_SPILLED))
- uint16_t mp_flags;
- union {
- uint32_t mp_pages; /* number of overflow pages */
- __anonymous_struct_extension__ struct {
- indx_t mp_lower; /* lower bound of free space */
- indx_t mp_upper; /* upper bound of free space */
- };
- };
- pgno_t mp_pgno; /* page number */
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- indx_t mp_ptrs[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_page;
-
-/* Size of the page header, excluding dynamic data at the end */
-#define PAGEHDRSZ ((unsigned)offsetof(MDBX_page, mp_ptrs))
-
-#pragma pack(pop)
-
-#if MDBX_ENABLE_PGOP_STAT
-/* Statistics of page operations overall of all (running, completed and aborted)
- * transactions */
-typedef struct {
- MDBX_atomic_uint64_t newly; /* Quantity of a new pages added */
- MDBX_atomic_uint64_t cow; /* Quantity of pages copied for update */
- MDBX_atomic_uint64_t clone; /* Quantity of parent's dirty pages clones
- for nested transactions */
- MDBX_atomic_uint64_t split; /* Page splits */
- MDBX_atomic_uint64_t merge; /* Page merges */
- MDBX_atomic_uint64_t spill; /* Quantity of spilled dirty pages */
- MDBX_atomic_uint64_t unspill; /* Quantity of unspilled/reloaded pages */
- MDBX_atomic_uint64_t
- wops; /* Number of explicit write operations (not a pages) to a disk */
-} MDBX_pgop_stat_t;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-#if MDBX_LOCKING == MDBX_LOCKING_WIN32FILES
-#define MDBX_CLOCK_SIGN UINT32_C(0xF10C)
-typedef void mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
-
-#define MDBX_CLOCK_SIGN UINT32_C(0xF18D)
-typedef mdbx_pid_t mdbx_ipclock_t;
-#ifndef EOWNERDEAD
-#define EOWNERDEAD MDBX_RESULT_TRUE
-#endif
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#define MDBX_CLOCK_SIGN UINT32_C(0x8017)
-typedef pthread_mutex_t mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-#define MDBX_CLOCK_SIGN UINT32_C(0xFC29)
-typedef sem_t mdbx_ipclock_t;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV && !defined(__cplusplus)
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc);
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc);
-#endif /* MDBX_LOCKING */
-
-/* Reader Lock Table
- *
- * Readers don't acquire any locks for their data access. Instead, they
- * simply record their transaction ID in the reader table. The reader
- * mutex is needed just to find an empty slot in the reader table. The
- * slot's address is saved in thread-specific data so that subsequent
- * read transactions started by the same thread need no further locking to
- * proceed.
- *
- * If MDBX_NOTLS is set, the slot address is not saved in thread-specific data.
- * No reader table is used if the database is on a read-only filesystem.
- *
- * Since the database uses multi-version concurrency control, readers don't
- * actually need any locking. This table is used to keep track of which
- * readers are using data from which old transactions, so that we'll know
- * when a particular old transaction is no longer in use. Old transactions
- * that have discarded any data pages can then have those pages reclaimed
- * for use by a later write transaction.
- *
- * The lock table is constructed such that reader slots are aligned with the
- * processor's cache line size. Any slot is only ever used by one thread.
- * This alignment guarantees that there will be no contention or cache
- * thrashing as threads update their own slot info, and also eliminates
- * any need for locking when accessing a slot.
- *
- * A writer thread will scan every slot in the table to determine the oldest
- * outstanding reader transaction. Any freed pages older than this will be
- * reclaimed by the writer. The writer doesn't use any locks when scanning
- * this table. This means that there's no guarantee that the writer will
- * see the most up-to-date reader info, but that's not required for correct
- * operation - all we need is to know the upper bound on the oldest reader,
- * we don't care at all about the newest reader. So the only consequence of
- * reading stale information here is that old pages might hang around a
- * while longer before being reclaimed. That's actually good anyway, because
- * the longer we delay reclaiming old pages, the more likely it is that a
- * string of contiguous pages can be found after coalescing old pages from
- * many old transactions together. */
-
-/* The actual reader record, with cacheline padding. */
-typedef struct MDBX_reader {
- /* Current Transaction ID when this transaction began, or (txnid_t)-1.
- * Multiple readers that start at the same time will probably have the
- * same ID here. Again, it's not important to exclude them from
- * anything; all we need to know is which version of the DB they
- * started from so we can avoid overwriting any data used in that
- * particular version. */
- MDBX_atomic_uint64_t /* txnid_t */ mr_txnid;
-
- /* The information we store in a single slot of the reader table.
- * In addition to a transaction ID, we also record the process and
- * thread ID that owns a slot, so that we can detect stale information,
- * e.g. threads or processes that went away without cleaning up.
- *
- * NOTE: We currently don't check for stale records.
- * We simply re-init the table when we know that we're the only process
- * opening the lock file. */
-
- /* The thread ID of the thread owning this txn. */
- MDBX_atomic_uint64_t mr_tid;
-
- /* The process ID of the process owning this reader txn. */
- MDBX_atomic_uint32_t mr_pid;
-
- /* The number of pages used in the reader's MVCC snapshot,
- * i.e. the value of meta->mm_geo.next and txn->mt_next_pgno */
- atomic_pgno_t mr_snapshot_pages_used;
- /* Number of retired pages at the time this reader starts transaction. So,
- * at any time the difference mm_pages_retired - mr_snapshot_pages_retired
- * will give the number of pages which this reader restraining from reuse. */
- MDBX_atomic_uint64_t mr_snapshot_pages_retired;
-} MDBX_reader;
-
-/* The header for the reader table (a memory-mapped lock file). */
-typedef struct MDBX_lockinfo {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with with MDBX_LOCK_VERSION. */
- uint64_t mti_magic_and_version;
-
- /* Format of this lock file. Must be set to MDBX_LOCK_FORMAT. */
- uint32_t mti_os_and_format;
-
- /* Flags which environment was opened. */
- MDBX_atomic_uint32_t mti_envmode;
-
- /* Threshold of un-synced-with-disk pages for auto-sync feature,
- * zero means no-threshold, i.e. auto-sync is disabled. */
- atomic_pgno_t mti_autosync_threshold;
-
- /* Low 32-bit of txnid with which meta-pages was synced,
- * i.e. for sync-polling in the MDBX_NOMETASYNC mode. */
- MDBX_atomic_uint32_t mti_meta_sync_txnid;
-
- /* Period for timed auto-sync feature, i.e. at the every steady checkpoint
- * the mti_unsynced_timeout sets to the current_time + mti_autosync_period.
- * The time value is represented in a suitable system-dependent form, for
- * example clock_gettime(CLOCK_BOOTTIME) or clock_gettime(CLOCK_MONOTONIC).
- * Zero means timed auto-sync is disabled. */
- MDBX_atomic_uint64_t mti_autosync_period;
-
- /* Marker to distinguish uniqueness of DB/CLK. */
- MDBX_atomic_uint64_t mti_bait_uniqueness;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
-#if MDBX_ENABLE_PGOP_STAT
- /* Statistics of costly ops of all (running, completed and aborted)
- * transactions */
- MDBX_pgop_stat_t mti_pgop_stat;
-#endif /* MDBX_ENABLE_PGOP_STAT*/
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Write transaction lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_wlock;
-#endif /* MDBX_LOCKING > 0 */
-
- atomic_txnid_t mti_oldest_reader;
-
- /* Timestamp of the last steady sync. Value is represented in a suitable
- * system-dependent form, for example clock_gettime(CLOCK_BOOTTIME) or
- * clock_gettime(CLOCK_MONOTONIC). */
- MDBX_atomic_uint64_t mti_sync_timestamp;
-
- /* Number un-synced-with-disk pages for auto-sync feature. */
- atomic_pgno_t mti_unsynced_pages;
-
- /* Number of page which was discarded last time by madvise(MADV_FREE). */
- atomic_pgno_t mti_discarded_tail;
-
- /* Timestamp of the last readers check. */
- MDBX_atomic_uint64_t mti_reader_check_timestamp;
-
- /* Shared anchor for tracking readahead edge and enabled/disabled status. */
- pgno_t mti_readahead_anchor;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Readeaders registration lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_rlock;
-#endif /* MDBX_LOCKING > 0 */
-
- /* The number of slots that have been used in the reader table.
- * This always records the maximum count, it is not decremented
- * when readers release their slots. */
- MDBX_atomic_uint32_t mti_numreaders;
- MDBX_atomic_uint32_t mti_readers_refresh_flag;
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
- MDBX_reader mti_readers[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_lockinfo;
-
-/* Lockfile format signature: version, features and field layout */
-#define MDBX_LOCK_FORMAT \
- (MDBX_CLOCK_SIGN * 27733 + (unsigned)sizeof(MDBX_reader) * 13 + \
- (unsigned)offsetof(MDBX_reader, mr_snapshot_pages_used) * 251 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_oldest_reader) * 83 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_numreaders) * 37 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_readers) * 29)
-
-#define MDBX_DATA_MAGIC \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + MDBX_DATA_VERSION)
-
-#define MDBX_DATA_MAGIC_LEGACY_COMPAT \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + 2)
-
-#define MDBX_DATA_MAGIC_LEGACY_DEVEL ((MDBX_MAGIC << 8) + 255)
-
-#define MDBX_LOCK_MAGIC ((MDBX_MAGIC << 8) + MDBX_LOCK_VERSION)
-
-/* The maximum size of a database page.
- *
- * It is 64K, but value-PAGEHDRSZ must fit in MDBX_page.mp_upper.
- *
- * MDBX will use database pages < OS pages if needed.
- * That causes more I/O in write transactions: The OS must
- * know (read) the whole page before writing a partial page.
- *
- * Note that we don't currently support Huge pages. On Linux,
- * regular data files cannot use Huge pages, and in general
- * Huge pages aren't actually pageable. We rely on the OS
- * demand-pager to read our data and page it out when memory
- * pressure from other processes is high. So until OSs have
- * actual paging support for Huge pages, they're not viable. */
-#define MAX_PAGESIZE MDBX_MAX_PAGESIZE
-#define MIN_PAGESIZE MDBX_MIN_PAGESIZE
-
-#define MIN_MAPSIZE (MIN_PAGESIZE * MIN_PAGENO)
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_MAPSIZE32 UINT32_C(0x38000000)
-#else
-#define MAX_MAPSIZE32 UINT32_C(0x7f000000)
-#endif
-#define MAX_MAPSIZE64 ((MAX_PAGENO + 1) * (uint64_t)MAX_PAGESIZE)
-
-#if MDBX_WORDBITS >= 64
-#define MAX_MAPSIZE MAX_MAPSIZE64
-#define MDBX_PGL_LIMIT ((size_t)MAX_PAGENO)
-#else
-#define MAX_MAPSIZE MAX_MAPSIZE32
-#define MDBX_PGL_LIMIT (MAX_MAPSIZE32 / MIN_PAGESIZE)
-#endif /* MDBX_WORDBITS */
-
-#define MDBX_READERS_LIMIT 32767
-#define MDBX_RADIXSORT_THRESHOLD 333
-
-/*----------------------------------------------------------------------------*/
-
-/* An PNL is an Page Number List, a sorted array of IDs.
- * The first element of the array is a counter for how many actual page-numbers
- * are in the list. By default PNLs are sorted in descending order, this allow
- * cut off a page with lowest pgno (at the tail) just truncating the list. The
- * sort order of PNLs is controlled by the MDBX_PNL_ASCENDING build option. */
-typedef pgno_t *MDBX_PNL;
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_ORDERED(first, last) ((first) < (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) >= (last))
-#else
-#define MDBX_PNL_ORDERED(first, last) ((first) > (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) <= (last))
-#endif
-
-/* List of txnid, only for MDBX_txn.tw.lifo_reclaimed */
-typedef txnid_t *MDBX_TXL;
-
-/* An Dirty-Page list item is an pgno/pointer pair. */
-typedef struct MDBX_dp {
- MDBX_page *ptr;
- pgno_t pgno;
- union {
- unsigned extra;
- __anonymous_struct_extension__ struct {
- unsigned multi : 1;
- unsigned lru : 31;
- };
- };
-} MDBX_dp;
-
-/* An DPL (dirty-page list) is a sorted array of MDBX_DPs. */
-typedef struct MDBX_dpl {
- unsigned sorted;
- unsigned length;
- unsigned detent; /* allocated size excluding the MDBX_DPL_RESERVE_GAP */
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_dp items[] /* dynamic size with holes at zero and after the last */;
-#endif
-} MDBX_dpl;
-
-/* PNL sizes */
-#define MDBX_PNL_GRANULATE 1024
-#define MDBX_PNL_INITIAL \
- (MDBX_PNL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(pgno_t))
-
-#define MDBX_TXL_GRANULATE 32
-#define MDBX_TXL_INITIAL \
- (MDBX_TXL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-#define MDBX_TXL_MAX \
- ((1u << 17) - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-
-#define MDBX_PNL_ALLOCLEN(pl) ((pl)[-1])
-#define MDBX_PNL_SIZE(pl) ((pl)[0])
-#define MDBX_PNL_FIRST(pl) ((pl)[1])
-#define MDBX_PNL_LAST(pl) ((pl)[MDBX_PNL_SIZE(pl)])
-#define MDBX_PNL_BEGIN(pl) (&(pl)[1])
-#define MDBX_PNL_END(pl) (&(pl)[MDBX_PNL_SIZE(pl) + 1])
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_FIRST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_LAST(pl)
-#else
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_LAST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_FIRST(pl)
-#endif
-
-#define MDBX_PNL_SIZEOF(pl) ((MDBX_PNL_SIZE(pl) + 1) * sizeof(pgno_t))
-#define MDBX_PNL_IS_EMPTY(pl) (MDBX_PNL_SIZE(pl) == 0)
-
-/*----------------------------------------------------------------------------*/
-/* Internal structures */
-
-/* Auxiliary DB info.
- * The information here is mostly static/read-only. There is
- * only a single copy of this record in the environment. */
-typedef struct MDBX_dbx {
- MDBX_val md_name; /* name of the database */
- MDBX_cmp_func *md_cmp; /* function for comparing keys */
- MDBX_cmp_func *md_dcmp; /* function for comparing data items */
- size_t md_klen_min, md_klen_max; /* min/max key length for the database */
- size_t md_vlen_min,
- md_vlen_max; /* min/max value/data length for the database */
-} MDBX_dbx;
-
-/* A database transaction.
- * Every operation requires a transaction handle. */
-struct MDBX_txn {
-#define MDBX_MT_SIGNATURE UINT32_C(0x93D53A31)
- uint32_t mt_signature;
-
- /* Transaction Flags */
- /* mdbx_txn_begin() flags */
-#define MDBX_TXN_RO_BEGIN_FLAGS (MDBX_TXN_RDONLY | MDBX_TXN_RDONLY_PREPARE)
-#define MDBX_TXN_RW_BEGIN_FLAGS \
- (MDBX_TXN_NOMETASYNC | MDBX_TXN_NOSYNC | MDBX_TXN_TRY)
- /* Additional flag for mdbx_sync_locked() */
-#define MDBX_SHRINK_ALLOWED UINT32_C(0x40000000)
-
- /* internal txn flags */
-#define MDBX_TXN_FINISHED 0x01 /* txn is finished or never began */
-#define MDBX_TXN_ERROR 0x02 /* txn is unusable after an error */
-#define MDBX_TXN_DIRTY 0x04 /* must write, even if dirty list is empty */
-#define MDBX_TXN_SPILLS 0x08 /* txn or a parent has spilled pages */
-#define MDBX_TXN_HAS_CHILD 0x10 /* txn has an MDBX_txn.mt_child */
- /* most operations on the txn are currently illegal */
-#define MDBX_TXN_BLOCKED \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_HAS_CHILD)
-
-#define TXN_FLAGS \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_DIRTY | MDBX_TXN_SPILLS | \
- MDBX_TXN_HAS_CHILD)
-
-#if (TXN_FLAGS & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS)) || \
- ((MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS | TXN_FLAGS) & \
- MDBX_SHRINK_ALLOWED)
-#error "Oops, some flags overlapped or wrong"
-#endif
- uint32_t mt_flags;
-
- MDBX_txn *mt_parent; /* parent of a nested txn */
- /* Nested txn under this txn, set together with flag MDBX_TXN_HAS_CHILD */
- MDBX_txn *mt_child;
- MDBX_geo mt_geo;
- /* next unallocated page */
-#define mt_next_pgno mt_geo.next
- /* corresponding to the current size of datafile */
-#define mt_end_pgno mt_geo.now
-
- /* The ID of this transaction. IDs are integers incrementing from 1.
- * Only committed write transactions increment the ID. If a transaction
- * aborts, the ID may be re-used by the next writer. */
- txnid_t mt_txnid;
- txnid_t mt_front;
-
- MDBX_env *mt_env; /* the DB environment */
- /* Array of records for each DB known in the environment. */
- MDBX_dbx *mt_dbxs;
- /* Array of MDBX_db records for each known DB */
- MDBX_db *mt_dbs;
- /* Array of sequence numbers for each DB handle */
- unsigned *mt_dbiseqs;
-
- /* Transaction DBI Flags */
-#define DBI_DIRTY MDBX_DBI_DIRTY /* DB was written in this txn */
-#define DBI_STALE MDBX_DBI_STALE /* Named-DB record is older than txnID */
-#define DBI_FRESH MDBX_DBI_FRESH /* Named-DB handle opened in this txn */
-#define DBI_CREAT MDBX_DBI_CREAT /* Named-DB handle created in this txn */
-#define DBI_VALID 0x10 /* DB handle is valid, see also DB_VALID */
-#define DBI_USRVALID 0x20 /* As DB_VALID, but not set for FREE_DBI */
-#define DBI_AUDITED 0x40 /* Internal flag for accounting during audit */
- /* Array of flags for each DB */
- uint8_t *mt_dbistate;
- /* Number of DB records in use, or 0 when the txn is finished.
- * This number only ever increments until the txn finishes; we
- * don't decrement it when individual DB handles are closed. */
- MDBX_dbi mt_numdbs;
- size_t mt_owner; /* thread ID that owns this transaction */
- MDBX_canary mt_canary;
- void *mt_userctx; /* User-settable context */
- MDBX_cursor **mt_cursors;
-
- union {
- struct {
- /* For read txns: This thread/txn's reader table slot, or NULL. */
- MDBX_reader *reader;
- } to;
- struct {
- /* In write txns, array of cursors for each DB */
- pgno_t *reclaimed_pglist; /* Reclaimed GC pages */
- txnid_t last_reclaimed; /* ID of last used record */
-#if MDBX_ENABLE_REFUND
- pgno_t loose_refund_wl /* FIXME: describe */;
-#endif /* MDBX_ENABLE_REFUND */
- /* dirtylist room: Dirty array size - dirty pages visible to this txn.
- * Includes ancestor txns' dirty pages not hidden by other txns'
- * dirty/spilled pages. Thus commit(nested txn) has room to merge
- * dirtylist into mt_parent after freeing hidden mt_parent pages. */
- unsigned dirtyroom;
- /* a sequence to spilling dirty page with LRU policy */
- unsigned dirtylru;
- /* For write txns: Modified pages. Sorted when not MDBX_WRITEMAP. */
- MDBX_dpl *dirtylist;
- /* The list of reclaimed txns from GC */
- MDBX_TXL lifo_reclaimed;
- /* The list of pages that became unused during this transaction. */
- MDBX_PNL retired_pages;
- /* The list of loose pages that became unused and may be reused
- * in this transaction, linked through `mp_next`. */
- MDBX_page *loose_pages;
- /* Number of loose pages (tw.loose_pages) */
- unsigned loose_count;
- /* The sorted list of dirty pages we temporarily wrote to disk
- * because the dirty list was full. page numbers in here are
- * shifted left by 1, deleted slots have the LSB set. */
- MDBX_PNL spill_pages;
- unsigned spill_least_removed;
- } tw;
- };
-};
-
-#if MDBX_WORDBITS >= 64
-#define CURSOR_STACK 32
-#else
-#define CURSOR_STACK 24
-#endif
-
-struct MDBX_xcursor;
-
-/* Cursors are used for all DB operations.
- * A cursor holds a path of (page pointer, key index) from the DB
- * root to a position in the DB, plus other state. MDBX_DUPSORT
- * cursors include an xcursor to the current data item. Write txns
- * track their cursors and keep them up to date when data moves.
- * Exception: An xcursor's pointer to a P_SUBP page can be stale.
- * (A node with F_DUPDATA but no F_SUBDATA contains a subpage). */
-struct MDBX_cursor {
-#define MDBX_MC_LIVE UINT32_C(0xFE05D5B1)
-#define MDBX_MC_READY4CLOSE UINT32_C(0x2817A047)
-#define MDBX_MC_WAIT4EOT UINT32_C(0x90E297A7)
- uint32_t mc_signature;
- /* The database handle this cursor operates on */
- MDBX_dbi mc_dbi;
- /* Next cursor on this DB in this txn */
- MDBX_cursor *mc_next;
- /* Backup of the original cursor if this cursor is a shadow */
- MDBX_cursor *mc_backup;
- /* Context used for databases with MDBX_DUPSORT, otherwise NULL */
- struct MDBX_xcursor *mc_xcursor;
- /* The transaction that owns this cursor */
- MDBX_txn *mc_txn;
- /* The database record for this cursor */
- MDBX_db *mc_db;
- /* The database auxiliary record for this cursor */
- MDBX_dbx *mc_dbx;
- /* The mt_dbistate for this database */
- uint8_t *mc_dbistate;
- unsigned mc_snum; /* number of pushed pages */
- unsigned mc_top; /* index of top page, normally mc_snum-1 */
-
- /* Cursor state flags. */
-#define C_INITIALIZED 0x01 /* cursor has been initialized and is valid */
-#define C_EOF 0x02 /* No more data */
-#define C_SUB 0x04 /* Cursor is a sub-cursor */
-#define C_DEL 0x08 /* last op was a cursor_del */
-#define C_UNTRACK 0x10 /* Un-track cursor when closing */
-#define C_RECLAIMING 0x20 /* GC lookup is prohibited */
-#define C_GCFREEZE 0x40 /* reclaimed_pglist must not be updated */
-
- /* Cursor checking flags. */
-#define C_COPYING 0x100 /* skip key-value length check (copying simplify) */
-#define C_UPDATING 0x200 /* update/rebalance pending */
-#define C_RETIRING 0x400 /* refs to child pages may be invalid */
-#define C_SKIPORD 0x800 /* don't check keys ordering */
-
- unsigned mc_flags; /* see mdbx_cursor */
- MDBX_page *mc_pg[CURSOR_STACK]; /* stack of pushed pages */
- indx_t mc_ki[CURSOR_STACK]; /* stack of page indices */
-};
-
-/* Context for sorted-dup records.
- * We could have gone to a fully recursive design, with arbitrarily
- * deep nesting of sub-databases. But for now we only handle these
- * levels - main DB, optional sub-DB, sorted-duplicate DB. */
-typedef struct MDBX_xcursor {
- /* A sub-cursor for traversing the Dup DB */
- MDBX_cursor mx_cursor;
- /* The database record for this Dup DB */
- MDBX_db mx_db;
- /* The auxiliary DB record for this Dup DB */
- MDBX_dbx mx_dbx;
-} MDBX_xcursor;
-
-typedef struct MDBX_cursor_couple {
- MDBX_cursor outer;
- void *mc_userctx; /* User-settable context */
- MDBX_xcursor inner;
-} MDBX_cursor_couple;
-
-/* The database environment. */
-struct MDBX_env {
- /* ----------------------------------------------------- mostly static part */
-#define MDBX_ME_SIGNATURE UINT32_C(0x9A899641)
- MDBX_atomic_uint32_t me_signature;
- /* Failed to update the meta page. Probably an I/O error. */
-#define MDBX_FATAL_ERROR UINT32_C(0x80000000)
- /* Some fields are initialized. */
-#define MDBX_ENV_ACTIVE UINT32_C(0x20000000)
- /* me_txkey is set */
-#define MDBX_ENV_TXKEY UINT32_C(0x10000000)
- /* Legacy MDBX_MAPASYNC (prior v0.9) */
-#define MDBX_DEPRECATED_MAPASYNC UINT32_C(0x100000)
-#define ENV_INTERNAL_FLAGS (MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE | MDBX_ENV_TXKEY)
- uint32_t me_flags;
- mdbx_mmap_t me_dxb_mmap; /* The main data file */
-#define me_map me_dxb_mmap.dxb
-#define me_lazy_fd me_dxb_mmap.fd
- mdbx_filehandle_t me_dsync_fd;
- mdbx_mmap_t me_lck_mmap; /* The lock file */
-#define me_lfd me_lck_mmap.fd
- struct MDBX_lockinfo *me_lck;
-
- unsigned me_psize; /* DB page size, initialized from me_os_psize */
- unsigned me_leaf_nodemax; /* max size of a leaf-node */
- uint8_t me_psize2log; /* log2 of DB page size */
- int8_t me_stuck_meta; /* recovery-only: target meta page or less that zero */
- uint16_t me_merge_threshold,
- me_merge_threshold_gc; /* pages emptier than this are candidates for
- merging */
- unsigned me_os_psize; /* OS page size, from mdbx_syspagesize() */
- unsigned me_maxreaders; /* size of the reader table */
- MDBX_dbi me_maxdbs; /* size of the DB table */
- uint32_t me_pid; /* process ID of this env */
- mdbx_thread_key_t me_txkey; /* thread-key for readers */
- char *me_pathname; /* path to the DB files */
- void *me_pbuf; /* scratch area for DUPSORT put() */
- MDBX_txn *me_txn0; /* preallocated write transaction */
-
- MDBX_dbx *me_dbxs; /* array of static DB info */
- uint16_t *me_dbflags; /* array of flags from MDBX_db.md_flags */
- unsigned *me_dbiseqs; /* array of dbi sequence numbers */
- unsigned
- me_maxgc_ov1page; /* Number of pgno_t fit in a single overflow page */
- uint32_t me_live_reader; /* have liveness lock in reader table */
- void *me_userctx; /* User-settable context */
- MDBX_hsr_func *me_hsr_callback; /* Callback for kicking laggard readers */
-
- struct {
- unsigned dp_reserve_limit;
- unsigned rp_augment_limit;
- unsigned dp_limit;
- unsigned dp_initial;
- uint8_t dp_loose_limit;
- uint8_t spill_max_denominator;
- uint8_t spill_min_denominator;
- uint8_t spill_parent4child_denominator;
- unsigned merge_threshold_16dot16_percent;
- union {
- unsigned all;
- /* tracks options with non-auto values but tuned by user */
- struct {
- unsigned dp_limit : 1;
- } non_auto;
- } flags;
- } me_options;
-
- /* struct me_dbgeo used for accepting db-geo params from user for the new
- * database creation, i.e. when mdbx_env_set_geometry() was called before
- * mdbx_env_open(). */
- struct {
- size_t lower; /* minimal size of datafile */
- size_t upper; /* maximal size of datafile */
- size_t now; /* current size of datafile */
- size_t grow; /* step to grow datafile */
- size_t shrink; /* threshold to shrink datafile */
- } me_dbgeo;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- union {
- key_t key;
- int semid;
- } me_sysv_ipc;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
- MDBX_env *me_lcklist_next;
-
- /* --------------------------------------------------- mostly volatile part */
-
- MDBX_txn *me_txn; /* current write transaction */
- mdbx_fastmutex_t me_dbi_lock;
- MDBX_dbi me_numdbs; /* number of DBs opened */
-
- MDBX_page *me_dp_reserve; /* list of malloc'ed blocks for re-use */
- unsigned me_dp_reserve_len;
- /* PNL of pages that became unused in a write txn */
- MDBX_PNL me_retired_pages;
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_srwlock me_remap_guard;
- /* Workaround for LockFileEx and WriteFile multithread bug */
- CRITICAL_SECTION me_windowsbug_lock;
-#else
- mdbx_fastmutex_t me_remap_guard;
-#endif
-
- /* -------------------------------------------------------------- debugging */
-
-#if MDBX_DEBUG
- MDBX_assert_func *me_assert_func; /* Callback for assertion failures */
-#endif
-#ifdef MDBX_USE_VALGRIND
- int me_valgrind_handle;
-#endif
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- pgno_t me_poison_edge;
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-#ifndef xMDBX_DEBUG_SPILLING
-#define xMDBX_DEBUG_SPILLING 0
-#endif
-#if xMDBX_DEBUG_SPILLING == 2
- unsigned debug_dirtied_est, debug_dirtied_act;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- /* ------------------------------------------------- stub for lck-less mode */
- MDBX_atomic_uint64_t
- x_lckless_stub[(sizeof(MDBX_lockinfo) + MDBX_CACHELINE_SIZE - 1) /
- sizeof(MDBX_atomic_uint64_t)];
-};
-
-#ifndef __cplusplus
-/*----------------------------------------------------------------------------*/
-/* Debug and Logging stuff */
-
-#define MDBX_RUNTIME_FLAGS_INIT \
- ((MDBX_DEBUG) > 0) * MDBX_DBG_ASSERT + ((MDBX_DEBUG) > 1) * MDBX_DBG_AUDIT
-
-extern uint8_t mdbx_runtime_flags;
-extern uint8_t mdbx_loglevel;
-extern MDBX_debug_func *mdbx_debug_logger;
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny) {
-#if MDBX_DEBUG
- if (MDBX_DBG_JITTER & mdbx_runtime_flags)
- mdbx_osal_jitter(tiny);
-#else
- (void)tiny;
-#endif
-}
-
-MDBX_INTERNAL_FUNC void MDBX_PRINTF_ARGS(4, 5)
- mdbx_debug_log(int level, const char *function, int line, const char *fmt,
- ...) MDBX_PRINTF_ARGS(4, 5);
-MDBX_INTERNAL_FUNC void mdbx_debug_log_va(int level, const char *function,
- int line, const char *fmt,
- va_list args);
-
-#if MDBX_DEBUG
-#define mdbx_log_enabled(msg) unlikely(msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() unlikely((mdbx_runtime_flags & MDBX_DBG_AUDIT))
-#else /* MDBX_DEBUG */
-#define mdbx_log_enabled(msg) (msg < MDBX_LOG_VERBOSE && msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() (0)
-#endif /* MDBX_DEBUG */
-
-#if MDBX_FORCE_ASSERTIONS
-#define mdbx_assert_enabled() (1)
-#elif MDBX_DEBUG
-#define mdbx_assert_enabled() likely((mdbx_runtime_flags & MDBX_DBG_ASSERT))
-#else
-#define mdbx_assert_enabled() (0)
-#endif /* assertions */
-
-#if !MDBX_DEBUG && defined(__ANDROID_API__)
-#define mdbx_assert_fail(env, msg, func, line) \
- __android_log_assert(msg, "mdbx", "%s:%u", func, line)
-#else
-void mdbx_assert_fail(const MDBX_env *env, const char *msg, const char *func,
- int line);
-#endif
-
-#define mdbx_debug_extra(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, __func__, __LINE__, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug_extra_print(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, NULL, 0, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_trace(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_TRACE)) \
- mdbx_debug_log(MDBX_LOG_TRACE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_DEBUG)) \
- mdbx_debug_log(MDBX_LOG_DEBUG, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_verbose(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_VERBOSE)) \
- mdbx_debug_log(MDBX_LOG_VERBOSE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_notice(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_NOTICE)) \
- mdbx_debug_log(MDBX_LOG_NOTICE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_warning(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_WARN)) \
- mdbx_debug_log(MDBX_LOG_WARN, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_error(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) \
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_fatal(fmt, ...) \
- mdbx_debug_log(MDBX_LOG_FATAL, __func__, __LINE__, fmt "\n", __VA_ARGS__);
-
-#define mdbx_ensure_msg(env, expr, msg) \
- do { \
- if (unlikely(!(expr))) \
- mdbx_assert_fail(env, msg, __func__, __LINE__); \
- } while (0)
-
-#define mdbx_ensure(env, expr) mdbx_ensure_msg(env, expr, #expr)
-
-/* assert(3) variant in environment context */
-#define mdbx_assert(env, expr) \
- do { \
- if (mdbx_assert_enabled()) \
- mdbx_ensure(env, expr); \
- } while (0)
-
-/* assert(3) variant in cursor context */
-#define mdbx_cassert(mc, expr) mdbx_assert((mc)->mc_txn->mt_env, expr)
-
-/* assert(3) variant in transaction context */
-#define mdbx_tassert(txn, expr) mdbx_assert((txn)->mt_env, expr)
-
-#ifndef xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-#undef assert
-#define assert(expr) mdbx_assert(NULL, expr)
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Cache coherence and mmap invalidation */
-
-#if MDBX_CPU_WRITEBACK_INCOHERENT
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_memory_barrier()
-#else
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_compiler_barrier()
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-MDBX_MAYBE_UNUSED static __inline void
-mdbx_flush_incoherent_mmap(void *addr, size_t nbytes, const intptr_t pagesize) {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- char *const begin = (char *)(-pagesize & (intptr_t)addr);
- char *const end =
- (char *)(-pagesize & (intptr_t)((char *)addr + nbytes + pagesize - 1));
- int err = msync(begin, end - begin, MS_SYNC | MS_INVALIDATE) ? errno : 0;
- mdbx_assert(nullptr, err == 0);
- (void)err;
-#else
- (void)pagesize;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#if MDBX_MMAP_INCOHERENT_CPU_CACHE
-#ifdef DCACHE
- /* MIPS has cache coherency issues.
- * Note: for any nbytes >= on-chip cache size, entire is flushed. */
- cacheflush(addr, nbytes, DCACHE);
-#else
-#error "Oops, cacheflush() not available"
-#endif /* DCACHE */
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#if !MDBX_MMAP_INCOHERENT_FILE_WRITE && !MDBX_MMAP_INCOHERENT_CPU_CACHE
- (void)addr;
- (void)nbytes;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* Internal prototypes */
-
-MDBX_INTERNAL_FUNC int mdbx_cleanup_dead_readers(MDBX_env *env, int rlocked,
- int *dead);
-MDBX_INTERNAL_FUNC int mdbx_rthc_alloc(mdbx_thread_key_t *key,
- MDBX_reader *begin, MDBX_reader *end);
-MDBX_INTERNAL_FUNC void mdbx_rthc_remove(const mdbx_thread_key_t key);
-
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_init(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_dtor(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_thread_dtor(void *ptr);
-
-#endif /* !__cplusplus */
-
-#define MDBX_IS_ERROR(rc) \
- ((rc) != MDBX_RESULT_TRUE && (rc) != MDBX_RESULT_FALSE)
-
-/* Internal error codes, not exposed outside libmdbx */
-#define MDBX_NO_ROOT (MDBX_LAST_ADDED_ERRCODE + 10)
-
-/* Debugging output value of a cursor DBI: Negative in a sub-cursor. */
-#define DDBI(mc) \
- (((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
-
-/* Key size which fits in a DKBUF (debug key buffer). */
-#define DKBUF_MAX 511
-#define DKBUF char _kbuf[DKBUF_MAX * 4 + 2]
-#define DKEY(x) mdbx_dump_val(x, _kbuf, DKBUF_MAX * 2 + 1)
-#define DVAL(x) mdbx_dump_val(x, _kbuf + DKBUF_MAX * 2 + 1, DKBUF_MAX * 2 + 1)
-
-#if MDBX_DEBUG
-#define DKBUF_DEBUG DKBUF
-#define DKEY_DEBUG(x) DKEY(x)
-#define DVAL_DEBUG(x) DVAL(x)
-#else
-#define DKBUF_DEBUG ((void)(0))
-#define DKEY_DEBUG(x) ("-")
-#define DVAL_DEBUG(x) ("-")
-#endif
-
-/* An invalid page number.
- * Mainly used to denote an empty tree. */
-#define P_INVALID (~(pgno_t)0)
-
-/* Test if the flags f are set in a flag word w. */
-#define F_ISSET(w, f) (((w) & (f)) == (f))
-
-/* Round n up to an even number. */
-#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
-
-/* Default size of memory map.
- * This is certainly too small for any actual applications. Apps should
- * always set the size explicitly using mdbx_env_set_geometry(). */
-#define DEFAULT_MAPSIZE MEGABYTE
-
-/* Number of slots in the reader table.
- * This value was chosen somewhat arbitrarily. The 61 is a prime number,
- * and such readers plus a couple mutexes fit into single 4KB page.
- * Applications should set the table size using mdbx_env_set_maxreaders(). */
-#define DEFAULT_READERS 61
-
-/* Test if a page is a leaf page */
-#define IS_LEAF(p) (((p)->mp_flags & P_LEAF) != 0)
-/* Test if a page is a LEAF2 page */
-#define IS_LEAF2(p) unlikely(((p)->mp_flags & P_LEAF2) != 0)
-/* Test if a page is a branch page */
-#define IS_BRANCH(p) (((p)->mp_flags & P_BRANCH) != 0)
-/* Test if a page is an overflow page */
-#define IS_OVERFLOW(p) unlikely(((p)->mp_flags & P_OVERFLOW) != 0)
-/* Test if a page is a sub page */
-#define IS_SUBP(p) (((p)->mp_flags & P_SUBP) != 0)
-
-#define PAGETYPE(p) ((p)->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW))
-
-/* Header for a single key/data pair within a page.
- * Used in pages of type P_BRANCH and P_LEAF without P_LEAF2.
- * We guarantee 2-byte alignment for 'MDBX_node's.
- *
- * Leaf node flags describe node contents. F_BIGDATA says the node's
- * data part is the page number of an overflow page with actual data.
- * F_DUPDATA and F_SUBDATA can be combined giving duplicate data in
- * a sub-page/sub-database, and named databases (just F_SUBDATA). */
-typedef struct MDBX_node {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- union {
- uint32_t mn_dsize;
- uint32_t mn_pgno32;
- };
- uint8_t mn_flags; /* see mdbx_node flags */
- uint8_t mn_extra;
- uint16_t mn_ksize; /* key size */
-#else
- uint16_t mn_ksize; /* key size */
- uint8_t mn_extra;
- uint8_t mn_flags; /* see mdbx_node flags */
- union {
- uint32_t mn_pgno32;
- uint32_t mn_dsize;
- };
-#endif /* __BYTE_ORDER__ */
-
- /* mdbx_node Flags */
-#define F_BIGDATA 0x01 /* data put on overflow page */
-#define F_SUBDATA 0x02 /* data is a sub-database */
-#define F_DUPDATA 0x04 /* data has duplicates */
-
- /* valid flags for mdbx_node_add() */
-#define NODE_ADD_FLAGS (F_DUPDATA | F_SUBDATA | MDBX_RESERVE | MDBX_APPEND)
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- uint8_t mn_data[] /* key and data are appended here */;
-#endif /* C99 */
-} MDBX_node;
-
-#define DB_PERSISTENT_FLAGS \
- (MDBX_REVERSEKEY | MDBX_DUPSORT | MDBX_INTEGERKEY | MDBX_DUPFIXED | \
- MDBX_INTEGERDUP | MDBX_REVERSEDUP)
-
-/* mdbx_dbi_open() flags */
-#define DB_USABLE_FLAGS (DB_PERSISTENT_FLAGS | MDBX_CREATE | MDBX_DB_ACCEDE)
-
-#define DB_VALID 0x8000 /* DB handle is valid, for me_dbflags */
-#define DB_INTERNAL_FLAGS DB_VALID
-
-#if DB_INTERNAL_FLAGS & DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-#if DB_PERSISTENT_FLAGS & ~DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-
-/* max number of pages to commit in one writev() call */
-#define MDBX_COMMIT_PAGES 64
-#if defined(IOV_MAX) && IOV_MAX < MDBX_COMMIT_PAGES /* sysconf(_SC_IOV_MAX) */
-#undef MDBX_COMMIT_PAGES
-#define MDBX_COMMIT_PAGES IOV_MAX
-#endif
-
-/*
- * /
- * | -1, a < b
- * CMP2INT(a,b) = < 0, a == b
- * | 1, a > b
- * \
- */
-#ifndef __e2k__
-/* LY: fast enough on most systems */
-#define CMP2INT(a, b) (((b) > (a)) ? -1 : (a) > (b))
-#else
-/* LY: more parallelable on VLIW Elbrus */
-#define CMP2INT(a, b) (((a) > (b)) - ((b) > (a)))
-#endif
-
-/* Do not spill pages to disk if txn is getting full, may fail instead */
-#define MDBX_NOSPILL 0x8000
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-int64pgno(int64_t i64) {
- if (likely(i64 >= (int64_t)MIN_PAGENO && i64 <= (int64_t)MAX_PAGENO + 1))
- return (pgno_t)i64;
- return (i64 < (int64_t)MIN_PAGENO) ? MIN_PAGENO : MAX_PAGENO;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_add(size_t base, size_t augend) {
- assert(base <= MAX_PAGENO + 1 && augend < MAX_PAGENO);
- return int64pgno(base + augend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_sub(size_t base, size_t subtrahend) {
- assert(base >= MIN_PAGENO && base <= MAX_PAGENO + 1 &&
- subtrahend < MAX_PAGENO);
- return int64pgno(base - subtrahend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline bool
-is_powerof2(size_t x) {
- return (x & (x - 1)) == 0;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-floor_powerof2(size_t value, size_t granularity) {
- assert(is_powerof2(granularity));
- return value & ~(granularity - 1);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-ceil_powerof2(size_t value, size_t granularity) {
- return floor_powerof2(value + granularity - 1, granularity);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static unsigned
-log2n_powerof2(size_t value) {
- assert(value > 0 && value < INT32_MAX && is_powerof2(value));
- assert((value & -(int32_t)value) == value);
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_ctzl)
- return __builtin_ctzl(value);
-#elif defined(_MSC_VER)
- unsigned long index;
- _BitScanForward(&index, (unsigned long)value);
- return index;
-#else
- static const uint8_t debruijn_ctz32[32] = {
- 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
- 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9};
- return debruijn_ctz32[(uint32_t)(value * 0x077CB531u) >> 27];
-#endif
-}
-
-/* Only a subset of the mdbx_env flags can be changed
- * at runtime. Changing other flags requires closing the
- * environment and re-opening it with the new flags. */
-#define ENV_CHANGEABLE_FLAGS \
- (MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC | MDBX_DEPRECATED_MAPASYNC | \
- MDBX_NOMEMINIT | MDBX_COALESCE | MDBX_PAGEPERTURB | MDBX_ACCEDE)
-#define ENV_CHANGELESS_FLAGS \
- (MDBX_NOSUBDIR | MDBX_RDONLY | MDBX_WRITEMAP | MDBX_NOTLS | MDBX_NORDAHEAD | \
- MDBX_LIFORECLAIM | MDBX_EXCLUSIVE)
-#define ENV_USABLE_FLAGS (ENV_CHANGEABLE_FLAGS | ENV_CHANGELESS_FLAGS)
-
-#if !defined(__cplusplus) || CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_MAYBE_UNUSED static void static_checks(void) {
- STATIC_ASSERT_MSG(INT16_MAX - CORE_DBS == MDBX_MAX_DBI,
- "Oops, MDBX_MAX_DBI or CORE_DBS?");
- STATIC_ASSERT_MSG((unsigned)(MDBX_DB_ACCEDE | MDBX_CREATE) ==
- ((DB_USABLE_FLAGS | DB_INTERNAL_FLAGS) &
- (ENV_USABLE_FLAGS | ENV_INTERNAL_FLAGS)),
- "Oops, some flags overlapped or wrong");
- STATIC_ASSERT_MSG((ENV_INTERNAL_FLAGS & ENV_USABLE_FLAGS) == 0,
- "Oops, some flags overlapped or wrong");
-}
-#endif /* Disabled for MSVC 19.0 (VisualStudio 2015) */
-
-#ifdef __cplusplus
-}
-#endif
-
-#define MDBX_ASAN_POISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("POISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_POISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#define MDBX_ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("UNPOISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_UNPOISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#include <ctype.h>
-
-#if defined(_WIN32) || defined(_WIN64)
-/*
- * POSIX getopt for Windows
- *
- * AT&T Public License
- *
- * Code given out at the 1985 UNIFORUM conference in Dallas.
- */
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif
-#endif /* _MSC_VER (warnings) */
-
-#include <stdio.h>
-#include <string.h>
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-/*----------------------------------------------------------------------------*/
-
-#ifndef NULL
-#define NULL 0
-#endif
-
-#ifndef EOF
-#define EOF (-1)
-#endif
-
-int optind = 1;
-int optopt;
-char *optarg;
-
-int getopt(int argc, char *const argv[], const char *opts) {
- static int sp = 1;
- int c;
- const char *cp;
-
- if (sp == 1) {
- if (optind >= argc || argv[optind][0] != '-' || argv[optind][1] == '\0')
- return EOF;
- else if (strcmp(argv[optind], "--") == 0) {
- optind++;
- return EOF;
- }
- }
- optopt = c = argv[optind][sp];
- if (c == ':' || (cp = strchr(opts, c)) == NULL) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "illegal option", c);
- if (argv[optind][++sp] == '\0') {
- optind++;
- sp = 1;
- }
- return '?';
- }
- if (*++cp == ':') {
- if (argv[optind][sp + 1] != '\0')
- optarg = &argv[optind++][sp + 1];
- else if (++optind >= argc) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "option requires an argument",
- c);
- sp = 1;
- return '?';
- } else
- optarg = argv[optind++];
- sp = 1;
- } else {
- if (argv[optind][++sp] == '\0') {
- sp = 1;
- optind++;
- }
- optarg = NULL;
- }
- return c;
-}
-
-static volatile BOOL user_break;
-static BOOL WINAPI ConsoleBreakHandlerRoutine(DWORD dwCtrlType) {
- (void)dwCtrlType;
- user_break = true;
- return true;
-}
-
-#else /* WINDOWS */
-
-static volatile sig_atomic_t user_break;
-static void signal_handler(int sig) {
- (void)sig;
- user_break = 1;
-}
-
-#endif /* !WINDOWS */
-
-static char *prog;
-static bool quiet = false;
-static size_t lineno;
-static void error(const char *func, int rc) {
- if (!quiet) {
- if (lineno)
- fprintf(stderr, "%s: at input line %" PRIiSIZE ": %s() error %d, %s\n",
- prog, lineno, func, rc, mdbx_strerror(rc));
- else
- fprintf(stderr, "%s: %s() error %d %s\n", prog, func, rc,
- mdbx_strerror(rc));
- }
-}
-
-static char *valstr(char *line, const char *item) {
- const size_t len = strlen(item);
- if (strncmp(line, item, len) != 0)
- return nullptr;
- if (line[len] != '=') {
- if (line[len] > ' ')
- return nullptr;
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": unexpected line format for '%s'\n", prog,
- lineno, item);
- exit(EXIT_FAILURE);
- }
- char *ptr = strchr(line, '\n');
- if (ptr)
- *ptr = '\0';
- return line + len + 1;
-}
-
-static bool valnum(char *line, const char *item, uint64_t *value) {
- char *str = valstr(line, item);
- if (!str)
- return false;
-
- char *end = nullptr;
- *value = strtoull(str, &end, 0);
- if (end && *end) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": unexpected number format for '%s'\n",
- prog, lineno, item);
- exit(EXIT_FAILURE);
- }
- return true;
-}
-
-static bool valbool(char *line, const char *item, bool *value) {
- uint64_t u64;
- if (!valnum(line, item, &u64))
- return false;
-
- if (u64 > 1) {
- if (!quiet)
- fprintf(stderr, "%s: line %" PRIiSIZE ": unexpected value for '%s'\n",
- prog, lineno, item);
- exit(EXIT_FAILURE);
- }
- *value = u64 != 0;
- return true;
-}
-
-/*----------------------------------------------------------------------------*/
-
-static char *subname = nullptr;
-static int dbi_flags;
-static txnid_t txnid;
-static uint64_t sequence;
-static MDBX_canary canary;
-static MDBX_envinfo envinfo;
-
-#define PRINT 1
-#define NOHDR 2
-#define GLOBAL 4
-static int mode = GLOBAL;
-
-static MDBX_val kbuf, dbuf;
-
-#define STRLENOF(s) (sizeof(s) - 1)
-
-typedef struct flagbit {
- unsigned bit;
- unsigned len;
- char *name;
-} flagbit;
-
-#define S(s) STRLENOF(s), s
-
-flagbit dbflags[] = {
- {MDBX_REVERSEKEY, S("reversekey")}, {MDBX_DUPSORT, S("duplicates")},
- {MDBX_DUPSORT, S("dupsort")}, {MDBX_INTEGERKEY, S("integerkey")},
- {MDBX_DUPFIXED, S("dupfixed")}, {MDBX_INTEGERDUP, S("integerdup")},
- {MDBX_REVERSEDUP, S("reversedup")}, {0, 0, nullptr}};
-
-static int readhdr(void) {
- /* reset parameters */
- if (subname) {
- free(subname);
- subname = nullptr;
- }
- dbi_flags = 0;
- txnid = 0;
- sequence = 0;
-
- while (true) {
- errno = 0;
- if (fgets(dbuf.iov_base, (int)dbuf.iov_len, stdin) == nullptr)
- return errno ? errno : EOF;
- if (user_break)
- return MDBX_EINTR;
-
- lineno++;
- uint64_t u64;
-
- if (valnum(dbuf.iov_base, "VERSION", &u64)) {
- if (u64 != 3) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": unsupported value %" PRIu64
- " for %s\n",
- prog, lineno, u64, "VERSION");
- exit(EXIT_FAILURE);
- }
- continue;
- }
-
- if (valnum(dbuf.iov_base, "db_pagesize", &u64)) {
- if (!(mode & GLOBAL) && envinfo.mi_dxb_pagesize != u64) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore value %" PRIu64
- " for '%s' in non-global context\n",
- prog, lineno, u64, "db_pagesize");
- } else if (u64 < MDBX_MIN_PAGESIZE || u64 > MDBX_MAX_PAGESIZE) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore unsupported value %" PRIu64
- " for %s\n",
- prog, lineno, u64, "db_pagesize");
- } else
- envinfo.mi_dxb_pagesize = (uint32_t)u64;
- continue;
- }
-
- char *str = valstr(dbuf.iov_base, "format");
- if (str) {
- if (strcmp(str, "print") == 0) {
- mode |= PRINT;
- continue;
- }
- if (strcmp(str, "bytevalue") == 0) {
- mode &= ~PRINT;
- continue;
- }
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": unsupported value '%s' for %s\n", prog,
- lineno, str, "format");
- exit(EXIT_FAILURE);
- }
-
- str = valstr(dbuf.iov_base, "database");
- if (str) {
- if (*str) {
- free(subname);
- subname = mdbx_strdup(str);
- if (!subname) {
- if (!quiet)
- perror("strdup()");
- exit(EXIT_FAILURE);
- }
- }
- continue;
- }
-
- str = valstr(dbuf.iov_base, "type");
- if (str) {
- if (strcmp(str, "btree") != 0) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": unsupported value '%s' for %s\n",
- prog, lineno, str, "type");
- free(subname);
- exit(EXIT_FAILURE);
- }
- continue;
- }
-
- if (valnum(dbuf.iov_base, "mapaddr", &u64)) {
- if (u64) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore unsupported value 0x%" PRIx64
- " for %s\n",
- prog, lineno, u64, "mapaddr");
- }
- continue;
- }
-
- if (valnum(dbuf.iov_base, "mapsize", &u64)) {
- if (!(mode & GLOBAL)) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore value %" PRIu64
- " for '%s' in non-global context\n",
- prog, lineno, u64, "mapsize");
- } else if (u64 < MIN_MAPSIZE || u64 > MAX_MAPSIZE64) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore unsupported value 0x%" PRIx64
- " for %s\n",
- prog, lineno, u64, "mapsize");
- } else
- envinfo.mi_mapsize = (size_t)u64;
- continue;
- }
-
- if (valnum(dbuf.iov_base, "maxreaders", &u64)) {
- if (!(mode & GLOBAL)) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore value %" PRIu64
- " for '%s' in non-global context\n",
- prog, lineno, u64, "maxreaders");
- } else if (u64 < 1 || u64 > MDBX_READERS_LIMIT) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore unsupported value 0x%" PRIx64
- " for %s\n",
- prog, lineno, u64, "maxreaders");
- } else
- envinfo.mi_maxreaders = (int)u64;
- continue;
- }
-
- if (valnum(dbuf.iov_base, "txnid", &u64)) {
- if (u64 < MIN_TXNID || u64 > MAX_TXNID) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore unsupported value 0x%" PRIx64
- " for %s\n",
- prog, lineno, u64, "txnid");
- } else
- txnid = u64;
- continue;
- }
-
- if (valnum(dbuf.iov_base, "sequence", &u64)) {
- sequence = u64;
- continue;
- }
-
- str = valstr(dbuf.iov_base, "geometry");
- if (str) {
- if (!(mode & GLOBAL)) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore values %s"
- " for '%s' in non-global context\n",
- prog, lineno, str, "geometry");
- } else if (sscanf(str,
- "l%" PRIu64 ",c%" PRIu64 ",u%" PRIu64 ",s%" PRIu64
- ",g%" PRIu64,
- &envinfo.mi_geo.lower, &envinfo.mi_geo.current,
- &envinfo.mi_geo.upper, &envinfo.mi_geo.shrink,
- &envinfo.mi_geo.grow) != 5) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": unexpected line format for '%s'\n",
- prog, lineno, "geometry");
- exit(EXIT_FAILURE);
- }
- continue;
- }
-
- str = valstr(dbuf.iov_base, "canary");
- if (str) {
- if (!(mode & GLOBAL)) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": ignore values %s"
- " for '%s' in non-global context\n",
- prog, lineno, str, "canary");
- } else if (sscanf(str, "v%" PRIu64 ",x%" PRIu64 ",y%" PRIu64 ",z%" PRIu64,
- &canary.v, &canary.x, &canary.y, &canary.z) != 4) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": unexpected line format for '%s'\n",
- prog, lineno, "canary");
- exit(EXIT_FAILURE);
- }
- continue;
- }
-
- for (int i = 0; dbflags[i].bit; i++) {
- bool value = false;
- if (valbool(dbuf.iov_base, dbflags[i].name, &value)) {
- if (value)
- dbi_flags |= dbflags[i].bit;
- else
- dbi_flags &= ~dbflags[i].bit;
- goto next;
- }
- }
-
- str = valstr(dbuf.iov_base, "HEADER");
- if (str) {
- if (strcmp(str, "END") == 0)
- return MDBX_SUCCESS;
- }
-
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": unrecognized keyword ignored: %s\n",
- prog, lineno, (char *)dbuf.iov_base);
- next:;
- }
- return EOF;
-}
-
-static int badend(void) {
- if (!quiet)
- fprintf(stderr, "%s: line %" PRIiSIZE ": unexpected end of input\n", prog,
- lineno);
- return errno ? errno : MDBX_ENODATA;
-}
-
-static __inline int unhex(unsigned char *c2) {
- int x, c;
- x = *c2++ & 0x4f;
- if (x & 0x40)
- x -= 55;
- c = x << 4;
- x = *c2 & 0x4f;
- if (x & 0x40)
- x -= 55;
- c |= x;
- return c;
-}
-
-__hot static int readline(MDBX_val *out, MDBX_val *buf) {
- unsigned char *c1, *c2, *end;
- size_t len, l2;
- int c;
-
- if (user_break)
- return MDBX_EINTR;
-
- errno = 0;
- if (!(mode & NOHDR)) {
- c = fgetc(stdin);
- if (c == EOF)
- return errno ? errno : EOF;
- if (c != ' ') {
- lineno++;
- errno = 0;
- if (fgets(buf->iov_base, (int)buf->iov_len, stdin)) {
- if (c == 'D' && !strncmp(buf->iov_base, "ATA=END", STRLENOF("ATA=END")))
- return EOF;
- }
- return badend();
- }
- }
- if (fgets(buf->iov_base, (int)buf->iov_len, stdin) == nullptr)
- return errno ? errno : EOF;
- lineno++;
-
- c1 = buf->iov_base;
- len = strlen((char *)c1);
- l2 = len;
-
- /* Is buffer too short? */
- while (c1[len - 1] != '\n') {
- buf->iov_base = mdbx_realloc(buf->iov_base, buf->iov_len * 2);
- if (!buf->iov_base) {
- if (!quiet)
- fprintf(stderr,
- "%s: line %" PRIiSIZE ": out of memory, line too long\n", prog,
- lineno);
- return MDBX_ENOMEM;
- }
- c1 = buf->iov_base;
- c1 += l2;
- errno = 0;
- if (fgets((char *)c1, (int)buf->iov_len + 1, stdin) == nullptr)
- return errno ? errno : EOF;
- buf->iov_len *= 2;
- len = strlen((char *)c1);
- l2 += len;
- }
- c1 = c2 = buf->iov_base;
- len = l2;
- c1[--len] = '\0';
- end = c1 + len;
-
- if (mode & PRINT) {
- while (c2 < end) {
- if (unlikely(*c2 == '\\')) {
- if (c2[1] == '\\') {
- *c1++ = '\\';
- } else {
- if (c2 + 3 > end || !isxdigit(c2[1]) || !isxdigit(c2[2]))
- return badend();
- *c1++ = (char)unhex(++c2);
- }
- c2 += 2;
- } else {
- /* copies are redundant when no escapes were used */
- *c1++ = *c2++;
- }
- }
- } else {
- /* odd length not allowed */
- if (len & 1)
- return badend();
- while (c2 < end) {
- if (!isxdigit(*c2) || !isxdigit(c2[1]))
- return badend();
- *c1++ = (char)unhex(c2);
- c2 += 2;
- }
- }
- c2 = out->iov_base = buf->iov_base;
- out->iov_len = c1 - c2;
-
- return MDBX_SUCCESS;
-}
-
-static void usage(void) {
- fprintf(stderr,
- "usage: %s "
- "[-V] [-q] [-a] [-f file] [-s name] [-N] [-p] [-T] [-r] [-n] dbpath\n"
- " -V\t\tprint version and exit\n"
- " -q\t\tbe quiet\n"
- " -a\t\tappend records in input order (required for custom "
- "comparators)\n"
- " -f file\tread from file instead of stdin\n"
- " -s name\tload into specified named subDB\n"
- " -N\t\tdon't overwrite existing records when loading, just skip "
- "ones\n"
- " -p\t\tpurge subDB before loading\n"
- " -T\t\tread plaintext\n"
- " -r\t\trescue mode (ignore errors to load corrupted DB dump)\n"
- " -n\t\tdon't use subdirectory for newly created database "
- "(MDBX_NOSUBDIR)\n",
- prog);
- exit(EXIT_FAILURE);
-}
-
-static int equal_or_greater(const MDBX_val *a, const MDBX_val *b) {
- return (a->iov_len == b->iov_len &&
- memcmp(a->iov_base, b->iov_base, a->iov_len) == 0)
- ? 0
- : 1;
-}
-
-int main(int argc, char *argv[]) {
- int i, rc;
- MDBX_env *env = nullptr;
- MDBX_txn *txn = nullptr;
- MDBX_cursor *mc = nullptr;
- MDBX_dbi dbi;
- char *envname = nullptr;
- int envflags = MDBX_SAFE_NOSYNC | MDBX_ACCEDE, putflags = MDBX_UPSERT;
- bool rescue = false;
- bool purge = false;
-
- prog = argv[0];
- if (argc < 2)
- usage();
-
- while ((i = getopt(argc, argv,
- "a"
- "f:"
- "n"
- "s:"
- "N"
- "p"
- "T"
- "V"
- "r"
- "q")) != EOF) {
- switch (i) {
- case 'V':
- printf("mdbx_load version %d.%d.%d.%d\n"
- " - source: %s %s, commit %s, tree %s\n"
- " - anchor: %s\n"
- " - build: %s for %s by %s\n"
- " - flags: %s\n"
- " - options: %s\n",
- mdbx_version.major, mdbx_version.minor, mdbx_version.release,
- mdbx_version.revision, mdbx_version.git.describe,
- mdbx_version.git.datetime, mdbx_version.git.commit,
- mdbx_version.git.tree, mdbx_sourcery_anchor, mdbx_build.datetime,
- mdbx_build.target, mdbx_build.compiler, mdbx_build.flags,
- mdbx_build.options);
- return EXIT_SUCCESS;
- case 'a':
- putflags |= MDBX_APPEND;
- break;
- case 'f':
- if (freopen(optarg, "r", stdin) == nullptr) {
- if (!quiet)
- fprintf(stderr, "%s: %s: open: %s\n", prog, optarg,
- mdbx_strerror(errno));
- exit(EXIT_FAILURE);
- }
- break;
- case 'n':
- envflags |= MDBX_NOSUBDIR;
- break;
- case 's':
- subname = mdbx_strdup(optarg);
- break;
- case 'N':
- putflags |= MDBX_NOOVERWRITE | MDBX_NODUPDATA;
- break;
- case 'p':
- purge = true;
- break;
- case 'T':
- mode |= NOHDR | PRINT;
- break;
- case 'q':
- quiet = true;
- break;
- case 'r':
- rescue = true;
- break;
- default:
- usage();
- }
- }
-
- if (optind != argc - 1)
- usage();
-
-#if defined(_WIN32) || defined(_WIN64)
- SetConsoleCtrlHandler(ConsoleBreakHandlerRoutine, true);
-#else
-#ifdef SIGPIPE
- signal(SIGPIPE, signal_handler);
-#endif
-#ifdef SIGHUP
- signal(SIGHUP, signal_handler);
-#endif
- signal(SIGINT, signal_handler);
- signal(SIGTERM, signal_handler);
-#endif /* !WINDOWS */
-
- envname = argv[optind];
- if (!quiet)
- printf("mdbx_load %s (%s, T-%s)\nRunning for %s...\n",
- mdbx_version.git.describe, mdbx_version.git.datetime,
- mdbx_version.git.tree, envname);
- fflush(nullptr);
-
- dbuf.iov_len = 4096;
- dbuf.iov_base = mdbx_malloc(dbuf.iov_len);
- if (!dbuf.iov_base) {
- rc = MDBX_ENOMEM;
- error("value-buffer", rc);
- goto env_close;
- }
-
- /* read first header for mapsize= */
- if (!(mode & NOHDR)) {
- rc = readhdr();
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc == EOF)
- rc = MDBX_ENODATA;
- error("readheader", rc);
- goto env_close;
- }
- }
-
- rc = mdbx_env_create(&env);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_create", rc);
- return EXIT_FAILURE;
- }
-
- mdbx_env_set_maxdbs(env, 2);
- if (envinfo.mi_maxreaders) {
- rc = mdbx_env_set_maxreaders(env, envinfo.mi_maxreaders);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_set_maxreaders", rc);
- goto env_close;
- }
- }
-
- if (envinfo.mi_geo.current | envinfo.mi_mapsize) {
- if (envinfo.mi_geo.current) {
- rc = mdbx_env_set_geometry(
- env, (intptr_t)envinfo.mi_geo.lower, (intptr_t)envinfo.mi_geo.current,
- (intptr_t)envinfo.mi_geo.upper, (intptr_t)envinfo.mi_geo.shrink,
- (intptr_t)envinfo.mi_geo.grow,
- envinfo.mi_dxb_pagesize ? (intptr_t)envinfo.mi_dxb_pagesize : -1);
- } else {
- if (envinfo.mi_mapsize > MAX_MAPSIZE) {
- if (!quiet)
- fprintf(
- stderr,
- "Database size is too large for current system (mapsize=%" PRIu64
- " is great than system-limit %zu)\n",
- envinfo.mi_mapsize, (size_t)MAX_MAPSIZE);
- goto env_close;
- }
- rc = mdbx_env_set_geometry(
- env, (intptr_t)envinfo.mi_mapsize, (intptr_t)envinfo.mi_mapsize,
- (intptr_t)envinfo.mi_mapsize, 0, 0,
- envinfo.mi_dxb_pagesize ? (intptr_t)envinfo.mi_dxb_pagesize : -1);
- }
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_set_geometry", rc);
- goto env_close;
- }
- }
-
- rc = mdbx_env_open(env, envname, envflags, 0664);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_open", rc);
- goto env_close;
- }
-
- kbuf.iov_len = mdbx_env_get_maxvalsize_ex(env, 0) + 1;
- if (kbuf.iov_len >= INTPTR_MAX / 2) {
- if (!quiet)
- fprintf(stderr, "mdbx_env_get_maxkeysize() failed, returns %zu\n",
- kbuf.iov_len);
- goto env_close;
- }
-
- kbuf.iov_base = malloc(kbuf.iov_len);
- if (!kbuf.iov_base) {
- rc = MDBX_ENOMEM;
- error("key-buffer", rc);
- goto env_close;
- }
-
- while (rc == MDBX_SUCCESS) {
- if (user_break) {
- rc = MDBX_EINTR;
- break;
- }
-
- rc = mdbx_txn_begin(env, nullptr, 0, &txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_begin", rc);
- goto env_close;
- }
-
- if (mode & GLOBAL) {
- mode -= GLOBAL;
- if (canary.v | canary.x | canary.y | canary.z) {
- rc = mdbx_canary_put(txn, &canary);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_canary_put", rc);
- goto txn_abort;
- }
- }
- }
-
- const char *const dbi_name = subname ? subname : "@MAIN";
- rc =
- mdbx_dbi_open_ex(txn, subname, dbi_flags | MDBX_CREATE, &dbi,
- (putflags & MDBX_APPEND) ? equal_or_greater : nullptr,
- (putflags & MDBX_APPEND) ? equal_or_greater : nullptr);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_open_ex", rc);
- goto txn_abort;
- }
-
- uint64_t present_sequence;
- rc = mdbx_dbi_sequence(txn, dbi, &present_sequence, 0);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_sequence", rc);
- goto txn_abort;
- }
- if (present_sequence > sequence) {
- if (!quiet)
- fprintf(stderr,
- "present sequence for '%s' value (%" PRIu64
- ") is greater than loaded (%" PRIu64 ")\n",
- dbi_name, present_sequence, sequence);
- rc = MDBX_RESULT_TRUE;
- goto txn_abort;
- }
- if (present_sequence < sequence) {
- rc = mdbx_dbi_sequence(txn, dbi, nullptr, sequence - present_sequence);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_sequence", rc);
- goto txn_abort;
- }
- }
-
- if (purge) {
- rc = mdbx_drop(txn, dbi, false);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_drop", rc);
- goto txn_abort;
- }
- }
-
- if (putflags & MDBX_APPEND)
- putflags = (dbi_flags & MDBX_DUPSORT) ? putflags | MDBX_APPENDDUP
- : putflags & ~MDBX_APPENDDUP;
-
- rc = mdbx_cursor_open(txn, dbi, &mc);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_cursor_open", rc);
- goto txn_abort;
- }
-
- int batch = 0;
- while (rc == MDBX_SUCCESS) {
- MDBX_val key, data;
- rc = readline(&key, &kbuf);
- if (rc == EOF)
- break;
-
- if (rc == MDBX_SUCCESS)
- rc = readline(&data, &dbuf);
- if (rc) {
- if (!quiet)
- fprintf(stderr, "%s: line %" PRIiSIZE ": failed to read key value\n",
- prog, lineno);
- goto txn_abort;
- }
-
- rc = mdbx_cursor_put(mc, &key, &data, putflags);
- if (rc == MDBX_KEYEXIST && putflags)
- continue;
- if (rc == MDBX_BAD_VALSIZE && rescue) {
- if (!quiet)
- fprintf(stderr, "%s: skip line %" PRIiSIZE ": due %s\n", prog, lineno,
- mdbx_strerror(rc));
- continue;
- }
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_cursor_put", rc);
- goto txn_abort;
- }
- batch++;
-
- MDBX_txn_info txn_info;
- rc = mdbx_txn_info(txn, &txn_info, false);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_info", rc);
- goto txn_abort;
- }
-
- if (batch == 10000 || txn_info.txn_space_dirty > MEGABYTE * 256) {
- rc = mdbx_txn_commit(txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_commit", rc);
- goto env_close;
- }
- batch = 0;
-
- rc = mdbx_txn_begin(env, nullptr, 0, &txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_begin", rc);
- goto env_close;
- }
- rc = mdbx_cursor_bind(txn, mc, dbi);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_cursor_bind", rc);
- goto txn_abort;
- }
- }
- }
-
- mdbx_cursor_close(mc);
- mc = nullptr;
- rc = mdbx_txn_commit(txn);
- txn = nullptr;
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_commit", rc);
- goto env_close;
- }
- if (subname) {
- assert(dbi != MAIN_DBI);
- rc = mdbx_dbi_close(env, dbi);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_close", rc);
- goto env_close;
- }
- } else {
- assert(dbi == MAIN_DBI);
- }
-
- /* try read next header */
- if (!(mode & NOHDR))
- rc = readhdr();
- else if (ferror(stdin) || feof(stdin))
- break;
- }
-
- switch (rc) {
- case EOF:
- rc = MDBX_SUCCESS;
- case MDBX_SUCCESS:
- break;
- case MDBX_EINTR:
- if (!quiet)
- fprintf(stderr, "Interrupted by signal/user\n");
- break;
- default:
- if (unlikely(rc != MDBX_SUCCESS))
- error("readline", rc);
- }
-
-txn_abort:
- mdbx_cursor_close(mc);
- mdbx_txn_abort(txn);
-env_close:
- mdbx_env_close(env);
- free(kbuf.iov_base);
- free(dbuf.iov_base);
-
- return rc ? EXIT_FAILURE : EXIT_SUCCESS;
-}
diff --git a/3rdparty/libmdbx/mdbx_stat.c b/3rdparty/libmdbx/mdbx_stat.c
deleted file mode 100644
index 68a46191eb4bc3d52b2e9290eea05253a074bc04..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/mdbx_stat.c
+++ /dev/null
@@ -1,4041 +0,0 @@
-/* mdbx_stat.c - memory-mapped database status tool */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#ifdef _MSC_VER
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#pragma warning(disable : 4996) /* The POSIX name is deprecated... */
-#endif /* _MSC_VER (warnings) */
-
-#define xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>. */
-
-#define MDBX_BUILD_SOURCERY c510b93d78b5848764f976ea67f06d278a7b99f141b4c98ceb095285f834f31c_v0_11_7_0_g40ec559c
-#ifdef MDBX_CONFIG_H
-#include MDBX_CONFIG_H
-#endif
-
-#define LIBMDBX_INTERNALS
-#ifdef xMDBX_TOOLS
-#define MDBX_DEPRECATED
-#endif /* xMDBX_TOOLS */
-
-#ifdef xMDBX_ALLOY
-/* Amalgamated build */
-#define MDBX_INTERNAL_FUNC static
-#define MDBX_INTERNAL_VAR static
-#else
-/* Non-amalgamated build */
-#define MDBX_INTERNAL_FUNC
-#define MDBX_INTERNAL_VAR extern
-#endif /* xMDBX_ALLOY */
-
-/** Disables using GNU/Linux libc extensions.
- * \ingroup build_option
- * \note This option couldn't be moved to the options.h since dependant
- * control macros/defined should be prepared before include the options.h */
-#ifndef MDBX_DISABLE_GNU_SOURCE
-#define MDBX_DISABLE_GNU_SOURCE 0
-#endif
-#if MDBX_DISABLE_GNU_SOURCE
-#undef _GNU_SOURCE
-#elif (defined(__linux__) || defined(__gnu_linux__)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif /* MDBX_DISABLE_GNU_SOURCE */
-
-/*----------------------------------------------------------------------------*/
-
-/* Should be defined before any includes */
-#if !defined(_FILE_OFFSET_BITS) && !defined(__ANDROID_API__) && \
- !defined(ANDROID)
-#define _FILE_OFFSET_BITS 64
-#endif
-
-#ifdef __APPLE__
-#define _DARWIN_C_SOURCE
-#endif
-
-#ifdef _MSC_VER
-#if _MSC_FULL_VER < 190024234
-/* Actually libmdbx was not tested with compilers older than 19.00.24234 (Visual
- * Studio 2015 Update 3). But you could remove this #error and try to continue
- * at your own risk. In such case please don't rise up an issues related ONLY to
- * old compilers.
- *
- * NOTE:
- * Unfortunately, there are several different builds of "Visual Studio" that
- * are called "Visual Studio 2015 Update 3".
- *
- * The 190024234 is used here because it is minimal version of Visual Studio
- * that was used for build and testing libmdbx in recent years. Soon this
- * value will be increased to 19.0.24241.7, since build and testing using
- * "Visual Studio 2015" will be performed only at https://ci.appveyor.com.
- *
- * Please ask Microsoft (but not us) for information about version differences
- * and how to and where you can obtain the latest "Visual Studio 2015" build
- * with all fixes.
- */
-#error \
- "At least \"Microsoft C/C++ Compiler\" version 19.00.24234 (Visual Studio 2015 Update 3) is required."
-#endif
-#ifndef _CRT_SECURE_NO_WARNINGS
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if _MSC_VER > 1800
-#pragma warning(disable : 4464) /* relative include path contains '..' */
-#endif
-#if _MSC_VER > 1913
-#pragma warning(disable : 5045) /* Compiler will insert Spectre mitigation... \
- */
-#endif
-#if _MSC_VER > 1914
-#pragma warning( \
- disable : 5105) /* winbase.h(9531): warning C5105: macro expansion \
- producing 'defined' has undefined behavior */
-#endif
-#pragma warning(disable : 4710) /* 'xyz': function not inlined */
-#pragma warning(disable : 4711) /* function 'xyz' selected for automatic \
- inline expansion */
-#pragma warning( \
- disable : 4201) /* nonstandard extension used : nameless struct / union */
-#pragma warning(disable : 4702) /* unreachable code */
-#pragma warning(disable : 4706) /* assignment within conditional expression */
-#pragma warning(disable : 4127) /* conditional expression is constant */
-#pragma warning(disable : 4324) /* 'xyz': structure was padded due to \
- alignment specifier */
-#pragma warning(disable : 4310) /* cast truncates constant value */
-#pragma warning( \
- disable : 4820) /* bytes padding added after data member for alignment */
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4366) /* the result of the unary '&' operator may be \
- unaligned */
-#pragma warning(disable : 4200) /* nonstandard extension used: zero-sized \
- array in struct/union */
-#pragma warning(disable : 4204) /* nonstandard extension used: non-constant \
- aggregate initializer */
-#pragma warning( \
- disable : 4505) /* unreferenced local function has been removed */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(__GNUC__) && __GNUC__ < 9
-#pragma GCC diagnostic ignored "-Wattributes"
-#endif /* GCC < 9 */
-
-#if (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* __USE_MINGW_ANSI_STDIO */
-
-#include "mdbx.h"
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-
-#ifndef __GNUC_PREREQ
-# if defined(__GNUC__) && defined(__GNUC_MINOR__)
-# define __GNUC_PREREQ(maj, min) \
- ((__GNUC__ << 16) + __GNUC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GNUC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GNUC_PREREQ */
-
-#ifndef __CLANG_PREREQ
-# ifdef __clang__
-# define __CLANG_PREREQ(maj,min) \
- ((__clang_major__ << 16) + __clang_minor__ >= ((maj) << 16) + (min))
-# else
-# define __CLANG_PREREQ(maj,min) (0)
-# endif
-#endif /* __CLANG_PREREQ */
-
-#ifndef __GLIBC_PREREQ
-# if defined(__GLIBC__) && defined(__GLIBC_MINOR__)
-# define __GLIBC_PREREQ(maj, min) \
- ((__GLIBC__ << 16) + __GLIBC_MINOR__ >= ((maj) << 16) + (min))
-# else
-# define __GLIBC_PREREQ(maj, min) (0)
-# endif
-#endif /* __GLIBC_PREREQ */
-
-#ifndef __has_warning
-# define __has_warning(x) (0)
-#endif
-
-#ifndef __has_include
-# define __has_include(x) (0)
-#endif
-
-#if __has_feature(thread_sanitizer)
-# define __SANITIZE_THREAD__ 1
-#endif
-
-#if __has_feature(address_sanitizer)
-# define __SANITIZE_ADDRESS__ 1
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __extern_C
-# ifdef __cplusplus
-# define __extern_C extern "C"
-# else
-# define __extern_C
-# endif
-#endif /* __extern_C */
-
-#if !defined(nullptr) && !defined(__cplusplus) || (__cplusplus < 201103L && !defined(_MSC_VER))
-# define nullptr NULL
-#endif
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef __always_inline
-# if defined(__GNUC__) || __has_attribute(__always_inline__)
-# define __always_inline __inline __attribute__((__always_inline__))
-# elif defined(_MSC_VER)
-# define __always_inline __forceinline
-# else
-# define __always_inline
-# endif
-#endif /* __always_inline */
-
-#ifndef __noinline
-# if defined(__GNUC__) || __has_attribute(__noinline__)
-# define __noinline __attribute__((__noinline__))
-# elif defined(_MSC_VER)
-# define __noinline __declspec(noinline)
-# else
-# define __noinline
-# endif
-#endif /* __noinline */
-
-#ifndef __must_check_result
-# if defined(__GNUC__) || __has_attribute(__warn_unused_result__)
-# define __must_check_result __attribute__((__warn_unused_result__))
-# else
-# define __must_check_result
-# endif
-#endif /* __must_check_result */
-
-#if !defined(__noop) && !defined(_MSC_VER)
-# define __noop(...) do {} while(0)
-#endif /* __noop */
-
-#if defined(__fallthrough) && \
- (defined(__MINGW__) || defined(__MINGW32__) || defined(__MINGW64__))
-#undef __fallthrough
-#endif /* __fallthrough workaround for MinGW */
-
-#ifndef __fallthrough
-# if defined(__cplusplus) && (__has_cpp_attribute(fallthrough) && \
- (!defined(__clang__) || __clang__ > 4)) || __cplusplus >= 201703L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(8, 0) && defined(__cplusplus) && __cplusplus >= 201103L
-# define __fallthrough [[fallthrough]]
-# elif __GNUC_PREREQ(7, 0) && \
- (!defined(__LCC__) || (__LCC__ == 124 && __LCC_MINOR__ >= 12) || \
- (__LCC__ == 125 && __LCC_MINOR__ >= 5) || (__LCC__ >= 126))
-# define __fallthrough __attribute__((__fallthrough__))
-# elif defined(__clang__) && defined(__cplusplus) && __cplusplus >= 201103L &&\
- __has_feature(cxx_attributes) && __has_warning("-Wimplicit-fallthrough")
-# define __fallthrough [[clang::fallthrough]]
-# else
-# define __fallthrough
-# endif
-#endif /* __fallthrough */
-
-#ifndef __unreachable
-# if __GNUC_PREREQ(4,5) || __has_builtin(__builtin_unreachable)
-# define __unreachable() __builtin_unreachable()
-# elif defined(_MSC_VER)
-# define __unreachable() __assume(0)
-# else
-# define __unreachable() __noop()
-# endif
-#endif /* __unreachable */
-
-#ifndef __prefetch
-# if defined(__GNUC__) || defined(__clang__) || __has_builtin(__builtin_prefetch)
-# define __prefetch(ptr) __builtin_prefetch(ptr)
-# else
-# define __prefetch(ptr) __noop(ptr)
-# endif
-#endif /* __prefetch */
-
-#ifndef __nothrow
-# if defined(__cplusplus)
-# if __cplusplus < 201703L
-# define __nothrow throw()
-# else
-# define __nothrow noexcept(true)
-# endif /* __cplusplus */
-# elif defined(__GNUC__) || __has_attribute(__nothrow__)
-# define __nothrow __attribute__((__nothrow__))
-# elif defined(_MSC_VER) && defined(__cplusplus)
-# define __nothrow __declspec(nothrow)
-# else
-# define __nothrow
-# endif
-#endif /* __nothrow */
-
-#ifndef __hidden
-# if defined(__GNUC__) || __has_attribute(__visibility__)
-# define __hidden __attribute__((__visibility__("hidden")))
-# else
-# define __hidden
-# endif
-#endif /* __hidden */
-
-#ifndef __optimize
-# if defined(__OPTIMIZE__)
-# if (defined(__GNUC__) && !defined(__clang__)) || __has_attribute(__optimize__)
-# define __optimize(ops) __attribute__((__optimize__(ops)))
-# else
-# define __optimize(ops)
-# endif
-# else
-# define __optimize(ops)
-# endif
-#endif /* __optimize */
-
-#ifndef __hot
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __hot __attribute__((__hot__)) __optimize(3)
-# elif defined(__clang__) && !__has_attribute(__hot_) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put frequently used functions in separate section */
-# define __hot __attribute__((__section__("text.hot"))) __optimize("O3")
-# elif defined(__GNUC__) || __has_attribute(__hot__)
-# define __hot __attribute__((__hot__)) __optimize("O3")
-# else
-# define __hot __optimize("O3")
-# endif
-# else
-# define __hot
-# endif
-#endif /* __hot */
-
-#ifndef __cold
-# if defined(__OPTIMIZE__)
-# if defined(__e2k__)
-# define __cold __attribute__((__cold__)) __optimize(1)
-# elif defined(__clang__) && !__has_attribute(cold) \
- && __has_attribute(__section__) && (defined(__linux__) || defined(__gnu_linux__))
- /* just put infrequently used functions in separate section */
-# define __cold __attribute__((__section__("text.unlikely"))) __optimize("Os")
-# elif defined(__GNUC__) || __has_attribute(cold)
-# define __cold __attribute__((__cold__)) __optimize("Os")
-# else
-# define __cold __optimize("Os")
-# endif
-# else
-# define __cold
-# endif
-#endif /* __cold */
-
-#ifndef __flatten
-# if defined(__OPTIMIZE__) && (defined(__GNUC__) || __has_attribute(__flatten__))
-# define __flatten __attribute__((__flatten__))
-# else
-# define __flatten
-# endif
-#endif /* __flatten */
-
-#ifndef likely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define likely(cond) __builtin_expect(!!(cond), 1)
-# else
-# define likely(x) (!!(x))
-# endif
-#endif /* likely */
-
-#ifndef unlikely
-# if (defined(__GNUC__) || __has_builtin(__builtin_expect)) && !defined(__COVERITY__)
-# define unlikely(cond) __builtin_expect(!!(cond), 0)
-# else
-# define unlikely(x) (!!(x))
-# endif
-#endif /* unlikely */
-
-#ifndef __anonymous_struct_extension__
-# if defined(__GNUC__)
-# define __anonymous_struct_extension__ __extension__
-# else
-# define __anonymous_struct_extension__
-# endif
-#endif /* __anonymous_struct_extension__ */
-
-#ifndef __Wpedantic_format_voidptr
- MDBX_MAYBE_UNUSED MDBX_PURE_FUNCTION static __inline const void*
- __Wpedantic_format_voidptr(const void* ptr) {return ptr;}
-# define __Wpedantic_format_voidptr(ARG) __Wpedantic_format_voidptr(ARG)
-#endif /* __Wpedantic_format_voidptr */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(MDBX_USE_VALGRIND)
-# include <valgrind/memcheck.h>
-# ifndef VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE
- /* LY: available since Valgrind 3.10 */
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# endif
-#elif !defined(RUNNING_ON_VALGRIND)
-# define VALGRIND_CREATE_MEMPOOL(h,r,z)
-# define VALGRIND_DESTROY_MEMPOOL(h)
-# define VALGRIND_MEMPOOL_TRIM(h,a,s)
-# define VALGRIND_MEMPOOL_ALLOC(h,a,s)
-# define VALGRIND_MEMPOOL_FREE(h,a)
-# define VALGRIND_MEMPOOL_CHANGE(h,a,b,s)
-# define VALGRIND_MAKE_MEM_NOACCESS(a,s)
-# define VALGRIND_MAKE_MEM_DEFINED(a,s)
-# define VALGRIND_MAKE_MEM_UNDEFINED(a,s)
-# define VALGRIND_DISABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_ENABLE_ADDR_ERROR_REPORTING_IN_RANGE(a,s)
-# define VALGRIND_CHECK_MEM_IS_ADDRESSABLE(a,s) (0)
-# define VALGRIND_CHECK_MEM_IS_DEFINED(a,s) (0)
-# define RUNNING_ON_VALGRIND (0)
-#endif /* MDBX_USE_VALGRIND */
-
-#ifdef __SANITIZE_ADDRESS__
-# include <sanitizer/asan_interface.h>
-#elif !defined(ASAN_POISON_MEMORY_REGION)
-# define ASAN_POISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-# define ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- ((void)(addr), (void)(size))
-#endif /* __SANITIZE_ADDRESS__ */
-
-/*----------------------------------------------------------------------------*/
-
-#ifndef ARRAY_LENGTH
-# ifdef __cplusplus
- template <typename T, size_t N>
- char (&__ArraySizeHelper(T (&array)[N]))[N];
-# define ARRAY_LENGTH(array) (sizeof(::__ArraySizeHelper(array)))
-# else
-# define ARRAY_LENGTH(array) (sizeof(array) / sizeof(array[0]))
-# endif
-#endif /* ARRAY_LENGTH */
-
-#ifndef ARRAY_END
-# define ARRAY_END(array) (&array[ARRAY_LENGTH(array)])
-#endif /* ARRAY_END */
-
-#define CONCAT(a,b) a##b
-#define XCONCAT(a,b) CONCAT(a,b)
-
-#ifndef offsetof
-# define offsetof(type, member) __builtin_offsetof(type, member)
-#endif /* offsetof */
-
-#ifndef container_of
-# define container_of(ptr, type, member) \
- ((type *)((char *)(ptr) - offsetof(type, member)))
-#endif /* container_of */
-
-#define MDBX_TETRAD(a, b, c, d) \
- ((uint32_t)(a) << 24 | (uint32_t)(b) << 16 | (uint32_t)(c) << 8 | (d))
-
-#define MDBX_STRING_TETRAD(str) MDBX_TETRAD(str[0], str[1], str[2], str[3])
-
-#define FIXME "FIXME: " __FILE__ ", " MDBX_STRINGIFY(__LINE__)
-
-#ifndef STATIC_ASSERT_MSG
-# if defined(static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) static_assert(expr, msg)
-# elif defined(_STATIC_ASSERT)
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif defined(_MSC_VER)
-# include <crtdbg.h>
-# define STATIC_ASSERT_MSG(expr, msg) _STATIC_ASSERT(expr)
-# elif (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) \
- || __has_feature(c_static_assert)
-# define STATIC_ASSERT_MSG(expr, msg) _Static_assert(expr, msg)
-# else
-# define STATIC_ASSERT_MSG(expr, msg) switch (0) {case 0:case (expr):;}
-# endif
-#endif /* STATIC_ASSERT */
-
-#ifndef STATIC_ASSERT
-# define STATIC_ASSERT(expr) STATIC_ASSERT_MSG(expr, #expr)
-#endif
-
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__GNUC__) && !__GNUC_PREREQ(4, 2)
-/* Actually libmdbx was not tested with compilers older than GCC 4.2.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required GCC >= 4.2"
-#endif
-
-#if defined(__clang__) && !__CLANG_PREREQ(3, 8)
-/* Actually libmdbx was not tested with CLANG older than 3.8.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old compilers.
- */
-#warning "libmdbx required CLANG >= 3.8"
-#endif
-
-#if defined(__GLIBC__) && !__GLIBC_PREREQ(2, 12)
-/* Actually libmdbx was not tested with something older than glibc 2.12.
- * But you could ignore this warning at your own risk.
- * In such case please don't rise up an issues related ONLY to old systems.
- */
-#warning "libmdbx was only tested with GLIBC >= 2.12."
-#endif
-
-#ifdef __SANITIZE_THREAD__
-#warning \
- "libmdbx don't compatible with ThreadSanitizer, you will get a lot of false-positive issues."
-#endif /* __SANITIZE_THREAD__ */
-
-#if __has_warning("-Wnested-anon-types")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wnested-anon-types"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wnested-anon-types"
-#else
-#pragma warning disable "nested-anon-types"
-#endif
-#endif /* -Wnested-anon-types */
-
-#if __has_warning("-Wconstant-logical-operand")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Wconstant-logical-operand"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Wconstant-logical-operand"
-#else
-#pragma warning disable "constant-logical-operand"
-#endif
-#endif /* -Wconstant-logical-operand */
-
-#if defined(__LCC__) && (__LCC__ <= 121)
-/* bug #2798 */
-#pragma diag_suppress alignment_reduction_ignored
-#elif defined(__ICC)
-#pragma warning(disable : 3453 1366)
-#elif __has_warning("-Walignment-reduction-ignored")
-#if defined(__clang__)
-#pragma clang diagnostic ignored "-Walignment-reduction-ignored"
-#elif defined(__GNUC__)
-#pragma GCC diagnostic ignored "-Walignment-reduction-ignored"
-#else
-#pragma warning disable "alignment-reduction-ignored"
-#endif
-#endif /* -Walignment-reduction-ignored */
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* https://en.wikipedia.org/wiki/Operating_system_abstraction_layer */
-
-/*
- * Copyright 2015-2022 Leonid Yuriev <leo@yuriev.ru>
- * and other libmdbx authors: please see AUTHORS file.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted only as authorized by the OpenLDAP
- * Public License.
- *
- * A copy of this license is available in the file LICENSE in the
- * top-level directory of the distribution or, alternatively, at
- * <http://www.OpenLDAP.org/license.html>.
- */
-
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#endif /* _MSC_VER (warnings) */
-
-#if defined(_WIN32) || defined(_WIN64)
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif /* _CRT_SECURE_NO_WARNINGS */
-#if !defined(_NO_CRT_STDIO_INLINE) && MDBX_BUILD_SHARED_LIBRARY && \
- !defined(xMDBX_TOOLS) && MDBX_WITHOUT_MSVC_CRT
-#define _NO_CRT_STDIO_INLINE
-#endif
-#elif !defined(_POSIX_C_SOURCE)
-#define _POSIX_C_SOURCE 200809L
-#endif /* Windows */
-
-/*----------------------------------------------------------------------------*/
-/* C99 includes */
-#include <inttypes.h>
-#include <stddef.h>
-#include <stdint.h>
-#include <stdlib.h>
-
-#include <assert.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <stdio.h>
-#include <string.h>
-#include <time.h>
-
-/* C11' alignas() */
-#if __has_include(<stdalign.h>)
-#include <stdalign.h>
-#endif
-#if defined(alignas) || defined(__cplusplus)
-#define MDBX_ALIGNAS(N) alignas(N)
-#elif defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L
-#define MDBX_ALIGNAS(N) _Alignas(N)
-#elif defined(_MSC_VER)
-#define MDBX_ALIGNAS(N) __declspec(align(N))
-#elif __has_attribute(__aligned__) || defined(__GNUC__)
-#define MDBX_ALIGNAS(N) __attribute__((__aligned__(N)))
-#else
-#error "FIXME: Required alignas() or equivalent."
-#endif /* MDBX_ALIGNAS */
-
-/*----------------------------------------------------------------------------*/
-/* Systems includes */
-
-#ifdef __APPLE__
-#include <TargetConditionals.h>
-#endif /* Apple OSX & iOS */
-
-#if defined(__FreeBSD__) || defined(__NetBSD__) || defined(__OpenBSD__) || \
- defined(__BSD__) || defined(__bsdi__) || defined(__DragonFly__) || \
- defined(__APPLE__) || defined(__MACH__)
-#include <sys/cdefs.h>
-#include <sys/mount.h>
-#include <sys/sysctl.h>
-#include <sys/types.h>
-#if defined(__FreeBSD__) || defined(__DragonFly__)
-#include <vm/vm_param.h>
-#elif defined(__OpenBSD__) || defined(__NetBSD__)
-#include <uvm/uvm_param.h>
-#else
-#define SYSCTL_LEGACY_NONCONST_MIB
-#endif
-#ifndef __MACH__
-#include <sys/vmmeter.h>
-#endif
-#else
-#include <malloc.h>
-#if !(defined(__sun) || defined(__SVR4) || defined(__svr4__) || \
- defined(_WIN32) || defined(_WIN64))
-#include <mntent.h>
-#endif /* !Solaris */
-#endif /* !xBSD */
-
-#if defined(__FreeBSD__) || __has_include(<malloc_np.h>)
-#include <malloc_np.h>
-#endif
-
-#if defined(__APPLE__) || defined(__MACH__) || __has_include(<malloc/malloc.h>)
-#include <malloc/malloc.h>
-#endif /* MacOS */
-
-#if defined(__MACH__)
-#include <mach/host_info.h>
-#include <mach/mach_host.h>
-#include <mach/mach_port.h>
-#include <uuid/uuid.h>
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-#include <sched.h>
-#include <sys/sendfile.h>
-#include <sys/statfs.h>
-#endif /* Linux */
-
-#ifndef _XOPEN_SOURCE
-#define _XOPEN_SOURCE 0
-#endif
-
-#ifndef _XOPEN_SOURCE_EXTENDED
-#define _XOPEN_SOURCE_EXTENDED 0
-#else
-#include <utmpx.h>
-#endif /* _XOPEN_SOURCE_EXTENDED */
-
-#if defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#include <kstat.h>
-#include <sys/mnttab.h>
-/* On Solaris, it's easier to add a missing prototype rather than find a
- * combination of #defines that break nothing. */
-__extern_C key_t ftok(const char *, int);
-#endif /* SunOS/Solaris */
-
-#if defined(_WIN32) || defined(_WIN64)
-#ifndef _WIN32_WINNT
-#define _WIN32_WINNT 0x0601 /* Windows 7 */
-#elif _WIN32_WINNT < 0x0500
-#error At least 'Windows 2000' API is required for libmdbx.
-#endif /* _WIN32_WINNT */
-#if (defined(__MINGW32__) || defined(__MINGW64__)) && \
- !defined(__USE_MINGW_ANSI_STDIO)
-#define __USE_MINGW_ANSI_STDIO 1
-#endif /* MinGW */
-#ifndef WIN32_LEAN_AND_MEAN
-#define WIN32_LEAN_AND_MEAN
-#endif /* WIN32_LEAN_AND_MEAN */
-#include <excpt.h>
-#include <tlhelp32.h>
-#include <windows.h>
-#include <winnt.h>
-#include <winternl.h>
-#define HAVE_SYS_STAT_H
-#define HAVE_SYS_TYPES_H
-typedef HANDLE mdbx_thread_t;
-typedef unsigned mdbx_thread_key_t;
-#define MAP_FAILED NULL
-#define HIGH_DWORD(v) ((DWORD)((sizeof(v) > 4) ? ((uint64_t)(v) >> 32) : 0))
-#define THREAD_CALL WINAPI
-#define THREAD_RESULT DWORD
-typedef struct {
- HANDLE mutex;
- HANDLE event[2];
-} mdbx_condpair_t;
-typedef CRITICAL_SECTION mdbx_fastmutex_t;
-
-#if !defined(_MSC_VER) && !defined(__try)
-/* *INDENT-OFF* */
-/* clang-format off */
-#define __try
-#define __except(COND) if(false)
-/* *INDENT-ON* */
-/* clang-format on */
-#endif /* stub for MSVC's __try/__except */
-
-#if MDBX_WITHOUT_MSVC_CRT
-
-#ifndef mdbx_malloc
-static inline void *mdbx_malloc(size_t bytes) {
- return HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_malloc */
-
-#ifndef mdbx_calloc
-static inline void *mdbx_calloc(size_t nelem, size_t size) {
- return HeapAlloc(GetProcessHeap(), HEAP_ZERO_MEMORY, nelem * size);
-}
-#endif /* mdbx_calloc */
-
-#ifndef mdbx_realloc
-static inline void *mdbx_realloc(void *ptr, size_t bytes) {
- return ptr ? HeapReAlloc(GetProcessHeap(), 0, ptr, bytes)
- : HeapAlloc(GetProcessHeap(), 0, bytes);
-}
-#endif /* mdbx_realloc */
-
-#ifndef mdbx_free
-static inline void mdbx_free(void *ptr) { HeapFree(GetProcessHeap(), 0, ptr); }
-#endif /* mdbx_free */
-
-#else /* MDBX_WITHOUT_MSVC_CRT */
-
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup _strdup
-
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-#ifndef snprintf
-#define snprintf _snprintf /* ntdll */
-#endif
-
-#ifndef vsnprintf
-#define vsnprintf _vsnprintf /* ntdll */
-#endif
-
-#else /*----------------------------------------------------------------------*/
-
-#include <unistd.h>
-#if !defined(_POSIX_MAPPED_FILES) || _POSIX_MAPPED_FILES < 1
-#error "libmdbx requires the _POSIX_MAPPED_FILES feature"
-#endif /* _POSIX_MAPPED_FILES */
-
-#include <pthread.h>
-#include <semaphore.h>
-#include <signal.h>
-#include <sys/file.h>
-#include <sys/ipc.h>
-#include <sys/mman.h>
-#include <sys/param.h>
-#include <sys/stat.h>
-#include <sys/statvfs.h>
-#include <sys/uio.h>
-typedef pthread_t mdbx_thread_t;
-typedef pthread_key_t mdbx_thread_key_t;
-#define INVALID_HANDLE_VALUE (-1)
-#define THREAD_CALL
-#define THREAD_RESULT void *
-typedef struct {
- pthread_mutex_t mutex;
- pthread_cond_t cond[2];
-} mdbx_condpair_t;
-typedef pthread_mutex_t mdbx_fastmutex_t;
-#define mdbx_malloc malloc
-#define mdbx_calloc calloc
-#define mdbx_realloc realloc
-#define mdbx_free free
-#define mdbx_strdup strdup
-#endif /* Platform */
-
-#if __GLIBC_PREREQ(2, 12) || defined(__FreeBSD__) || defined(malloc_usable_size)
-/* malloc_usable_size() already provided */
-#elif defined(__APPLE__)
-#define malloc_usable_size(ptr) malloc_size(ptr)
-#elif defined(_MSC_VER) && !MDBX_WITHOUT_MSVC_CRT
-#define malloc_usable_size(ptr) _msize(ptr)
-#endif /* malloc_usable_size */
-
-#ifdef __ANDROID_API__
-#include <android/log.h>
-#if __ANDROID_API__ >= 21
-#include <sys/sendfile.h>
-#endif
-#endif /* Android */
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(HAVE_SYS_STAT_H) || __has_include(<sys/stat.h>)
-#include <sys/stat.h>
-#endif
-#if defined(HAVE_SYS_TYPES_H) || __has_include(<sys/types.h>)
-#include <sys/types.h>
-#endif
-#if defined(HAVE_SYS_FILE_H) || __has_include(<sys/file.h>)
-#include <sys/file.h>
-#endif
-/* *INDENT-ON* */
-/* clang-format on */
-
-#ifndef SSIZE_MAX
-#define SSIZE_MAX INTPTR_MAX
-#endif
-
-#if !defined(MADV_DODUMP) && defined(MADV_CORE)
-#define MADV_DODUMP MADV_CORE
-#endif /* MADV_CORE -> MADV_DODUMP */
-
-#if !defined(MADV_DONTDUMP) && defined(MADV_NOCORE)
-#define MADV_DONTDUMP MADV_NOCORE
-#endif /* MADV_NOCORE -> MADV_DONTDUMP */
-
-#if defined(i386) || defined(__386) || defined(__i386) || defined(__i386__) || \
- defined(i486) || defined(__i486) || defined(__i486__) || \
- defined(i586) | defined(__i586) || defined(__i586__) || defined(i686) || \
- defined(__i686) || defined(__i686__) || defined(_M_IX86) || \
- defined(_X86_) || defined(__THW_INTEL__) || defined(__I86__) || \
- defined(__INTEL__) || defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64__) || defined(__amd64) || defined(_M_X64) || \
- defined(_M_AMD64) || defined(__IA32__) || defined(__INTEL__)
-#ifndef __ia32__
-/* LY: define neutral __ia32__ for x86 and x86-64 */
-#define __ia32__ 1
-#endif /* __ia32__ */
-#if !defined(__amd64__) && (defined(__x86_64) || defined(__x86_64__) || \
- defined(__amd64) || defined(_M_X64))
-/* LY: define trusty __amd64__ for all AMD64/x86-64 arch */
-#define __amd64__ 1
-#endif /* __amd64__ */
-#endif /* all x86 */
-
-#if (-6 & 5) || CHAR_BIT != 8 || UINT_MAX < 0xffffffff || ULONG_MAX % 0xFFFF
-#error \
- "Sanity checking failed: Two's complement, reasonably sized integer types"
-#endif
-
-#if UINTPTR_MAX > 0xffffFFFFul || ULONG_MAX > 0xffffFFFFul
-#define MDBX_WORDBITS 64
-#else
-#define MDBX_WORDBITS 32
-#endif /* MDBX_WORDBITS */
-
-#if defined(__ANDROID_API__) || defined(ANDROID)
-#if defined(_FILE_OFFSET_BITS) && _FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "_FILE_OFFSET_BITS != MDBX_WORDBITS" (_FILE_OFFSET_BITS != MDBX_WORDBITS)
-#elif defined(__FILE_OFFSET_BITS) && __FILE_OFFSET_BITS != MDBX_WORDBITS
-#error "__FILE_OFFSET_BITS != MDBX_WORDBITS" (__FILE_OFFSET_BITS != MDBX_WORDBITS)
-#endif
-#endif /* Android */
-
-/*----------------------------------------------------------------------------*/
-/* Compiler's includes for builtins/intrinsics */
-
-#if defined(_MSC_VER) || defined(__INTEL_COMPILER)
-#include <intrin.h>
-#elif __GNUC_PREREQ(4, 4) || defined(__clang__)
-#if defined(__ia32__) || defined(__e2k__)
-#include <x86intrin.h>
-#endif /* __ia32__ */
-#if defined(__ia32__)
-#include <cpuid.h>
-#endif /* __ia32__ */
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
-#include <mbarrier.h>
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
-#include <machine/sys/inline.h>
-#elif defined(__IBMC__) && defined(__powerpc)
-#include <atomic.h>
-#elif defined(_AIX)
-#include <builtins.h>
-#include <sys/atomic_op.h>
-#elif (defined(__osf__) && defined(__DECC)) || defined(__alpha)
-#include <c_asm.h>
-#include <machine/builtins.h>
-#elif defined(__MWERKS__)
-/* CodeWarrior - troubles ? */
-#pragma gcc_extensions
-#elif defined(__SNC__)
-/* Sony PS3 - troubles ? */
-#elif defined(__hppa__) || defined(__hppa)
-#include <machine/inline.h>
-#else
-#error Unsupported C compiler, please use GNU C 4.4 or newer
-#endif /* Compiler */
-
-/*----------------------------------------------------------------------------*/
-/* Byteorder */
-
-#if !defined(__BYTE_ORDER__) || !defined(__ORDER_LITTLE_ENDIAN__) || \
- !defined(__ORDER_BIG_ENDIAN__)
-
-/* *INDENT-OFF* */
-/* clang-format off */
-#if defined(__GLIBC__) || defined(__GNU_LIBRARY__) || defined(__ANDROID_API__) || \
- defined(HAVE_ENDIAN_H) || __has_include(<endian.h>)
-#include <endian.h>
-#elif defined(__APPLE__) || defined(__MACH__) || defined(__OpenBSD__) || \
- defined(HAVE_MACHINE_ENDIAN_H) || __has_include(<machine/endian.h>)
-#include <machine/endian.h>
-#elif defined(HAVE_SYS_ISA_DEFS_H) || __has_include(<sys/isa_defs.h>)
-#include <sys/isa_defs.h>
-#elif (defined(HAVE_SYS_TYPES_H) && defined(HAVE_SYS_ENDIAN_H)) || \
- (__has_include(<sys/types.h>) && __has_include(<sys/endian.h>))
-#include <sys/endian.h>
-#include <sys/types.h>
-#elif defined(__bsdi__) || defined(__DragonFly__) || defined(__FreeBSD__) || \
- defined(__NetBSD__) || \
- defined(HAVE_SYS_PARAM_H) || __has_include(<sys/param.h>)
-#include <sys/param.h>
-#endif /* OS */
-/* *INDENT-ON* */
-/* clang-format on */
-
-#if defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && defined(__BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ __LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ __BIG_ENDIAN
-#define __BYTE_ORDER__ __BYTE_ORDER
-#elif defined(_BYTE_ORDER) && defined(_LITTLE_ENDIAN) && defined(_BIG_ENDIAN)
-#define __ORDER_LITTLE_ENDIAN__ _LITTLE_ENDIAN
-#define __ORDER_BIG_ENDIAN__ _BIG_ENDIAN
-#define __BYTE_ORDER__ _BYTE_ORDER
-#else
-#define __ORDER_LITTLE_ENDIAN__ 1234
-#define __ORDER_BIG_ENDIAN__ 4321
-
-#if defined(__LITTLE_ENDIAN__) || \
- (defined(_LITTLE_ENDIAN) && !defined(_BIG_ENDIAN)) || \
- defined(__ARMEL__) || defined(__THUMBEL__) || defined(__AARCH64EL__) || \
- defined(__MIPSEL__) || defined(_MIPSEL) || defined(__MIPSEL) || \
- defined(_M_ARM) || defined(_M_ARM64) || defined(__e2k__) || \
- defined(__elbrus_4c__) || defined(__elbrus_8c__) || defined(__bfin__) || \
- defined(__BFIN__) || defined(__ia64__) || defined(_IA64) || \
- defined(__IA64__) || defined(__ia64) || defined(_M_IA64) || \
- defined(__itanium__) || defined(__ia32__) || defined(__CYGWIN__) || \
- defined(_WIN64) || defined(_WIN32) || defined(__TOS_WIN__) || \
- defined(__WINDOWS__)
-#define __BYTE_ORDER__ __ORDER_LITTLE_ENDIAN__
-
-#elif defined(__BIG_ENDIAN__) || \
- (defined(_BIG_ENDIAN) && !defined(_LITTLE_ENDIAN)) || \
- defined(__ARMEB__) || defined(__THUMBEB__) || defined(__AARCH64EB__) || \
- defined(__MIPSEB__) || defined(_MIPSEB) || defined(__MIPSEB) || \
- defined(__m68k__) || defined(M68000) || defined(__hppa__) || \
- defined(__hppa) || defined(__HPPA__) || defined(__sparc__) || \
- defined(__sparc) || defined(__370__) || defined(__THW_370__) || \
- defined(__s390__) || defined(__s390x__) || defined(__SYSC_ZARCH__)
-#define __BYTE_ORDER__ __ORDER_BIG_ENDIAN__
-
-#else
-#error __BYTE_ORDER__ should be defined.
-#endif /* Arch */
-
-#endif
-#endif /* __BYTE_ORDER__ || __ORDER_LITTLE_ENDIAN__ || __ORDER_BIG_ENDIAN__ */
-
-/* Get the size of a memory page for the system.
- * This is the basic size that the platform's memory manager uses, and is
- * fundamental to the use of memory-mapped files. */
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline size_t
-mdbx_syspagesize(void) {
-#if defined(_WIN32) || defined(_WIN64)
- SYSTEM_INFO si;
- GetSystemInfo(&si);
- return si.dwPageSize;
-#else
- return sysconf(_SC_PAGE_SIZE);
-#endif
-}
-
-typedef struct mdbx_mmap_param {
- union {
- void *address;
- uint8_t *dxb;
- struct MDBX_lockinfo *lck;
- };
- mdbx_filehandle_t fd;
- size_t limit; /* mapping length, but NOT a size of file nor DB */
- size_t current; /* mapped region size, i.e. the size of file and DB */
- uint64_t filesize /* in-process cache of a file size */;
-#if defined(_WIN32) || defined(_WIN64)
- HANDLE section; /* memory-mapped section handle */
-#endif
-} mdbx_mmap_t;
-
-typedef union bin128 {
- __anonymous_struct_extension__ struct { uint64_t x, y; };
- __anonymous_struct_extension__ struct { uint32_t a, b, c, d; };
-} bin128_t;
-
-#if defined(_WIN32) || defined(_WIN64)
-typedef union MDBX_srwlock {
- __anonymous_struct_extension__ struct {
- long volatile readerCount;
- long volatile writerCount;
- };
- RTL_SRWLOCK native;
-} MDBX_srwlock;
-#endif /* Windows */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC void mdbx_osal_jitter(bool tiny);
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny);
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-#if defined(__cplusplus) && !defined(__STDC_NO_ATOMICS__) && (__has_include(<cstdatomic>) || __has_extension(cxx_atomic))
-#include <cstdatomic>
-#define MDBX_HAVE_C11ATOMICS
-#elif !defined(__cplusplus) && \
- (__STDC_VERSION__ >= 201112L || __has_extension(c_atomic)) && \
- !defined(__STDC_NO_ATOMICS__) && \
- (__GNUC_PREREQ(4, 9) || __CLANG_PREREQ(3, 8) || \
- !(defined(__GNUC__) || defined(__clang__)))
-#include <stdatomic.h>
-#define MDBX_HAVE_C11ATOMICS
-#elif defined(__GNUC__) || defined(__clang__)
-#elif defined(_MSC_VER)
-#pragma warning(disable : 4163) /* 'xyz': not available as an intrinsic */
-#pragma warning(disable : 4133) /* 'function': incompatible types - from \
- 'size_t' to 'LONGLONG' */
-#pragma warning(disable : 4244) /* 'return': conversion from 'LONGLONG' to \
- 'std::size_t', possible loss of data */
-#pragma warning(disable : 4267) /* 'function': conversion from 'size_t' to \
- 'long', possible loss of data */
-#pragma intrinsic(_InterlockedExchangeAdd, _InterlockedCompareExchange)
-#pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedCompareExchange64)
-#elif defined(__APPLE__)
-#include <libkern/OSAtomic.h>
-#else
-#error FIXME atomic-ops
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Memory/Compiler barriers, cache coherence */
-
-#if __has_include(<sys/cachectl.h>)
-#include <sys/cachectl.h>
-#elif defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS should have explicit cache control */
-#include <sys/cachectl.h>
-#endif
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_compiler_barrier(void) {
-#if defined(__clang__) || defined(__GNUC__)
- __asm__ __volatile__("" ::: "memory");
-#elif defined(_MSC_VER)
- _ReadWriteBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
- __memory_barrier();
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __compiler_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_sched_fence(/* LY: no-arg meaning 'all expect ALU', e.g. 0x3D3D */);
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __fence();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_memory_barrier(void) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(memory_order_seq_cst);
-#elif defined(__ATOMIC_SEQ_CST)
-#ifdef __clang__
- __c11_atomic_thread_fence(__ATOMIC_SEQ_CST);
-#else
- __atomic_thread_fence(__ATOMIC_SEQ_CST);
-#endif
-#elif defined(__clang__) || defined(__GNUC__)
- __sync_synchronize();
-#elif defined(_WIN32) || defined(_WIN64)
- MemoryBarrier();
-#elif defined(__INTEL_COMPILER) /* LY: Intel Compiler may mimic GCC and MSC */
-#if defined(__ia32__)
- _mm_mfence();
-#else
- __mf();
-#endif
-#elif defined(__SUNPRO_C) || defined(__sun) || defined(sun)
- __machine_rw_barrier();
-#elif (defined(_HPUX_SOURCE) || defined(__hpux) || defined(__HP_aCC)) && \
- (defined(HP_IA64) || defined(__ia64))
- _Asm_mf();
-#elif defined(_AIX) || defined(__ppc__) || defined(__powerpc__) || \
- defined(__ppc64__) || defined(__powerpc64__)
- __lwsync();
-#else
-#error "Could not guess the kind of compiler, please report to us."
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* libc compatibility stuff */
-
-#if (!defined(__GLIBC__) && __GLIBC_PREREQ(2, 1)) && \
- (defined(_GNU_SOURCE) || defined(_BSD_SOURCE))
-#define mdbx_asprintf asprintf
-#define mdbx_vasprintf vasprintf
-#else
-MDBX_MAYBE_UNUSED MDBX_INTERNAL_FUNC
- MDBX_PRINTF_ARGS(2, 3) int mdbx_asprintf(char **strp, const char *fmt, ...);
-MDBX_INTERNAL_FUNC int mdbx_vasprintf(char **strp, const char *fmt, va_list ap);
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* OS abstraction layer stuff */
-
-/* max bytes to write in one call */
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_WRITE UINT32_C(0x01000000)
-#else
-#define MAX_WRITE UINT32_C(0x3fff0000)
-#endif
-
-#if defined(__linux__) || defined(__gnu_linux__)
-MDBX_INTERNAL_VAR uint32_t mdbx_linux_kernel_version;
-MDBX_INTERNAL_VAR bool mdbx_RunningOnWSL1 /* Windows Subsystem 1 for Linux */;
-#endif /* Linux */
-
-#ifndef mdbx_strdup
-LIBMDBX_API char *mdbx_strdup(const char *str);
-#endif
-
-MDBX_MAYBE_UNUSED static __inline int mdbx_get_errno(void) {
-#if defined(_WIN32) || defined(_WIN64)
- DWORD rc = GetLastError();
-#else
- int rc = errno;
-#endif
- return rc;
-}
-
-#ifndef mdbx_memalign_alloc
-MDBX_INTERNAL_FUNC int mdbx_memalign_alloc(size_t alignment, size_t bytes,
- void **result);
-#endif
-#ifndef mdbx_memalign_free
-MDBX_INTERNAL_FUNC void mdbx_memalign_free(void *ptr);
-#endif
-
-MDBX_INTERNAL_FUNC int mdbx_condpair_init(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_lock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_unlock(mdbx_condpair_t *condpair);
-MDBX_INTERNAL_FUNC int mdbx_condpair_signal(mdbx_condpair_t *condpair,
- bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_wait(mdbx_condpair_t *condpair, bool part);
-MDBX_INTERNAL_FUNC int mdbx_condpair_destroy(mdbx_condpair_t *condpair);
-
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_init(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_acquire(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_release(mdbx_fastmutex_t *fastmutex);
-MDBX_INTERNAL_FUNC int mdbx_fastmutex_destroy(mdbx_fastmutex_t *fastmutex);
-
-MDBX_INTERNAL_FUNC int mdbx_pwritev(mdbx_filehandle_t fd, struct iovec *iov,
- int iovcnt, uint64_t offset,
- size_t expected_written);
-MDBX_INTERNAL_FUNC int mdbx_pread(mdbx_filehandle_t fd, void *buf, size_t count,
- uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_pwrite(mdbx_filehandle_t fd, const void *buf,
- size_t count, uint64_t offset);
-MDBX_INTERNAL_FUNC int mdbx_write(mdbx_filehandle_t fd, const void *buf,
- size_t count);
-
-MDBX_INTERNAL_FUNC int
-mdbx_thread_create(mdbx_thread_t *thread,
- THREAD_RESULT(THREAD_CALL *start_routine)(void *),
- void *arg);
-MDBX_INTERNAL_FUNC int mdbx_thread_join(mdbx_thread_t thread);
-
-enum mdbx_syncmode_bits {
- MDBX_SYNC_NONE = 0,
- MDBX_SYNC_DATA = 1,
- MDBX_SYNC_SIZE = 2,
- MDBX_SYNC_IODQ = 4
-};
-
-MDBX_INTERNAL_FUNC int mdbx_fsync(mdbx_filehandle_t fd,
- const enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_ftruncate(mdbx_filehandle_t fd, uint64_t length);
-MDBX_INTERNAL_FUNC int mdbx_fseek(mdbx_filehandle_t fd, uint64_t pos);
-MDBX_INTERNAL_FUNC int mdbx_filesize(mdbx_filehandle_t fd, uint64_t *length);
-
-enum mdbx_openfile_purpose {
- MDBX_OPEN_DXB_READ = 0,
- MDBX_OPEN_DXB_LAZY = 1,
- MDBX_OPEN_DXB_DSYNC = 2,
- MDBX_OPEN_LCK = 3,
- MDBX_OPEN_COPY = 4,
- MDBX_OPEN_DELETE = 5
-};
-
-MDBX_INTERNAL_FUNC int mdbx_openfile(const enum mdbx_openfile_purpose purpose,
- const MDBX_env *env, const char *pathname,
- mdbx_filehandle_t *fd,
- mdbx_mode_t unix_mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_closefile(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_removefile(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_removedirectory(const char *pathname);
-MDBX_INTERNAL_FUNC int mdbx_is_pipe(mdbx_filehandle_t fd);
-MDBX_INTERNAL_FUNC int mdbx_lockfile(mdbx_filehandle_t fd, bool wait);
-
-#define MMAP_OPTION_TRUNCATE 1
-#define MMAP_OPTION_SEMAPHORE 2
-MDBX_INTERNAL_FUNC int mdbx_mmap(const int flags, mdbx_mmap_t *map,
- const size_t must, const size_t limit,
- const unsigned options);
-MDBX_INTERNAL_FUNC int mdbx_munmap(mdbx_mmap_t *map);
-#define MDBX_MRESIZE_MAY_MOVE 0x00000100
-#define MDBX_MRESIZE_MAY_UNMAP 0x00000200
-MDBX_INTERNAL_FUNC int mdbx_mresize(const int flags, mdbx_mmap_t *map,
- size_t size, size_t limit);
-#if defined(_WIN32) || defined(_WIN64)
-typedef struct {
- unsigned limit, count;
- HANDLE handles[31];
-} mdbx_handle_array_t;
-MDBX_INTERNAL_FUNC int
-mdbx_suspend_threads_before_remap(MDBX_env *env, mdbx_handle_array_t **array);
-MDBX_INTERNAL_FUNC int
-mdbx_resume_threads_after_remap(mdbx_handle_array_t *array);
-#endif /* Windows */
-MDBX_INTERNAL_FUNC int mdbx_msync(mdbx_mmap_t *map, size_t offset,
- size_t length,
- enum mdbx_syncmode_bits mode_bits);
-MDBX_INTERNAL_FUNC int mdbx_check_fs_rdonly(mdbx_filehandle_t handle,
- const char *pathname, int err);
-
-MDBX_MAYBE_UNUSED static __inline uint32_t mdbx_getpid(void) {
- STATIC_ASSERT(sizeof(mdbx_pid_t) <= sizeof(uint32_t));
-#if defined(_WIN32) || defined(_WIN64)
- return GetCurrentProcessId();
-#else
- STATIC_ASSERT(sizeof(pid_t) <= sizeof(uint32_t));
- return getpid();
-#endif
-}
-
-MDBX_MAYBE_UNUSED static __inline uintptr_t mdbx_thread_self(void) {
- mdbx_tid_t thunk;
- STATIC_ASSERT(sizeof(uintptr_t) >= sizeof(thunk));
-#if defined(_WIN32) || defined(_WIN64)
- thunk = GetCurrentThreadId();
-#else
- thunk = pthread_self();
-#endif
- return (uintptr_t)thunk;
-}
-
-#if !defined(_WIN32) && !defined(_WIN64)
-#if defined(__ANDROID_API__) || defined(ANDROID) || defined(BIONIC)
-MDBX_INTERNAL_FUNC int mdbx_check_tid4bionic(void);
-#else
-static __inline int mdbx_check_tid4bionic(void) { return 0; }
-#endif /* __ANDROID_API__ || ANDROID) || BIONIC */
-
-MDBX_MAYBE_UNUSED static __inline int
-mdbx_pthread_mutex_lock(pthread_mutex_t *mutex) {
- int err = mdbx_check_tid4bionic();
- return unlikely(err) ? err : pthread_mutex_lock(mutex);
-}
-#endif /* !Windows */
-
-MDBX_INTERNAL_FUNC uint64_t mdbx_osal_monotime(void);
-MDBX_INTERNAL_FUNC uint64_t
-mdbx_osal_16dot16_to_monotime(uint32_t seconds_16dot16);
-MDBX_INTERNAL_FUNC uint32_t mdbx_osal_monotime_to_16dot16(uint64_t monotime);
-
-MDBX_INTERNAL_FUNC bin128_t mdbx_osal_bootid(void);
-/*----------------------------------------------------------------------------*/
-/* lck stuff */
-
-/// \brief Initialization of synchronization primitives linked with MDBX_env
-/// instance both in LCK-file and within the current process.
-/// \param
-/// global_uniqueness_flag = true - denotes that there are no other processes
-/// working with DB and LCK-file. Thus the function MUST initialize
-/// shared synchronization objects in memory-mapped LCK-file.
-/// global_uniqueness_flag = false - denotes that at least one process is
-/// already working with DB and LCK-file, including the case when DB
-/// has already been opened in the current process. Thus the function
-/// MUST NOT initialize shared synchronization objects in memory-mapped
-/// LCK-file that are already in use.
-/// \return Error code or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_init(MDBX_env *env,
- MDBX_env *inprocess_neighbor,
- int global_uniqueness_flag);
-
-/// \brief Disconnects from shared interprocess objects and destructs
-/// synchronization objects linked with MDBX_env instance
-/// within the current process.
-/// \param
-/// inprocess_neighbor = NULL - if the current process does not have other
-/// instances of MDBX_env linked with the DB being closed.
-/// Thus the function MUST check for other processes working with DB or
-/// LCK-file, and keep or destroy shared synchronization objects in
-/// memory-mapped LCK-file depending on the result.
-/// inprocess_neighbor = not-NULL - pointer to another instance of MDBX_env
-/// (anyone of there is several) working with DB or LCK-file within the
-/// current process. Thus the function MUST NOT try to acquire exclusive
-/// lock and/or try to destruct shared synchronization objects linked with
-/// DB or LCK-file. Moreover, the implementation MUST ensure correct work
-/// of other instances of MDBX_env within the current process, e.g.
-/// restore POSIX-fcntl locks after the closing of file descriptors.
-/// \return Error code (MDBX_PANIC) or zero on success.
-MDBX_INTERNAL_FUNC int mdbx_lck_destroy(MDBX_env *env,
- MDBX_env *inprocess_neighbor);
-
-/// \brief Connects to shared interprocess locking objects and tries to acquire
-/// the maximum lock level (shared if exclusive is not available)
-/// Depending on implementation or/and platform (Windows) this function may
-/// acquire the non-OS super-level lock (e.g. for shared synchronization
-/// objects initialization), which will be downgraded to OS-exclusive or
-/// shared via explicit calling of mdbx_lck_downgrade().
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if an exclusive lock was acquired and thus
-/// the current process is the first and only after the last use of DB.
-/// MDBX_RESULT_FALSE (0) - if a shared lock was acquired and thus
-/// DB has already been opened and now is used by other processes.
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_lck_seize(MDBX_env *env);
-
-/// \brief Downgrades the level of initially acquired lock to
-/// operational level specified by argument. The reson for such downgrade:
-/// - unblocking of other processes that are waiting for access, i.e.
-/// if (env->me_flags & MDBX_EXCLUSIVE) != 0, then other processes
-/// should be made aware that access is unavailable rather than
-/// wait for it.
-/// - freeing locks that interfere file operation (especially for Windows)
-/// (env->me_flags & MDBX_EXCLUSIVE) == 0 - downgrade to shared lock.
-/// (env->me_flags & MDBX_EXCLUSIVE) != 0 - downgrade to exclusive
-/// operational lock.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_lck_downgrade(MDBX_env *env);
-
-/// \brief Locks LCK-file or/and table of readers for (de)registering.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rdt_lock(MDBX_env *env);
-
-/// \brief Unlocks LCK-file or/and table of readers after (de)registering.
-MDBX_INTERNAL_FUNC void mdbx_rdt_unlock(MDBX_env *env);
-
-/// \brief Acquires lock for DB change (on writing transaction start)
-/// Reading transactions will not be blocked.
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-/// \return Error code or zero on success
-LIBMDBX_API int mdbx_txn_lock(MDBX_env *env, bool dont_wait);
-
-/// \brief Releases lock once DB changes is made (after writing transaction
-/// has finished).
-/// Declared as LIBMDBX_API because it is used in mdbx_chk.
-LIBMDBX_API void mdbx_txn_unlock(MDBX_env *env);
-
-/// \brief Sets alive-flag of reader presence (indicative lock) for PID of
-/// the current process. The function does no more than needed for
-/// the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_set(MDBX_env *env);
-
-/// \brief Resets alive-flag of reader presence (indicative lock)
-/// for PID of the current process. The function does no more than needed
-/// for the correct working of mdbx_rpid_check() in other processes.
-/// \return Error code or zero on success
-MDBX_INTERNAL_FUNC int mdbx_rpid_clear(MDBX_env *env);
-
-/// \brief Checks for reading process status with the given pid with help of
-/// alive-flag of presence (indicative lock) or using another way.
-/// \return
-/// MDBX_RESULT_TRUE (-1) - if the reader process with the given PID is alive
-/// and working with DB (indicative lock is present).
-/// MDBX_RESULT_FALSE (0) - if the reader process with the given PID is absent
-/// or not working with DB (indicative lock is not present).
-/// Otherwise (not 0 and not -1) - error code.
-MDBX_INTERNAL_FUNC int mdbx_rpid_check(MDBX_env *env, uint32_t pid);
-
-#if defined(_WIN32) || defined(_WIN64)
-
-typedef void(WINAPI *MDBX_srwlock_function)(MDBX_srwlock *);
-MDBX_INTERNAL_VAR MDBX_srwlock_function mdbx_srwlock_Init,
- mdbx_srwlock_AcquireShared, mdbx_srwlock_ReleaseShared,
- mdbx_srwlock_AcquireExclusive, mdbx_srwlock_ReleaseExclusive;
-
-#if _WIN32_WINNT < 0x0600 /* prior to Windows Vista */
-typedef enum _FILE_INFO_BY_HANDLE_CLASS {
- FileBasicInfo,
- FileStandardInfo,
- FileNameInfo,
- FileRenameInfo,
- FileDispositionInfo,
- FileAllocationInfo,
- FileEndOfFileInfo,
- FileStreamInfo,
- FileCompressionInfo,
- FileAttributeTagInfo,
- FileIdBothDirectoryInfo,
- FileIdBothDirectoryRestartInfo,
- FileIoPriorityHintInfo,
- FileRemoteProtocolInfo,
- MaximumFileInfoByHandleClass
-} FILE_INFO_BY_HANDLE_CLASS,
- *PFILE_INFO_BY_HANDLE_CLASS;
-
-typedef struct _FILE_END_OF_FILE_INFO {
- LARGE_INTEGER EndOfFile;
-} FILE_END_OF_FILE_INFO, *PFILE_END_OF_FILE_INFO;
-
-#define REMOTE_PROTOCOL_INFO_FLAG_LOOPBACK 0x00000001
-#define REMOTE_PROTOCOL_INFO_FLAG_OFFLINE 0x00000002
-
-typedef struct _FILE_REMOTE_PROTOCOL_INFO {
- USHORT StructureVersion;
- USHORT StructureSize;
- DWORD Protocol;
- USHORT ProtocolMajorVersion;
- USHORT ProtocolMinorVersion;
- USHORT ProtocolRevision;
- USHORT Reserved;
- DWORD Flags;
- struct {
- DWORD Reserved[8];
- } GenericReserved;
- struct {
- DWORD Reserved[16];
- } ProtocolSpecificReserved;
-} FILE_REMOTE_PROTOCOL_INFO, *PFILE_REMOTE_PROTOCOL_INFO;
-
-#endif /* _WIN32_WINNT < 0x0600 (prior to Windows Vista) */
-
-typedef BOOL(WINAPI *MDBX_GetFileInformationByHandleEx)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_GetFileInformationByHandleEx
- mdbx_GetFileInformationByHandleEx;
-
-typedef BOOL(WINAPI *MDBX_GetVolumeInformationByHandleW)(
- _In_ HANDLE hFile, _Out_opt_ LPWSTR lpVolumeNameBuffer,
- _In_ DWORD nVolumeNameSize, _Out_opt_ LPDWORD lpVolumeSerialNumber,
- _Out_opt_ LPDWORD lpMaximumComponentLength,
- _Out_opt_ LPDWORD lpFileSystemFlags,
- _Out_opt_ LPWSTR lpFileSystemNameBuffer, _In_ DWORD nFileSystemNameSize);
-MDBX_INTERNAL_VAR MDBX_GetVolumeInformationByHandleW
- mdbx_GetVolumeInformationByHandleW;
-
-typedef DWORD(WINAPI *MDBX_GetFinalPathNameByHandleW)(_In_ HANDLE hFile,
- _Out_ LPWSTR lpszFilePath,
- _In_ DWORD cchFilePath,
- _In_ DWORD dwFlags);
-MDBX_INTERNAL_VAR MDBX_GetFinalPathNameByHandleW mdbx_GetFinalPathNameByHandleW;
-
-typedef BOOL(WINAPI *MDBX_SetFileInformationByHandle)(
- _In_ HANDLE hFile, _In_ FILE_INFO_BY_HANDLE_CLASS FileInformationClass,
- _Out_ LPVOID lpFileInformation, _In_ DWORD dwBufferSize);
-MDBX_INTERNAL_VAR MDBX_SetFileInformationByHandle
- mdbx_SetFileInformationByHandle;
-
-typedef NTSTATUS(NTAPI *MDBX_NtFsControlFile)(
- IN HANDLE FileHandle, IN OUT HANDLE Event,
- IN OUT PVOID /* PIO_APC_ROUTINE */ ApcRoutine, IN OUT PVOID ApcContext,
- OUT PIO_STATUS_BLOCK IoStatusBlock, IN ULONG FsControlCode,
- IN OUT PVOID InputBuffer, IN ULONG InputBufferLength,
- OUT OPTIONAL PVOID OutputBuffer, IN ULONG OutputBufferLength);
-MDBX_INTERNAL_VAR MDBX_NtFsControlFile mdbx_NtFsControlFile;
-
-typedef uint64_t(WINAPI *MDBX_GetTickCount64)(void);
-MDBX_INTERNAL_VAR MDBX_GetTickCount64 mdbx_GetTickCount64;
-
-#if !defined(_WIN32_WINNT_WIN8) || _WIN32_WINNT < _WIN32_WINNT_WIN8
-typedef struct _WIN32_MEMORY_RANGE_ENTRY {
- PVOID VirtualAddress;
- SIZE_T NumberOfBytes;
-} WIN32_MEMORY_RANGE_ENTRY, *PWIN32_MEMORY_RANGE_ENTRY;
-#endif /* Windows 8.x */
-
-typedef BOOL(WINAPI *MDBX_PrefetchVirtualMemory)(
- HANDLE hProcess, ULONG_PTR NumberOfEntries,
- PWIN32_MEMORY_RANGE_ENTRY VirtualAddresses, ULONG Flags);
-MDBX_INTERNAL_VAR MDBX_PrefetchVirtualMemory mdbx_PrefetchVirtualMemory;
-
-typedef enum _SECTION_INHERIT { ViewShare = 1, ViewUnmap = 2 } SECTION_INHERIT;
-
-typedef NTSTATUS(NTAPI *MDBX_NtExtendSection)(IN HANDLE SectionHandle,
- IN PLARGE_INTEGER NewSectionSize);
-MDBX_INTERNAL_VAR MDBX_NtExtendSection mdbx_NtExtendSection;
-
-static __inline bool mdbx_RunningUnderWine(void) {
- return !mdbx_NtExtendSection;
-}
-
-typedef LSTATUS(WINAPI *MDBX_RegGetValueA)(HKEY hkey, LPCSTR lpSubKey,
- LPCSTR lpValue, DWORD dwFlags,
- LPDWORD pdwType, PVOID pvData,
- LPDWORD pcbData);
-MDBX_INTERNAL_VAR MDBX_RegGetValueA mdbx_RegGetValueA;
-
-NTSYSAPI ULONG RtlRandomEx(PULONG Seed);
-
-#endif /* Windows */
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-
-#if defined(_MSC_VER) && _MSC_VER >= 1900
-/* LY: MSVC 2015/2017/2019 has buggy/inconsistent PRIuPTR/PRIxPTR macros
- * for internal format-args checker. */
-#undef PRIuPTR
-#undef PRIiPTR
-#undef PRIdPTR
-#undef PRIxPTR
-#define PRIuPTR "Iu"
-#define PRIiPTR "Ii"
-#define PRIdPTR "Id"
-#define PRIxPTR "Ix"
-#define PRIuSIZE "zu"
-#define PRIiSIZE "zi"
-#define PRIdSIZE "zd"
-#define PRIxSIZE "zx"
-#endif /* fix PRI*PTR for _MSC_VER */
-
-#ifndef PRIuSIZE
-#define PRIuSIZE PRIuPTR
-#define PRIiSIZE PRIiPTR
-#define PRIdSIZE PRIdPTR
-#define PRIxSIZE PRIxPTR
-#endif /* PRI*SIZE macros for MSVC */
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-
-#define mdbx_sourcery_anchor XCONCAT(mdbx_sourcery_, MDBX_BUILD_SOURCERY)
-#if defined(xMDBX_TOOLS)
-extern LIBMDBX_API const char *const mdbx_sourcery_anchor;
-#endif
-
-/*******************************************************************************
- *******************************************************************************
- *******************************************************************************
- *
- *
- * #### ##### ##### # #### # # ####
- * # # # # # # # # ## # #
- * # # # # # # # # # # # ####
- * # # ##### # # # # # # # #
- * # # # # # # # # ## # #
- * #### # # # #### # # ####
- *
- *
- */
-
-/** \defgroup build_option Build options
- * The libmdbx build options.
- @{ */
-
-/** Using fcntl(F_FULLFSYNC) with 5-10 times slowdown */
-#define MDBX_OSX_WANNA_DURABILITY 0
-/** Using fsync() with chance of data lost on power failure */
-#define MDBX_OSX_WANNA_SPEED 1
-
-#ifndef MDBX_OSX_SPEED_INSTEADOF_DURABILITY
-/** Choices \ref MDBX_OSX_WANNA_DURABILITY or \ref MDBX_OSX_WANNA_SPEED
- * for OSX & iOS */
-#define MDBX_OSX_SPEED_INSTEADOF_DURABILITY MDBX_OSX_WANNA_DURABILITY
-#endif /* MDBX_OSX_SPEED_INSTEADOF_DURABILITY */
-
-/** Controls checking PID against reuse DB environment after the fork() */
-#ifndef MDBX_ENV_CHECKPID
-#if defined(MADV_DONTFORK) || defined(_WIN32) || defined(_WIN64)
-/* PID check could be omitted:
- * - on Linux when madvise(MADV_DONTFORK) is available, i.e. after the fork()
- * mapped pages will not be available for child process.
- * - in Windows where fork() not available. */
-#define MDBX_ENV_CHECKPID 0
-#else
-#define MDBX_ENV_CHECKPID 1
-#endif
-#define MDBX_ENV_CHECKPID_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#else
-#define MDBX_ENV_CHECKPID_CONFIG MDBX_STRINGIFY(MDBX_ENV_CHECKPID)
-#endif /* MDBX_ENV_CHECKPID */
-
-/** Controls checking transaction owner thread against misuse transactions from
- * other threads. */
-#ifndef MDBX_TXN_CHECKOWNER
-#define MDBX_TXN_CHECKOWNER 1
-#define MDBX_TXN_CHECKOWNER_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#else
-#define MDBX_TXN_CHECKOWNER_CONFIG MDBX_STRINGIFY(MDBX_TXN_CHECKOWNER)
-#endif /* MDBX_TXN_CHECKOWNER */
-
-/** Does a system have battery-backed Real-Time Clock or just a fake. */
-#ifndef MDBX_TRUST_RTC
-#if defined(__linux__) || defined(__gnu_linux__) || defined(__NetBSD__) || \
- defined(__OpenBSD__)
-#define MDBX_TRUST_RTC 0 /* a lot of embedded systems have a fake RTC */
-#else
-#define MDBX_TRUST_RTC 1
-#endif
-#define MDBX_TRUST_RTC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#else
-#define MDBX_TRUST_RTC_CONFIG MDBX_STRINGIFY(MDBX_TRUST_RTC)
-#endif /* MDBX_TRUST_RTC */
-
-/** Controls online database auto-compactification during write-transactions. */
-#ifndef MDBX_ENABLE_REFUND
-#define MDBX_ENABLE_REFUND 1
-#elif !(MDBX_ENABLE_REFUND == 0 || MDBX_ENABLE_REFUND == 1)
-#error MDBX_ENABLE_REFUND must be defined as 0 or 1
-#endif /* MDBX_ENABLE_REFUND */
-
-/** Controls gathering statistics for page operations. */
-#ifndef MDBX_ENABLE_PGOP_STAT
-#define MDBX_ENABLE_PGOP_STAT 1
-#elif !(MDBX_ENABLE_PGOP_STAT == 0 || MDBX_ENABLE_PGOP_STAT == 1)
-#error MDBX_ENABLE_PGOP_STAT must be defined as 0 or 1
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-/** Controls use of POSIX madvise() hints and friends. */
-#ifndef MDBX_ENABLE_MADVISE
-#define MDBX_ENABLE_MADVISE 1
-#elif !(MDBX_ENABLE_MADVISE == 0 || MDBX_ENABLE_MADVISE == 1)
-#error MDBX_ENABLE_MADVISE must be defined as 0 or 1
-#endif /* MDBX_ENABLE_MADVISE */
-
-/** Disable some checks to reduce an overhead and detection probability of
- * database corruption to a values closer to the LMDB. */
-#ifndef MDBX_DISABLE_PAGECHECKS
-#define MDBX_DISABLE_PAGECHECKS 0
-#elif !(MDBX_DISABLE_PAGECHECKS == 0 || MDBX_DISABLE_PAGECHECKS == 1)
-#error MDBX_DISABLE_PAGECHECKS must be defined as 0 or 1
-#endif /* MDBX_DISABLE_PAGECHECKS */
-
-#ifndef MDBX_PNL_PREALLOC_FOR_RADIXSORT
-#define MDBX_PNL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_PNL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_PNL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_PNL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_PNL_PREALLOC_FOR_RADIXSORT */
-
-#ifndef MDBX_DPL_PREALLOC_FOR_RADIXSORT
-#define MDBX_DPL_PREALLOC_FOR_RADIXSORT 1
-#elif !(MDBX_DPL_PREALLOC_FOR_RADIXSORT == 0 || \
- MDBX_DPL_PREALLOC_FOR_RADIXSORT == 1)
-#error MDBX_DPL_PREALLOC_FOR_RADIXSORT must be defined as 0 or 1
-#endif /* MDBX_DPL_PREALLOC_FOR_RADIXSORT */
-
-/* Basically, this build-option is for TODO. Guess it should be replaced
- * with MDBX_ENABLE_WRITEMAP_SPILLING with the three variants:
- * 0/OFF = Don't track dirty pages at all and don't spilling ones.
- * This should be by-default on Linux and may-be other systems
- * (not sure: Darwin/OSX, FreeBSD, Windows 10) where kernel provides
- * properly LRU tracking and async writing on-demand.
- * 1/ON = Lite tracking of dirty pages but with LRU labels and explicit
- * spilling with msync(MS_ASYNC). */
-#ifndef MDBX_FAKE_SPILL_WRITEMAP
-#if defined(__linux__) || defined(__gnu_linux__)
-#define MDBX_FAKE_SPILL_WRITEMAP 1 /* msync(MS_ASYNC) is no-op on Linux */
-#else
-#define MDBX_FAKE_SPILL_WRITEMAP 0
-#endif
-#elif !(MDBX_FAKE_SPILL_WRITEMAP == 0 || MDBX_FAKE_SPILL_WRITEMAP == 1)
-#error MDBX_FAKE_SPILL_WRITEMAP must be defined as 0 or 1
-#endif /* MDBX_FAKE_SPILL_WRITEMAP */
-
-/** Controls sort order of internal page number lists.
- * This mostly experimental/advanced option with not for regular MDBX users.
- * \warning The database format depend on this option and libmdbx builded with
- * different option value are incompatible. */
-#ifndef MDBX_PNL_ASCENDING
-#define MDBX_PNL_ASCENDING 0
-#elif !(MDBX_PNL_ASCENDING == 0 || MDBX_PNL_ASCENDING == 1)
-#error MDBX_PNL_ASCENDING must be defined as 0 or 1
-#endif /* MDBX_PNL_ASCENDING */
-
-/** Avoid dependence from MSVC CRT and use ntdll.dll instead. */
-#ifndef MDBX_WITHOUT_MSVC_CRT
-#define MDBX_WITHOUT_MSVC_CRT 1
-#elif !(MDBX_WITHOUT_MSVC_CRT == 0 || MDBX_WITHOUT_MSVC_CRT == 1)
-#error MDBX_WITHOUT_MSVC_CRT must be defined as 0 or 1
-#endif /* MDBX_WITHOUT_MSVC_CRT */
-
-/** Size of buffer used during copying a environment/database file. */
-#ifndef MDBX_ENVCOPY_WRITEBUF
-#define MDBX_ENVCOPY_WRITEBUF 1048576u
-#elif MDBX_ENVCOPY_WRITEBUF < 65536u || MDBX_ENVCOPY_WRITEBUF > 1073741824u || \
- MDBX_ENVCOPY_WRITEBUF % 65536u
-#error MDBX_ENVCOPY_WRITEBUF must be defined in range 65536..1073741824 and be multiple of 65536
-#endif /* MDBX_ENVCOPY_WRITEBUF */
-
-/** Forces assertion checking */
-#ifndef MDBX_FORCE_ASSERTIONS
-#define MDBX_FORCE_ASSERTIONS 0
-#elif !(MDBX_FORCE_ASSERTIONS == 0 || MDBX_FORCE_ASSERTIONS == 1)
-#error MDBX_FORCE_ASSERTIONS must be defined as 0 or 1
-#endif /* MDBX_FORCE_ASSERTIONS */
-
-/** Presumed malloc size overhead for each allocation
- * to adjust allocations to be more aligned. */
-#ifndef MDBX_ASSUME_MALLOC_OVERHEAD
-#ifdef __SIZEOF_POINTER__
-#define MDBX_ASSUME_MALLOC_OVERHEAD (__SIZEOF_POINTER__ * 2u)
-#else
-#define MDBX_ASSUME_MALLOC_OVERHEAD (sizeof(void *) * 2u)
-#endif
-#elif MDBX_ASSUME_MALLOC_OVERHEAD < 0 || MDBX_ASSUME_MALLOC_OVERHEAD > 64 || \
- MDBX_ASSUME_MALLOC_OVERHEAD % 4
-#error MDBX_ASSUME_MALLOC_OVERHEAD must be defined in range 0..64 and be multiple of 4
-#endif /* MDBX_ASSUME_MALLOC_OVERHEAD */
-
-/** In case the MDBX_DEBUG is undefined set it corresponding to NDEBUG */
-#ifndef MDBX_DEBUG
-#ifdef NDEBUG
-#define MDBX_DEBUG 0
-#else
-#define MDBX_DEBUG 1
-#endif
-#endif /* MDBX_DEBUG */
-
-/** If defined then enables integration with Valgrind,
- * a memory analyzing tool. */
-#ifndef MDBX_USE_VALGRIND
-#endif /* MDBX_USE_VALGRIND */
-
-/** If defined then enables use C11 atomics,
- * otherwise detects ones availability automatically. */
-#ifndef MDBX_HAVE_C11ATOMICS
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-//------------------------------------------------------------------------------
-
-/** Win32 File Locking API for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_WIN32FILES -1
-
-/** SystemV IPC semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_SYSV 5
-
-/** POSIX-1 Shared anonymous semaphores for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX1988 1988
-
-/** POSIX-2001 Shared Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2001 2001
-
-/** POSIX-2008 Robust Mutexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_POSIX2008 2008
-
-/** BeOS Benaphores, aka Futexes for \ref MDBX_LOCKING */
-#define MDBX_LOCKING_BENAPHORE 1995
-
-/** Advanced: Choices the locking implementation (autodetection by default). */
-#if defined(_WIN32) || defined(_WIN64)
-#define MDBX_LOCKING MDBX_LOCKING_WIN32FILES
-#else
-#ifndef MDBX_LOCKING
-#if defined(_POSIX_THREAD_PROCESS_SHARED) && \
- _POSIX_THREAD_PROCESS_SHARED >= 200112L && !defined(__FreeBSD__)
-
-/* Some platforms define the EOWNERDEAD error code even though they
- * don't support Robust Mutexes. If doubt compile with -MDBX_LOCKING=2001. */
-#if defined(EOWNERDEAD) && _POSIX_THREAD_PROCESS_SHARED >= 200809L && \
- ((defined(_POSIX_THREAD_ROBUST_PRIO_INHERIT) && \
- _POSIX_THREAD_ROBUST_PRIO_INHERIT > 0) || \
- (defined(_POSIX_THREAD_ROBUST_PRIO_PROTECT) && \
- _POSIX_THREAD_ROBUST_PRIO_PROTECT > 0) || \
- defined(PTHREAD_MUTEX_ROBUST) || defined(PTHREAD_MUTEX_ROBUST_NP)) && \
- (!defined(__GLIBC__) || \
- __GLIBC_PREREQ(2, 10) /* troubles with Robust mutexes before 2.10 */)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2008
-#else
-#define MDBX_LOCKING MDBX_LOCKING_POSIX2001
-#endif
-#elif defined(__sun) || defined(__SVR4) || defined(__svr4__)
-#define MDBX_LOCKING MDBX_LOCKING_POSIX1988
-#else
-#define MDBX_LOCKING MDBX_LOCKING_SYSV
-#endif
-#define MDBX_LOCKING_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_LOCKING)
-#else
-#define MDBX_LOCKING_CONFIG MDBX_STRINGIFY(MDBX_LOCKING)
-#endif /* MDBX_LOCKING */
-#endif /* !Windows */
-
-/** Advanced: Using POSIX OFD-locks (autodetection by default). */
-#ifndef MDBX_USE_OFDLOCKS
-#if defined(F_OFD_SETLK) && defined(F_OFD_SETLKW) && defined(F_OFD_GETLK) && \
- !defined(MDBX_SAFE4QEMU) && \
- !defined(__sun) /* OFD-lock are broken on Solaris */
-#define MDBX_USE_OFDLOCKS 1
-#else
-#define MDBX_USE_OFDLOCKS 0
-#endif
-#define MDBX_USE_OFDLOCKS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#else
-#define MDBX_USE_OFDLOCKS_CONFIG MDBX_STRINGIFY(MDBX_USE_OFDLOCKS)
-#endif /* MDBX_USE_OFDLOCKS */
-
-/** Advanced: Using sendfile() syscall (autodetection by default). */
-#ifndef MDBX_USE_SENDFILE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 21)
-#define MDBX_USE_SENDFILE 1
-#else
-#define MDBX_USE_SENDFILE 0
-#endif
-#endif /* MDBX_USE_SENDFILE */
-
-/** Advanced: Using copy_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_COPYFILERANGE
-#if __GLIBC_PREREQ(2, 27) && defined(_GNU_SOURCE)
-#define MDBX_USE_COPYFILERANGE 1
-#else
-#define MDBX_USE_COPYFILERANGE 0
-#endif
-#endif /* MDBX_USE_COPYFILERANGE */
-
-/** Advanced: Using sync_file_range() syscall (autodetection by default). */
-#ifndef MDBX_USE_SYNCFILERANGE
-#if ((defined(__linux__) || defined(__gnu_linux__)) && \
- defined(SYNC_FILE_RANGE_WRITE) && !defined(__ANDROID_API__)) || \
- (defined(__ANDROID_API__) && __ANDROID_API__ >= 26)
-#define MDBX_USE_SYNCFILERANGE 1
-#else
-#define MDBX_USE_SYNCFILERANGE 0
-#endif
-#endif /* MDBX_USE_SYNCFILERANGE */
-
-//------------------------------------------------------------------------------
-
-#ifndef MDBX_CPU_WRITEBACK_INCOHERENT
-#if defined(__ia32__) || defined(__e2k__) || defined(__hppa) || \
- defined(__hppa__) || defined(DOXYGEN)
-#define MDBX_CPU_WRITEBACK_INCOHERENT 0
-#else
-#define MDBX_CPU_WRITEBACK_INCOHERENT 1
-#endif
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-#ifndef MDBX_MMAP_INCOHERENT_FILE_WRITE
-#ifdef __OpenBSD__
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 1
-#else
-#define MDBX_MMAP_INCOHERENT_FILE_WRITE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#ifndef MDBX_MMAP_INCOHERENT_CPU_CACHE
-#if defined(__mips) || defined(__mips__) || defined(__mips64) || \
- defined(__mips64__) || defined(_M_MRX000) || defined(_MIPS_) || \
- defined(__MWERKS__) || defined(__sgi)
-/* MIPS has cache coherency issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 1
-#else
-/* LY: assume no relevant mmap/dcache issues. */
-#define MDBX_MMAP_INCOHERENT_CPU_CACHE 0
-#endif
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#ifndef MDBX_64BIT_ATOMIC
-#if MDBX_WORDBITS >= 64 || defined(DOXYGEN)
-#define MDBX_64BIT_ATOMIC 1
-#else
-#define MDBX_64BIT_ATOMIC 0
-#endif
-#define MDBX_64BIT_ATOMIC_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#else
-#define MDBX_64BIT_ATOMIC_CONFIG MDBX_STRINGIFY(MDBX_64BIT_ATOMIC)
-#endif /* MDBX_64BIT_ATOMIC */
-
-#ifndef MDBX_64BIT_CAS
-#if defined(ATOMIC_LLONG_LOCK_FREE)
-#if ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__GCC_ATOMIC_LLONG_LOCK_FREE)
-#if __GCC_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(__CLANG_ATOMIC_LLONG_LOCK_FREE)
-#if __CLANG_ATOMIC_LLONG_LOCK_FREE > 1
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS 0
-#endif
-#elif defined(_MSC_VER) || defined(__APPLE__) || defined(DOXYGEN)
-#define MDBX_64BIT_CAS 1
-#else
-#define MDBX_64BIT_CAS MDBX_64BIT_ATOMIC
-#endif
-#define MDBX_64BIT_CAS_CONFIG "AUTO=" MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#else
-#define MDBX_64BIT_CAS_CONFIG MDBX_STRINGIFY(MDBX_64BIT_CAS)
-#endif /* MDBX_64BIT_CAS */
-
-#ifndef MDBX_UNALIGNED_OK
-#if defined(__ALIGNED__) || defined(__SANITIZE_UNDEFINED__)
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#elif defined(__ARM_FEATURE_UNALIGNED)
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#elif __CLANG_PREREQ(5, 0) || __GNUC_PREREQ(5, 0)
-/* expecting an optimization will well done, also this
- * hushes false-positives from UBSAN (undefined behaviour sanitizer) */
-#define MDBX_UNALIGNED_OK 0
-#elif defined(__e2k__) || defined(__elbrus__)
-#if __iset__ > 4
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 4 /* ok unaligned for 32-bit words */
-#endif
-#elif defined(__ia32__)
-#define MDBX_UNALIGNED_OK 8 /* ok unaligned for 64-bit words */
-#else
-#define MDBX_UNALIGNED_OK 0 /* no unaligned access allowed */
-#endif
-#elif MDBX_UNALIGNED_OK == 1
-#undef MDBX_UNALIGNED_OK
-#define MDBX_UNALIGNED_OK 32 /* any unaligned access allowed */
-#endif /* MDBX_UNALIGNED_OK */
-
-#ifndef MDBX_CACHELINE_SIZE
-#if defined(SYSTEM_CACHE_ALIGNMENT_SIZE)
-#define MDBX_CACHELINE_SIZE SYSTEM_CACHE_ALIGNMENT_SIZE
-#elif defined(__ia64__) || defined(__ia64) || defined(_M_IA64)
-#define MDBX_CACHELINE_SIZE 128
-#else
-#define MDBX_CACHELINE_SIZE 64
-#endif
-#endif /* MDBX_CACHELINE_SIZE */
-
-/** @} end of build options */
-/*******************************************************************************
- *******************************************************************************
- ******************************************************************************/
-
-#ifdef DOXYGEN
-/* !!! Actually this is a fake definitions !!!
- * !!! for documentation generation by Doxygen !!! */
-
-/** Controls enabling of debugging features.
- *
- * - `MDBX_DEBUG = 0` (by default) Disables any debugging features at all,
- * including logging and assertion controls.
- * Logging level and corresponding debug flags changing
- * by \ref mdbx_setup_debug() will not have effect.
- * - `MDBX_DEBUG > 0` Enables code for the debugging features (logging,
- * assertions checking and internal audit).
- * Simultaneously sets the default logging level
- * to the `MDBX_DEBUG` value.
- * Also enables \ref MDBX_DBG_AUDIT if `MDBX_DEBUG >= 2`.
- *
- * \ingroup build_option */
-#define MDBX_DEBUG 0...7
-
-/** Disables using of GNU libc extensions. */
-#define MDBX_DISABLE_GNU_SOURCE 0 or 1
-
-#endif /* DOXYGEN */
-
-/* Undefine the NDEBUG if debugging is enforced by MDBX_DEBUG */
-#if MDBX_DEBUG
-#undef NDEBUG
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Atomics */
-
-enum MDBX_memory_order {
- mo_Relaxed,
- mo_AcquireRelease,
- mo_SequentialConsistency
-};
-
-typedef union {
- volatile uint32_t weak;
-#ifdef MDBX_HAVE_C11ATOMICS
- volatile _Atomic uint32_t c11a;
-#endif /* MDBX_HAVE_C11ATOMICS */
-} MDBX_atomic_uint32_t;
-
-typedef union {
- volatile uint64_t weak;
-#if defined(MDBX_HAVE_C11ATOMICS) && (MDBX_64BIT_CAS || MDBX_64BIT_ATOMIC)
- volatile _Atomic uint64_t c11a;
-#endif
-#if !defined(MDBX_HAVE_C11ATOMICS) || !MDBX_64BIT_CAS || !MDBX_64BIT_ATOMIC
- __anonymous_struct_extension__ struct {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- MDBX_atomic_uint32_t low, high;
-#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
- MDBX_atomic_uint32_t high, low;
-#else
-#error "FIXME: Unsupported byte order"
-#endif /* __BYTE_ORDER__ */
- };
-#endif
-} MDBX_atomic_uint64_t;
-
-#ifdef MDBX_HAVE_C11ATOMICS
-
-/* Crutches for C11 atomic compiler's bugs */
-#if defined(__e2k__) && defined(__LCC__) && __LCC__ < /* FIXME */ 127
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->weak)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->weak)
-#elif defined(__clang__) && __clang__ < 8
-#define MDBX_c11a_ro(type, ptr) ((volatile _Atomic(type) *)&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#else
-#define MDBX_c11a_ro(type, ptr) (&(ptr)->c11a)
-#define MDBX_c11a_rw(type, ptr) (&(ptr)->c11a)
-#endif /* Crutches for C11 atomic compiler's bugs */
-
-static __always_inline memory_order mo_c11_store(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_release;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-
-static __always_inline memory_order mo_c11_load(enum MDBX_memory_order fence) {
- switch (fence) {
- default:
- assert(false);
- __unreachable();
- case mo_Relaxed:
- return memory_order_relaxed;
- case mo_AcquireRelease:
- return memory_order_acquire;
- case mo_SequentialConsistency:
- return memory_order_seq_cst;
- }
-}
-#endif /* MDBX_HAVE_C11ATOMICS */
-
-#ifndef __cplusplus
-
-MDBX_MAYBE_UNUSED static __always_inline void
-mdbx_memory_fence(enum MDBX_memory_order order, bool write) {
-#ifdef MDBX_HAVE_C11ATOMICS
- atomic_thread_fence(write ? mo_c11_store(order) : mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_compiler_barrier();
- if (write &&
- order > (MDBX_CPU_WRITEBACK_INCOHERENT ? mo_Relaxed : mo_AcquireRelease))
- mdbx_memory_barrier();
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_store32(MDBX_atomic_uint32_t *p, const uint32_t value,
- enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_rw(uint32_t, p)));
- atomic_store_explicit(MDBX_c11a_rw(uint32_t, p), value, mo_c11_store(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- p->weak = value;
- mdbx_memory_fence(order, true);
-#endif /* MDBX_HAVE_C11ATOMICS */
- return value;
-}
-
-MDBX_MAYBE_UNUSED static __always_inline uint32_t
-atomic_load32(const MDBX_atomic_uint32_t *p, enum MDBX_memory_order order) {
- STATIC_ASSERT(sizeof(MDBX_atomic_uint32_t) == 4);
-#ifdef MDBX_HAVE_C11ATOMICS
- assert(atomic_is_lock_free(MDBX_c11a_ro(uint32_t, p)));
- return atomic_load_explicit(MDBX_c11a_ro(uint32_t, p), mo_c11_load(order));
-#else /* MDBX_HAVE_C11ATOMICS */
- mdbx_memory_fence(order, false);
- const uint32_t value = p->weak;
- if (order != mo_Relaxed)
- mdbx_compiler_barrier();
- return value;
-#endif /* MDBX_HAVE_C11ATOMICS */
-}
-
-#endif /* !__cplusplus */
-
-/*----------------------------------------------------------------------------*/
-/* Basic constants and types */
-
-/* A stamp that identifies a file as an MDBX file.
- * There's nothing special about this value other than that it is easily
- * recognizable, and it will reflect any byte order mismatches. */
-#define MDBX_MAGIC UINT64_C(/* 56-bit prime */ 0x59659DBDEF4C11)
-
-/* FROZEN: The version number for a database's datafile format. */
-#define MDBX_DATA_VERSION 3
-/* The version number for a database's lockfile format. */
-#define MDBX_LOCK_VERSION 4
-
-/* handle for the DB used to track free pages. */
-#define FREE_DBI 0
-/* handle for the default DB. */
-#define MAIN_DBI 1
-/* Number of DBs in metapage (free and main) - also hardcoded elsewhere */
-#define CORE_DBS 2
-
-/* Number of meta pages - also hardcoded elsewhere */
-#define NUM_METAS 3
-
-/* A page number in the database.
- *
- * MDBX uses 32 bit for page numbers. This limits database
- * size up to 2^44 bytes, in case of 4K pages. */
-typedef uint32_t pgno_t;
-typedef MDBX_atomic_uint32_t atomic_pgno_t;
-#define PRIaPGNO PRIu32
-#define MAX_PAGENO UINT32_C(0x7FFFffff)
-#define MIN_PAGENO NUM_METAS
-
-#define SAFE64_INVALID_THRESHOLD UINT64_C(0xffffFFFF00000000)
-
-/* A transaction ID. */
-typedef uint64_t txnid_t;
-typedef MDBX_atomic_uint64_t atomic_txnid_t;
-#define PRIaTXN PRIi64
-#define MIN_TXNID UINT64_C(1)
-#define MAX_TXNID (SAFE64_INVALID_THRESHOLD - 1)
-#define INITIAL_TXNID (MIN_TXNID + NUM_METAS - 1)
-#define INVALID_TXNID UINT64_MAX
-/* LY: for testing non-atomic 64-bit txnid on 32-bit arches.
- * #define xMDBX_TXNID_STEP (UINT32_MAX / 3) */
-#ifndef xMDBX_TXNID_STEP
-#if MDBX_64BIT_CAS
-#define xMDBX_TXNID_STEP 1u
-#else
-#define xMDBX_TXNID_STEP 2u
-#endif
-#endif /* xMDBX_TXNID_STEP */
-
-/* Used for offsets within a single page.
- * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
- * this is plenty. */
-typedef uint16_t indx_t;
-
-#define MEGABYTE ((size_t)1 << 20)
-
-/*----------------------------------------------------------------------------*/
-/* Core structures for database and shared memory (i.e. format definition) */
-#pragma pack(push, 4)
-
-/* Information about a single database in the environment. */
-typedef struct MDBX_db {
- uint16_t md_flags; /* see mdbx_dbi_open */
- uint16_t md_depth; /* depth of this tree */
- uint32_t md_xsize; /* key-size for MDBX_DUPFIXED (LEAF2 pages) */
- pgno_t md_root; /* the root page of this tree */
- pgno_t md_branch_pages; /* number of internal pages */
- pgno_t md_leaf_pages; /* number of leaf pages */
- pgno_t md_overflow_pages; /* number of overflow pages */
- uint64_t md_seq; /* table sequence counter */
- uint64_t md_entries; /* number of data items */
- uint64_t md_mod_txnid; /* txnid of last committed modification */
-} MDBX_db;
-
-/* database size-related parameters */
-typedef struct MDBX_geo {
- uint16_t grow_pv; /* datafile growth step as a 16-bit packed (exponential
- quantized) value */
- uint16_t shrink_pv; /* datafile shrink threshold as a 16-bit packed
- (exponential quantized) value */
- pgno_t lower; /* minimal size of datafile in pages */
- pgno_t upper; /* maximal size of datafile in pages */
- pgno_t now; /* current size of datafile in pages */
- pgno_t next; /* first unused page in the datafile,
- but actually the file may be shorter. */
-} MDBX_geo;
-
-/* Meta page content.
- * A meta page is the start point for accessing a database snapshot.
- * Pages 0-1 are meta pages. Transaction N writes meta page (N % 2). */
-typedef struct MDBX_meta {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with MDBX_DATA_VERSION. */
- uint32_t mm_magic_and_version[2];
-
- /* txnid that committed this page, the first of a two-phase-update pair */
- uint32_t mm_txnid_a[2];
-
- uint16_t mm_extra_flags; /* extra DB flags, zero (nothing) for now */
- uint8_t mm_validator_id; /* ID of checksum and page validation method,
- * zero (nothing) for now */
- uint8_t mm_extra_pagehdr; /* extra bytes in the page header,
- * zero (nothing) for now */
-
- MDBX_geo mm_geo; /* database size-related parameters */
-
- MDBX_db mm_dbs[CORE_DBS]; /* first is free space, 2nd is main db */
- /* The size of pages used in this DB */
-#define mm_psize mm_dbs[FREE_DBI].md_xsize
- MDBX_canary mm_canary;
-
-#define MDBX_DATASIGN_NONE 0u
-#define MDBX_DATASIGN_WEAK 1u
-#define SIGN_IS_STEADY(sign) ((sign) > MDBX_DATASIGN_WEAK)
-#define META_IS_STEADY(meta) \
- SIGN_IS_STEADY(unaligned_peek_u64_volatile(4, (meta)->mm_datasync_sign))
- uint32_t mm_datasync_sign[2];
-
- /* txnid that committed this page, the second of a two-phase-update pair */
- uint32_t mm_txnid_b[2];
-
- /* Number of non-meta pages which were put in GC after COW. May be 0 in case
- * DB was previously handled by libmdbx without corresponding feature.
- * This value in couple with mr_snapshot_pages_retired allows fast estimation
- * of "how much reader is restraining GC recycling". */
- uint32_t mm_pages_retired[2];
-
- /* The analogue /proc/sys/kernel/random/boot_id or similar to determine
- * whether the system was rebooted after the last use of the database files.
- * If there was no reboot, but there is no need to rollback to the last
- * steady sync point. Zeros mean that no relevant information is available
- * from the system. */
- bin128_t mm_bootid;
-
-} MDBX_meta;
-
-#pragma pack(1)
-
-/* Common header for all page types. The page type depends on mp_flags.
- *
- * P_BRANCH and P_LEAF pages have unsorted 'MDBX_node's at the end, with
- * sorted mp_ptrs[] entries referring to them. Exception: P_LEAF2 pages
- * omit mp_ptrs and pack sorted MDBX_DUPFIXED values after the page header.
- *
- * P_OVERFLOW records occupy one or more contiguous pages where only the
- * first has a page header. They hold the real data of F_BIGDATA nodes.
- *
- * P_SUBP sub-pages are small leaf "pages" with duplicate data.
- * A node with flag F_DUPDATA but not F_SUBDATA contains a sub-page.
- * (Duplicate data can also go in sub-databases, which use normal pages.)
- *
- * P_META pages contain MDBX_meta, the start point of an MDBX snapshot.
- *
- * Each non-metapage up to MDBX_meta.mm_last_pg is reachable exactly once
- * in the snapshot: Either used by a database or listed in a GC record. */
-typedef struct MDBX_page {
- union {
-#define IS_FROZEN(txn, p) ((p)->mp_txnid < (txn)->mt_txnid)
-#define IS_SPILLED(txn, p) ((p)->mp_txnid == (txn)->mt_txnid)
-#define IS_SHADOWED(txn, p) ((p)->mp_txnid > (txn)->mt_txnid)
-#define IS_VALID(txn, p) ((p)->mp_txnid <= (txn)->mt_front)
-#define IS_MODIFIABLE(txn, p) ((p)->mp_txnid == (txn)->mt_front)
- uint64_t mp_txnid;
- struct MDBX_page *mp_next; /* for in-memory list of freed pages */
- };
- uint16_t mp_leaf2_ksize; /* key size if this is a LEAF2 page */
-#define P_BRANCH 0x01 /* branch page */
-#define P_LEAF 0x02 /* leaf page */
-#define P_OVERFLOW 0x04 /* overflow page */
-#define P_META 0x08 /* meta page */
-#define P_BAD 0x10 /* explicit flag for invalid/bad page */
-#define P_LEAF2 0x20 /* for MDBX_DUPFIXED records */
-#define P_SUBP 0x40 /* for MDBX_DUPSORT sub-pages */
-#define P_SPILLED 0x2000 /* spilled in parent txn */
-#define P_LOOSE 0x4000 /* page was dirtied then freed, can be reused */
-#define P_FROZEN 0x8000 /* used for retire page with known status */
-#define P_ILL_BITS (~(P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW | P_SPILLED))
- uint16_t mp_flags;
- union {
- uint32_t mp_pages; /* number of overflow pages */
- __anonymous_struct_extension__ struct {
- indx_t mp_lower; /* lower bound of free space */
- indx_t mp_upper; /* upper bound of free space */
- };
- };
- pgno_t mp_pgno; /* page number */
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- indx_t mp_ptrs[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_page;
-
-/* Size of the page header, excluding dynamic data at the end */
-#define PAGEHDRSZ ((unsigned)offsetof(MDBX_page, mp_ptrs))
-
-#pragma pack(pop)
-
-#if MDBX_ENABLE_PGOP_STAT
-/* Statistics of page operations overall of all (running, completed and aborted)
- * transactions */
-typedef struct {
- MDBX_atomic_uint64_t newly; /* Quantity of a new pages added */
- MDBX_atomic_uint64_t cow; /* Quantity of pages copied for update */
- MDBX_atomic_uint64_t clone; /* Quantity of parent's dirty pages clones
- for nested transactions */
- MDBX_atomic_uint64_t split; /* Page splits */
- MDBX_atomic_uint64_t merge; /* Page merges */
- MDBX_atomic_uint64_t spill; /* Quantity of spilled dirty pages */
- MDBX_atomic_uint64_t unspill; /* Quantity of unspilled/reloaded pages */
- MDBX_atomic_uint64_t
- wops; /* Number of explicit write operations (not a pages) to a disk */
-} MDBX_pgop_stat_t;
-#endif /* MDBX_ENABLE_PGOP_STAT */
-
-#if MDBX_LOCKING == MDBX_LOCKING_WIN32FILES
-#define MDBX_CLOCK_SIGN UINT32_C(0xF10C)
-typedef void mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_SYSV
-
-#define MDBX_CLOCK_SIGN UINT32_C(0xF18D)
-typedef mdbx_pid_t mdbx_ipclock_t;
-#ifndef EOWNERDEAD
-#define EOWNERDEAD MDBX_RESULT_TRUE
-#endif
-
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX2001 || \
- MDBX_LOCKING == MDBX_LOCKING_POSIX2008
-#define MDBX_CLOCK_SIGN UINT32_C(0x8017)
-typedef pthread_mutex_t mdbx_ipclock_t;
-#elif MDBX_LOCKING == MDBX_LOCKING_POSIX1988
-#define MDBX_CLOCK_SIGN UINT32_C(0xFC29)
-typedef sem_t mdbx_ipclock_t;
-#else
-#error "FIXME"
-#endif /* MDBX_LOCKING */
-
-#if MDBX_LOCKING > MDBX_LOCKING_SYSV && !defined(__cplusplus)
-MDBX_INTERNAL_FUNC int mdbx_ipclock_stub(mdbx_ipclock_t *ipc);
-MDBX_INTERNAL_FUNC int mdbx_ipclock_destroy(mdbx_ipclock_t *ipc);
-#endif /* MDBX_LOCKING */
-
-/* Reader Lock Table
- *
- * Readers don't acquire any locks for their data access. Instead, they
- * simply record their transaction ID in the reader table. The reader
- * mutex is needed just to find an empty slot in the reader table. The
- * slot's address is saved in thread-specific data so that subsequent
- * read transactions started by the same thread need no further locking to
- * proceed.
- *
- * If MDBX_NOTLS is set, the slot address is not saved in thread-specific data.
- * No reader table is used if the database is on a read-only filesystem.
- *
- * Since the database uses multi-version concurrency control, readers don't
- * actually need any locking. This table is used to keep track of which
- * readers are using data from which old transactions, so that we'll know
- * when a particular old transaction is no longer in use. Old transactions
- * that have discarded any data pages can then have those pages reclaimed
- * for use by a later write transaction.
- *
- * The lock table is constructed such that reader slots are aligned with the
- * processor's cache line size. Any slot is only ever used by one thread.
- * This alignment guarantees that there will be no contention or cache
- * thrashing as threads update their own slot info, and also eliminates
- * any need for locking when accessing a slot.
- *
- * A writer thread will scan every slot in the table to determine the oldest
- * outstanding reader transaction. Any freed pages older than this will be
- * reclaimed by the writer. The writer doesn't use any locks when scanning
- * this table. This means that there's no guarantee that the writer will
- * see the most up-to-date reader info, but that's not required for correct
- * operation - all we need is to know the upper bound on the oldest reader,
- * we don't care at all about the newest reader. So the only consequence of
- * reading stale information here is that old pages might hang around a
- * while longer before being reclaimed. That's actually good anyway, because
- * the longer we delay reclaiming old pages, the more likely it is that a
- * string of contiguous pages can be found after coalescing old pages from
- * many old transactions together. */
-
-/* The actual reader record, with cacheline padding. */
-typedef struct MDBX_reader {
- /* Current Transaction ID when this transaction began, or (txnid_t)-1.
- * Multiple readers that start at the same time will probably have the
- * same ID here. Again, it's not important to exclude them from
- * anything; all we need to know is which version of the DB they
- * started from so we can avoid overwriting any data used in that
- * particular version. */
- MDBX_atomic_uint64_t /* txnid_t */ mr_txnid;
-
- /* The information we store in a single slot of the reader table.
- * In addition to a transaction ID, we also record the process and
- * thread ID that owns a slot, so that we can detect stale information,
- * e.g. threads or processes that went away without cleaning up.
- *
- * NOTE: We currently don't check for stale records.
- * We simply re-init the table when we know that we're the only process
- * opening the lock file. */
-
- /* The thread ID of the thread owning this txn. */
- MDBX_atomic_uint64_t mr_tid;
-
- /* The process ID of the process owning this reader txn. */
- MDBX_atomic_uint32_t mr_pid;
-
- /* The number of pages used in the reader's MVCC snapshot,
- * i.e. the value of meta->mm_geo.next and txn->mt_next_pgno */
- atomic_pgno_t mr_snapshot_pages_used;
- /* Number of retired pages at the time this reader starts transaction. So,
- * at any time the difference mm_pages_retired - mr_snapshot_pages_retired
- * will give the number of pages which this reader restraining from reuse. */
- MDBX_atomic_uint64_t mr_snapshot_pages_retired;
-} MDBX_reader;
-
-/* The header for the reader table (a memory-mapped lock file). */
-typedef struct MDBX_lockinfo {
- /* Stamp identifying this as an MDBX file.
- * It must be set to MDBX_MAGIC with with MDBX_LOCK_VERSION. */
- uint64_t mti_magic_and_version;
-
- /* Format of this lock file. Must be set to MDBX_LOCK_FORMAT. */
- uint32_t mti_os_and_format;
-
- /* Flags which environment was opened. */
- MDBX_atomic_uint32_t mti_envmode;
-
- /* Threshold of un-synced-with-disk pages for auto-sync feature,
- * zero means no-threshold, i.e. auto-sync is disabled. */
- atomic_pgno_t mti_autosync_threshold;
-
- /* Low 32-bit of txnid with which meta-pages was synced,
- * i.e. for sync-polling in the MDBX_NOMETASYNC mode. */
- MDBX_atomic_uint32_t mti_meta_sync_txnid;
-
- /* Period for timed auto-sync feature, i.e. at the every steady checkpoint
- * the mti_unsynced_timeout sets to the current_time + mti_autosync_period.
- * The time value is represented in a suitable system-dependent form, for
- * example clock_gettime(CLOCK_BOOTTIME) or clock_gettime(CLOCK_MONOTONIC).
- * Zero means timed auto-sync is disabled. */
- MDBX_atomic_uint64_t mti_autosync_period;
-
- /* Marker to distinguish uniqueness of DB/CLK. */
- MDBX_atomic_uint64_t mti_bait_uniqueness;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
-#if MDBX_ENABLE_PGOP_STAT
- /* Statistics of costly ops of all (running, completed and aborted)
- * transactions */
- MDBX_pgop_stat_t mti_pgop_stat;
-#endif /* MDBX_ENABLE_PGOP_STAT*/
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Write transaction lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_wlock;
-#endif /* MDBX_LOCKING > 0 */
-
- atomic_txnid_t mti_oldest_reader;
-
- /* Timestamp of the last steady sync. Value is represented in a suitable
- * system-dependent form, for example clock_gettime(CLOCK_BOOTTIME) or
- * clock_gettime(CLOCK_MONOTONIC). */
- MDBX_atomic_uint64_t mti_sync_timestamp;
-
- /* Number un-synced-with-disk pages for auto-sync feature. */
- atomic_pgno_t mti_unsynced_pages;
-
- /* Number of page which was discarded last time by madvise(MADV_FREE). */
- atomic_pgno_t mti_discarded_tail;
-
- /* Timestamp of the last readers check. */
- MDBX_atomic_uint64_t mti_reader_check_timestamp;
-
- /* Shared anchor for tracking readahead edge and enabled/disabled status. */
- pgno_t mti_readahead_anchor;
-
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
-
- /* Readeaders registration lock. */
-#if MDBX_LOCKING > 0
- mdbx_ipclock_t mti_rlock;
-#endif /* MDBX_LOCKING > 0 */
-
- /* The number of slots that have been used in the reader table.
- * This always records the maximum count, it is not decremented
- * when readers release their slots. */
- MDBX_atomic_uint32_t mti_numreaders;
- MDBX_atomic_uint32_t mti_readers_refresh_flag;
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_ALIGNAS(MDBX_CACHELINE_SIZE) /* cacheline ----------------------------*/
- MDBX_reader mti_readers[] /* dynamic size */;
-#endif /* C99 */
-} MDBX_lockinfo;
-
-/* Lockfile format signature: version, features and field layout */
-#define MDBX_LOCK_FORMAT \
- (MDBX_CLOCK_SIGN * 27733 + (unsigned)sizeof(MDBX_reader) * 13 + \
- (unsigned)offsetof(MDBX_reader, mr_snapshot_pages_used) * 251 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_oldest_reader) * 83 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_numreaders) * 37 + \
- (unsigned)offsetof(MDBX_lockinfo, mti_readers) * 29)
-
-#define MDBX_DATA_MAGIC \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + MDBX_DATA_VERSION)
-
-#define MDBX_DATA_MAGIC_LEGACY_COMPAT \
- ((MDBX_MAGIC << 8) + MDBX_PNL_ASCENDING * 64 + 2)
-
-#define MDBX_DATA_MAGIC_LEGACY_DEVEL ((MDBX_MAGIC << 8) + 255)
-
-#define MDBX_LOCK_MAGIC ((MDBX_MAGIC << 8) + MDBX_LOCK_VERSION)
-
-/* The maximum size of a database page.
- *
- * It is 64K, but value-PAGEHDRSZ must fit in MDBX_page.mp_upper.
- *
- * MDBX will use database pages < OS pages if needed.
- * That causes more I/O in write transactions: The OS must
- * know (read) the whole page before writing a partial page.
- *
- * Note that we don't currently support Huge pages. On Linux,
- * regular data files cannot use Huge pages, and in general
- * Huge pages aren't actually pageable. We rely on the OS
- * demand-pager to read our data and page it out when memory
- * pressure from other processes is high. So until OSs have
- * actual paging support for Huge pages, they're not viable. */
-#define MAX_PAGESIZE MDBX_MAX_PAGESIZE
-#define MIN_PAGESIZE MDBX_MIN_PAGESIZE
-
-#define MIN_MAPSIZE (MIN_PAGESIZE * MIN_PAGENO)
-#if defined(_WIN32) || defined(_WIN64)
-#define MAX_MAPSIZE32 UINT32_C(0x38000000)
-#else
-#define MAX_MAPSIZE32 UINT32_C(0x7f000000)
-#endif
-#define MAX_MAPSIZE64 ((MAX_PAGENO + 1) * (uint64_t)MAX_PAGESIZE)
-
-#if MDBX_WORDBITS >= 64
-#define MAX_MAPSIZE MAX_MAPSIZE64
-#define MDBX_PGL_LIMIT ((size_t)MAX_PAGENO)
-#else
-#define MAX_MAPSIZE MAX_MAPSIZE32
-#define MDBX_PGL_LIMIT (MAX_MAPSIZE32 / MIN_PAGESIZE)
-#endif /* MDBX_WORDBITS */
-
-#define MDBX_READERS_LIMIT 32767
-#define MDBX_RADIXSORT_THRESHOLD 333
-
-/*----------------------------------------------------------------------------*/
-
-/* An PNL is an Page Number List, a sorted array of IDs.
- * The first element of the array is a counter for how many actual page-numbers
- * are in the list. By default PNLs are sorted in descending order, this allow
- * cut off a page with lowest pgno (at the tail) just truncating the list. The
- * sort order of PNLs is controlled by the MDBX_PNL_ASCENDING build option. */
-typedef pgno_t *MDBX_PNL;
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_ORDERED(first, last) ((first) < (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) >= (last))
-#else
-#define MDBX_PNL_ORDERED(first, last) ((first) > (last))
-#define MDBX_PNL_DISORDERED(first, last) ((first) <= (last))
-#endif
-
-/* List of txnid, only for MDBX_txn.tw.lifo_reclaimed */
-typedef txnid_t *MDBX_TXL;
-
-/* An Dirty-Page list item is an pgno/pointer pair. */
-typedef struct MDBX_dp {
- MDBX_page *ptr;
- pgno_t pgno;
- union {
- unsigned extra;
- __anonymous_struct_extension__ struct {
- unsigned multi : 1;
- unsigned lru : 31;
- };
- };
-} MDBX_dp;
-
-/* An DPL (dirty-page list) is a sorted array of MDBX_DPs. */
-typedef struct MDBX_dpl {
- unsigned sorted;
- unsigned length;
- unsigned detent; /* allocated size excluding the MDBX_DPL_RESERVE_GAP */
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- MDBX_dp items[] /* dynamic size with holes at zero and after the last */;
-#endif
-} MDBX_dpl;
-
-/* PNL sizes */
-#define MDBX_PNL_GRANULATE 1024
-#define MDBX_PNL_INITIAL \
- (MDBX_PNL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(pgno_t))
-
-#define MDBX_TXL_GRANULATE 32
-#define MDBX_TXL_INITIAL \
- (MDBX_TXL_GRANULATE - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-#define MDBX_TXL_MAX \
- ((1u << 17) - 2 - MDBX_ASSUME_MALLOC_OVERHEAD / sizeof(txnid_t))
-
-#define MDBX_PNL_ALLOCLEN(pl) ((pl)[-1])
-#define MDBX_PNL_SIZE(pl) ((pl)[0])
-#define MDBX_PNL_FIRST(pl) ((pl)[1])
-#define MDBX_PNL_LAST(pl) ((pl)[MDBX_PNL_SIZE(pl)])
-#define MDBX_PNL_BEGIN(pl) (&(pl)[1])
-#define MDBX_PNL_END(pl) (&(pl)[MDBX_PNL_SIZE(pl) + 1])
-
-#if MDBX_PNL_ASCENDING
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_FIRST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_LAST(pl)
-#else
-#define MDBX_PNL_LEAST(pl) MDBX_PNL_LAST(pl)
-#define MDBX_PNL_MOST(pl) MDBX_PNL_FIRST(pl)
-#endif
-
-#define MDBX_PNL_SIZEOF(pl) ((MDBX_PNL_SIZE(pl) + 1) * sizeof(pgno_t))
-#define MDBX_PNL_IS_EMPTY(pl) (MDBX_PNL_SIZE(pl) == 0)
-
-/*----------------------------------------------------------------------------*/
-/* Internal structures */
-
-/* Auxiliary DB info.
- * The information here is mostly static/read-only. There is
- * only a single copy of this record in the environment. */
-typedef struct MDBX_dbx {
- MDBX_val md_name; /* name of the database */
- MDBX_cmp_func *md_cmp; /* function for comparing keys */
- MDBX_cmp_func *md_dcmp; /* function for comparing data items */
- size_t md_klen_min, md_klen_max; /* min/max key length for the database */
- size_t md_vlen_min,
- md_vlen_max; /* min/max value/data length for the database */
-} MDBX_dbx;
-
-/* A database transaction.
- * Every operation requires a transaction handle. */
-struct MDBX_txn {
-#define MDBX_MT_SIGNATURE UINT32_C(0x93D53A31)
- uint32_t mt_signature;
-
- /* Transaction Flags */
- /* mdbx_txn_begin() flags */
-#define MDBX_TXN_RO_BEGIN_FLAGS (MDBX_TXN_RDONLY | MDBX_TXN_RDONLY_PREPARE)
-#define MDBX_TXN_RW_BEGIN_FLAGS \
- (MDBX_TXN_NOMETASYNC | MDBX_TXN_NOSYNC | MDBX_TXN_TRY)
- /* Additional flag for mdbx_sync_locked() */
-#define MDBX_SHRINK_ALLOWED UINT32_C(0x40000000)
-
- /* internal txn flags */
-#define MDBX_TXN_FINISHED 0x01 /* txn is finished or never began */
-#define MDBX_TXN_ERROR 0x02 /* txn is unusable after an error */
-#define MDBX_TXN_DIRTY 0x04 /* must write, even if dirty list is empty */
-#define MDBX_TXN_SPILLS 0x08 /* txn or a parent has spilled pages */
-#define MDBX_TXN_HAS_CHILD 0x10 /* txn has an MDBX_txn.mt_child */
- /* most operations on the txn are currently illegal */
-#define MDBX_TXN_BLOCKED \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_HAS_CHILD)
-
-#define TXN_FLAGS \
- (MDBX_TXN_FINISHED | MDBX_TXN_ERROR | MDBX_TXN_DIRTY | MDBX_TXN_SPILLS | \
- MDBX_TXN_HAS_CHILD)
-
-#if (TXN_FLAGS & (MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS)) || \
- ((MDBX_TXN_RW_BEGIN_FLAGS | MDBX_TXN_RO_BEGIN_FLAGS | TXN_FLAGS) & \
- MDBX_SHRINK_ALLOWED)
-#error "Oops, some flags overlapped or wrong"
-#endif
- uint32_t mt_flags;
-
- MDBX_txn *mt_parent; /* parent of a nested txn */
- /* Nested txn under this txn, set together with flag MDBX_TXN_HAS_CHILD */
- MDBX_txn *mt_child;
- MDBX_geo mt_geo;
- /* next unallocated page */
-#define mt_next_pgno mt_geo.next
- /* corresponding to the current size of datafile */
-#define mt_end_pgno mt_geo.now
-
- /* The ID of this transaction. IDs are integers incrementing from 1.
- * Only committed write transactions increment the ID. If a transaction
- * aborts, the ID may be re-used by the next writer. */
- txnid_t mt_txnid;
- txnid_t mt_front;
-
- MDBX_env *mt_env; /* the DB environment */
- /* Array of records for each DB known in the environment. */
- MDBX_dbx *mt_dbxs;
- /* Array of MDBX_db records for each known DB */
- MDBX_db *mt_dbs;
- /* Array of sequence numbers for each DB handle */
- unsigned *mt_dbiseqs;
-
- /* Transaction DBI Flags */
-#define DBI_DIRTY MDBX_DBI_DIRTY /* DB was written in this txn */
-#define DBI_STALE MDBX_DBI_STALE /* Named-DB record is older than txnID */
-#define DBI_FRESH MDBX_DBI_FRESH /* Named-DB handle opened in this txn */
-#define DBI_CREAT MDBX_DBI_CREAT /* Named-DB handle created in this txn */
-#define DBI_VALID 0x10 /* DB handle is valid, see also DB_VALID */
-#define DBI_USRVALID 0x20 /* As DB_VALID, but not set for FREE_DBI */
-#define DBI_AUDITED 0x40 /* Internal flag for accounting during audit */
- /* Array of flags for each DB */
- uint8_t *mt_dbistate;
- /* Number of DB records in use, or 0 when the txn is finished.
- * This number only ever increments until the txn finishes; we
- * don't decrement it when individual DB handles are closed. */
- MDBX_dbi mt_numdbs;
- size_t mt_owner; /* thread ID that owns this transaction */
- MDBX_canary mt_canary;
- void *mt_userctx; /* User-settable context */
- MDBX_cursor **mt_cursors;
-
- union {
- struct {
- /* For read txns: This thread/txn's reader table slot, or NULL. */
- MDBX_reader *reader;
- } to;
- struct {
- /* In write txns, array of cursors for each DB */
- pgno_t *reclaimed_pglist; /* Reclaimed GC pages */
- txnid_t last_reclaimed; /* ID of last used record */
-#if MDBX_ENABLE_REFUND
- pgno_t loose_refund_wl /* FIXME: describe */;
-#endif /* MDBX_ENABLE_REFUND */
- /* dirtylist room: Dirty array size - dirty pages visible to this txn.
- * Includes ancestor txns' dirty pages not hidden by other txns'
- * dirty/spilled pages. Thus commit(nested txn) has room to merge
- * dirtylist into mt_parent after freeing hidden mt_parent pages. */
- unsigned dirtyroom;
- /* a sequence to spilling dirty page with LRU policy */
- unsigned dirtylru;
- /* For write txns: Modified pages. Sorted when not MDBX_WRITEMAP. */
- MDBX_dpl *dirtylist;
- /* The list of reclaimed txns from GC */
- MDBX_TXL lifo_reclaimed;
- /* The list of pages that became unused during this transaction. */
- MDBX_PNL retired_pages;
- /* The list of loose pages that became unused and may be reused
- * in this transaction, linked through `mp_next`. */
- MDBX_page *loose_pages;
- /* Number of loose pages (tw.loose_pages) */
- unsigned loose_count;
- /* The sorted list of dirty pages we temporarily wrote to disk
- * because the dirty list was full. page numbers in here are
- * shifted left by 1, deleted slots have the LSB set. */
- MDBX_PNL spill_pages;
- unsigned spill_least_removed;
- } tw;
- };
-};
-
-#if MDBX_WORDBITS >= 64
-#define CURSOR_STACK 32
-#else
-#define CURSOR_STACK 24
-#endif
-
-struct MDBX_xcursor;
-
-/* Cursors are used for all DB operations.
- * A cursor holds a path of (page pointer, key index) from the DB
- * root to a position in the DB, plus other state. MDBX_DUPSORT
- * cursors include an xcursor to the current data item. Write txns
- * track their cursors and keep them up to date when data moves.
- * Exception: An xcursor's pointer to a P_SUBP page can be stale.
- * (A node with F_DUPDATA but no F_SUBDATA contains a subpage). */
-struct MDBX_cursor {
-#define MDBX_MC_LIVE UINT32_C(0xFE05D5B1)
-#define MDBX_MC_READY4CLOSE UINT32_C(0x2817A047)
-#define MDBX_MC_WAIT4EOT UINT32_C(0x90E297A7)
- uint32_t mc_signature;
- /* The database handle this cursor operates on */
- MDBX_dbi mc_dbi;
- /* Next cursor on this DB in this txn */
- MDBX_cursor *mc_next;
- /* Backup of the original cursor if this cursor is a shadow */
- MDBX_cursor *mc_backup;
- /* Context used for databases with MDBX_DUPSORT, otherwise NULL */
- struct MDBX_xcursor *mc_xcursor;
- /* The transaction that owns this cursor */
- MDBX_txn *mc_txn;
- /* The database record for this cursor */
- MDBX_db *mc_db;
- /* The database auxiliary record for this cursor */
- MDBX_dbx *mc_dbx;
- /* The mt_dbistate for this database */
- uint8_t *mc_dbistate;
- unsigned mc_snum; /* number of pushed pages */
- unsigned mc_top; /* index of top page, normally mc_snum-1 */
-
- /* Cursor state flags. */
-#define C_INITIALIZED 0x01 /* cursor has been initialized and is valid */
-#define C_EOF 0x02 /* No more data */
-#define C_SUB 0x04 /* Cursor is a sub-cursor */
-#define C_DEL 0x08 /* last op was a cursor_del */
-#define C_UNTRACK 0x10 /* Un-track cursor when closing */
-#define C_RECLAIMING 0x20 /* GC lookup is prohibited */
-#define C_GCFREEZE 0x40 /* reclaimed_pglist must not be updated */
-
- /* Cursor checking flags. */
-#define C_COPYING 0x100 /* skip key-value length check (copying simplify) */
-#define C_UPDATING 0x200 /* update/rebalance pending */
-#define C_RETIRING 0x400 /* refs to child pages may be invalid */
-#define C_SKIPORD 0x800 /* don't check keys ordering */
-
- unsigned mc_flags; /* see mdbx_cursor */
- MDBX_page *mc_pg[CURSOR_STACK]; /* stack of pushed pages */
- indx_t mc_ki[CURSOR_STACK]; /* stack of page indices */
-};
-
-/* Context for sorted-dup records.
- * We could have gone to a fully recursive design, with arbitrarily
- * deep nesting of sub-databases. But for now we only handle these
- * levels - main DB, optional sub-DB, sorted-duplicate DB. */
-typedef struct MDBX_xcursor {
- /* A sub-cursor for traversing the Dup DB */
- MDBX_cursor mx_cursor;
- /* The database record for this Dup DB */
- MDBX_db mx_db;
- /* The auxiliary DB record for this Dup DB */
- MDBX_dbx mx_dbx;
-} MDBX_xcursor;
-
-typedef struct MDBX_cursor_couple {
- MDBX_cursor outer;
- void *mc_userctx; /* User-settable context */
- MDBX_xcursor inner;
-} MDBX_cursor_couple;
-
-/* The database environment. */
-struct MDBX_env {
- /* ----------------------------------------------------- mostly static part */
-#define MDBX_ME_SIGNATURE UINT32_C(0x9A899641)
- MDBX_atomic_uint32_t me_signature;
- /* Failed to update the meta page. Probably an I/O error. */
-#define MDBX_FATAL_ERROR UINT32_C(0x80000000)
- /* Some fields are initialized. */
-#define MDBX_ENV_ACTIVE UINT32_C(0x20000000)
- /* me_txkey is set */
-#define MDBX_ENV_TXKEY UINT32_C(0x10000000)
- /* Legacy MDBX_MAPASYNC (prior v0.9) */
-#define MDBX_DEPRECATED_MAPASYNC UINT32_C(0x100000)
-#define ENV_INTERNAL_FLAGS (MDBX_FATAL_ERROR | MDBX_ENV_ACTIVE | MDBX_ENV_TXKEY)
- uint32_t me_flags;
- mdbx_mmap_t me_dxb_mmap; /* The main data file */
-#define me_map me_dxb_mmap.dxb
-#define me_lazy_fd me_dxb_mmap.fd
- mdbx_filehandle_t me_dsync_fd;
- mdbx_mmap_t me_lck_mmap; /* The lock file */
-#define me_lfd me_lck_mmap.fd
- struct MDBX_lockinfo *me_lck;
-
- unsigned me_psize; /* DB page size, initialized from me_os_psize */
- unsigned me_leaf_nodemax; /* max size of a leaf-node */
- uint8_t me_psize2log; /* log2 of DB page size */
- int8_t me_stuck_meta; /* recovery-only: target meta page or less that zero */
- uint16_t me_merge_threshold,
- me_merge_threshold_gc; /* pages emptier than this are candidates for
- merging */
- unsigned me_os_psize; /* OS page size, from mdbx_syspagesize() */
- unsigned me_maxreaders; /* size of the reader table */
- MDBX_dbi me_maxdbs; /* size of the DB table */
- uint32_t me_pid; /* process ID of this env */
- mdbx_thread_key_t me_txkey; /* thread-key for readers */
- char *me_pathname; /* path to the DB files */
- void *me_pbuf; /* scratch area for DUPSORT put() */
- MDBX_txn *me_txn0; /* preallocated write transaction */
-
- MDBX_dbx *me_dbxs; /* array of static DB info */
- uint16_t *me_dbflags; /* array of flags from MDBX_db.md_flags */
- unsigned *me_dbiseqs; /* array of dbi sequence numbers */
- unsigned
- me_maxgc_ov1page; /* Number of pgno_t fit in a single overflow page */
- uint32_t me_live_reader; /* have liveness lock in reader table */
- void *me_userctx; /* User-settable context */
- MDBX_hsr_func *me_hsr_callback; /* Callback for kicking laggard readers */
-
- struct {
- unsigned dp_reserve_limit;
- unsigned rp_augment_limit;
- unsigned dp_limit;
- unsigned dp_initial;
- uint8_t dp_loose_limit;
- uint8_t spill_max_denominator;
- uint8_t spill_min_denominator;
- uint8_t spill_parent4child_denominator;
- unsigned merge_threshold_16dot16_percent;
- union {
- unsigned all;
- /* tracks options with non-auto values but tuned by user */
- struct {
- unsigned dp_limit : 1;
- } non_auto;
- } flags;
- } me_options;
-
- /* struct me_dbgeo used for accepting db-geo params from user for the new
- * database creation, i.e. when mdbx_env_set_geometry() was called before
- * mdbx_env_open(). */
- struct {
- size_t lower; /* minimal size of datafile */
- size_t upper; /* maximal size of datafile */
- size_t now; /* current size of datafile */
- size_t grow; /* step to grow datafile */
- size_t shrink; /* threshold to shrink datafile */
- } me_dbgeo;
-
-#if MDBX_LOCKING == MDBX_LOCKING_SYSV
- union {
- key_t key;
- int semid;
- } me_sysv_ipc;
-#endif /* MDBX_LOCKING == MDBX_LOCKING_SYSV */
-
- MDBX_env *me_lcklist_next;
-
- /* --------------------------------------------------- mostly volatile part */
-
- MDBX_txn *me_txn; /* current write transaction */
- mdbx_fastmutex_t me_dbi_lock;
- MDBX_dbi me_numdbs; /* number of DBs opened */
-
- MDBX_page *me_dp_reserve; /* list of malloc'ed blocks for re-use */
- unsigned me_dp_reserve_len;
- /* PNL of pages that became unused in a write txn */
- MDBX_PNL me_retired_pages;
-
-#if defined(_WIN32) || defined(_WIN64)
- MDBX_srwlock me_remap_guard;
- /* Workaround for LockFileEx and WriteFile multithread bug */
- CRITICAL_SECTION me_windowsbug_lock;
-#else
- mdbx_fastmutex_t me_remap_guard;
-#endif
-
- /* -------------------------------------------------------------- debugging */
-
-#if MDBX_DEBUG
- MDBX_assert_func *me_assert_func; /* Callback for assertion failures */
-#endif
-#ifdef MDBX_USE_VALGRIND
- int me_valgrind_handle;
-#endif
-#if defined(MDBX_USE_VALGRIND) || defined(__SANITIZE_ADDRESS__)
- pgno_t me_poison_edge;
-#endif /* MDBX_USE_VALGRIND || __SANITIZE_ADDRESS__ */
-
-#ifndef xMDBX_DEBUG_SPILLING
-#define xMDBX_DEBUG_SPILLING 0
-#endif
-#if xMDBX_DEBUG_SPILLING == 2
- unsigned debug_dirtied_est, debug_dirtied_act;
-#endif /* xMDBX_DEBUG_SPILLING */
-
- /* ------------------------------------------------- stub for lck-less mode */
- MDBX_atomic_uint64_t
- x_lckless_stub[(sizeof(MDBX_lockinfo) + MDBX_CACHELINE_SIZE - 1) /
- sizeof(MDBX_atomic_uint64_t)];
-};
-
-#ifndef __cplusplus
-/*----------------------------------------------------------------------------*/
-/* Debug and Logging stuff */
-
-#define MDBX_RUNTIME_FLAGS_INIT \
- ((MDBX_DEBUG) > 0) * MDBX_DBG_ASSERT + ((MDBX_DEBUG) > 1) * MDBX_DBG_AUDIT
-
-extern uint8_t mdbx_runtime_flags;
-extern uint8_t mdbx_loglevel;
-extern MDBX_debug_func *mdbx_debug_logger;
-
-MDBX_MAYBE_UNUSED static __inline void mdbx_jitter4testing(bool tiny) {
-#if MDBX_DEBUG
- if (MDBX_DBG_JITTER & mdbx_runtime_flags)
- mdbx_osal_jitter(tiny);
-#else
- (void)tiny;
-#endif
-}
-
-MDBX_INTERNAL_FUNC void MDBX_PRINTF_ARGS(4, 5)
- mdbx_debug_log(int level, const char *function, int line, const char *fmt,
- ...) MDBX_PRINTF_ARGS(4, 5);
-MDBX_INTERNAL_FUNC void mdbx_debug_log_va(int level, const char *function,
- int line, const char *fmt,
- va_list args);
-
-#if MDBX_DEBUG
-#define mdbx_log_enabled(msg) unlikely(msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() unlikely((mdbx_runtime_flags & MDBX_DBG_AUDIT))
-#else /* MDBX_DEBUG */
-#define mdbx_log_enabled(msg) (msg < MDBX_LOG_VERBOSE && msg <= mdbx_loglevel)
-#define mdbx_audit_enabled() (0)
-#endif /* MDBX_DEBUG */
-
-#if MDBX_FORCE_ASSERTIONS
-#define mdbx_assert_enabled() (1)
-#elif MDBX_DEBUG
-#define mdbx_assert_enabled() likely((mdbx_runtime_flags & MDBX_DBG_ASSERT))
-#else
-#define mdbx_assert_enabled() (0)
-#endif /* assertions */
-
-#if !MDBX_DEBUG && defined(__ANDROID_API__)
-#define mdbx_assert_fail(env, msg, func, line) \
- __android_log_assert(msg, "mdbx", "%s:%u", func, line)
-#else
-void mdbx_assert_fail(const MDBX_env *env, const char *msg, const char *func,
- int line);
-#endif
-
-#define mdbx_debug_extra(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, __func__, __LINE__, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug_extra_print(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_EXTRA)) \
- mdbx_debug_log(MDBX_LOG_EXTRA, NULL, 0, fmt, __VA_ARGS__); \
- } while (0)
-
-#define mdbx_trace(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_TRACE)) \
- mdbx_debug_log(MDBX_LOG_TRACE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_debug(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_DEBUG)) \
- mdbx_debug_log(MDBX_LOG_DEBUG, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_verbose(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_VERBOSE)) \
- mdbx_debug_log(MDBX_LOG_VERBOSE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_notice(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_NOTICE)) \
- mdbx_debug_log(MDBX_LOG_NOTICE, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_warning(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_WARN)) \
- mdbx_debug_log(MDBX_LOG_WARN, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_error(fmt, ...) \
- do { \
- if (mdbx_log_enabled(MDBX_LOG_ERROR)) \
- mdbx_debug_log(MDBX_LOG_ERROR, __func__, __LINE__, fmt "\n", \
- __VA_ARGS__); \
- } while (0)
-
-#define mdbx_fatal(fmt, ...) \
- mdbx_debug_log(MDBX_LOG_FATAL, __func__, __LINE__, fmt "\n", __VA_ARGS__);
-
-#define mdbx_ensure_msg(env, expr, msg) \
- do { \
- if (unlikely(!(expr))) \
- mdbx_assert_fail(env, msg, __func__, __LINE__); \
- } while (0)
-
-#define mdbx_ensure(env, expr) mdbx_ensure_msg(env, expr, #expr)
-
-/* assert(3) variant in environment context */
-#define mdbx_assert(env, expr) \
- do { \
- if (mdbx_assert_enabled()) \
- mdbx_ensure(env, expr); \
- } while (0)
-
-/* assert(3) variant in cursor context */
-#define mdbx_cassert(mc, expr) mdbx_assert((mc)->mc_txn->mt_env, expr)
-
-/* assert(3) variant in transaction context */
-#define mdbx_tassert(txn, expr) mdbx_assert((txn)->mt_env, expr)
-
-#ifndef xMDBX_TOOLS /* Avoid using internal mdbx_assert() */
-#undef assert
-#define assert(expr) mdbx_assert(NULL, expr)
-#endif
-
-/*----------------------------------------------------------------------------*/
-/* Cache coherence and mmap invalidation */
-
-#if MDBX_CPU_WRITEBACK_INCOHERENT
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_memory_barrier()
-#else
-#define mdbx_flush_incoherent_cpu_writeback() mdbx_compiler_barrier()
-#endif /* MDBX_CPU_WRITEBACK_INCOHERENT */
-
-MDBX_MAYBE_UNUSED static __inline void
-mdbx_flush_incoherent_mmap(void *addr, size_t nbytes, const intptr_t pagesize) {
-#if MDBX_MMAP_INCOHERENT_FILE_WRITE
- char *const begin = (char *)(-pagesize & (intptr_t)addr);
- char *const end =
- (char *)(-pagesize & (intptr_t)((char *)addr + nbytes + pagesize - 1));
- int err = msync(begin, end - begin, MS_SYNC | MS_INVALIDATE) ? errno : 0;
- mdbx_assert(nullptr, err == 0);
- (void)err;
-#else
- (void)pagesize;
-#endif /* MDBX_MMAP_INCOHERENT_FILE_WRITE */
-
-#if MDBX_MMAP_INCOHERENT_CPU_CACHE
-#ifdef DCACHE
- /* MIPS has cache coherency issues.
- * Note: for any nbytes >= on-chip cache size, entire is flushed. */
- cacheflush(addr, nbytes, DCACHE);
-#else
-#error "Oops, cacheflush() not available"
-#endif /* DCACHE */
-#endif /* MDBX_MMAP_INCOHERENT_CPU_CACHE */
-
-#if !MDBX_MMAP_INCOHERENT_FILE_WRITE && !MDBX_MMAP_INCOHERENT_CPU_CACHE
- (void)addr;
- (void)nbytes;
-#endif
-}
-
-/*----------------------------------------------------------------------------*/
-/* Internal prototypes */
-
-MDBX_INTERNAL_FUNC int mdbx_cleanup_dead_readers(MDBX_env *env, int rlocked,
- int *dead);
-MDBX_INTERNAL_FUNC int mdbx_rthc_alloc(mdbx_thread_key_t *key,
- MDBX_reader *begin, MDBX_reader *end);
-MDBX_INTERNAL_FUNC void mdbx_rthc_remove(const mdbx_thread_key_t key);
-
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_init(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_global_dtor(void);
-MDBX_INTERNAL_FUNC void mdbx_rthc_thread_dtor(void *ptr);
-
-#endif /* !__cplusplus */
-
-#define MDBX_IS_ERROR(rc) \
- ((rc) != MDBX_RESULT_TRUE && (rc) != MDBX_RESULT_FALSE)
-
-/* Internal error codes, not exposed outside libmdbx */
-#define MDBX_NO_ROOT (MDBX_LAST_ADDED_ERRCODE + 10)
-
-/* Debugging output value of a cursor DBI: Negative in a sub-cursor. */
-#define DDBI(mc) \
- (((mc)->mc_flags & C_SUB) ? -(int)(mc)->mc_dbi : (int)(mc)->mc_dbi)
-
-/* Key size which fits in a DKBUF (debug key buffer). */
-#define DKBUF_MAX 511
-#define DKBUF char _kbuf[DKBUF_MAX * 4 + 2]
-#define DKEY(x) mdbx_dump_val(x, _kbuf, DKBUF_MAX * 2 + 1)
-#define DVAL(x) mdbx_dump_val(x, _kbuf + DKBUF_MAX * 2 + 1, DKBUF_MAX * 2 + 1)
-
-#if MDBX_DEBUG
-#define DKBUF_DEBUG DKBUF
-#define DKEY_DEBUG(x) DKEY(x)
-#define DVAL_DEBUG(x) DVAL(x)
-#else
-#define DKBUF_DEBUG ((void)(0))
-#define DKEY_DEBUG(x) ("-")
-#define DVAL_DEBUG(x) ("-")
-#endif
-
-/* An invalid page number.
- * Mainly used to denote an empty tree. */
-#define P_INVALID (~(pgno_t)0)
-
-/* Test if the flags f are set in a flag word w. */
-#define F_ISSET(w, f) (((w) & (f)) == (f))
-
-/* Round n up to an even number. */
-#define EVEN(n) (((n) + 1U) & -2) /* sign-extending -2 to match n+1U */
-
-/* Default size of memory map.
- * This is certainly too small for any actual applications. Apps should
- * always set the size explicitly using mdbx_env_set_geometry(). */
-#define DEFAULT_MAPSIZE MEGABYTE
-
-/* Number of slots in the reader table.
- * This value was chosen somewhat arbitrarily. The 61 is a prime number,
- * and such readers plus a couple mutexes fit into single 4KB page.
- * Applications should set the table size using mdbx_env_set_maxreaders(). */
-#define DEFAULT_READERS 61
-
-/* Test if a page is a leaf page */
-#define IS_LEAF(p) (((p)->mp_flags & P_LEAF) != 0)
-/* Test if a page is a LEAF2 page */
-#define IS_LEAF2(p) unlikely(((p)->mp_flags & P_LEAF2) != 0)
-/* Test if a page is a branch page */
-#define IS_BRANCH(p) (((p)->mp_flags & P_BRANCH) != 0)
-/* Test if a page is an overflow page */
-#define IS_OVERFLOW(p) unlikely(((p)->mp_flags & P_OVERFLOW) != 0)
-/* Test if a page is a sub page */
-#define IS_SUBP(p) (((p)->mp_flags & P_SUBP) != 0)
-
-#define PAGETYPE(p) ((p)->mp_flags & (P_BRANCH | P_LEAF | P_LEAF2 | P_OVERFLOW))
-
-/* Header for a single key/data pair within a page.
- * Used in pages of type P_BRANCH and P_LEAF without P_LEAF2.
- * We guarantee 2-byte alignment for 'MDBX_node's.
- *
- * Leaf node flags describe node contents. F_BIGDATA says the node's
- * data part is the page number of an overflow page with actual data.
- * F_DUPDATA and F_SUBDATA can be combined giving duplicate data in
- * a sub-page/sub-database, and named databases (just F_SUBDATA). */
-typedef struct MDBX_node {
-#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
- union {
- uint32_t mn_dsize;
- uint32_t mn_pgno32;
- };
- uint8_t mn_flags; /* see mdbx_node flags */
- uint8_t mn_extra;
- uint16_t mn_ksize; /* key size */
-#else
- uint16_t mn_ksize; /* key size */
- uint8_t mn_extra;
- uint8_t mn_flags; /* see mdbx_node flags */
- union {
- uint32_t mn_pgno32;
- uint32_t mn_dsize;
- };
-#endif /* __BYTE_ORDER__ */
-
- /* mdbx_node Flags */
-#define F_BIGDATA 0x01 /* data put on overflow page */
-#define F_SUBDATA 0x02 /* data is a sub-database */
-#define F_DUPDATA 0x04 /* data has duplicates */
-
- /* valid flags for mdbx_node_add() */
-#define NODE_ADD_FLAGS (F_DUPDATA | F_SUBDATA | MDBX_RESERVE | MDBX_APPEND)
-
-#if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L) || \
- (!defined(__cplusplus) && defined(_MSC_VER))
- uint8_t mn_data[] /* key and data are appended here */;
-#endif /* C99 */
-} MDBX_node;
-
-#define DB_PERSISTENT_FLAGS \
- (MDBX_REVERSEKEY | MDBX_DUPSORT | MDBX_INTEGERKEY | MDBX_DUPFIXED | \
- MDBX_INTEGERDUP | MDBX_REVERSEDUP)
-
-/* mdbx_dbi_open() flags */
-#define DB_USABLE_FLAGS (DB_PERSISTENT_FLAGS | MDBX_CREATE | MDBX_DB_ACCEDE)
-
-#define DB_VALID 0x8000 /* DB handle is valid, for me_dbflags */
-#define DB_INTERNAL_FLAGS DB_VALID
-
-#if DB_INTERNAL_FLAGS & DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-#if DB_PERSISTENT_FLAGS & ~DB_USABLE_FLAGS
-#error "Oops, some flags overlapped or wrong"
-#endif
-
-/* max number of pages to commit in one writev() call */
-#define MDBX_COMMIT_PAGES 64
-#if defined(IOV_MAX) && IOV_MAX < MDBX_COMMIT_PAGES /* sysconf(_SC_IOV_MAX) */
-#undef MDBX_COMMIT_PAGES
-#define MDBX_COMMIT_PAGES IOV_MAX
-#endif
-
-/*
- * /
- * | -1, a < b
- * CMP2INT(a,b) = < 0, a == b
- * | 1, a > b
- * \
- */
-#ifndef __e2k__
-/* LY: fast enough on most systems */
-#define CMP2INT(a, b) (((b) > (a)) ? -1 : (a) > (b))
-#else
-/* LY: more parallelable on VLIW Elbrus */
-#define CMP2INT(a, b) (((a) > (b)) - ((b) > (a)))
-#endif
-
-/* Do not spill pages to disk if txn is getting full, may fail instead */
-#define MDBX_NOSPILL 0x8000
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-int64pgno(int64_t i64) {
- if (likely(i64 >= (int64_t)MIN_PAGENO && i64 <= (int64_t)MAX_PAGENO + 1))
- return (pgno_t)i64;
- return (i64 < (int64_t)MIN_PAGENO) ? MIN_PAGENO : MAX_PAGENO;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_add(size_t base, size_t augend) {
- assert(base <= MAX_PAGENO + 1 && augend < MAX_PAGENO);
- return int64pgno(base + augend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __inline pgno_t
-pgno_sub(size_t base, size_t subtrahend) {
- assert(base >= MIN_PAGENO && base <= MAX_PAGENO + 1 &&
- subtrahend < MAX_PAGENO);
- return int64pgno(base - subtrahend);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline bool
-is_powerof2(size_t x) {
- return (x & (x - 1)) == 0;
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-floor_powerof2(size_t value, size_t granularity) {
- assert(is_powerof2(granularity));
- return value & ~(granularity - 1);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static __always_inline size_t
-ceil_powerof2(size_t value, size_t granularity) {
- return floor_powerof2(value + granularity - 1, granularity);
-}
-
-MDBX_MAYBE_UNUSED MDBX_NOTHROW_CONST_FUNCTION static unsigned
-log2n_powerof2(size_t value) {
- assert(value > 0 && value < INT32_MAX && is_powerof2(value));
- assert((value & -(int32_t)value) == value);
-#if __GNUC_PREREQ(4, 1) || __has_builtin(__builtin_ctzl)
- return __builtin_ctzl(value);
-#elif defined(_MSC_VER)
- unsigned long index;
- _BitScanForward(&index, (unsigned long)value);
- return index;
-#else
- static const uint8_t debruijn_ctz32[32] = {
- 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
- 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9};
- return debruijn_ctz32[(uint32_t)(value * 0x077CB531u) >> 27];
-#endif
-}
-
-/* Only a subset of the mdbx_env flags can be changed
- * at runtime. Changing other flags requires closing the
- * environment and re-opening it with the new flags. */
-#define ENV_CHANGEABLE_FLAGS \
- (MDBX_SAFE_NOSYNC | MDBX_NOMETASYNC | MDBX_DEPRECATED_MAPASYNC | \
- MDBX_NOMEMINIT | MDBX_COALESCE | MDBX_PAGEPERTURB | MDBX_ACCEDE)
-#define ENV_CHANGELESS_FLAGS \
- (MDBX_NOSUBDIR | MDBX_RDONLY | MDBX_WRITEMAP | MDBX_NOTLS | MDBX_NORDAHEAD | \
- MDBX_LIFORECLAIM | MDBX_EXCLUSIVE)
-#define ENV_USABLE_FLAGS (ENV_CHANGEABLE_FLAGS | ENV_CHANGELESS_FLAGS)
-
-#if !defined(__cplusplus) || CONSTEXPR_ENUM_FLAGS_OPERATIONS
-MDBX_MAYBE_UNUSED static void static_checks(void) {
- STATIC_ASSERT_MSG(INT16_MAX - CORE_DBS == MDBX_MAX_DBI,
- "Oops, MDBX_MAX_DBI or CORE_DBS?");
- STATIC_ASSERT_MSG((unsigned)(MDBX_DB_ACCEDE | MDBX_CREATE) ==
- ((DB_USABLE_FLAGS | DB_INTERNAL_FLAGS) &
- (ENV_USABLE_FLAGS | ENV_INTERNAL_FLAGS)),
- "Oops, some flags overlapped or wrong");
- STATIC_ASSERT_MSG((ENV_INTERNAL_FLAGS & ENV_USABLE_FLAGS) == 0,
- "Oops, some flags overlapped or wrong");
-}
-#endif /* Disabled for MSVC 19.0 (VisualStudio 2015) */
-
-#ifdef __cplusplus
-}
-#endif
-
-#define MDBX_ASAN_POISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("POISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_POISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#define MDBX_ASAN_UNPOISON_MEMORY_REGION(addr, size) \
- do { \
- mdbx_trace("UNPOISON_MEMORY_REGION(%p, %zu) at %u", (void *)(addr), \
- (size_t)(size), __LINE__); \
- ASAN_UNPOISON_MEMORY_REGION(addr, size); \
- } while (0)
-
-#if defined(_WIN32) || defined(_WIN64)
-/*
- * POSIX getopt for Windows
- *
- * AT&T Public License
- *
- * Code given out at the 1985 UNIFORUM conference in Dallas.
- */
-
-/*----------------------------------------------------------------------------*/
-/* Microsoft compiler generates a lot of warning for self includes... */
-
-#ifdef _MSC_VER
-#pragma warning(push, 1)
-#pragma warning(disable : 4548) /* expression before comma has no effect; \
- expected expression with side - effect */
-#pragma warning(disable : 4530) /* C++ exception handler used, but unwind \
- * semantics are not enabled. Specify /EHsc */
-#pragma warning(disable : 4577) /* 'noexcept' used with no exception handling \
- * mode specified; termination on exception is \
- * not guaranteed. Specify /EHsc */
-#if !defined(_CRT_SECURE_NO_WARNINGS)
-#define _CRT_SECURE_NO_WARNINGS
-#endif
-#endif /* _MSC_VER (warnings) */
-
-#include <stdio.h>
-#include <string.h>
-
-#ifdef _MSC_VER
-#pragma warning(pop)
-#endif
-/*----------------------------------------------------------------------------*/
-
-#ifndef NULL
-#define NULL 0
-#endif
-
-#ifndef EOF
-#define EOF (-1)
-#endif
-
-int optind = 1;
-int optopt;
-char *optarg;
-
-int getopt(int argc, char *const argv[], const char *opts) {
- static int sp = 1;
- int c;
- const char *cp;
-
- if (sp == 1) {
- if (optind >= argc || argv[optind][0] != '-' || argv[optind][1] == '\0')
- return EOF;
- else if (strcmp(argv[optind], "--") == 0) {
- optind++;
- return EOF;
- }
- }
- optopt = c = argv[optind][sp];
- if (c == ':' || (cp = strchr(opts, c)) == NULL) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "illegal option", c);
- if (argv[optind][++sp] == '\0') {
- optind++;
- sp = 1;
- }
- return '?';
- }
- if (*++cp == ':') {
- if (argv[optind][sp + 1] != '\0')
- optarg = &argv[optind++][sp + 1];
- else if (++optind >= argc) {
- fprintf(stderr, "%s: %s -- %c\n", argv[0], "option requires an argument",
- c);
- sp = 1;
- return '?';
- } else
- optarg = argv[optind++];
- sp = 1;
- } else {
- if (argv[optind][++sp] == '\0') {
- sp = 1;
- optind++;
- }
- optarg = NULL;
- }
- return c;
-}
-
-static volatile BOOL user_break;
-static BOOL WINAPI ConsoleBreakHandlerRoutine(DWORD dwCtrlType) {
- (void)dwCtrlType;
- user_break = true;
- return true;
-}
-
-#else /* WINDOWS */
-
-static volatile sig_atomic_t user_break;
-static void signal_handler(int sig) {
- (void)sig;
- user_break = 1;
-}
-
-#endif /* !WINDOWS */
-
-static void print_stat(MDBX_stat *ms) {
- printf(" Pagesize: %u\n", ms->ms_psize);
- printf(" Tree depth: %u\n", ms->ms_depth);
- printf(" Branch pages: %" PRIu64 "\n", ms->ms_branch_pages);
- printf(" Leaf pages: %" PRIu64 "\n", ms->ms_leaf_pages);
- printf(" Overflow pages: %" PRIu64 "\n", ms->ms_overflow_pages);
- printf(" Entries: %" PRIu64 "\n", ms->ms_entries);
-}
-
-static void usage(const char *prog) {
- fprintf(stderr,
- "usage: %s [-V] [-q] [-e] [-f[f[f]]] [-r[r]] [-a|-s name] dbpath\n"
- " -V\t\tprint version and exit\n"
- " -q\t\tbe quiet\n"
- " -p\t\tshow statistics of page operations for current session\n"
- " -e\t\tshow whole DB info\n"
- " -f\t\tshow GC info\n"
- " -r\t\tshow readers\n"
- " -a\t\tprint stat of main DB and all subDBs\n"
- " -s name\tprint stat of only the specified named subDB\n"
- " \t\tby default print stat of only the main DB\n",
- prog);
- exit(EXIT_FAILURE);
-}
-
-static int reader_list_func(void *ctx, int num, int slot, mdbx_pid_t pid,
- mdbx_tid_t thread, uint64_t txnid, uint64_t lag,
- size_t bytes_used, size_t bytes_retained) {
- (void)ctx;
- if (num == 1)
- printf("Reader Table\n"
- " #\tslot\t%6s %*s %20s %10s %13s %13s\n",
- "pid", (int)sizeof(size_t) * 2, "thread", "txnid", "lag", "used",
- "retained");
-
- printf(" %3d)\t[%d]\t%6" PRIdSIZE " %*" PRIxPTR, num, slot, (size_t)pid,
- (int)sizeof(size_t) * 2, (uintptr_t)thread);
- if (txnid)
- printf(" %20" PRIu64 " %10" PRIu64 " %12.1fM %12.1fM\n", txnid, lag,
- bytes_used / 1048576.0, bytes_retained / 1048576.0);
- else
- printf(" %20s %10s %13s %13s\n", "-", "0", "0", "0");
-
- return user_break ? MDBX_RESULT_TRUE : MDBX_RESULT_FALSE;
-}
-
-const char *prog;
-bool quiet = false;
-static void error(const char *func, int rc) {
- if (!quiet)
- fprintf(stderr, "%s: %s() error %d %s\n", prog, func, rc,
- mdbx_strerror(rc));
-}
-
-int main(int argc, char *argv[]) {
- int opt, rc;
- MDBX_env *env;
- MDBX_txn *txn;
- MDBX_dbi dbi;
- MDBX_envinfo mei;
- prog = argv[0];
- char *envname;
- char *subname = nullptr;
- bool alldbs = false, envinfo = false, pgop = false;
- int freinfo = 0, rdrinfo = 0;
-
- if (argc < 2)
- usage(prog);
-
- while ((opt = getopt(argc, argv,
- "V"
- "q"
- "p"
- "a"
- "e"
- "f"
- "n"
- "r"
- "s:")) != EOF) {
- switch (opt) {
- case 'V':
- printf("mdbx_stat version %d.%d.%d.%d\n"
- " - source: %s %s, commit %s, tree %s\n"
- " - anchor: %s\n"
- " - build: %s for %s by %s\n"
- " - flags: %s\n"
- " - options: %s\n",
- mdbx_version.major, mdbx_version.minor, mdbx_version.release,
- mdbx_version.revision, mdbx_version.git.describe,
- mdbx_version.git.datetime, mdbx_version.git.commit,
- mdbx_version.git.tree, mdbx_sourcery_anchor, mdbx_build.datetime,
- mdbx_build.target, mdbx_build.compiler, mdbx_build.flags,
- mdbx_build.options);
- return EXIT_SUCCESS;
- case 'q':
- quiet = true;
- break;
- case 'p':
- pgop = true;
- break;
- case 'a':
- if (subname)
- usage(prog);
- alldbs = true;
- break;
- case 'e':
- envinfo = true;
- break;
- case 'f':
- freinfo += 1;
- break;
- case 'n':
- break;
- case 'r':
- rdrinfo += 1;
- break;
- case 's':
- if (alldbs)
- usage(prog);
- subname = optarg;
- break;
- default:
- usage(prog);
- }
- }
-
- if (optind != argc - 1)
- usage(prog);
-
-#if defined(_WIN32) || defined(_WIN64)
- SetConsoleCtrlHandler(ConsoleBreakHandlerRoutine, true);
-#else
-#ifdef SIGPIPE
- signal(SIGPIPE, signal_handler);
-#endif
-#ifdef SIGHUP
- signal(SIGHUP, signal_handler);
-#endif
- signal(SIGINT, signal_handler);
- signal(SIGTERM, signal_handler);
-#endif /* !WINDOWS */
-
- envname = argv[optind];
- envname = argv[optind];
- if (!quiet) {
- printf("mdbx_stat %s (%s, T-%s)\nRunning for %s...\n",
- mdbx_version.git.describe, mdbx_version.git.datetime,
- mdbx_version.git.tree, envname);
- fflush(nullptr);
- }
-
- rc = mdbx_env_create(&env);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_create", rc);
- return EXIT_FAILURE;
- }
-
- if (alldbs || subname) {
- rc = mdbx_env_set_maxdbs(env, 2);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_set_maxdbs", rc);
- goto env_close;
- }
- }
-
- rc = mdbx_env_open(env, envname, MDBX_RDONLY, 0);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_open", rc);
- goto env_close;
- }
-
- rc = mdbx_txn_begin(env, nullptr, MDBX_TXN_RDONLY, &txn);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_txn_begin", rc);
- goto txn_abort;
- }
-
- if (envinfo || freinfo || pgop) {
- rc = mdbx_env_info_ex(env, txn, &mei, sizeof(mei));
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_env_info_ex", rc);
- goto txn_abort;
- }
- } else {
- /* LY: zap warnings from gcc */
- memset(&mei, 0, sizeof(mei));
- }
-
- if (pgop) {
- printf("Page Operations (for current session):\n");
- printf(" New: %8" PRIu64 "\t// quantity of a new pages added\n",
- mei.mi_pgop_stat.newly);
- printf(" CoW: %8" PRIu64
- "\t// quantity of pages copied for altering\n",
- mei.mi_pgop_stat.cow);
- printf(" Clone: %8" PRIu64 "\t// quantity of parent's dirty pages "
- "clones for nested transactions\n",
- mei.mi_pgop_stat.clone);
- printf(" Split: %8" PRIu64
- "\t// page splits during insertions or updates\n",
- mei.mi_pgop_stat.split);
- printf(" Merge: %8" PRIu64
- "\t// page merges during deletions or updates\n",
- mei.mi_pgop_stat.merge);
- printf(" Spill: %8" PRIu64 "\t// quantity of spilled/ousted `dirty` "
- "pages during large transactions\n",
- mei.mi_pgop_stat.spill);
- printf(" Unspill: %8" PRIu64 "\t// quantity of unspilled/redone `dirty` "
- "pages during large transactions\n",
- mei.mi_pgop_stat.unspill);
- printf(" WOP: %8" PRIu64
- "\t// number of explicit write operations (not a pages) to a disk\n",
- mei.mi_pgop_stat.wops);
- }
-
- if (envinfo) {
- printf("Environment Info\n");
- printf(" Pagesize: %u\n", mei.mi_dxb_pagesize);
- if (mei.mi_geo.lower != mei.mi_geo.upper) {
- printf(" Dynamic datafile: %" PRIu64 "..%" PRIu64 " bytes (+%" PRIu64
- "/-%" PRIu64 "), %" PRIu64 "..%" PRIu64 " pages (+%" PRIu64
- "/-%" PRIu64 ")\n",
- mei.mi_geo.lower, mei.mi_geo.upper, mei.mi_geo.grow,
- mei.mi_geo.shrink, mei.mi_geo.lower / mei.mi_dxb_pagesize,
- mei.mi_geo.upper / mei.mi_dxb_pagesize,
- mei.mi_geo.grow / mei.mi_dxb_pagesize,
- mei.mi_geo.shrink / mei.mi_dxb_pagesize);
- printf(" Current mapsize: %" PRIu64 " bytes, %" PRIu64 " pages \n",
- mei.mi_mapsize, mei.mi_mapsize / mei.mi_dxb_pagesize);
- printf(" Current datafile: %" PRIu64 " bytes, %" PRIu64 " pages\n",
- mei.mi_geo.current, mei.mi_geo.current / mei.mi_dxb_pagesize);
-#if defined(_WIN32) || defined(_WIN64)
- if (mei.mi_geo.shrink && mei.mi_geo.current != mei.mi_geo.upper)
- printf(" WARNING: Due Windows system limitations a "
- "file couldn't\n be truncated while database "
- "is opened. So, the size of\n database file "
- "may by large than the database itself,\n "
- "until it will be closed or reopened in read-write mode.\n");
-#endif
- } else {
- printf(" Fixed datafile: %" PRIu64 " bytes, %" PRIu64 " pages\n",
- mei.mi_geo.current, mei.mi_geo.current / mei.mi_dxb_pagesize);
- }
- printf(" Last transaction ID: %" PRIu64 "\n", mei.mi_recent_txnid);
- printf(" Latter reader transaction ID: %" PRIu64 " (%" PRIi64 ")\n",
- mei.mi_latter_reader_txnid,
- mei.mi_latter_reader_txnid - mei.mi_recent_txnid);
- printf(" Max readers: %u\n", mei.mi_maxreaders);
- printf(" Number of reader slots uses: %u\n", mei.mi_numreaders);
- }
-
- if (rdrinfo) {
- rc = mdbx_reader_list(env, reader_list_func, nullptr);
- if (MDBX_IS_ERROR(rc)) {
- error("mdbx_reader_list", rc);
- goto txn_abort;
- }
- if (rc == MDBX_RESULT_TRUE)
- printf("Reader Table is empty\n");
- else if (rc == MDBX_SUCCESS && rdrinfo > 1) {
- int dead;
- rc = mdbx_reader_check(env, &dead);
- if (MDBX_IS_ERROR(rc)) {
- error("mdbx_reader_check", rc);
- goto txn_abort;
- }
- if (rc == MDBX_RESULT_TRUE) {
- printf(" %d stale readers cleared.\n", dead);
- rc = mdbx_reader_list(env, reader_list_func, nullptr);
- if (rc == MDBX_RESULT_TRUE)
- printf(" Now Reader Table is empty\n");
- } else
- printf(" No stale readers.\n");
- }
- if (!(subname || alldbs || freinfo))
- goto txn_abort;
- }
-
- if (freinfo) {
- printf("Garbage Collection\n");
- dbi = 0;
- MDBX_cursor *cursor;
- rc = mdbx_cursor_open(txn, dbi, &cursor);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_cursor_open", rc);
- goto txn_abort;
- }
-
- MDBX_stat mst;
- rc = mdbx_dbi_stat(txn, dbi, &mst, sizeof(mst));
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_stat", rc);
- goto txn_abort;
- }
- print_stat(&mst);
-
- pgno_t pages = 0, *iptr;
- pgno_t reclaimable = 0;
- MDBX_val key, data;
- while (MDBX_SUCCESS ==
- (rc = mdbx_cursor_get(cursor, &key, &data, MDBX_NEXT))) {
- if (user_break) {
- rc = MDBX_EINTR;
- break;
- }
- iptr = data.iov_base;
- const pgno_t number = *iptr++;
-
- pages += number;
- if (envinfo && mei.mi_latter_reader_txnid > *(txnid_t *)key.iov_base)
- reclaimable += number;
-
- if (freinfo > 1) {
- char *bad = "";
- pgno_t prev =
- MDBX_PNL_ASCENDING ? NUM_METAS - 1 : (pgno_t)mei.mi_last_pgno + 1;
- pgno_t span = 1;
- for (unsigned i = 0; i < number; ++i) {
- pgno_t pg = iptr[i];
- if (MDBX_PNL_DISORDERED(prev, pg))
- bad = " [bad sequence]";
- prev = pg;
- while (i + span < number &&
- iptr[i + span] == (MDBX_PNL_ASCENDING ? pgno_add(pg, span)
- : pgno_sub(pg, span)))
- ++span;
- }
- printf(" Transaction %" PRIaTXN ", %" PRIaPGNO
- " pages, maxspan %" PRIaPGNO "%s\n",
- *(txnid_t *)key.iov_base, number, span, bad);
- if (freinfo > 2) {
- for (unsigned i = 0; i < number; i += span) {
- const pgno_t pg = iptr[i];
- for (span = 1;
- i + span < number &&
- iptr[i + span] == (MDBX_PNL_ASCENDING ? pgno_add(pg, span)
- : pgno_sub(pg, span));
- ++span)
- ;
- if (span > 1)
- printf(" %9" PRIaPGNO "[%" PRIaPGNO "]\n", pg, span);
- else
- printf(" %9" PRIaPGNO "\n", pg);
- }
- }
- }
- }
- mdbx_cursor_close(cursor);
- cursor = nullptr;
-
- switch (rc) {
- case MDBX_SUCCESS:
- case MDBX_NOTFOUND:
- break;
- case MDBX_EINTR:
- if (!quiet)
- fprintf(stderr, "Interrupted by signal/user\n");
- goto txn_abort;
- default:
- error("mdbx_cursor_get", rc);
- goto txn_abort;
- }
-
- if (envinfo) {
- uint64_t value = mei.mi_mapsize / mei.mi_dxb_pagesize;
- double percent = value / 100.0;
- printf("Page Usage\n");
- printf(" Total: %" PRIu64 " 100%%\n", value);
-
- value = mei.mi_geo.current / mei.mi_dxb_pagesize;
- printf(" Backed: %" PRIu64 " %.1f%%\n", value, value / percent);
-
- value = mei.mi_last_pgno + 1;
- printf(" Allocated: %" PRIu64 " %.1f%%\n", value, value / percent);
-
- value = mei.mi_mapsize / mei.mi_dxb_pagesize - (mei.mi_last_pgno + 1);
- printf(" Remained: %" PRIu64 " %.1f%%\n", value, value / percent);
-
- value = mei.mi_last_pgno + 1 - pages;
- printf(" Used: %" PRIu64 " %.1f%%\n", value, value / percent);
-
- value = pages;
- printf(" GC: %" PRIu64 " %.1f%%\n", value, value / percent);
-
- value = pages - reclaimable;
- printf(" Retained: %" PRIu64 " %.1f%%\n", value, value / percent);
-
- value = reclaimable;
- printf(" Reclaimable: %" PRIu64 " %.1f%%\n", value, value / percent);
-
- value = mei.mi_mapsize / mei.mi_dxb_pagesize - (mei.mi_last_pgno + 1) +
- reclaimable;
- printf(" Available: %" PRIu64 " %.1f%%\n", value, value / percent);
- } else
- printf(" GC: %" PRIaPGNO " pages\n", pages);
- }
-
- rc = mdbx_dbi_open(txn, subname, MDBX_DB_ACCEDE, &dbi);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_open", rc);
- goto txn_abort;
- }
-
- MDBX_stat mst;
- rc = mdbx_dbi_stat(txn, dbi, &mst, sizeof(mst));
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_stat", rc);
- goto txn_abort;
- }
- printf("Status of %s\n", subname ? subname : "Main DB");
- print_stat(&mst);
-
- if (alldbs) {
- MDBX_cursor *cursor;
- rc = mdbx_cursor_open(txn, dbi, &cursor);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_cursor_open", rc);
- goto txn_abort;
- }
-
- MDBX_val key;
- while (MDBX_SUCCESS ==
- (rc = mdbx_cursor_get(cursor, &key, nullptr, MDBX_NEXT_NODUP))) {
- MDBX_dbi subdbi;
- if (memchr(key.iov_base, '\0', key.iov_len))
- continue;
- subname = mdbx_malloc(key.iov_len + 1);
- memcpy(subname, key.iov_base, key.iov_len);
- subname[key.iov_len] = '\0';
- rc = mdbx_dbi_open(txn, subname, MDBX_DB_ACCEDE, &subdbi);
- if (rc == MDBX_SUCCESS)
- printf("Status of %s\n", subname);
- mdbx_free(subname);
- if (unlikely(rc != MDBX_SUCCESS)) {
- if (rc == MDBX_INCOMPATIBLE)
- continue;
- error("mdbx_dbi_open", rc);
- goto txn_abort;
- }
-
- rc = mdbx_dbi_stat(txn, subdbi, &mst, sizeof(mst));
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_stat", rc);
- goto txn_abort;
- }
- print_stat(&mst);
-
- rc = mdbx_dbi_close(env, subdbi);
- if (unlikely(rc != MDBX_SUCCESS)) {
- error("mdbx_dbi_close", rc);
- goto txn_abort;
- }
- }
- mdbx_cursor_close(cursor);
- cursor = nullptr;
- }
-
- switch (rc) {
- case MDBX_SUCCESS:
- case MDBX_NOTFOUND:
- break;
- case MDBX_EINTR:
- if (!quiet)
- fprintf(stderr, "Interrupted by signal/user\n");
- break;
- default:
- if (unlikely(rc != MDBX_SUCCESS))
- error("mdbx_cursor_get", rc);
- }
-
- mdbx_dbi_close(env, dbi);
-txn_abort:
- mdbx_txn_abort(txn);
-env_close:
- mdbx_env_close(env);
-
- return rc ? EXIT_FAILURE : EXIT_SUCCESS;
-}
diff --git a/3rdparty/libmdbx/ntdll.def b/3rdparty/libmdbx/ntdll.def
deleted file mode 100644
index f974ea63803120da94ef5ea6882cfed8ce01c713..0000000000000000000000000000000000000000
--- a/3rdparty/libmdbx/ntdll.def
+++ /dev/null
@@ -1,1246 +0,0 @@
-LIBRARY ntdll
-EXPORTS
-CsrAllocateCaptureBuffer
-CsrAllocateMessagePointer
-CsrCaptureMessageBuffer
-CsrCaptureMessageMultiUnicodeStringsInPlace
-CsrCaptureMessageString
-CsrCaptureTimeout
-CsrClientCallServer
-CsrClientConnectToServer
-CsrFreeCaptureBuffer
-CsrGetProcessId
-CsrIdentifyAlertableThread
-CsrSetPriorityClass
-DbgBreakPoint
-DbgPrint
-DbgPrintEx
-DbgPrintReturnControlC
-DbgPrompt
-DbgQueryDebugFilterState
-DbgSetDebugFilterState
-DbgUiConnectToDbg
-DbgUiContinue
-DbgUiConvertStateChangeStructure
-DbgUiDebugActiveProcess
-DbgUiGetThreadDebugObject
-DbgUiIssueRemoteBreakin
-DbgUiRemoteBreakin
-DbgUiSetThreadDebugObject
-DbgUiStopDebugging
-DbgUiWaitStateChange
-DbgUserBreakPoint
-KiRaiseUserExceptionDispatcher
-KiUserApcDispatcher
-KiUserCallbackDispatcher
-KiUserExceptionDispatcher
-LdrAccessResource
-LdrAddRefDll
-LdrDisableThreadCalloutsForDll
-LdrEnumResources
-LdrEnumerateLoadedModules
-LdrFindEntryForAddress
-LdrFindResourceDirectory_U
-LdrFindResourceEx_U
-LdrFindResource_U
-LdrFlushAlternateResourceModules
-LdrGetDllHandle
-LdrGetDllHandleEx
-LdrGetProcedureAddress
-LdrInitShimEngineDynamic
-LdrInitializeThunk
-LdrLoadAlternateResourceModule
-LdrLoadDll
-LdrLockLoaderLock
-LdrProcessRelocationBlock
-LdrQueryImageFileExecutionOptions
-LdrQueryProcessModuleInformation
-LdrSetAppCompatDllRedirectionCallback
-LdrSetDllManifestProber
-LdrShutdownProcess
-LdrShutdownThread
-LdrUnloadAlternateResourceModule
-LdrUnloadDll
-LdrUnlockLoaderLock
-LdrVerifyImageMatchesChecksum
-NlsAnsiCodePage
-NlsMbCodePageTag
-NlsMbOemCodePageTag
-NtAcceptConnectPort
-NtAccessCheck
-NtAccessCheckAndAuditAlarm
-NtAccessCheckByType
-NtAccessCheckByTypeAndAuditAlarm
-NtAccessCheckByTypeResultList
-NtAccessCheckByTypeResultListAndAuditAlarm
-NtAccessCheckByTypeResultListAndAuditAlarmByHandle
-NtAddAtom
-NtAddBootEntry
-NtAdjustGroupsToken
-NtAdjustPrivilegesToken
-NtAlertResumeThread
-NtAlertThread
-NtAllocateLocallyUniqueId
-NtAllocateUserPhysicalPages
-NtAllocateUuids
-NtAllocateVirtualMemory
-NtAreMappedFilesTheSame
-NtAssignProcessToJobObject
-NtCallbackReturn
-NtCancelIoFile
-NtCancelTimer
-NtClearEvent
-NtClose
-NtCloseObjectAuditAlarm
-NtCompactKeys
-NtCompareTokens
-NtCompleteConnectPort
-NtCompressKey
-NtConnectPort
-NtContinue
-NtCreateDebugObject
-NtCreateDirectoryObject
-NtCreateEvent
-NtCreateEventPair
-NtCreateFile
-NtCreateIoCompletion
-NtCreateJobObject
-NtCreateJobSet
-NtCreateKey
-NtCreateKeyedEvent
-NtCreateMailslotFile
-NtCreateMutant
-NtCreateNamedPipeFile
-NtCreatePagingFile
-NtCreatePort
-NtCreateProcess
-NtCreateProcessEx
-NtCreateProfile
-NtCreateSection
-NtCreateSemaphore
-NtCreateSymbolicLinkObject
-NtCreateThread
-NtCreateTimer
-NtCreateToken
-NtCreateWaitablePort
-NtDebugActiveProcess
-NtDebugContinue
-NtDelayExecution
-NtDeleteAtom
-NtDeleteBootEntry
-NtDeleteFile
-NtDeleteKey
-NtDeleteObjectAuditAlarm
-NtDeleteValueKey
-NtDeviceIoControlFile
-NtDisplayString
-NtDuplicateObject
-NtDuplicateToken
-NtEnumerateBootEntries
-NtEnumerateKey
-NtEnumerateSystemEnvironmentValuesEx
-NtEnumerateValueKey
-NtExtendSection
-NtFilterToken
-NtFindAtom
-NtFlushBuffersFile
-NtFlushInstructionCache
-NtFlushKey
-NtFlushVirtualMemory
-NtFlushWriteBuffer
-NtFreeUserPhysicalPages
-NtFreeVirtualMemory
-NtFsControlFile
-NtGetContextThread
-NtGetDevicePowerState
-NtGetWriteWatch
-NtImpersonateAnonymousToken
-NtImpersonateClientOfPort
-NtImpersonateThread
-NtInitializeRegistry
-NtInitiatePowerAction
-NtIsProcessInJob
-NtIsSystemResumeAutomatic
-NtListenPort
-NtLoadDriver
-NtLoadKey
-NtLoadKey2
-NtLockFile
-NtLockProductActivationKeys
-NtLockRegistryKey
-NtLockVirtualMemory
-NtMakePermanentObject
-NtMakeTemporaryObject
-NtMapUserPhysicalPages
-NtMapUserPhysicalPagesScatter
-NtMapViewOfSection
-NtModifyBootEntry
-NtNotifyChangeDirectoryFile
-NtNotifyChangeKey
-NtNotifyChangeMultipleKeys
-NtOpenDirectoryObject
-NtOpenEvent
-NtOpenEventPair
-NtOpenFile
-NtOpenIoCompletion
-NtOpenJobObject
-NtOpenKey
-NtOpenKeyedEvent
-NtOpenMutant
-NtOpenObjectAuditAlarm
-NtOpenProcess
-NtOpenProcessToken
-NtOpenProcessTokenEx
-NtOpenSection
-NtOpenSemaphore
-NtOpenSymbolicLinkObject
-NtOpenThread
-NtOpenThreadToken
-NtOpenThreadTokenEx
-NtOpenTimer
-NtPlugPlayControl
-NtPowerInformation
-NtPrivilegeCheck
-NtPrivilegeObjectAuditAlarm
-NtPrivilegedServiceAuditAlarm
-NtProtectVirtualMemory
-NtPulseEvent
-NtQueryAttributesFile
-NtQueryBootEntryOrder
-NtQueryBootOptions
-NtQueryDebugFilterState
-NtQueryDefaultLocale
-NtQueryDefaultUILanguage
-NtQueryDirectoryFile
-NtQueryDirectoryObject
-NtQueryEaFile
-NtQueryEvent
-NtQueryFullAttributesFile
-NtQueryInformationAtom
-NtQueryInformationFile
-NtQueryInformationJobObject
-NtQueryInformationPort
-NtQueryInformationProcess
-NtQueryInformationThread
-NtQueryInformationToken
-NtQueryInstallUILanguage
-NtQueryIntervalProfile
-NtQueryIoCompletion
-NtQueryKey
-NtQueryMultipleValueKey
-NtQueryMutant
-NtQueryObject
-NtQueryOpenSubKeys
-NtQueryPerformanceCounter
-NtQueryPortInformationProcess
-NtQueryQuotaInformationFile
-NtQuerySection
-NtQuerySecurityObject
-NtQuerySemaphore
-NtQuerySymbolicLinkObject
-NtQuerySystemEnvironmentValue
-NtQuerySystemEnvironmentValueEx
-NtQuerySystemInformation
-NtQuerySystemTime
-NtQueryTimer
-NtQueryTimerResolution
-NtQueryValueKey
-NtQueryVirtualMemory
-NtQueryVolumeInformationFile
-NtQueueApcThread
-NtRaiseException
-NtRaiseHardError
-NtReadFile
-NtReadFileScatter
-NtReadRequestData
-NtReadVirtualMemory
-NtRegisterThreadTerminatePort
-NtReleaseKeyedEvent
-NtReleaseMutant
-NtReleaseSemaphore
-NtRemoveIoCompletion
-NtRemoveProcessDebug
-NtRenameKey
-NtReplaceKey
-NtReplyPort
-NtReplyWaitReceivePort
-NtReplyWaitReceivePortEx
-NtReplyWaitReplyPort
-NtRequestPort
-NtRequestWaitReplyPort
-NtResetEvent
-NtResetWriteWatch
-NtRestoreKey
-NtResumeProcess
-NtResumeThread
-NtSaveKey
-NtSaveKeyEx
-NtSaveMergedKeys
-NtSecureConnectPort
-NtSetBootEntryOrder
-NtSetBootOptions
-NtSetContextThread
-NtSetDebugFilterState
-NtSetDefaultHardErrorPort
-NtSetDefaultLocale
-NtSetDefaultUILanguage
-NtSetEaFile
-NtSetEvent
-NtSetEventBoostPriority
-NtSetHighEventPair
-NtSetHighWaitLowEventPair
-NtSetInformationDebugObject
-NtSetInformationFile
-NtSetInformationJobObject
-NtSetInformationKey
-NtSetInformationObject
-NtSetInformationProcess
-NtSetInformationThread
-NtSetInformationToken
-NtSetIntervalProfile
-NtSetIoCompletion
-NtSetLdtEntries
-NtSetLowEventPair
-NtSetLowWaitHighEventPair
-NtSetQuotaInformationFile
-NtSetSecurityObject
-NtSetSystemEnvironmentValue
-NtSetSystemEnvironmentValueEx
-NtSetSystemInformation
-NtSetSystemPowerState
-NtSetSystemTime
-NtSetThreadExecutionState
-NtSetTimer
-NtSetTimerResolution
-NtSetUuidSeed
-NtSetValueKey
-NtSetVolumeInformationFile
-NtShutdownSystem
-NtSignalAndWaitForSingleObject
-NtStartProfile
-NtStopProfile
-NtSuspendProcess
-NtSuspendThread
-NtSystemDebugControl
-NtTerminateJobObject
-NtTerminateProcess
-NtTerminateThread
-NtTestAlert
-NtTraceEvent
-NtTranslateFilePath
-NtUnloadDriver
-NtUnloadKey
-NtUnloadKeyEx
-NtUnlockFile
-NtUnlockVirtualMemory
-NtUnmapViewOfSection
-NtVdmControl
-NtWaitForDebugEvent
-NtWaitForKeyedEvent
-NtWaitForMultipleObjects
-NtWaitForSingleObject
-NtWaitHighEventPair
-NtWaitLowEventPair
-NtWriteFile
-NtWriteFileGather
-NtWriteRequestData
-NtWriteVirtualMemory
-NtYieldExecution
-PfxFindPrefix
-PfxInitialize
-PfxInsertPrefix
-PfxRemovePrefix
-RtlAbortRXact
-RtlAbsoluteToSelfRelativeSD
-RtlAcquirePebLock
-RtlAcquireResourceExclusive
-RtlAcquireResourceShared
-RtlActivateActivationContext
-RtlActivateActivationContextEx
-RtlActivateActivationContextUnsafeFast
-RtlAddAccessAllowedAce
-RtlAddAccessAllowedAceEx
-RtlAddAccessAllowedObjectAce
-RtlAddAccessDeniedAce
-RtlAddAccessDeniedAceEx
-RtlAddAccessDeniedObjectAce
-RtlAddAce
-RtlAddActionToRXact
-RtlAddAtomToAtomTable
-RtlAddAttributeActionToRXact
-RtlAddAuditAccessAce
-RtlAddAuditAccessAceEx
-RtlAddAuditAccessObjectAce
-RtlAddCompoundAce
-RtlAddRefActivationContext
-RtlAddRefMemoryStream
-RtlAddVectoredExceptionHandler
-RtlAddressInSectionTable
-RtlAdjustPrivilege
-RtlAllocateAndInitializeSid
-RtlAllocateHandle
-RtlAllocateHeap
-RtlAnsiCharToUnicodeChar
-RtlAnsiStringToUnicodeSize
-RtlAnsiStringToUnicodeString
-RtlAppendAsciizToString
-RtlAppendPathElement
-RtlAppendStringToString
-RtlAppendUnicodeStringToString
-RtlAppendUnicodeToString
-RtlApplicationVerifierStop
-RtlApplyRXact
-RtlApplyRXactNoFlush
-RtlAreAllAccessesGranted
-RtlAreAnyAccessesGranted
-RtlAreBitsClear
-RtlAreBitsSet
-RtlAssert
-RtlCancelTimer
-RtlCaptureContext
-RtlCaptureStackBackTrace
-RtlCharToInteger
-RtlCheckForOrphanedCriticalSections
-RtlCheckRegistryKey
-RtlClearAllBits
-RtlClearBits
-RtlCloneMemoryStream
-RtlCommitMemoryStream
-RtlCompactHeap
-RtlCompareMemory
-RtlCompareMemoryUlong
-RtlCompareString
-RtlCompareUnicodeString
-RtlCompressBuffer
-RtlComputeCrc32
-RtlComputeImportTableHash
-RtlComputePrivatizedDllName_U
-RtlConsoleMultiByteToUnicodeN
-RtlConvertExclusiveToShared
-RtlConvertSharedToExclusive
-RtlConvertSidToUnicodeString
-RtlConvertToAutoInheritSecurityObject
-RtlCopyLuid
-RtlCopyLuidAndAttributesArray
-RtlCopyMemoryStreamTo
-RtlCopyOutOfProcessMemoryStreamTo
-RtlCopySecurityDescriptor
-RtlCopySid
-RtlCopySidAndAttributesArray
-RtlCopyString
-RtlCopyUnicodeString
-RtlCreateAcl
-RtlCreateActivationContext
-RtlCreateAndSetSD
-RtlCreateAtomTable
-RtlCreateBootStatusDataFile
-RtlCreateEnvironment
-RtlCreateHeap
-RtlCreateProcessParameters
-RtlCreateQueryDebugBuffer
-RtlCreateRegistryKey
-RtlCreateSecurityDescriptor
-RtlCreateServiceSid
-RtlCreateSystemVolumeInformationFolder
-RtlCreateTagHeap
-RtlCreateTimer
-RtlCreateTimerQueue
-RtlCreateUnicodeString
-RtlCreateUnicodeStringFromAsciiz
-RtlCreateUserProcess
-RtlCreateUserSecurityObject
-RtlCreateUserThread
-RtlCustomCPToUnicodeN
-RtlCutoverTimeToSystemTime
-RtlDeNormalizeProcessParams
-RtlDeactivateActivationContext
-RtlDeactivateActivationContextUnsafeFast
-RtlDebugPrintTimes
-RtlDecodePointer
-RtlDecodeSystemPointer
-RtlDecompressBuffer
-RtlDecompressFragment
-RtlDefaultNpAcl
-RtlDelete
-RtlDeleteAce
-RtlDeleteAtomFromAtomTable
-RtlDeleteCriticalSection
-RtlDeleteElementGenericTable
-RtlDeleteElementGenericTableAvl
-RtlDeleteNoSplay
-RtlDeleteRegistryValue
-RtlDeleteResource
-RtlDeleteSecurityObject
-RtlDeleteTimer
-RtlDeleteTimerQueue
-RtlDeleteTimerQueueEx
-RtlDeregisterWait
-RtlDeregisterWaitEx
-RtlDestroyAtomTable
-RtlDestroyEnvironment
-RtlDestroyHandleTable
-RtlDestroyHeap
-RtlDestroyProcessParameters
-RtlDestroyQueryDebugBuffer
-RtlDetermineDosPathNameType_U
-RtlDllShutdownInProgress
-RtlDnsHostNameToComputerName
-RtlDoesFileExists_U
-RtlDosApplyFileIsolationRedirection_Ustr
-RtlDosPathNameToNtPathName_U
-RtlDosSearchPath_U
-RtlDosSearchPath_Ustr
-RtlDowncaseUnicodeChar
-RtlDowncaseUnicodeString
-RtlDumpResource
-RtlDuplicateUnicodeString
-RtlEmptyAtomTable
-RtlEnableEarlyCriticalSectionEventCreation
-RtlEncodePointer
-RtlEncodeSystemPointer
-RtlEnterCriticalSection
-RtlEnumProcessHeaps
-RtlEnumerateGenericTable
-RtlEnumerateGenericTableAvl
-RtlEnumerateGenericTableLikeADirectory
-RtlEnumerateGenericTableWithoutSplaying
-RtlEnumerateGenericTableWithoutSplayingAvl
-RtlEqualComputerName
-RtlEqualDomainName
-RtlEqualLuid
-RtlEqualPrefixSid
-RtlEqualSid
-RtlEqualString
-RtlEqualUnicodeString
-RtlEraseUnicodeString
-RtlExitUserThread
-RtlExpandEnvironmentStrings_U
-RtlFillMemory
-RtlFinalReleaseOutOfProcessMemoryStream
-RtlFindActivationContextSectionGuid
-RtlFindActivationContextSectionString
-RtlFindCharInUnicodeString
-RtlFindClearBits
-RtlFindClearBitsAndSet
-RtlFindClearRuns
-RtlFindLastBackwardRunClear
-RtlFindLeastSignificantBit
-RtlFindLongestRunClear
-RtlFindMessage
-RtlFindMostSignificantBit
-RtlFindNextForwardRunClear
-RtlFindSetBits
-RtlFindSetBitsAndClear
-RtlFirstEntrySList
-RtlFirstFreeAce
-RtlFlushSecureMemoryCache
-RtlFormatCurrentUserKeyPath
-RtlFormatMessage
-RtlFreeAnsiString
-RtlFreeHandle
-RtlFreeHeap
-RtlFreeOemString
-RtlFreeSid
-RtlFreeThreadActivationContextStack
-RtlFreeUnicodeString
-RtlGUIDFromString
-RtlGenerate8dot3Name
-RtlGetAce
-RtlGetActiveActivationContext
-RtlGetCallersAddress
-RtlGetCompressionWorkSpaceSize
-RtlGetControlSecurityDescriptor
-RtlGetCurrentDirectory_U
-RtlGetCurrentPeb
-RtlGetDaclSecurityDescriptor
-RtlGetElementGenericTable
-RtlGetElementGenericTableAvl
-RtlGetFrame
-RtlGetFullPathName_U
-RtlGetGroupSecurityDescriptor
-RtlGetLastNtStatus
-RtlGetLastWin32Error
-RtlGetLengthWithoutLastFullDosOrNtPathElement
-RtlGetLengthWithoutTrailingPathSeperators
-RtlGetLongestNtPathLength
-RtlGetNativeSystemInformation
-RtlGetNtGlobalFlags
-RtlGetNtProductType
-RtlGetNtVersionNumbers
-RtlGetOwnerSecurityDescriptor
-RtlGetProcessHeaps
-RtlGetSaclSecurityDescriptor
-RtlGetSecurityDescriptorRMControl
-RtlGetSetBootStatusData
-RtlGetUnloadEventTrace
-RtlGetUserInfoHeap
-RtlGetVersion
-RtlHashUnicodeString
-RtlIdentifierAuthoritySid
-RtlImageDirectoryEntryToData
-RtlImageNtHeader
-RtlImageRvaToSection
-RtlImageRvaToVa
-RtlImpersonateSelf
-RtlInitAnsiString
-RtlInitCodePageTable
-RtlInitMemoryStream
-RtlInitNlsTables
-RtlInitOutOfProcessMemoryStream
-RtlInitString
-RtlInitUnicodeString
-RtlInitUnicodeStringEx
-RtlInitializeAtomPackage
-RtlInitializeBitMap
-RtlInitializeContext
-RtlInitializeCriticalSection
-RtlInitializeCriticalSectionAndSpinCount
-RtlInitializeGenericTable
-RtlInitializeGenericTableAvl
-RtlInitializeHandleTable
-RtlInitializeRXact
-RtlInitializeResource
-RtlInitializeSListHead
-RtlInitializeSid
-RtlInsertElementGenericTable
-RtlInsertElementGenericTableAvl
-RtlInt64ToUnicodeString
-RtlIntegerToChar
-RtlIntegerToUnicodeString
-RtlInterlockedFlushSList
-RtlInterlockedPopEntrySList
-RtlInterlockedPushEntrySList
-RtlInterlockedPushListSList
-RtlIpv4AddressToStringA
-RtlIpv4AddressToStringExA
-RtlIpv4AddressToStringExW
-RtlIpv4AddressToStringW
-RtlIpv4StringToAddressA
-RtlIpv4StringToAddressExA
-RtlIpv4StringToAddressExW
-RtlIpv4StringToAddressW
-RtlIpv6AddressToStringA
-RtlIpv6AddressToStringExA
-RtlIpv6AddressToStringExW
-RtlIpv6AddressToStringW
-RtlIpv6StringToAddressA
-RtlIpv6StringToAddressExA
-RtlIpv6StringToAddressExW
-RtlIpv6StringToAddressW
-RtlIsActivationContextActive
-RtlIsDosDeviceName_U
-RtlIsGenericTableEmpty
-RtlIsGenericTableEmptyAvl
-RtlIsNameLegalDOS8Dot3
-RtlIsTextUnicode
-RtlIsThreadWithinLoaderCallout
-RtlIsValidHandle
-RtlIsValidIndexHandle
-RtlLargeIntegerToChar
-RtlLeaveCriticalSection
-RtlLengthRequiredSid
-RtlLengthSecurityDescriptor
-RtlLengthSid
-RtlLocalTimeToSystemTime
-RtlLockBootStatusData
-RtlLockHeap
-RtlLockMemoryStreamRegion
-RtlLogStackBackTrace
-RtlLookupAtomInAtomTable
-RtlLookupElementGenericTable
-RtlLookupElementGenericTableAvl
-RtlMakeSelfRelativeSD
-RtlMapGenericMask
-RtlMapSecurityErrorToNtStatus
-RtlMoveMemory
-RtlMultiAppendUnicodeStringBuffer
-RtlMultiByteToUnicodeN
-RtlMultiByteToUnicodeSize
-RtlNewInstanceSecurityObject
-RtlNewSecurityGrantedAccess
-RtlNewSecurityObject
-RtlNewSecurityObjectEx
-RtlNewSecurityObjectWithMultipleInheritance
-RtlNormalizeProcessParams
-RtlNtPathNameToDosPathName
-RtlNtStatusToDosError
-RtlNtStatusToDosErrorNoTeb
-RtlNumberGenericTableElements
-RtlNumberGenericTableElementsAvl
-RtlNumberOfClearBits
-RtlNumberOfSetBits
-RtlOemStringToUnicodeSize
-RtlOemStringToUnicodeString
-RtlOemToUnicodeN
-RtlOpenCurrentUser
-RtlPcToFileHeader
-RtlPinAtomInAtomTable
-RtlPopFrame
-RtlPrefixString
-RtlPrefixUnicodeString
-RtlProtectHeap
-RtlPushFrame
-RtlQueryAtomInAtomTable
-RtlQueryDepthSList
-RtlQueryEnvironmentVariable_U
-RtlQueryHeapInformation
-RtlQueryInformationAcl
-RtlQueryInformationActivationContext
-RtlQueryInformationActiveActivationContext
-RtlQueryInterfaceMemoryStream
-RtlQueryProcessBackTraceInformation
-RtlQueryProcessDebugInformation
-RtlQueryProcessHeapInformation
-RtlQueryProcessLockInformation
-RtlQueryRegistryValues
-RtlQuerySecurityObject
-RtlQueryTagHeap
-RtlQueryTimeZoneInformation
-RtlQueueApcWow64Thread
-RtlQueueWorkItem
-RtlRaiseException
-RtlRaiseStatus
-RtlRandom
-RtlRandomEx
-RtlReAllocateHeap
-RtlReadMemoryStream
-RtlReadOutOfProcessMemoryStream
-RtlRealPredecessor
-RtlRealSuccessor
-RtlRegisterSecureMemoryCacheCallback
-RtlRegisterWait
-RtlReleaseActivationContext
-RtlReleaseMemoryStream
-RtlReleasePebLock
-RtlReleaseResource
-RtlRemoteCall
-RtlRemoveVectoredExceptionHandler
-RtlResetRtlTranslations
-RtlRestoreLastWin32Error
-RtlRevertMemoryStream
-RtlRunDecodeUnicodeString
-RtlRunEncodeUnicodeString
-RtlSecondsSince1970ToTime
-RtlSecondsSince1980ToTime
-RtlSeekMemoryStream
-RtlSelfRelativeToAbsoluteSD
-RtlSelfRelativeToAbsoluteSD2
-RtlSetAllBits
-RtlSetAttributesSecurityDescriptor
-RtlSetBits
-RtlSetControlSecurityDescriptor
-RtlSetCriticalSectionSpinCount
-RtlSetCurrentDirectory_U
-RtlSetCurrentEnvironment
-RtlSetDaclSecurityDescriptor
-RtlSetEnvironmentVariable
-RtlSetGroupSecurityDescriptor
-RtlSetHeapInformation
-RtlSetInformationAcl
-RtlSetIoCompletionCallback
-RtlSetLastWin32Error
-RtlSetLastWin32ErrorAndNtStatusFromNtStatus
-RtlSetMemoryStreamSize
-RtlSetOwnerSecurityDescriptor
-RtlSetProcessIsCritical
-RtlSetSaclSecurityDescriptor
-RtlSetSecurityDescriptorRMControl
-RtlSetSecurityObject
-RtlSetSecurityObjectEx
-RtlSetThreadIsCritical
-RtlSetThreadPoolStartFunc
-RtlSetTimeZoneInformation
-RtlSetTimer
-RtlSetUserFlagsHeap
-RtlSetUserValueHeap
-RtlSizeHeap
-RtlSplay
-RtlStartRXact
-RtlStatMemoryStream
-RtlStringFromGUID
-RtlSubAuthorityCountSid
-RtlSubAuthoritySid
-RtlSubtreePredecessor
-RtlSubtreeSuccessor
-RtlSystemTimeToLocalTime
-RtlTimeFieldsToTime
-RtlTimeToElapsedTimeFields
-RtlTimeToSecondsSince1970
-RtlTimeToSecondsSince1980
-RtlTimeToTimeFields
-RtlTraceDatabaseAdd
-RtlTraceDatabaseCreate
-RtlTraceDatabaseDestroy
-RtlTraceDatabaseEnumerate
-RtlTraceDatabaseFind
-RtlTraceDatabaseLock
-RtlTraceDatabaseUnlock
-RtlTraceDatabaseValidate
-RtlTryEnterCriticalSection
-RtlUnhandledExceptionFilter
-RtlUnhandledExceptionFilter2
-RtlUnicodeStringToAnsiSize
-RtlUnicodeStringToAnsiString
-RtlUnicodeStringToCountedOemString
-RtlUnicodeStringToInteger
-RtlUnicodeStringToOemSize
-RtlUnicodeStringToOemString
-RtlUnicodeToCustomCPN
-RtlUnicodeToMultiByteN
-RtlUnicodeToMultiByteSize
-RtlUnicodeToOemN
-RtlUniform
-RtlUnlockBootStatusData
-RtlUnlockHeap
-RtlUnlockMemoryStreamRegion
-RtlUnwind
-RtlUpcaseUnicodeChar
-RtlUpcaseUnicodeString
-RtlUpcaseUnicodeStringToAnsiString
-RtlUpcaseUnicodeStringToCountedOemString
-RtlUpcaseUnicodeStringToOemString
-RtlUpcaseUnicodeToCustomCPN
-RtlUpcaseUnicodeToMultiByteN
-RtlUpcaseUnicodeToOemN
-RtlUpdateTimer
-RtlUpperChar
-RtlUpperString
-RtlValidAcl
-RtlValidRelativeSecurityDescriptor
-RtlValidSecurityDescriptor
-RtlValidSid
-RtlValidateHeap
-RtlValidateProcessHeaps
-RtlValidateUnicodeString
-RtlVerifyVersionInfo
-RtlWalkFrameChain
-RtlWalkHeap
-RtlWriteMemoryStream
-RtlWriteRegistryValue
-RtlZeroHeap
-RtlZeroMemory
-RtlZombifyActivationContext
-RtlpApplyLengthFunction
-RtlpEnsureBufferSize
-RtlpNotOwnerCriticalSection
-RtlpNtCreateKey
-RtlpNtEnumerateSubKey
-RtlpNtMakeTemporaryKey
-RtlpNtOpenKey
-RtlpNtQueryValueKey
-RtlpNtSetValueKey
-RtlpUnWaitCriticalSection
-RtlpWaitForCriticalSection
-RtlxAnsiStringToUnicodeSize
-RtlxOemStringToUnicodeSize
-RtlxUnicodeStringToAnsiSize
-RtlxUnicodeStringToOemSize
-VerSetConditionMask
-ZwAcceptConnectPort
-ZwAccessCheck
-ZwAccessCheckAndAuditAlarm
-ZwAccessCheckByType
-ZwAccessCheckByTypeAndAuditAlarm
-ZwAccessCheckByTypeResultList
-ZwAccessCheckByTypeResultListAndAuditAlarm
-ZwAccessCheckByTypeResultListAndAuditAlarmByHandle
-ZwAddAtom
-ZwAddBootEntry
-ZwAdjustGroupsToken
-ZwAdjustPrivilegesToken
-ZwAlertResumeThread
-ZwAlertThread
-ZwAllocateLocallyUniqueId
-ZwAllocateUserPhysicalPages
-ZwAllocateUuids
-ZwAllocateVirtualMemory
-ZwAreMappedFilesTheSame
-ZwAssignProcessToJobObject
-ZwCallbackReturn
-ZwCancelIoFile
-ZwCancelTimer
-ZwClearEvent
-ZwClose
-ZwCloseObjectAuditAlarm
-ZwCompactKeys
-ZwCompareTokens
-ZwCompleteConnectPort
-ZwCompressKey
-ZwConnectPort
-ZwContinue
-ZwCreateDebugObject
-ZwCreateDirectoryObject
-ZwCreateEvent
-ZwCreateEventPair
-ZwCreateFile
-ZwCreateIoCompletion
-ZwCreateJobObject
-ZwCreateJobSet
-ZwCreateKey
-ZwCreateKeyedEvent
-ZwCreateMailslotFile
-ZwCreateMutant
-ZwCreateNamedPipeFile
-ZwCreatePagingFile
-ZwCreatePort
-ZwCreateProcess
-ZwCreateProcessEx
-ZwCreateProfile
-ZwCreateSection
-ZwCreateSemaphore
-ZwCreateSymbolicLinkObject
-ZwCreateThread
-ZwCreateTimer
-ZwCreateToken
-ZwCreateWaitablePort
-ZwDebugActiveProcess
-ZwDebugContinue
-ZwDelayExecution
-ZwDeleteAtom
-ZwDeleteBootEntry
-ZwDeleteFile
-ZwDeleteKey
-ZwDeleteObjectAuditAlarm
-ZwDeleteValueKey
-ZwDeviceIoControlFile
-ZwDisplayString
-ZwDuplicateObject
-ZwDuplicateToken
-ZwEnumerateBootEntries
-ZwEnumerateKey
-ZwEnumerateSystemEnvironmentValuesEx
-ZwEnumerateValueKey
-ZwExtendSection
-ZwFilterToken
-ZwFindAtom
-ZwFlushBuffersFile
-ZwFlushInstructionCache
-ZwFlushKey
-ZwFlushVirtualMemory
-ZwFlushWriteBuffer
-ZwFreeUserPhysicalPages
-ZwFreeVirtualMemory
-ZwFsControlFile
-ZwGetContextThread
-ZwGetDevicePowerState
-ZwGetWriteWatch
-ZwImpersonateAnonymousToken
-ZwImpersonateClientOfPort
-ZwImpersonateThread
-ZwInitializeRegistry
-ZwInitiatePowerAction
-ZwIsProcessInJob
-ZwIsSystemResumeAutomatic
-ZwListenPort
-ZwLoadDriver
-ZwLoadKey
-ZwLoadKey2
-ZwLockFile
-ZwLockProductActivationKeys
-ZwLockRegistryKey
-ZwLockVirtualMemory
-ZwMakePermanentObject
-ZwMakeTemporaryObject
-ZwMapUserPhysicalPages
-ZwMapUserPhysicalPagesScatter
-ZwMapViewOfSection
-ZwModifyBootEntry
-ZwNotifyChangeDirectoryFile
-ZwNotifyChangeKey
-ZwNotifyChangeMultipleKeys
-ZwOpenDirectoryObject
-ZwOpenEvent
-ZwOpenEventPair
-ZwOpenFile
-ZwOpenIoCompletion
-ZwOpenJobObject
-ZwOpenKey
-ZwOpenKeyedEvent
-ZwOpenMutant
-ZwOpenObjectAuditAlarm
-ZwOpenProcess
-ZwOpenProcessToken
-ZwOpenProcessTokenEx
-ZwOpenSection
-ZwOpenSemaphore
-ZwOpenSymbolicLinkObject
-ZwOpenThread
-ZwOpenThreadToken
-ZwOpenThreadTokenEx
-ZwOpenTimer
-ZwPlugPlayControl
-ZwPowerInformation
-ZwPrivilegeCheck
-ZwPrivilegeObjectAuditAlarm
-ZwPrivilegedServiceAuditAlarm
-ZwProtectVirtualMemory
-ZwPulseEvent
-ZwQueryAttributesFile
-ZwQueryBootEntryOrder
-ZwQueryBootOptions
-ZwQueryDebugFilterState
-ZwQueryDefaultLocale
-ZwQueryDefaultUILanguage
-ZwQueryDirectoryFile
-ZwQueryDirectoryObject
-ZwQueryEaFile
-ZwQueryEvent
-ZwQueryFullAttributesFile
-ZwQueryInformationAtom
-ZwQueryInformationFile
-ZwQueryInformationJobObject
-ZwQueryInformationPort
-ZwQueryInformationProcess
-ZwQueryInformationThread
-ZwQueryInformationToken
-ZwQueryInstallUILanguage
-ZwQueryIntervalProfile
-ZwQueryIoCompletion
-ZwQueryKey
-ZwQueryMultipleValueKey
-ZwQueryMutant
-ZwQueryObject
-ZwQueryOpenSubKeys
-ZwQueryPerformanceCounter
-ZwQueryPortInformationProcess
-ZwQueryQuotaInformationFile
-ZwQuerySection
-ZwQuerySecurityObject
-ZwQuerySemaphore
-ZwQuerySymbolicLinkObject
-ZwQuerySystemEnvironmentValue
-ZwQuerySystemEnvironmentValueEx
-ZwQuerySystemInformation
-ZwQuerySystemTime
-ZwQueryTimer
-ZwQueryTimerResolution
-ZwQueryValueKey
-ZwQueryVirtualMemory
-ZwQueryVolumeInformationFile
-ZwQueueApcThread
-ZwRaiseException
-ZwRaiseHardError
-ZwReadFile
-ZwReadFileScatter
-ZwReadRequestData
-ZwReadVirtualMemory
-ZwRegisterThreadTerminatePort
-ZwReleaseKeyedEvent
-ZwReleaseMutant
-ZwReleaseSemaphore
-ZwRemoveIoCompletion
-ZwRemoveProcessDebug
-ZwRenameKey
-ZwReplaceKey
-ZwReplyPort
-ZwReplyWaitReceivePort
-ZwReplyWaitReceivePortEx
-ZwReplyWaitReplyPort
-ZwRequestPort
-ZwRequestWaitReplyPort
-ZwResetEvent
-ZwResetWriteWatch
-ZwRestoreKey
-ZwResumeProcess
-ZwResumeThread
-ZwSaveKey
-ZwSaveKeyEx
-ZwSaveMergedKeys
-ZwSecureConnectPort
-ZwSetBootEntryOrder
-ZwSetBootOptions
-ZwSetContextThread
-ZwSetDebugFilterState
-ZwSetDefaultHardErrorPort
-ZwSetDefaultLocale
-ZwSetDefaultUILanguage
-ZwSetEaFile
-ZwSetEvent
-ZwSetEventBoostPriority
-ZwSetHighEventPair
-ZwSetHighWaitLowEventPair
-ZwSetInformationDebugObject
-ZwSetInformationFile
-ZwSetInformationJobObject
-ZwSetInformationKey
-ZwSetInformationObject
-ZwSetInformationProcess
-ZwSetInformationThread
-ZwSetInformationToken
-ZwSetIntervalProfile
-ZwSetIoCompletion
-ZwSetLdtEntries
-ZwSetLowEventPair
-ZwSetLowWaitHighEventPair
-ZwSetQuotaInformationFile
-ZwSetSecurityObject
-ZwSetSystemEnvironmentValue
-ZwSetSystemEnvironmentValueEx
-ZwSetSystemInformation
-ZwSetSystemPowerState
-ZwSetSystemTime
-ZwSetThreadExecutionState
-ZwSetTimer
-ZwSetTimerResolution
-ZwSetUuidSeed
-ZwSetValueKey
-ZwSetVolumeInformationFile
-ZwShutdownSystem
-ZwSignalAndWaitForSingleObject
-ZwStartProfile
-ZwStopProfile
-ZwSuspendProcess
-ZwSuspendThread
-ZwSystemDebugControl
-ZwTerminateJobObject
-ZwTerminateProcess
-ZwTerminateThread
-ZwTestAlert
-ZwTraceEvent
-ZwTranslateFilePath
-ZwUnloadDriver
-ZwUnloadKey
-ZwUnloadKeyEx
-ZwUnlockFile
-ZwUnlockVirtualMemory
-ZwUnmapViewOfSection
-ZwVdmControl
-ZwWaitForDebugEvent
-ZwWaitForKeyedEvent
-ZwWaitForMultipleObjects
-ZwWaitForSingleObject
-ZwWaitHighEventPair
-ZwWaitLowEventPair
-ZwWriteFile
-ZwWriteFileGather
-ZwWriteRequestData
-ZwWriteVirtualMemory
-ZwYieldExecution
-__isascii
-__iscsym
-__iscsymf
-__toascii
-_alldiv
-_alldvrm
-_allmul
-_allrem
-_allshl
-_allshr
-_atoi64
-_aulldiv
-_aulldvrm
-_aullrem
-_aullshr
-_fltused
-_i64toa
-_i64tow
-_itoa
-_itow
-_lfind
-_ltoa
-_ltow
-_memccpy
-_memicmp
-_snprintf
-_snwprintf
-_splitpath
-_strcmpi
-_stricmp
-_strlwr
-_strnicmp
-_strupr
-_ui64toa
-_ui64tow
-_ultoa
-_ultow
-_vsnprintf
-_vsnwprintf
-_wcsicmp
-_wcslwr
-_wcsnicmp
-_wcsupr
-_wtoi
-_wtoi64
-_wtol
-abs
-atan
-atoi
-atol
-bsearch
-ceil
-cos
-fabs
-floor
-isalnum
-isalpha
-iscntrl
-isdigit
-isgraph
-islower
-isprint
-ispunct
-isspace
-isupper
-iswalpha
-iswctype
-iswdigit
-iswlower
-iswspace
-iswxdigit
-isxdigit
-labs
-log
-mbstowcs
-memchr
-memcmp
-memcpy
-memmove
-memset
-pow
-qsort
-sin
-sprintf
-sqrt
-sscanf
-strcat
-strchr
-strcmp
-strcpy
-strcspn
-strlen
-strncat
-strncmp
-strncpy
-strpbrk
-strrchr
-strspn
-strstr
-strtol
-strtoul
-swprintf
-tan
-tolower
-toupper
-towlower
-towupper
-vDbgPrintEx
-vDbgPrintExWithPrefix
-vsprintf
-wcscat
-wcschr
-wcscmp
-wcscpy
-wcscspn
-wcslen
-wcsncat
-wcsncmp
-wcsncpy
-wcspbrk
-wcsrchr
-wcsspn
-wcsstr
-wcstol
-wcstombs
-wcstoul
-__C_specific_handler
-_except_handler4_common
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 3d922c8554160c15fbfa5f3bce24a387badf9b8f..887d8d16536e6c00bf68e4690040cdb310ea2b56 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -66,12 +66,9 @@ if (CELLFRAME_MODULES MATCHES "test-framework")
set(CELLFRAME_LIBS ${CELLFRAME_LIBS} dap_test)
endif()
-
- add_subdirectory(3rdparty/crc32c_adler) # https://github.com/fonic/crc32c_adler
- include_directories(3rdparty/crc32c_adler)
- set(CELLFRAME_LIBS ${CELLFRAME_LIBS} crc32c_adler)
-
-
+add_subdirectory(3rdparty/crc32c_adler) # https://github.com/fonic/crc32c_adler
+include_directories(3rdparty/crc32c_adler)
+set(CELLFRAME_LIBS ${CELLFRAME_LIBS} crc32c_adler)
if(BUILD_TESTS)
include(cmake/OS_Detection.cmake)
@@ -106,7 +103,7 @@ endif()
# General chain libs
if (CELLFRAME_MODULES MATCHES "chains")
message("[+] Module 'chains'")
- set(CELLFRAME_LIBS ${CELLFRAME_LIBS} dap_chain_common dap_chain dap_chain_global_db dap_chain_wallet)
+ set(CELLFRAME_LIBS ${CELLFRAME_LIBS} dap_chain_common dap_chain dap_global_db dap_chain_wallet)
endif()
# Networking
@@ -237,12 +234,4 @@ endif()
set(CELLFRAME_LIBS ${CELLFRAME_LIBS} crc32c_adler)
-
-
-target_link_libraries(${PROJECT_NAME} ${CELLFRAME_LIBS} crc32c_adler)
-#if(BUILD_DAP_TESTS)
-# file(COPY ${CMAKE_CURRENT_SOURCE_DIR}/test/main_test.py
-# DESTINATION ${CMAKE_CURRENT_BINARY_DIR}/)
-# enable_testing()
- #add_subdirectory(test)
-#endif()
+target_link_libraries(${PROJECT_NAME} ${CELLFRAME_LIBS})
diff --git a/dap-sdk b/dap-sdk
index 70151d1c362d4e9669358991c5ac70ad24e33a9a..2e12edb4fff37fcadadd6414643e9fbeb3e8a574 160000
--- a/dap-sdk
+++ b/dap-sdk
@@ -1 +1 @@
-Subproject commit 70151d1c362d4e9669358991c5ac70ad24e33a9a
+Subproject commit 2e12edb4fff37fcadadd6414643e9fbeb3e8a574
diff --git a/modules/CMakeLists.txt b/modules/CMakeLists.txt
index 8c5b31bc5fd794240c41594d178e94329c764e43..5b2c9751855e2352bf26fe28f0ef0a2926c16a69 100644
--- a/modules/CMakeLists.txt
+++ b/modules/CMakeLists.txt
@@ -8,7 +8,6 @@ if (CELLFRAME_MODULES MATCHES "chains")
add_subdirectory(chain)
add_subdirectory(chain/btc_rpc)
add_subdirectory(wallet)
- add_subdirectory(global-db)
endif()
# Distributed Hash Tables (DHT)
diff --git a/modules/chain/CMakeLists.txt b/modules/chain/CMakeLists.txt
index e3fc6f7f2c7e18c8d543ade7fc944d5f07839067..3602ee72433be5b9898892d5c8f738eb61f44855 100644
--- a/modules/chain/CMakeLists.txt
+++ b/modules/chain/CMakeLists.txt
@@ -14,7 +14,7 @@ endif()
#find_package(PkgConfig REQUIRED)
#pkg_search_module(GLIB REQUIRED glib-2.0)
-target_link_libraries(${PROJECT_NAME} dap_chain_common dap_chain_global_db ${GLIB_LDFLAGS})
+target_link_libraries(${PROJECT_NAME} dap_chain_common dap_global_db ${GLIB_LDFLAGS})
target_include_directories(${PROJECT_NAME} INTERFACE . include/ ${GLIB_INCLUDE_DIRS})
target_include_directories(${PROJECT_NAME} PUBLIC include)
target_include_directories(${PROJECT_NAME} PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../3rdparty/uthash/src)
diff --git a/modules/chain/dap_chain_ledger.c b/modules/chain/dap_chain_ledger.c
index 06002250774057809a3703f22cd50a762d6ec3e1..c039306961f1d54958508ce9a2bfbd0da048543c 100644
--- a/modules/chain/dap_chain_ledger.c
+++ b/modules/chain/dap_chain_ledger.c
@@ -1539,10 +1539,9 @@ static void s_threshold_txs_proc( dap_ledger_t *a_ledger)
* @param a_arg
*/
static bool s_load_cache_gdb_loaded_balances_callback(dap_global_db_context_t * a_global_db_context,int a_rc,
- const char * a_group, const char * a_key,
- const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values,
- void * a_arg)
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_ledger_t * l_ledger = (dap_ledger_t*) a_arg;
dap_ledger_private_t * l_ledger_pvt = PVT(l_ledger);
@@ -1580,11 +1579,10 @@ static bool s_load_cache_gdb_loaded_balances_callback(dap_global_db_context_t *
* @param a_values
* @param a_arg
*/
-static bool s_load_cache_gdb_loaded_spent_txs_callback(dap_global_db_context_t * a_global_db_context,int a_rc,
- const char * a_group, const char * a_key,
- const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values,
- void * a_arg)
+static bool s_load_cache_gdb_loaded_spent_txs_callback(dap_global_db_context_t *a_global_db_context, int a_rc,
+ const char *a_group, const char *a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_ledger_t * l_ledger = (dap_ledger_t*) a_arg;
dap_ledger_private_t * l_ledger_pvt = PVT(l_ledger);
@@ -1614,11 +1612,10 @@ static bool s_load_cache_gdb_loaded_spent_txs_callback(dap_global_db_context_t *
* @param a_values
* @param a_arg
*/
-static bool s_load_cache_gdb_loaded_txs_callback(dap_global_db_context_t * a_global_db_context,int a_rc,
- const char * a_group, const char * a_key,
- const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values,
- void * a_arg)
+static bool s_load_cache_gdb_loaded_txs_callback(dap_global_db_context_t *a_global_db_context, int a_rc,
+ const char *a_group, const char *a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_ledger_t * l_ledger = (dap_ledger_t*) a_arg;
dap_ledger_private_t * l_ledger_pvt = PVT(l_ledger);
@@ -1638,10 +1635,9 @@ static bool s_load_cache_gdb_loaded_txs_callback(dap_global_db_context_t * a_glo
}
static bool s_load_cache_gdb_loaded_stake_lock_callback(dap_global_db_context_t * a_global_db_context,int a_rc,
- const char * a_group, const char * a_key,
- const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values,
- void * a_arg)
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_ledger_t *l_ledger = (dap_ledger_t *) a_arg;
dap_ledger_private_t *l_ledger_pvt = PVT(l_ledger);
@@ -1680,10 +1676,9 @@ static bool s_load_cache_gdb_loaded_stake_lock_callback(dap_global_db_context_t
* @return Always true thats means to clear up a_values
*/
static bool s_load_cache_gdb_loaded_emissions_callback(dap_global_db_context_t * a_global_db_context,int a_rc,
- const char * a_group, const char * a_key,
- const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values,
- void * a_arg)
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_ledger_t * l_ledger = (dap_ledger_t*) a_arg;
dap_ledger_private_t * l_ledger_pvt = PVT(l_ledger);
@@ -1710,7 +1705,7 @@ static bool s_load_cache_gdb_loaded_emissions_callback(dap_global_db_context_t *
}
char* l_gdb_group = dap_chain_ledger_get_gdb_group(l_ledger, DAP_CHAIN_LEDGER_STAKE_LOCK_STR);
- dap_global_db_get_all(l_gdb_group,0, s_load_cache_gdb_loaded_txs_callback, l_ledger);
+ dap_global_db_get_all(l_gdb_group,0, s_load_cache_gdb_loaded_stake_lock_callback, l_ledger);
DAP_DELETE(l_gdb_group);
return true;
@@ -1730,10 +1725,9 @@ static bool s_load_cache_gdb_loaded_emissions_callback(dap_global_db_context_t *
* @param a_arg
*/
static bool s_load_cache_gdb_loaded_tokens_callback(dap_global_db_context_t * a_global_db_context,int a_rc,
- const char * a_group, const char * a_key,
- const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values,
- void * a_arg)
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_ledger_t * l_ledger = (dap_ledger_t*) a_arg;
dap_ledger_private_t * l_ledger_pvt = PVT(l_ledger);
@@ -1776,9 +1770,7 @@ static bool s_load_cache_gdb_loaded_tokens_callback(dap_global_db_context_t * a_
void dap_chain_ledger_load_cache(dap_ledger_t *a_ledger)
{
dap_ledger_private_t *l_ledger_pvt = PVT(a_ledger);
-
char *l_gdb_group = dap_chain_ledger_get_gdb_group(a_ledger, DAP_CHAIN_LEDGER_TOKENS_STR);
- size_t l_objs_count = 0;
pthread_mutex_lock(& l_ledger_pvt->load_mutex);
dap_global_db_get_all(l_gdb_group,0,s_load_cache_gdb_loaded_tokens_callback, a_ledger);
@@ -2703,8 +2695,6 @@ int dap_chain_ledger_tx_cache_check(dap_ledger_t *a_ledger, dap_chain_datum_tx_t
if (a_list_tx_out)
*a_list_tx_out = l_list_tx_out;
- dap_list_t* l_list_tx_out_cond = NULL;
-
// sum of values in 'out' items from the previous transactions
dap_chain_ledger_tokenizer_t *l_values_from_prev_tx = NULL, *l_values_from_cur_tx = NULL,
*l_value_cur = NULL, *l_tmp = NULL, *l_res = NULL;
@@ -2941,8 +2931,9 @@ int dap_chain_ledger_tx_cache_check(dap_ledger_t *a_ledger, dap_chain_datum_tx_t
l_err_num = DAP_CHAIN_CS_VERIFY_CODE_TX_NO_PREVIOUS;
break;
}
- if (s_debug_more && !a_from_threshold)
- log_it(L_INFO,"Previous transaction was found for hash %s",l_tx_prev_hash_str);
+ bound_item->item_out = l_item_out;
+ l_token = l_item_out->cache_data.token_ticker;
+ debug_if(s_debug_more && !a_from_threshold, L_INFO, "Previous transaction was found for hash %s",l_tx_prev_hash_str);
bound_item->tx_prev = l_tx_prev;
// 2. Check if out in previous transaction has spent
diff --git a/modules/channel/chain-net/dap_stream_ch_chain_net.c b/modules/channel/chain-net/dap_stream_ch_chain_net.c
index 335eef1d45485778cafcb3b3296d4b5ab9ab4947..04a6991b8bbed305cd4c7e17c6c0bb24fb7d7266 100644
--- a/modules/channel/chain-net/dap_stream_ch_chain_net.c
+++ b/modules/channel/chain-net/dap_stream_ch_chain_net.c
@@ -47,7 +47,7 @@
#include "uthash.h"
#include "dap_http_client.h"
#include "dap_global_db.h"
-#include "dap_chain_global_db_remote.h"
+#include "dap_global_db_remote.h"
#include "dap_stream.h"
#include "dap_stream_ch_pkt.h"
#include "dap_stream_ch_proc.h"
diff --git a/modules/channel/chain-voting/include/dap_stream_ch_chain_voting.h b/modules/channel/chain-voting/include/dap_stream_ch_chain_voting.h
index acb4cd61391cfc997f85e819375bf0de95d8bbe5..30163c46ae8e9c0ee07fcc2e6ad90efab7018075 100644
--- a/modules/channel/chain-voting/include/dap_stream_ch_chain_voting.h
+++ b/modules/channel/chain-voting/include/dap_stream_ch_chain_voting.h
@@ -4,7 +4,7 @@
#include "dap_chain_common.h"
#include "dap_chain.h"
-#include "dap_chain_global_db_remote.h"
+#include "dap_global_db_remote.h"
#include "dap_chain_node_client.h"
#include "dap_list.h"
// #include "dap_stream_ch_chain_pkt.h"
diff --git a/modules/channel/chain/CMakeLists.txt b/modules/channel/chain/CMakeLists.txt
index fa08c680949d1096948850602083007537f2fa01..9114276651bb4885574331e01d35115a61737480 100644
--- a/modules/channel/chain/CMakeLists.txt
+++ b/modules/channel/chain/CMakeLists.txt
@@ -6,7 +6,7 @@ file(GLOB DAP_STREAM_CH_CHAIN_HDRS include/*.h)
add_library(${PROJECT_NAME} STATIC ${DAP_STREAM_CH_CHAIN_SRCS} ${DAP_STREAM_CH_CHAIN_HDRS})
-target_link_libraries(${PROJECT_NAME} dap_core dap_crypto dap_chain dap_chain_global_db dap_stream dap_stream_ch)
+target_link_libraries(${PROJECT_NAME} dap_core dap_crypto dap_chain dap_global_db dap_stream dap_stream_ch)
target_include_directories(${PROJECT_NAME} INTERFACE .)
target_include_directories(${PROJECT_NAME} PUBLIC include)
diff --git a/modules/channel/chain/dap_stream_ch_chain.c b/modules/channel/chain/dap_stream_ch_chain.c
index 5203deaab1d82a03e72d505176dd9363cfa84390..69e3d419cf68e05f3efd94ccf29c7849dcd26606 100644
--- a/modules/channel/chain/dap_stream_ch_chain.c
+++ b/modules/channel/chain/dap_stream_ch_chain.c
@@ -57,7 +57,7 @@
#include "dap_chain_cell.h"
#include "dap_global_db.h"
-#include "dap_chain_global_db_remote.h"
+#include "dap_global_db_remote.h"
#include "dap_stream.h"
#include "dap_stream_pkt.h"
@@ -118,9 +118,10 @@ static bool s_sync_out_gdb_proc_callback(dap_proc_thread_t *a_thread, void *a_ar
static bool s_sync_in_chains_callback(dap_proc_thread_t *a_thread, void *a_arg);
static bool s_gdb_in_pkt_proc_callback(dap_proc_thread_t *a_thread, void *a_arg);
-static bool s_gdb_in_pkt_proc_set_raw_callback(dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_current, const size_t a_values_shift,
- const size_t a_values_count, dap_store_obj_t * a_values, void * a_arg);
-
+static bool s_gdb_in_pkt_proc_set_raw_callback(dap_global_db_context_t *a_global_db_context, int a_rc,
+ const char *a_group, const char *a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_store_obj_t *a_values, void *a_arg);
static void s_gdb_in_pkt_error_worker_callback(dap_worker_t *a_thread, void *a_arg);
static void s_free_log_list_gdb ( dap_stream_ch_chain_t * a_ch_chain);
@@ -312,8 +313,8 @@ static bool s_sync_out_chains_proc_callback(dap_proc_thread_t *a_thread, void *a
//pthread_rwlock_rdlock(&l_chain->atoms_rwlock);
l_sync_request->chain.request_atom_iter = l_chain->callback_atom_iter_create(l_chain, l_sync_request->request_hdr.cell_id, 1);
size_t l_first_size = 0;
- dap_chain_atom_ptr_t l_iter = l_chain->callback_atom_iter_get_first(l_sync_request->chain.request_atom_iter, &l_first_size);
- if (l_iter && l_first_size) {
+ dap_chain_atom_ptr_t l_atom = l_chain->callback_atom_iter_get_first(l_sync_request->chain.request_atom_iter, &l_first_size);
+ if (l_atom && l_first_size) {
// first packet
dap_chain_hash_fast_t l_hash_from = l_sync_request->request.hash_from;
if (!dap_hash_fast_is_blank(&l_hash_from)) {
@@ -321,7 +322,6 @@ static bool s_sync_out_chains_proc_callback(dap_proc_thread_t *a_thread, void *a
&l_hash_from, &l_first_size);
}
//pthread_rwlock_unlock(&l_chain->atoms_rwlock);
- l_sync_request->chain.request_atom_iter = NULL;
dap_proc_thread_worker_exec_callback_inter(a_thread, l_sync_request->worker->id, s_sync_out_chains_first_worker_callback, l_sync_request );
} else {
//pthread_rwlock_unlock(&l_chain->atoms_rwlock);
@@ -824,8 +824,8 @@ static bool s_gdb_in_pkt_proc_callback(dap_proc_thread_t *a_thread, void *a_arg)
// Record is pinned or not
bool l_is_pinned_cur = false;
- if (dap_chain_global_db_driver_is(l_obj->group, l_obj->key)) {
- dap_store_obj_t *l_read_obj = dap_chain_global_db_driver_read(l_obj->group, l_obj->key, NULL);
+ if (dap_global_db_driver_is(l_obj->group, l_obj->key)) {
+ dap_store_obj_t *l_read_obj = dap_global_db_driver_read(l_obj->group, l_obj->key, NULL);
if (l_read_obj) {
l_sync_request->timestamp_cur = l_read_obj->timestamp;
l_is_pinned_cur = l_read_obj->flags & RECORD_PINNED;
@@ -874,8 +874,10 @@ static bool s_gdb_in_pkt_proc_callback(dap_proc_thread_t *a_thread, void *a_arg)
* @param a_values
* @param a_arg
*/
-static bool s_gdb_in_pkt_proc_set_raw_callback(dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_current, const size_t a_values_shift,
- const size_t a_values_count, dap_store_obj_t * a_values, void * a_arg)
+static bool s_gdb_in_pkt_proc_set_raw_callback(dap_global_db_context_t *a_global_db_context, int a_rc,
+ const char *a_group, const char *a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_store_obj_t *a_values, void *a_arg)
{
struct sync_request *l_sync_req = (struct sync_request*) a_arg;
diff --git a/modules/channel/chain/include/dap_stream_ch_chain.h b/modules/channel/chain/include/dap_stream_ch_chain.h
index 519381ef606cad56a891988c7c882b3a9eb73841..f6e681a2741e9ee0c7d573ce04bbfdcece058b08 100644
--- a/modules/channel/chain/include/dap_stream_ch_chain.h
+++ b/modules/channel/chain/include/dap_stream_ch_chain.h
@@ -32,7 +32,7 @@
#include "dap_list.h"
#include "dap_stream_ch_chain_pkt.h"
#include "uthash.h"
-#include "dap_chain_global_db_remote.h"
+#include "dap_global_db_remote.h"
typedef struct dap_stream_ch_chain dap_stream_ch_chain_t;
typedef void (*dap_stream_ch_chain_callback_packet_t)(dap_stream_ch_chain_t*, uint8_t a_pkt_type,
diff --git a/modules/common/dap_chain_datum_tx_items.c b/modules/common/dap_chain_datum_tx_items.c
index ddbd99d53b56f1879b8135d13dbf1a31877f38f5..f82d50fffaf4dea89671ffa7d7d0e8d1cf34f11e 100644
--- a/modules/common/dap_chain_datum_tx_items.c
+++ b/modules/common/dap_chain_datum_tx_items.c
@@ -451,7 +451,9 @@ uint8_t* dap_chain_datum_tx_item_get( dap_chain_datum_tx_t *a_tx, int *a_item_id
(a_type == TX_ITEM_TYPE_OUT_ALL && l_type == TX_ITEM_TYPE_OUT) ||
(a_type == TX_ITEM_TYPE_OUT_ALL && l_type == TX_ITEM_TYPE_OUT_OLD) ||
(a_type == TX_ITEM_TYPE_OUT_ALL && l_type == TX_ITEM_TYPE_OUT_COND) ||
- (a_type == TX_ITEM_TYPE_OUT_ALL && l_type == TX_ITEM_TYPE_OUT_EXT)) {
+ (a_type == TX_ITEM_TYPE_OUT_ALL && l_type == TX_ITEM_TYPE_OUT_EXT) ||
+ (a_type == TX_ITEM_TYPE_IN_ALL && l_type == TX_ITEM_TYPE_IN) ||
+ (a_type == TX_ITEM_TYPE_IN_ALL && l_type == TX_ITEM_TYPE_IN_COND)) {
if(a_item_idx)
*a_item_idx = l_item_idx;
if(a_item_out_size)
diff --git a/modules/consensus/block-ton/dap_chain_cs_block_ton.c b/modules/consensus/block-ton/dap_chain_cs_block_ton.c
index e49adde197cb06184f206caf4b47a6907838f9d6..ad9e989cc1d7414f71744b4bec7b432ebc914693 100644
--- a/modules/consensus/block-ton/dap_chain_cs_block_ton.c
+++ b/modules/consensus/block-ton/dap_chain_cs_block_ton.c
@@ -412,8 +412,10 @@ static bool s_session_send_votefor(s_session_send_votefor_data_t *a_data){
* @param a_values
* @param a_arg
*/
-static bool s_session_round_start_callback_load_session_store_coordinator_state_proc(dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg)
+static bool s_session_round_start_callback_load_session_store_coordinator_state_proc(dap_global_db_context_t * a_global_db_context,int a_rc,
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_chain_cs_block_ton_session_t * l_session = (dap_chain_cs_block_ton_session_t *) a_arg;
dap_list_t *l_list_candidate = NULL;
@@ -852,8 +854,10 @@ static void s_session_round_clear(dap_chain_cs_block_ton_session_t *a_session)
* @param a_values
* @param a_arg
*/
-static bool s_session_round_finish_callback_load_store(dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg)
+static bool s_session_round_finish_callback_load_store(dap_global_db_context_t * a_global_db_context,int a_rc,
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_chain_cs_block_ton_session_t *l_session = (dap_chain_cs_block_ton_session_t *)a_arg;
@@ -980,8 +984,10 @@ struct vote_for_load_store_args
* @param a_values
* @param a_arg
*/
-static bool s_session_packet_in_callback_vote_for_load_store (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg)
+static bool s_session_packet_in_callback_vote_for_load_store(dap_global_db_context_t * a_global_db_context,int a_rc,
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_chain_cs_block_ton_session_t * l_session = ((struct vote_for_load_store_args *) a_arg)->session;
dap_chain_hash_fast_t * l_candidate_hash = &((struct vote_for_load_store_args *) a_arg)->candidate_hash;
diff --git a/modules/consensus/dag-poa/dap_chain_cs_dag_poa.c b/modules/consensus/dag-poa/dap_chain_cs_dag_poa.c
index 986cd1ee38545ecac54fa71db3031c35cbae89fc..8d377c630b77066522c357b98b03dfed5bb48497 100644
--- a/modules/consensus/dag-poa/dap_chain_cs_dag_poa.c
+++ b/modules/consensus/dag-poa/dap_chain_cs_dag_poa.c
@@ -42,7 +42,7 @@
#include "dap_chain_node_cli.h"
#include "dap_chain_node_cli_cmd.h"
#include "dap_global_db.h"
-#include "dap_chain_global_db_driver.h"
+#include "dap_global_db_driver.h"
#include "dap_chain_cs.h"
#include "dap_chain_cs_dag.h"
#include "dap_chain_cs_dag_event.h"
@@ -393,8 +393,10 @@ static int s_callback_new(dap_chain_t * a_chain, dap_config_t * a_chain_cfg)
* @param a_values
* @param a_arg
*/
-static bool s_poa_round_check_callback_round_clean(dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg)
+static bool s_poa_round_check_callback_round_clean(dap_global_db_context_t * a_global_db_context,int a_rc,
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
dap_chain_cs_dag_t *l_dag = (dap_chain_cs_dag_t *)a_arg;
dap_chain_cs_dag_poa_t *l_poa = DAP_CHAIN_CS_DAG_POA(l_dag);
@@ -518,11 +520,10 @@ static dap_chain_cs_dag_event_round_item_t *s_round_event_choose_dup(dap_list_t
* @param a_is_pinned
* @param a_arg
*/
-static bool s_callback_round_event_to_chain_callback_get_round_item(dap_global_db_context_t *a_global_db_context, int a_rc,
- const char *a_group, const char *a_key,
- const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_store_obj_t *a_values,
- void *a_arg)
+static bool s_callback_round_event_to_chain_callback_get_round_item(dap_global_db_context_t * a_global_db_context,int a_rc,
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_store_obj_t *a_values, void *a_arg)
{
if (a_rc != DAP_GLOBAL_DB_RC_SUCCESS)
return true;
diff --git a/modules/consensus/none/CMakeLists.txt b/modules/consensus/none/CMakeLists.txt
index ddf73b014ea1143a5a1c663582a517c424137c7c..cd45b3e9cdb5dfb5e77b5ce5aaf5d2c93a5068db 100644
--- a/modules/consensus/none/CMakeLists.txt
+++ b/modules/consensus/none/CMakeLists.txt
@@ -6,6 +6,6 @@ file(GLOB DAP_CHAIN_CS_NO_CONSENSUS_HDR include/*.h)
add_library(${PROJECT_NAME} STATIC ${DAP_CHAIN_CS_NO_CONSENSUS_SRC} ${DAP_CHAIN_CS_NO_CONSENSUS_HDR})
-target_link_libraries(${PROJECT_NAME} dap_core dap_chain dap_chain_global_db dap_chain_mempool dap_chain_net)
+target_link_libraries(${PROJECT_NAME} dap_core dap_chain dap_global_db dap_chain_mempool dap_chain_net)
target_include_directories(${PROJECT_NAME} INTERFACE .)
target_include_directories(${PROJECT_NAME} PUBLIC include)
diff --git a/modules/consensus/none/dap_chain_cs_none.c b/modules/consensus/none/dap_chain_cs_none.c
index c7c5e58d1f3bcdf8ce91bd0928630de0186a4dcc..38025b81041a3d9be7d8b48e02651d7dbcf571b8 100644
--- a/modules/consensus/none/dap_chain_cs_none.c
+++ b/modules/consensus/none/dap_chain_cs_none.c
@@ -2,7 +2,7 @@
* Authors:
* Alexander Lysikov <alexander.lysikov@demlabs.net>
* DeM Labs Inc. https://demlabs.net
- * Kelvin Project https://github.com/kelvindap_chain_global_dbblockchain
+ * Kelvin Project https://github.com/kelvindap_global_dbblockchain
* Copyright (c) 2019
* All rights reserved.
@@ -36,7 +36,7 @@
#include "dap_chain_cell.h"
#include "dap_chain_ledger.h"
#include "dap_global_db.h"
-#include "dap_chain_global_db_driver.h"
+#include "dap_global_db_driver.h"
#include "dap_chain_cs.h"
#include "dap_chain_cs_none.h"
@@ -184,7 +184,7 @@ int dap_chain_gdb_new(dap_chain_t * a_chain, dap_config_t * a_chain_cfg)
}
// Add group prefix that will be tracking all changes
- dap_chain_global_db_add_sync_group(l_net->pub.name, "chain-gdb", s_history_callback_notify, l_gdb);
+ dap_global_db_add_sync_group(l_net->pub.name, "chain-gdb", s_history_callback_notify, l_gdb);
// load ledger
l_gdb_priv->is_load_mode = true;
@@ -281,8 +281,10 @@ const char* dap_chain_gdb_get_group(dap_chain_t * a_chain)
* @param a_values
* @param a_arg
*/
-static bool s_ledger_load_callback(dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg)
+static bool s_ledger_load_callback(dap_global_db_context_t * a_global_db_context,int a_rc,
+ const char * a_group, const char * a_key,
+ const size_t a_values_total, const size_t a_values_count,
+ dap_global_db_obj_t *a_values, void *a_arg)
{
assert(a_arg);
dap_chain_t * l_chain = (dap_chain_t *) a_arg;
diff --git a/modules/global-db/CMakeLists.txt b/modules/global-db/CMakeLists.txt
deleted file mode 100644
index 43fc3fb4c4dc4af5f829a92e8f85feb9e5994d44..0000000000000000000000000000000000000000
--- a/modules/global-db/CMakeLists.txt
+++ /dev/null
@@ -1,51 +0,0 @@
-cmake_minimum_required(VERSION 3.1)
-project (dap_chain_global_db C)
-
-file(GLOB DAP_CHAIN_GLOBAL_DB_SRC
- dap_global_db.c
- dap_global_db_sync.c
- dap_chain_global_db_driver_cdb.c
- dap_chain_global_db_driver_sqlite.c
- dap_chain_global_db_remote.c
- dap_chain_global_db_driver.c
- dap_chain_global_db_driver_pgsql.c
-)
-file(GLOB DAP_CHAIN_GLOBAL_DB_HDR
- include/dap_global_db.h
- include/dap_global_db_sync.h
- include/dap_global_db.h
- include/dap_chain_global_db_remote.h
- include/dap_chain_global_db_driver.h
- include/dap_chain_global_db_driver_pgsql.h
- include/dap_chain_global_db_driver_cdb.h
- include/dap_chain_global_db_driver_sqlite.h
-)
-
-add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/../../3rdparty/cuttdb ${CMAKE_CURRENT_BINARY_DIR}/../../3rdparty/cuttdb)
-if (BUILD_WITH_GDB_DRIVER_MDBX)
-add_subdirectory(${CMAKE_CURRENT_SOURCE_DIR}/../../3rdparty/libmdbx ${CMAKE_CURRENT_BINARY_DIR}/../../3rdparty/libmdbx EXCLUDE_FROM_ALL)
-endif()
-
-set(DAP_CHAIN_GLOBAL_DB_LIBS dap_core dap_io dap_crypto dap_chain dap_chain_net sqlite3 dap_cuttdb json-c)
-
-if(BUILD_WITH_GDB_DRIVER_MDBX)
- file(GLOB DAP_CHAIN_GLOBAL_DB_SRC ${DAP_CHAIN_GLOBAL_DB_SRC} dap_chain_global_db_driver_mdbx.c)
- file(GLOB DAP_CHAIN_GLOBAL_DB_HRR ${DAP_CHAIN_GLOBAL_DB_HDR} include/dap_chain_global_db_driver_mdbx.h)
-
- set(DAP_CHAIN_GLOBAL_DB_LIBS ${DAP_CHAIN_GLOBAL_DB_LIBS} mdbx-static)
- add_definitions ("-DDAP_CHAIN_GDB_ENGINE_MDBX")
-endif()
-
-if(BUILD_WITH_GDB_DRIVER_PGSQL)
- set(DAP_CHAIN_GLOBAL_DB_LIBS ${DAP_CHAIN_GLOBAL_DB_LIBS} pq)
- add_definitions ("-DDAP_CHAIN_GDB_ENGINE_PGSQL")
-endif()
-
-add_library(${PROJECT_NAME} STATIC ${DAP_CHAIN_GLOBAL_DB_SRC} ${DAP_CHAIN_GLOBAL_DB_HDR})
-
-target_link_libraries(${PROJECT_NAME} ${DAP_CHAIN_GLOBAL_DB_LIBS})
-
-target_include_directories(${PROJECT_NAME} INTERFACE .)
-target_include_directories(${PROJECT_NAME} PUBLIC include)
-target_include_directories(${PROJECT_NAME} PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../3rdparty/uthash/src)
-
diff --git a/modules/global-db/dap_chain_global_db_driver.c b/modules/global-db/dap_chain_global_db_driver.c
deleted file mode 100644
index cbfb270450ec65dcf327577d85e59de956a2440f..0000000000000000000000000000000000000000
--- a/modules/global-db/dap_chain_global_db_driver.c
+++ /dev/null
@@ -1,493 +0,0 @@
-/*
- * Authors:
- * Alexander Lysikov <alexander.lysikov@demlabs.net>
- * DeM Labs Inc. https://demlabs.net
- * Kelvin Project https://github.com/kelvinblockchain
- * Copyright (c) 2019
- * All rights reserved.
-
- This file is part of DAP (Deus Applications Prototypes) the open source project
-
- DAP (Deus Applicaions Prototypes) is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP based project. If not, see <http://www.gnu.org/licenses/>.
- *
- * MODIFICATION HISTORY:
- *
- * 24-FEB-2022 RRL Added Async I/O functionality for DB request processing
- *
- * 15-MAR-2022 RRL Some cosmetic changes to reduce a diagnostic output.
- */
-
-#include <stddef.h>
-#include <errno.h>
-#include <stdint.h>
-#include <string.h>
-#include <pthread.h>
-#include <unistd.h>
-
-#include "dap_worker.h"
-#include "dap_file_utils.h"
-#include "dap_strfuncs.h"
-#include "dap_hash.h"
-#include "dap_proc_queue.h"
-#include "dap_events.h"
-#include "dap_list.h"
-#include "dap_common.h"
-#include "dap_global_db.h"
-
-#include "dap_chain_global_db_driver_sqlite.h"
-#include "dap_chain_global_db_driver_cdb.h"
-
-#ifdef DAP_CHAIN_GDB_ENGINE_MDBX
-#include "dap_chain_global_db_driver_mdbx.h"
-#endif
-
-#ifdef DAP_CHAIN_GDB_ENGINE_PGSQL
-#include "dap_chain_global_db_driver_pgsql.h"
-#endif
-
-#include "dap_chain_global_db_driver.h"
-
-#define LOG_TAG "db_driver"
-
-// A selected database driver.
-static char s_used_driver [32]; /* Name of the driver */
-
-
-static dap_db_driver_callbacks_t s_drv_callback; /* A set of interface routines for the selected
- DB Driver at startup time */
-
-int s_db_drvmode_async = 0; /* Set a kind of processing requests to DB:
- <> 0 - Async mode should be used */
-
-static pthread_mutex_t s_db_reqs_list_lock = PTHREAD_MUTEX_INITIALIZER; /* Lock to coordinate access to the <s_db_reqs_queue> */
-static dap_slist_t s_db_reqs_list = {0}; /* A queue of request to DB - maintained in */
-
-
-/**
- * @brief Initializes a database driver.
- * @note You should Call this function before using the driver.
- * @param driver_name a string determining a type of database driver:
- * "Ñdb", "sqlite" ("sqlite3") or "pgsql"
- * @param a_filename_db a path to a database file
- * @return Returns 0, if successful; otherwise <0.
- */
-int dap_db_driver_init(const char *a_driver_name, const char *a_filename_db, int a_mode_async)
-{
-int l_ret = -1;
-
- if (s_used_driver[0] )
- dap_db_driver_deinit();
-
- s_db_drvmode_async = a_mode_async;
-
- // Fill callbacks with zeros
- memset(&s_drv_callback, 0, sizeof(dap_db_driver_callbacks_t));
-
- if ( s_db_drvmode_async ) /* Set a kind of processing requests to DB: <> 0 - Async mode should be used */
- {
- s_db_reqs_list.head = s_db_reqs_list.tail = NULL;
- s_db_reqs_list.nr = 0;
- }
-
- // Setup driver name
- strncpy( s_used_driver, a_driver_name, sizeof(s_used_driver) - 1);
-
- dap_mkdir_with_parents(a_filename_db);
-
- // Compose path
- char l_db_path_ext[strlen(a_driver_name) + strlen(a_filename_db) + 6];
- dap_snprintf(l_db_path_ext, sizeof(l_db_path_ext), "%s/gdb-%s", a_filename_db, a_driver_name);
-
- // Check for engine
- if(!dap_strcmp(s_used_driver, "ldb"))
- l_ret = -1;
- else if(!dap_strcmp(s_used_driver, "sqlite") || !dap_strcmp(s_used_driver, "sqlite3") )
- l_ret = dap_db_driver_sqlite_init(l_db_path_ext, &s_drv_callback);
- else if(!dap_strcmp(s_used_driver, "cdb"))
- l_ret = dap_db_driver_cdb_init(l_db_path_ext, &s_drv_callback);
-#ifdef DAP_CHAIN_GDB_ENGINE_MDBX
- else if(!dap_strcmp(s_used_driver, "mdbx"))
- l_ret = dap_db_driver_mdbx_init(l_db_path_ext, &s_drv_callback);
-#endif
-
-#ifdef DAP_CHAIN_GDB_ENGINE_PGSQL
- else if(!dap_strcmp(s_used_driver, "pgsql"))
- l_ret = dap_db_driver_pgsql_init(l_db_path_ext, &s_drv_callback);
-#endif
- else
- log_it(L_ERROR, "Unknown global_db driver \"%s\"", a_driver_name);
-
- return l_ret;
-}
-
-/**
- * @brief Deinitializes a database driver.
- * @note You should call this function after using the driver.
- * @return (none)
- */
-void dap_db_driver_deinit(void)
-{
- log_it(L_NOTICE, "DeInit for %s ...", s_used_driver);
-
- if ( s_db_drvmode_async ) /* Let's finishing outstanding DB request ... */
- {
- for ( int i = 7; i-- && s_db_reqs_list.nr; )
- {
- log_it(L_WARNING, "Let's finished outstanding DB requests (%d) ... ", s_db_reqs_list.nr);
- for ( int j = 3; (j = sleep(j)); ); /* Hibernate for 3 seconds ... */
- }
-
- log_it(L_INFO, "Number of outstanding DB requests: %d", s_db_reqs_list.nr);
- }
-
- // deinit driver
- if(s_drv_callback.deinit)
- s_drv_callback.deinit();
-
- s_used_driver [ 0 ] = '\0';
-}
-
-/**
- * @brief Flushes a database cahce to disk.
- * @return Returns 0, if successful; otherwise <0.
- */
-int dap_db_driver_flush(void)
-{
- return s_db_drvmode_async ? 0 : s_drv_callback.flush();
-}
-
-/**
- * @brief Copies objects from a_store_obj.
- * @param a_store_obj a pointer to the source objects
- * @param a_store_count a number of objects
- * @return A pointer to the copied objects.
- */
-dap_store_obj_t* dap_store_obj_copy(dap_store_obj_t *a_store_obj, size_t a_store_count)
-{
-dap_store_obj_t *l_store_obj, *l_store_obj_dst, *l_store_obj_src;
-
- if(!a_store_obj || !a_store_count)
- return NULL;
-
- if ( !(l_store_obj = DAP_NEW_SIZE(dap_store_obj_t, sizeof(dap_store_obj_t) * a_store_count)) )
- return NULL;
-
- l_store_obj_dst = l_store_obj;
- l_store_obj_src = a_store_obj;
-
- for( int i = a_store_count; i--; l_store_obj_dst++, l_store_obj_src++)
- {
- memcpy(l_store_obj_dst, l_store_obj_src, sizeof(*l_store_obj_dst));
-
- l_store_obj_dst->group = dap_strdup(l_store_obj_src->group);
- l_store_obj_dst->key = dap_strdup(l_store_obj_src->key);
- if(l_store_obj_src->value &&l_store_obj_src->value_len){
- l_store_obj_dst->value = DAP_DUP_SIZE(l_store_obj_src->value, l_store_obj_src->value_len);
- }
-
- // Why to do if we did memcpy() before?
- //l_store_obj_dst->callback_proc_thread = l_store_obj_src->callback_proc_thread;
- //l_store_obj_dst->callback_proc_thread_arg = l_store_obj_src->callback_proc_thread_arg;
- }
-
- return l_store_obj;
-}
-
-/**
- * @brief Deallocates memory of objects.
- * @param a_store_obj a pointer to objects
- * @param a_store_count a number of objects
- * @return (none)
- */
-void dap_store_obj_free(dap_store_obj_t *a_store_obj, size_t a_store_count)
-{
- if(!a_store_obj || !a_store_count)
- return;
-
- dap_store_obj_t *l_store_obj_cur = a_store_obj;
-
- for ( ; a_store_count--; l_store_obj_cur++ ) {
- DAP_DEL_Z(l_store_obj_cur->group);
- DAP_DEL_Z(l_store_obj_cur->key);
- DAP_DEL_Z(l_store_obj_cur->value);
- }
- DAP_DEL_Z(a_store_obj);
-}
-
-/**
- * @brief Applies objects to database.
- * @param a_store an pointer to the objects
- * @param a_store_count a number of objectss
- * @return Returns 0, if successful.
- */
-static inline int s_dap_chain_global_db_driver_apply_do(dap_store_obj_t *a_store_obj, size_t a_store_count)
-{
-int l_ret;
-dap_store_obj_t *l_store_obj_cur;
-
- if(!a_store_obj || !a_store_count)
- return -1;
-
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "[%p] Process DB Request ...", a_store_obj);
-
- l_store_obj_cur = a_store_obj; /* We have to use a power of the address's incremental arithmetic */
- l_ret = 0; /* Preset return code to OK */
-
- if (a_store_count > 1 && s_drv_callback.transaction_start)
- s_drv_callback.transaction_start();
-
- if(s_drv_callback.apply_store_obj)
- for(int i = a_store_count; (!l_ret) && (i--); l_store_obj_cur++) {
- if ( 1 == (l_ret = s_drv_callback.apply_store_obj(l_store_obj_cur)) )
- log_it(L_INFO, "[%p] Item is missing (may be already deleted) %s/%s", a_store_obj, l_store_obj_cur->group, l_store_obj_cur->key);
- else if (l_ret < 0)
- log_it(L_ERROR, "[%p] Can't write item %s/%s (code %d)", a_store_obj, l_store_obj_cur->group, l_store_obj_cur->key, l_ret);
- }
-
- if(a_store_count > 1 && s_drv_callback.transaction_end)
- s_drv_callback.transaction_end();
-
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "[%p] Finished DB Request (code %d)", a_store_obj, l_ret);
- return l_ret;
-}
-
-static bool s_dap_driver_req_exec (struct dap_proc_thread *a_dap_thd __attribute__((unused)),
- void *arg __attribute__((unused)) )
-{
-int l_ret;
-dap_store_obj_t *l_store_obj_cur;
-dap_worker_t *l_dap_worker;
-size_t l_store_obj_cnt;
-
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Entering, %d entries in the queue ...", s_db_reqs_list.nr);
-
- if ( (l_ret = pthread_mutex_lock(&s_db_reqs_list_lock)) ) /* Get exclusive access to the request list */
- return log_it(L_ERROR, "Cannot lock request queue, errno=%d",l_ret), 0;
-
- if ( !s_db_reqs_list.nr ) /* Nothing to do ?! Just exit */
- {
- pthread_mutex_unlock(&s_db_reqs_list_lock);
- return 1; /* 1 - Don't call it again */
- }
-
- if ( (l_ret = dap_slist_get4head (&s_db_reqs_list, (void **) &l_store_obj_cur, &l_store_obj_cnt)) )
- {
- pthread_mutex_unlock(&s_db_reqs_list_lock);
- log_it(L_ERROR, "DB Request list is in incosistence state (code %d)", l_ret);
- return 1; /* 1 - Don't call it again */
- }
-
- /* So at this point we are ready to do work in the DB */
- s_dap_chain_global_db_driver_apply_do(l_store_obj_cur, l_store_obj_cnt);
-
- pthread_mutex_unlock(&s_db_reqs_list_lock);
-
-
- /* Is there a callback ? */
- if ( s_db_drvmode_async && l_store_obj_cur->callback_proc_thread )
- {
- /* Enqueue "Exec Complete" callback routine */
- l_dap_worker = dap_events_worker_get_auto ();
-
- if ( (l_ret = dap_proc_queue_add_callback(l_dap_worker,
- l_store_obj_cur->callback_proc_thread,
- l_store_obj_cur->callback_proc_thread_arg ) ) )
- log_it(L_ERROR, "[%p] Enqueue completion callback for item %s/%s (code %d)", l_store_obj_cur,
- l_store_obj_cur->group, l_store_obj_cur->key, l_ret);
- }
-
- dap_store_obj_free (l_store_obj_cur, l_store_obj_cnt); /* Release a memory !!! */
-
- return 1; /* 1 - Don't call it again */
-}
-
-
-/**
- * @brief Applies objects to database.
- * @param a_store an pointer to the objects
- * @param a_store_count a number of objectss
- * @return Returns 0, if successful.
- */
-int dap_chain_global_db_driver_apply(dap_store_obj_t *a_store_obj, size_t a_store_count)
-{
-int l_ret;
-dap_store_obj_t *l_store_obj_cur;
-dap_worker_t *l_dap_worker;
-
- if(!a_store_obj || !a_store_count)
- return -1;
-
- if ( !s_db_drvmode_async )
- return s_dap_chain_global_db_driver_apply_do(a_store_obj, a_store_count);
-
-
-
-
-
- /* Async mode - put request into the list for deffered processing */
- l_ret = -ENOMEM; /* Preset return code to non-OK */
-
- pthread_mutex_lock(&s_db_reqs_list_lock); /* Get exclusive access to the request list */
-
- if ( !(l_store_obj_cur = dap_store_obj_copy(a_store_obj, a_store_count)) )
- l_ret = - ENOMEM, log_it(L_ERROR, "[%p] No memory for DB Request for item %s/%s", a_store_obj, a_store_obj->group, a_store_obj->key);
- else if ( (l_ret = dap_slist_add2tail (&s_db_reqs_list, l_store_obj_cur, a_store_count)) )
- log_it(L_ERROR, "[%p] Can't enqueue DB request for item %s/%s (code %d)", a_store_obj, a_store_obj->group, a_store_obj->key, l_ret);
-
- pthread_mutex_unlock(&s_db_reqs_list_lock);
-
- if ( !l_ret )
- { /* So finaly enqueue an execution routine */
- if ( !(l_dap_worker = dap_events_worker_get_auto ()) )
- l_ret = -EBUSY, log_it(L_ERROR, "[%p] Error process DB request for %s/%s, dap_events_worker_get_auto()->NULL", a_store_obj, l_store_obj_cur->group, l_store_obj_cur->key);
- else l_ret = dap_proc_queue_add_callback(l_dap_worker, s_dap_driver_req_exec, NULL);
- }
-
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "[%p] DB Request has been enqueued (code %d)", l_store_obj_cur, l_ret);
-
- return l_ret;
-}
-
-
-
-
-
-
-/**
- * @brief Adds objects to a database.
- * @param a_store_obj objects to be added
- * @param a_store_count a number of added objects
- * @return Returns 0 if sucseesful.
- */
-int dap_chain_global_db_driver_add(pdap_store_obj_t a_store_obj, size_t a_store_count)
-{
-dap_store_obj_t *l_store_obj_cur = a_store_obj;
-
- for(int i = a_store_count; i--; l_store_obj_cur++)
- l_store_obj_cur->type = DAP_DB$K_OPTYPE_ADD;
-
- return dap_chain_global_db_driver_apply(a_store_obj, a_store_count);
-}
-
-/**
- * @brief Deletes objects from a database.
- * @param a_store_obj objects to be deleted
- * @param a_store_count a number of deleted objects
- * @return Returns 0 if sucseesful.
- */
-int dap_chain_global_db_driver_delete(pdap_store_obj_t a_store_obj, size_t a_store_count)
-{
-dap_store_obj_t *l_store_obj_cur = a_store_obj;
-
- for(int i = a_store_count; i--; l_store_obj_cur++)
- l_store_obj_cur->type = DAP_DB$K_OPTYPE_DEL;
-
- return dap_chain_global_db_driver_apply(a_store_obj, a_store_count);
-}
-
-/**
- * @brief Gets a number of stored objects in a database by a_group and id.
- * @param a_group the group name string
- * @param a_id id
- * @return Returns a number of objects.
- */
-size_t dap_chain_global_db_driver_count(const char *a_group, uint64_t id)
-{
- size_t l_count_out = 0;
- // read the number of items
- if(s_drv_callback.read_count_store)
- l_count_out = s_drv_callback.read_count_store(a_group, id);
- return l_count_out;
-}
-
-/**
- * @brief Gets a list of group names matching the pattern.
- * Check whether the groups match the pattern a_group_mask, which is a shell wildcard pattern
- * patterns: [] {} [!] * ? https://en.wikipedia.org/wiki/Glob_(programming).
- * @param a_group_mask the group mask string
- * @return If successful, returns the list of group names, otherwise NULL.
- */
-dap_list_t *dap_chain_global_db_driver_get_groups_by_mask(const char *a_group_mask)
-{
- dap_list_t *l_list = NULL;
- if(s_drv_callback.get_groups_by_mask)
- l_list = s_drv_callback.get_groups_by_mask(a_group_mask);
- return l_list;
-}
-
-
-/**
- * @brief Reads last object in the database.
- * @param a_group the group name
- * @return If successful, a pointer to the object, otherwise NULL.
- */
-dap_store_obj_t* dap_chain_global_db_driver_read_last(const char *a_group)
-{
- dap_store_obj_t *l_ret = NULL;
- // read records using the selected database engine
- if(s_drv_callback.read_last_store_obj)
- l_ret = s_drv_callback.read_last_store_obj(a_group);
- return l_ret;
-}
-
-/**
- * @brief Reads several objects from a database by a_group and id.
- * @param a_group the group name string
- * @param a_id id
- * @param a_count_out[in] a number of objects to be read, if 0 - no limits
- * @param a_count_out[out] a count of objects that were read
- * @return If successful, a pointer to an objects, otherwise NULL.
- */
-dap_store_obj_t* dap_chain_global_db_driver_cond_read(const char *a_group, uint64_t id, size_t *a_count_out)
-{
- dap_store_obj_t *l_ret = NULL;
- // read records using the selected database engine
- if(s_drv_callback.read_cond_store_obj)
- l_ret = s_drv_callback.read_cond_store_obj(a_group, id, a_count_out);
- return l_ret;
-}
-
-/**
- * @brief Reads several objects from a database by a_group and a_key.
- * If a_key is NULL, reads whole group.
- * @param a_group a group name string
- * @param a_key an object key string. If equal NULL, it means reading the whole group
- * @param a_count_out[in] a number of objects to be read, if 0 - no limits
- * @param a_count_out[out] a number of objects that were read
- * @return If successful, a pointer to an objects, otherwise NULL.
- */
-dap_store_obj_t* dap_chain_global_db_driver_read(const char *a_group, const char *a_key, size_t *a_count_out)
-{
- dap_store_obj_t *l_ret = NULL;
- // read records using the selected database engine
- if(s_drv_callback.read_store_obj)
- l_ret = s_drv_callback.read_store_obj(a_group, a_key, a_count_out);
- return l_ret;
-}
-
-/**
- * @brief Checks if an object is in a database by a_group and a_key.
- * @param a_group a group name string
- * @param a_key a object key string
- * @return Returns true if it is, false otherwise.
- */
-bool dap_chain_global_db_driver_is(const char *a_group, const char *a_key)
-{
- bool l_ret = NULL;
- // read records using the selected database engine
- if(s_drv_callback.is_obj && a_group && a_key)
- return s_drv_callback.is_obj(a_group, a_key);
- else
- return false;
-}
diff --git a/modules/global-db/dap_chain_global_db_driver_cdb.c b/modules/global-db/dap_chain_global_db_driver_cdb.c
deleted file mode 100644
index 7a1a55dfae7a49f15b7695c76913a7dba612b34b..0000000000000000000000000000000000000000
--- a/modules/global-db/dap_chain_global_db_driver_cdb.c
+++ /dev/null
@@ -1,651 +0,0 @@
-/*
- * Authors:
- * Konstantin Papizh <konstantin.papizh@demlabs.net>
- * DeM Labs Inc. https://demlabs.net
- * Kelvin Project https://github.com/kelvinblockchain
- * Copyright (c) 2019
- * All rights reserved.
-
- This file is part of DAP (Deus Applications Prototypes) the open source project
-
- DAP (Deus Applicaions Prototypes) is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP based project. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#include <stddef.h>
-#include <string.h>
-#include <dirent.h>
-#include <pthread.h>
-#include <sys/stat.h>
-#include <uthash.h>
-#include <pthread.h>
-#include <errno.h>
-#include <stdatomic.h>
-
-#define _GNU_SOURCE
-
-#include "dap_global_db.h"
-#include "dap_chain_global_db_driver_cdb.h"
-#include "dap_common.h"
-#include "dap_hash.h"
-#include "dap_strfuncs.h" // #include <dap_fnmatch.h>
-#include "dap_file_utils.h"
-
-#define LOG_TAG "dap_chain_global_db_cdb"
-
-/** Struct for a item */
-typedef struct _obj_arg {
- pdap_store_obj_t o;
- uint64_t q;
- uint64_t n;
- uint64_t id;
-} obj_arg, *pobj_arg;
-
-/** Struct for a CDB instanse */
-typedef struct _cdb_instance {
- CDB *cdb;
- char *local_group;
- uint64_t id;
- UT_hash_handle hh;
-} cdb_instance, *pcdb_instance;
-
-/** A path to a CDB file. */
-static char s_cdb_path[MAX_PATH];
-
-/** A pointer to a CDB instance. */
-static pcdb_instance s_cdb = NULL;
-/** A mutex for working with a CDB instanse. */
-static pthread_mutex_t cdb_mutex = PTHREAD_MUTEX_INITIALIZER;
-/** A read-write lock for working with a CDB instanse. */
-static pthread_rwlock_t cdb_rwlock = PTHREAD_RWLOCK_INITIALIZER;
-
-/**
- * @brief Serialize key and val to a item
- * key -> key
- * val[0..8] => id
- * val[..] => value_len
- * val[..] => value
- * val[..] => timestamp
- * @param a_obj a pointer to a item
- * @param key a key
- * @param val a serialize string
- */
-static void cdb_serialize_val_to_dap_store_obj(pdap_store_obj_t a_obj, const char *key, const char *val) {
- if (!key) {
- return;
- }
- int offset = 0;
- a_obj->key = dap_strdup(key);
- a_obj->id = dap_hex_to_uint(val, sizeof(uint64_t));
- offset += sizeof(uint64_t);
- a_obj->flags = dap_hex_to_uint(val + offset, sizeof(uint8_t));
- offset += sizeof(uint8_t);
- a_obj->value_len = dap_hex_to_uint(val + offset, sizeof(uint64_t));
- offset += sizeof(uint64_t);
- if (a_obj->value_len) {
- a_obj->value = DAP_NEW_SIZE(uint8_t, a_obj->value_len);
- memcpy((byte_t *)a_obj->value, val + offset, a_obj->value_len);
- }
- offset += a_obj->value_len;
- a_obj->timestamp = dap_hex_to_uint(val + offset, sizeof(uint64_t));
-}
-
-/** A callback function designed for finding a last item */
-bool dap_cdb_get_last_obj_iter_callback(void *arg, const char *key, int ksize, const char *val, int vsize, uint32_t expire, uint64_t oid) {
- UNUSED(ksize);
- UNUSED(val);
- UNUSED(vsize);
- UNUSED(expire);
- UNUSED(oid);
-
- if (--((pobj_arg)arg)->q == 0) {
- cdb_serialize_val_to_dap_store_obj((pdap_store_obj_t)(((pobj_arg)arg)->o), key, val);
- return false;
- }
- return true;
-}
-
-//** A callback function designed for finding a some items */
-bool dap_cdb_get_some_obj_iter_callback(void *arg, const char *key, int ksize, const char *val, int vsize, uint32_t expire, uint64_t oid) {
- UNUSED(ksize);
- UNUSED(val);
- UNUSED(vsize);
- UNUSED(expire);
- UNUSED(oid);
-
- pdap_store_obj_t l_obj = (pdap_store_obj_t)((pobj_arg)arg)->o;
- cdb_serialize_val_to_dap_store_obj(&l_obj[((pobj_arg)arg)->n - ((pobj_arg)arg)->q], key, val);
- if (--((pobj_arg)arg)->q == 0) {
- return false;
- }
- return true;
-}
-
-//** A callback function designed for finding a some items by conditionals */
-bool dap_cdb_get_cond_obj_iter_callback(void *arg, const char *key, int ksize, const char *val, int vsize, uint32_t expire, uint64_t oid) {
- UNUSED(ksize);
- UNUSED(val);
- UNUSED(vsize);
- UNUSED(expire);
- UNUSED(oid);
-
- if (dap_hex_to_uint(val, sizeof(uint64_t)) < ((pobj_arg)arg)->id) {
- return true;
- }
- pdap_store_obj_t l_obj = (pdap_store_obj_t)((pobj_arg)arg)->o;
- cdb_serialize_val_to_dap_store_obj(&l_obj[((pobj_arg)arg)->n - ((pobj_arg)arg)->q], key, val);
- if (--((pobj_arg)arg)->q == 0) {
- return false;
- }
- return true;
-}
-
-//** A callback function designed for counting items*/
-bool dap_cdb_get_count_iter_callback(void *arg, const char *key, int ksize, const char *val, int vsize, uint32_t expire, uint64_t oid) {
- UNUSED(ksize);
- UNUSED(val);
- UNUSED(vsize);
- UNUSED(expire);
- UNUSED(oid);
- UNUSED(key);
-
- if (dap_hex_to_uint(val, sizeof(uint64_t)) < ((pobj_arg)arg)->id) {
- return true;
- }
- if (--((pobj_arg)arg)->q == 0) {
- return false;
- }
- return true;
-}
-
-/**
- * @brief Initiates a CDB with main hash table size: 1000000,
- * record cache: 128Mb, index page cache: 1024Mb.
- * @param a_group a group name
- * @param a_flags should be combination of CDB_CREAT / CDB_TRUNC / CDB_PAGEWARMUP
- CDB_PAGEWARMUP
- * @return A pointer to CDB, if success. NULL, if error.
- */
-pcdb_instance dap_cdb_init_group(const char *a_group, int a_flags) {
- pcdb_instance l_cdb_i = NULL;
- char l_cdb_path[MAX_PATH + 2];
-
- pthread_mutex_lock(&cdb_mutex);
- HASH_FIND_STR(s_cdb, a_group, l_cdb_i);
- if (a_flags & CDB_TRUNC) {
- if (!l_cdb_i)
- goto FIN;
- cdb_close(l_cdb_i->cdb);
- }
- if (!(a_flags & CDB_TRUNC)) {
- if (l_cdb_i)
- goto FIN;
- l_cdb_i = DAP_NEW(cdb_instance);
- l_cdb_i->local_group = dap_strdup(a_group);
- l_cdb_i->cdb = cdb_new();
- }
- memset(l_cdb_path, '\0', sizeof(l_cdb_path));
- dap_snprintf(l_cdb_path, sizeof(l_cdb_path), "%s/%s", s_cdb_path, a_group);
- cdb_options l_opts = { 100000, 256, 1024 };
- if (cdb_option(l_cdb_i->cdb, l_opts.hsize, l_opts.pcacheMB, l_opts.rcacheMB) != CDB_SUCCESS) {
- log_it(L_ERROR, "Options are inacceptable: \"%s\"", cdb_errmsg(cdb_errno(l_cdb_i->cdb)));
- goto ERR;
- }
- if (cdb_open(l_cdb_i->cdb, l_cdb_path, a_flags) != CDB_SUCCESS) {
- log_it(L_ERROR, "An error occured while opening CDB: \"%s\"", cdb_errmsg(cdb_errno(l_cdb_i->cdb)));
- goto ERR;
- }
- if (!(a_flags & CDB_TRUNC)) {
- CDBSTAT l_cdb_stat;
- cdb_stat(l_cdb_i->cdb, &l_cdb_stat);
- if (l_cdb_stat.rnum > 0 || !(a_flags & CDB_CREAT)) {
- void *l_iter = cdb_iterate_new(l_cdb_i->cdb, 0);
- obj_arg l_arg;
- l_arg.o = DAP_NEW_Z(dap_store_obj_t);
- l_arg.q = l_cdb_stat.rnum;
- cdb_iterate(l_cdb_i->cdb, dap_cdb_get_last_obj_iter_callback, (void*)&l_arg, l_iter);
- cdb_iterate_destroy(l_cdb_i->cdb, l_iter);
- l_cdb_i->id = l_arg.o->id;
- //log_it(L_INFO, "Group \"%s\" found" , l_cdb_i->local_group);
- //log_it(L_INFO, "Records: %-24u" , l_cdb_stat.rnum);
- //log_it(L_INFO, "Average read latency: %-24u" , l_cdb_stat.rlatcy);
- //log_it(L_INFO, "Average write latency: %-24u" , l_cdb_stat.wlatcy);
- //log_it(L_INFO, "Last id: %-24u" , l_cdb_i->id);
- dap_store_obj_free_one(l_arg.o);
- } else {
- log_it(L_INFO, "Group \"%s\" created", l_cdb_i->local_group);
- l_cdb_i->id = 0;
- }
- HASH_ADD_KEYPTR(hh, s_cdb, l_cdb_i->local_group, strlen(l_cdb_i->local_group), l_cdb_i);
- } else {
- log_it(L_INFO, "Group \"%s\" truncated", l_cdb_i->local_group);
- DAP_DELETE(l_cdb_i->local_group);
- cdb_destroy(l_cdb_i->cdb);
- HASH_DEL(s_cdb, l_cdb_i);
- DAP_DEL_Z(l_cdb_i);
- }
-
-FIN:
- pthread_mutex_unlock(&cdb_mutex);
- return l_cdb_i;
-ERR:
- cdb_destroy(l_cdb_i->cdb);
- DAP_DELETE(l_cdb_i->local_group);
- DAP_DELETE(l_cdb_i);
- pthread_mutex_unlock(&cdb_mutex);
- return NULL;
-}
-
-/**
- * @brief Initiates a CDB with callback fuctions.
- * @param a_cdb_path a path to CDB. Saved in s_cdb_path
- * @param a_drv_callback a struct for callback functions
- * @param a_db_drvmode_async
- * @return 0 if success, -1 if Ñouldn't open db directory, -2 if dap_cdb_init_group() returns NULL.
- */
-int dap_db_driver_cdb_init(const char *a_cdb_path, dap_db_driver_callbacks_t *a_drv_callback)
-{
-int l_rc;
-struct dirent *d;
-DIR *dir;
-
- for (l_rc = 0; *a_cdb_path && ((size_t)l_rc < sizeof(s_cdb_path)); )
- s_cdb_path[l_rc++] = *(a_cdb_path++); /* Copy path to CDB root to local storage, calculate length */
-
- if(s_cdb_path[l_rc - 1] == '/') /* Truncate terminal '/' if need */
- s_cdb_path[--l_rc] = '\0';
-
- dap_mkdir_with_parents(s_cdb_path);
-
- if ( !(dir = opendir(s_cdb_path)) )
- return log_it(L_ERROR, "Couldn't open db directory '%s', errno=%d", s_cdb_path, errno), -errno;
-
- while ( (d = readdir(dir))) {
-#ifdef _DIRENT_HAVE_D_TYPE
- if (d->d_type != DT_DIR)
- continue;
-#elif defined(DAP_OS_LINUX)
- struct _stat buf;
- int res = _stat(d->d_name, &buf);
- if (!S_ISDIR(buf.st_mode) || !res) {
- continue;
- }
-#elif defined (DAP_OS_BSD)
- struct stat buf;
- int res = stat(d->d_name, &buf);
- if (!S_ISDIR(buf.st_mode) || !res) {
- continue;
- }
-#endif
- if ( (d->d_name[0] == '.') || !dap_strcmp(d->d_name, ".."))
- continue;
-
- pcdb_instance l_cdb_i = dap_cdb_init_group(d->d_name, CDB_PAGEWARMUP);
- if (!l_cdb_i) {
- dap_db_driver_cdb_deinit();
- closedir(dir);
- return -2;
- }
- }
-
- a_drv_callback->read_last_store_obj = dap_db_driver_cdb_read_last_store_obj;
- a_drv_callback->apply_store_obj = dap_db_driver_cdb_apply_store_obj;
- a_drv_callback->read_store_obj = dap_db_driver_cdb_read_store_obj;
- a_drv_callback->read_cond_store_obj = dap_db_driver_cdb_read_cond_store_obj;
- a_drv_callback->read_count_store = dap_db_driver_cdb_read_count_store;
- a_drv_callback->get_groups_by_mask = dap_db_driver_cdb_get_groups_by_mask;
- a_drv_callback->is_obj = dap_db_driver_cdb_is_obj;
- a_drv_callback->deinit = dap_db_driver_cdb_deinit;
- a_drv_callback->flush = dap_db_driver_cdb_flush;
-
- closedir(dir);
-
- return CDB_SUCCESS;
-}
-
-/**
- * @brief Gets CDB by a_group.
- * @param a_group a group name
- * @return if CDB is found, a pointer to CDB, otherwise NULL.
- */
-pcdb_instance dap_cdb_get_db_by_group(const char *a_group) {
- pcdb_instance l_cdb_i = NULL;
- pthread_rwlock_rdlock(&cdb_rwlock);
- HASH_FIND_STR(s_cdb, a_group, l_cdb_i);
- pthread_rwlock_unlock(&cdb_rwlock);
- return l_cdb_i;
-}
-
-/**
- * @brief Creates a directory on the path s_cdb_path/a_group.
- * @param a_group the group name
- * @return 0
- */
-int dap_cdb_add_group(const char *a_group) {
- char l_cdb_path[MAX_PATH + 2];
- memset(l_cdb_path, '\0', sizeof(l_cdb_path));
- dap_snprintf(l_cdb_path, sizeof(l_cdb_path), "%s/%s", s_cdb_path, a_group);
-#ifdef _WIN32
- mkdir(l_cdb_path);
-#else
- mkdir(l_cdb_path, 0755);
-#endif
- return 0;
-}
-
-/**
- * @brief Deinitialize CDB.
- * @return 0
- */
-int dap_db_driver_cdb_deinit() {
- pcdb_instance cur_cdb, tmp;
- pthread_rwlock_wrlock(&cdb_rwlock);
- HASH_ITER(hh, s_cdb, cur_cdb, tmp) {
- DAP_DELETE(cur_cdb->local_group);
- cdb_destroy(cur_cdb->cdb);
- HASH_DEL(s_cdb, cur_cdb);
- DAP_DELETE(cur_cdb);
- }
- pthread_rwlock_unlock(&cdb_rwlock);
-
- return CDB_SUCCESS;
-}
-
-/**
- * @brief Flushing CDB to the disk.
- * @return 0
- */
-int dap_db_driver_cdb_flush(void) {
- int ret = 0;
- log_it(L_DEBUG, "Flushing CDB to disk");
- cdb_instance *cur_cdb, *tmp;
- pthread_rwlock_rdlock(&cdb_rwlock);
- HASH_ITER(hh, s_cdb, cur_cdb, tmp) {
- cdb_flushalldpage(cur_cdb->cdb);
- }
- pthread_rwlock_unlock(&cdb_rwlock);
- log_it(L_DEBUG, "All data dumped");
- return ret;
-}
-
-/**
- * @brief Read last store item from CDB.
- * @param a_group a group name
- * @return If successful, a pointer to item, otherwise NULL.
- */
-dap_store_obj_t *dap_db_driver_cdb_read_last_store_obj(const char* a_group) {
- if (!a_group) {
- return NULL;
- }
- pcdb_instance l_cdb_i = dap_cdb_get_db_by_group(a_group);
- if (!l_cdb_i) {
- return NULL;
- }
- CDB *l_cdb = l_cdb_i->cdb;
- CDBSTAT l_cdb_stat;
- cdb_stat(l_cdb, &l_cdb_stat);
- if (!l_cdb_stat.rnum)
- return NULL;
- void *l_iter = cdb_iterate_new(l_cdb, 0);
- obj_arg l_arg;
- l_arg.o = DAP_NEW_Z(dap_store_obj_t);
- l_arg.q = l_cdb_stat.rnum;
- cdb_iterate(l_cdb, dap_cdb_get_last_obj_iter_callback, (void*)&l_arg, l_iter);
- cdb_iterate_destroy(l_cdb, l_iter);
- l_arg.o->group = dap_strdup(a_group);
- return l_arg.o;
-}
-
-/**
- * @brief Checks if CDB has a_key
- * @param a_group the group name
- * @param a_key the key
- * @return true or false
- */
-bool dap_db_driver_cdb_is_obj(const char *a_group, const char *a_key)
-{
- bool l_ret = false;
- if(!a_group) {
- return false;
- }
- pcdb_instance l_cdb_i = dap_cdb_get_db_by_group(a_group);
- if(!l_cdb_i) {
- return false;
- }
- CDB *l_cdb = l_cdb_i->cdb;
- if(a_key) {
- //int l_vsize;
- if(!cdb_is(l_cdb, a_key, (int) dap_strlen(a_key)))
- l_ret = true;
- }
- return l_ret;
-}
-
-/**
- * @brief Gets items from CDB by a_group and a_key. If a_key=NULL then gets a_count_out items.
- * @param a_group the group name
- * @param a_key the key or NULL
- * @param a_count_out IN. Count of read items. OUT Count of items was read
- * @return If successful, pointer to items; otherwise NULL.
- */
-dap_store_obj_t *dap_db_driver_cdb_read_store_obj(const char *a_group, const char *a_key, size_t *a_count_out) {
- if (!a_group) {
- return NULL;
- }
- pcdb_instance l_cdb_i = dap_cdb_get_db_by_group(a_group);
- if (!l_cdb_i) {
- return NULL;
- }
- CDB *l_cdb = l_cdb_i->cdb;
- dap_store_obj_t *l_obj = NULL;
- if (a_key) {
- char *l_value;
- int l_vsize;
- cdb_get(l_cdb, a_key, (int)strlen(a_key), (void**)&l_value, &l_vsize);
- if (!l_value) {
- return NULL;
- }
- l_obj = DAP_NEW_Z(dap_store_obj_t);
- cdb_serialize_val_to_dap_store_obj(l_obj, a_key, l_value);
- l_obj->group = dap_strdup(a_group);
- cdb_free_val((void**)&l_value);
- if(a_count_out) {
- *a_count_out = 1;
- }
- } else {
- uint64_t l_count_out = 0;
- if(a_count_out) {
- l_count_out = *a_count_out;
- }
- CDBSTAT l_cdb_stat;
- cdb_stat(l_cdb, &l_cdb_stat);
- if ((l_count_out == 0) || (l_count_out > l_cdb_stat.rnum)) {
- l_count_out = l_cdb_stat.rnum;
- }
- obj_arg l_arg;
- l_arg.o = DAP_NEW_Z_SIZE(dap_store_obj_t, l_count_out * sizeof(dap_store_obj_t));
- l_arg.q = l_count_out;
- l_arg.n = l_count_out;
- void *l_iter = cdb_iterate_new(l_cdb, 0);
- /*l_count_out = */cdb_iterate(l_cdb, dap_cdb_get_some_obj_iter_callback, (void*)&l_arg, l_iter);
- cdb_iterate_destroy(l_cdb, l_iter);
- if(a_count_out) {
- *a_count_out = l_count_out;
- }
- for (uint64_t i = 0; i < l_count_out; ++i) {
- l_arg.o[i].group = dap_strdup(a_group);
- }
- l_obj = l_arg.o;
- }
- return l_obj;
-}
-
-/**
- * @brief Gets items from CDB by a_group and a_id.
- * @param a_group the group name
- * @param a_id id
- * @param a_count_out[in] a count of items
- * @param a_count[out] a count of items were got
- * @return If successful, pointer to items, otherwise NULL.
- */
-dap_store_obj_t* dap_db_driver_cdb_read_cond_store_obj(const char *a_group, uint64_t a_id, size_t *a_count_out) {
- if (!a_group) {
- return NULL;
- }
- pcdb_instance l_cdb_i = dap_cdb_get_db_by_group(a_group);
- if (!l_cdb_i) {
- return NULL;
- }
- CDB *l_cdb = l_cdb_i->cdb;
- uint64_t l_count_out = 0;
- if(a_count_out) {
- l_count_out = *a_count_out;
- }
- CDBSTAT l_cdb_stat;
- cdb_stat(l_cdb, &l_cdb_stat);
-
- if (l_count_out == 0 || l_count_out > l_cdb_stat.rnum) {
- l_count_out = l_cdb_stat.rnum;
- }
- obj_arg l_arg;
- l_arg.o = DAP_NEW_Z_SIZE(dap_store_obj_t, l_count_out * sizeof(dap_store_obj_t));
- l_arg.n = l_count_out;
- l_arg.q = l_count_out;
- l_arg.id = a_id;
- void *l_iter = cdb_iterate_new(l_cdb, 0);
- /*l_count_out = */cdb_iterate(l_cdb, dap_cdb_get_cond_obj_iter_callback, (void*)&l_arg, l_iter);
- cdb_iterate_destroy(l_cdb, l_iter);
- if (l_arg.q > 0) {
- l_count_out = l_arg.n - l_arg.q;
- void *tmp = DAP_REALLOC(l_arg.o, l_count_out * sizeof(dap_store_obj_t));
- if (!tmp && l_count_out) {
- log_it(L_CRITICAL, "Couldn't re-allocate memory for portion of store objects!");
- DAP_DELETE(l_arg.o);
- return NULL;
- }
- l_arg.o = tmp;
- }
- if(a_count_out) {
- *a_count_out = l_count_out;
- }
- for (uint64_t i = 0; i < l_count_out; ++i) {
- l_arg.o[i].group = dap_strdup(a_group);
- }
- return l_arg.o;
-}
-
-/**
- * @brief Reads count of items in CDB by a_group and a_id.
- * @param a_group the group name
- * @param a_id id
- * @return If successful, count of store items; otherwise 0.
- */
-size_t dap_db_driver_cdb_read_count_store(const char *a_group, uint64_t a_id)
-{
- if (!a_group) {
- return 0;
- }
- pcdb_instance l_cdb_i = dap_cdb_get_db_by_group(a_group);
- if (!l_cdb_i) {
- return 0;
- }
- CDB *l_cdb = l_cdb_i->cdb;
- CDBSTAT l_cdb_stat;
- cdb_stat(l_cdb, &l_cdb_stat);
- obj_arg l_arg;
- l_arg.q = l_cdb_stat.rnum;
- l_arg.id = a_id;
- void *l_iter = cdb_iterate_new(l_cdb, 0);
- cdb_iterate(l_cdb, dap_cdb_get_count_iter_callback, (void*)&l_arg, l_iter);
- cdb_iterate_destroy(l_cdb, l_iter);
- return l_cdb_stat.rnum - l_arg.q;
-}
-
-/**
- * @brief Check whether the groups match the pattern a_group_mask, which is a shell wildcard pattern.
- * @param a_group_mask the mask
- * @return If successful, pointer to dap_list with group names; otherwise NULL.
- */
-dap_list_t* dap_db_driver_cdb_get_groups_by_mask(const char *a_group_mask)
-{
- dap_list_t *l_ret_list = NULL;
- if(!a_group_mask)
- return NULL;
- cdb_instance *cur_cdb, *tmp;
- pthread_rwlock_rdlock(&cdb_rwlock);
- HASH_ITER(hh, s_cdb, cur_cdb, tmp) {
- char *l_table_name = cur_cdb->local_group;
- if(!dap_fnmatch(a_group_mask, l_table_name, 0))
- l_ret_list = dap_list_prepend(l_ret_list, dap_strdup(l_table_name));
- }
- pthread_rwlock_unlock(&cdb_rwlock);
- return l_ret_list;
-}
-
-/**
- * @brief Adds or deletes item in CDB depending on a_store_obj->type.
- * @param a_store_obj a pointer to the item
- * @return 0 if success, <0 error.
- */
-int dap_db_driver_cdb_apply_store_obj(pdap_store_obj_t a_store_obj) {
- if(!a_store_obj || !a_store_obj->group) {
- return -1;
- }
- int ret = 0;
- pcdb_instance l_cdb_i = dap_cdb_get_db_by_group(a_store_obj->group);
- if (!l_cdb_i) {
- dap_cdb_add_group(a_store_obj->group);
- l_cdb_i = dap_cdb_init_group(a_store_obj->group, CDB_CREAT | CDB_PAGEWARMUP);
- }
- if (!l_cdb_i) {
- return -1;
- }
- if(a_store_obj->type == DAP_DB$K_OPTYPE_ADD) {
- if(!a_store_obj->key) {
- return -2;
- };
- int offset = 0;
- char *l_val = DAP_NEW_Z_SIZE(char, sizeof(uint64_t) + sizeof(uint8_t) + sizeof(uint64_t) + a_store_obj->value_len + sizeof(uint64_t));
- dap_uint_to_hex(l_val, ++l_cdb_i->id, sizeof(uint64_t));
- offset += sizeof(uint64_t);
- // Add flags
- dap_uint_to_hex(l_val + offset, a_store_obj->flags, sizeof(uint8_t));
- offset += sizeof(uint8_t);
- // Add length of value
- dap_uint_to_hex(l_val + offset, a_store_obj->value_len, sizeof(uint64_t));
- offset += sizeof(uint64_t);
- // Add value
- if(a_store_obj->value && a_store_obj->value_len){
- memcpy(l_val + offset, a_store_obj->value, a_store_obj->value_len);
- }
- offset += a_store_obj->value_len;
- // Add a timestamp
- dap_uint_to_hex(l_val + offset, a_store_obj->timestamp, sizeof(uint64_t));
- offset += sizeof(uint64_t);
- if (cdb_set2(l_cdb_i->cdb, a_store_obj->key, (int)strlen(a_store_obj->key), l_val, offset, CDB_INSERTCACHE | CDB_OVERWRITE, 0) != CDB_SUCCESS) {
- log_it(L_ERROR, "Couldn't add record with key [%s] to CDB: \"%s\"", a_store_obj->key, cdb_errmsg(cdb_errno(l_cdb_i->cdb)));
- ret = -1;
- }
- cdb_flushalldpage(l_cdb_i->cdb);
- DAP_DELETE(l_val);
- } else if(a_store_obj->type == DAP_DB$K_OPTYPE_DEL) {
- if(a_store_obj->key) {
- if(cdb_del(l_cdb_i->cdb, a_store_obj->key, (int) strlen(a_store_obj->key)) == -3)
- ret = 1;
- } else {
- dap_cdb_init_group(a_store_obj->group, CDB_TRUNC | CDB_PAGEWARMUP);
- ret = 0;
- }
- }
- return ret;
-}
diff --git a/modules/global-db/dap_chain_global_db_driver_mdbx.c b/modules/global-db/dap_chain_global_db_driver_mdbx.c
deleted file mode 100644
index e307141db085afc17ce151622881b99f9954e1ed..0000000000000000000000000000000000000000
--- a/modules/global-db/dap_chain_global_db_driver_mdbx.c
+++ /dev/null
@@ -1,1320 +0,0 @@
-/*
- * AUTHORS:
- * Ruslan R. (The BadAss SysMan) Laishev <ruslan.laishev@demlabs.net>
- * DeM Labs Ltd. https://demlabs.net
- * Copyright (c) 2022
- * All rights reserved.
-
- This file is part of DAP SDK the open source project
-
- DAP SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP based project. If not, see <http://www.gnu.org/licenses/>.
-
-
- DESCRIPTION: A database driver module provide an interface to MDBX API.
- https://gitflic.ru/project/erthink/libmdbx
- TG group: @libmdbx
-
-
- MODIFICATION HISTORY:
-
- 4-MAY-2022 RRL Developing for actual version of the LibMDBX
-
- 12-MAY-2022 RRL Finished developing of preliminary version
-
- 19-MAY-2022 RRL Added routines' decsriptions
-
- 1-JUN-2022 RRL Introduced dap_assert()
-
- 18-JUL-2022 RRL Fixed unitialized l_obj_arr in the s_db_mdbx_read_store_obj()
- */
-
-#include <stddef.h>
-#include <string.h>
-#include <dirent.h>
-#include <pthread.h>
-#include <sys/stat.h>
-#include <errno.h>
-#include <uthash.h>
-#include <stdatomic.h>
-
-#define _GNU_SOURCE
-
-#include "dap_global_db.h"
-#include "dap_config.h"
-#include "dap_chain_global_db_driver_mdbx.h"
-#include "dap_hash.h"
-#include "dap_strfuncs.h"
-#include "dap_file_utils.h"
-#include "dap_common.h"
-
-#include "mdbx.h" /* LibMDBX API */
-#define LOG_TAG "dap_chain_global_db_mdbx"
-
-
-/** Struct for a MDBX DB context */
-typedef struct __db_ctx__ {
- size_t namelen; /* Group name length */
- char name[DAP_GLOBAL_DB_GROUP_NAME_SIZE_MAX + 1]; /* Group's name */
-
- pthread_mutex_t dbi_mutex; /* Coordinate access the MDBX's <dbi> */
- MDBX_dbi dbi; /* MDBX's internal context id */
- MDBX_txn *txn; /* Current MDBX's transaction */
-
- UT_hash_handle hh;
-} dap_db_ctx_t;
-
-static pthread_mutex_t s_db_ctx_mutex = PTHREAD_MUTEX_INITIALIZER; /* A mutex for working with a DB context */
-
-
-
-static dap_db_ctx_t *s_db_ctxs = NULL; /* A hash table of <group/subDB/table> == <MDBX DB context> */
-static pthread_rwlock_t s_db_ctxs_rwlock = PTHREAD_RWLOCK_INITIALIZER; /* A read-write lock for working with a <s_db_ctxs>. */
-
-static char s_db_path[MAX_PATH]; /* A root directory for the MDBX files */
-
-/* Forward declarations of action routines */
-static int s_db_mdbx_deinit();
-static int s_db_mdbx_flush(void);
-static int s_db_mdbx_apply_store_obj (dap_store_obj_t *a_store_obj);
-static dap_store_obj_t *s_db_mdbx_read_last_store_obj(const char* a_group);
-static bool s_db_mdbx_is_obj(const char *a_group, const char *a_key);
-static dap_store_obj_t *s_db_mdbx_read_store_obj(const char *a_group, const char *a_key, size_t *a_count_out);
-static dap_store_obj_t *s_db_mdbx_read_cond_store_obj(const char *a_group, uint64_t a_id, size_t *a_count_out);
-static size_t s_db_mdbx_read_count_store(const char *a_group, uint64_t a_id);
-static dap_list_t *s_db_mdbx_get_groups_by_mask(const char *a_group_mask);
-
-
-static MDBX_env *s_mdbx_env; /* MDBX's context area */
-static char s_subdir [] = ""; /* Name of subdir for the MDBX's database files */
-
-static char s_db_master_tbl [] = "MDBX$MASTER"; /* A name of master table in the MDBX
- to keep and maintains application level information */
-static MDBX_dbi s_db_master_dbi; /* A handle of the MDBX' DBI of the master subDB */
-
-
-/*
- * Suffix structure is supposed to be added at end of MDBX record, so :
- * <value> + <suffix>
- */
-struct __record_suffix__ {
- uint64_t mbz; /* Must Be Zero ! */
- uint64_t id; /* An uniqe-like Id of the record - internaly created and maintained */
- uint64_t flags; /* Flag of the record : see RECORD_FLAGS enums */
- dap_time_t ts; /* Timestamp of the record */
-};
-
-
-
-
-/*
- * DESCRIPTION: A kind of replacement of the C RTL assert()
- *
- * INPUTS:
- *
- * OUTPUTS:
- *
- * RETURNS:
- * NONE
- */
-static inline void s_dap_assert_fail(int a_condition, const char *a_expr, const char *a_file, int a_line)
-{
-char buf[255];
-int buflen;
-
- if ( a_condition )
- return;
-
- buflen = snprintf(buf, sizeof(buf), "\n[%s:%d] <%s> expresion return false\n", a_file, a_line, a_expr);
- write(STDOUT_FILENO, buf, buflen);
- abort();
-}
-
-#define dap_assert(expr) s_dap_assert_fail( (bool) (expr), #expr, __FILE__, __LINE__)
-
-
-#if DAP_SYS_DEBUG /* cmake ../ -DCMAKE_BUILD_TYPE=Debug -DCMAKE_C_FLAGS="-DDAP_SYS_DEBUG=1" */
-/*
- * DESCRIPTION: Dump all records from the table . Is supposed to be used at debug time.
- *
- * INPUTS:
- * a_db_ctx: DB context
- *
- * OUTPUTS:
- * NONE:
- *
- * RETURNS:
- * NONE
- */
-static void s_db_dump (dap_db_ctx_t *a_db_ctx)
-{
-int l_rc;
-MDBX_val l_key_iov, l_data_iov;
-MDBX_cursor *l_cursor;
-char l_buf[1024] = {0};
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, MDBX_TXN_RDONLY, &a_db_ctx->txn)) )
- log_it(L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc));
- else if ( MDBX_SUCCESS != (l_rc = mdbx_cursor_open(a_db_ctx->txn, a_db_ctx->dbi, &l_cursor)) )
- log_it(L_ERROR, "mdbx_cursor_open: (%d) %s", l_rc, mdbx_strerror(l_rc));
- else {
- while ( !(l_rc = mdbx_cursor_get (l_cursor, &l_key_iov, &l_data_iov, MDBX_NEXT )) )
- {
- l_rc = dap_bin2hex (l_buf, l_data_iov.iov_base, min(l_data_iov.iov_len, 72) );
-
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "[0:%zu]: '%.*s' = [0:%zu]: '%.*s'",
- l_key_iov.iov_len, (int) l_key_iov.iov_len, (char *) l_key_iov.iov_base,
- l_data_iov.iov_len, l_rc, l_buf);
- }
- }
-
- if (l_cursor)
- mdbx_cursor_close(l_cursor);
-
- if (a_db_ctx->txn)
- mdbx_txn_abort(a_db_ctx->txn);
-}
-#endif /* DAP_SYS_DEBUG */
-
-/*
- * DESCRIPTION: Open or create (if a_flag=MDBX_CREATE) a DB context for a given group.
- * Initialize an MDBX's internal context for the subDB (== a_group);
- * Add new group/table name into the special MDBX subDB named MDBX$MASTER.
- *
- * INPUTS:
- * a_group: A group name (in terms of MDBX it's subDB), ASCIZ
- * a_flag: A flag
- *
- * IMPLICITE OUTPUTS:
- *
- * s_db_ctxs: Add new DB context into the hash table
- *
- * OUTPUTS:
- * NONE
- *
- * RETURNS:
- * A has been created DB Context
- * NULL in case of error
- *
- */
-static dap_db_ctx_t *s_cre_db_ctx_for_group(const char *a_group, int a_flags)
-{
-int l_rc;
-dap_db_ctx_t *l_db_ctx, *l_db_ctx2;
-uint64_t l_seq;
-MDBX_val l_key_iov, l_data_iov;
-
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Init group/table '%s', flags: %#x ...", a_group, a_flags);
-
-
- dap_assert( !pthread_rwlock_rdlock(&s_db_ctxs_rwlock) ); /* Get RD lock for lookup only */
- HASH_FIND_STR(s_db_ctxs, a_group, l_db_ctx); /* Is there exist context for the group ? */
- dap_assert( !pthread_rwlock_unlock(&s_db_ctxs_rwlock) );
-
- if ( l_db_ctx ) /* Found! Good job - return DB context */
- return log_it(L_INFO, "Found DB context: %p for group: '%s'", l_db_ctx, a_group), l_db_ctx;
-
-
- /* So , at this point we are going to create (if not exist) 'table' for new group */
-
- if ( (l_rc = strlen(a_group)) >(int) DAP_GLOBAL_DB_GROUP_NAME_SIZE_MAX ) /* Check length of the group name */
- return log_it(L_ERROR, "Group name '%s' is too long (%d>%lu)", a_group, l_rc, DAP_GLOBAL_DB_GROUP_NAME_SIZE_MAX), NULL;
-
- if ( !(l_db_ctx = DAP_NEW_Z(dap_db_ctx_t)) ) /* Allocate zeroed memory for new DB context */
- return log_it(L_ERROR, "Cannot allocate DB context for '%s', errno=%d", a_group, errno), NULL;
-
- memcpy(l_db_ctx->name, a_group, l_db_ctx->namelen = l_rc); /* Store group name in the DB context */
- dap_assert ( !pthread_mutex_init(&l_db_ctx->dbi_mutex, NULL));
-
- /*
- ** Start transaction, create table, commit.
- */
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, 0, &l_db_ctx->txn)) )
- return log_it(L_CRITICAL, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), NULL;
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_dbi_open(l_db_ctx->txn, a_group, a_flags, &l_db_ctx->dbi)) )
- return log_it(L_CRITICAL, "mdbx_dbi_open: (%d) %s", l_rc, mdbx_strerror(l_rc)), NULL;
-
- /* MDBX sequence is started from zero, zero is not so good for our case,
- * so we just increment a current (may be is not zero) sequence for <dbi>
- */
- mdbx_dbi_sequence (l_db_ctx->txn, l_db_ctx->dbi, &l_seq, 1);
-
- /*
- * Save new subDB name into the master table
- */
- l_data_iov.iov_base = l_key_iov.iov_base = l_db_ctx->name;
- l_data_iov.iov_len = l_key_iov.iov_len = l_db_ctx->namelen;
- l_data_iov.iov_len += 1; /* Count '\0' */
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_put(l_db_ctx->txn, s_db_master_dbi, &l_key_iov, &l_data_iov, MDBX_NOOVERWRITE ))
- && (l_rc != MDBX_KEYEXIST) )
- {
- log_it (L_ERROR, "mdbx_put: (%d) %s", l_rc, mdbx_strerror(l_rc));
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_abort(l_db_ctx->txn)) )
- return log_it(L_CRITICAL, "mdbx_txn_abort: (%d) %s", l_rc, mdbx_strerror(l_rc)), NULL;
- }
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_commit(l_db_ctx->txn)) )
- return log_it(L_CRITICAL, "mdbx_txn_commit: (%d) %s", l_rc, mdbx_strerror(l_rc)), NULL;
-
- /*
- ** Add new DB Context for the group into the hash for quick access
- */
- dap_assert( !pthread_rwlock_wrlock(&s_db_ctxs_rwlock) ); /* Get WR lock for the hash-table */
-
- l_db_ctx2 = NULL;
- HASH_FIND_STR(s_db_ctxs, a_group, l_db_ctx2); /* Check for existence of group again!!! */
-
- if ( !l_db_ctx2) /* Still not exist - fine, add new record */
- HASH_ADD_KEYPTR(hh, s_db_ctxs, l_db_ctx->name, l_db_ctx->namelen, l_db_ctx);
-
- dap_assert( !pthread_rwlock_unlock(&s_db_ctxs_rwlock) );
-
- if ( l_db_ctx2 ) /* Release unnecessary new context */
- DAP_DEL_Z(l_db_ctx);
-
- return l_db_ctx2 ? l_db_ctx2 : l_db_ctx;
-}
-
-
-
-/*
- * DESCRIPTION: Action routine - cleanup this module's internal contexts, DB contexts hash table,
- * close MDBX context. After call this routine any DB operation of this module is impossible.
- * You must/can perfroms initialization.
- *
- * INPUTS:
- * NONE
- *
- * OUTPUTS:
- * NONE
- *
- * RETURNS:
- * 0 - SUCCESS
- */
-
-static int s_db_mdbx_deinit(void)
-{
- dap_db_ctx_t *l_db_ctx = NULL, *l_tmp;
-
- dap_assert ( !pthread_rwlock_wrlock(&s_db_ctxs_rwlock) ); /* Prelock for WR */
-
- HASH_ITER(hh, s_db_ctxs, l_db_ctx, l_tmp) /* run over the hash table of the DB contexts */
- {
-
- dap_assert( !pthread_mutex_lock(&l_db_ctx->dbi_mutex) );
- if (l_db_ctx->txn) /* Commit, close table */
- mdbx_txn_commit(l_db_ctx->txn);
-
- if (l_db_ctx->dbi)
- mdbx_dbi_close(s_mdbx_env, l_db_ctx->dbi);
-
- dap_assert( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- HASH_DEL(s_db_ctxs, l_db_ctx); /* Delete DB context from the hash-table */
- DAP_DELETE(l_db_ctx); /* Release memory of DB context area */
- }
-
- if (s_mdbx_env)
- mdbx_env_close(s_mdbx_env); /* Finaly close MDBX DB */
-
- dap_assert ( !pthread_rwlock_unlock(&s_db_ctxs_rwlock) );
-
- return 0;
-}
-
-/*
- * DESCRIPTION: Performs an initial module internal context creation and setup,
- * Fill dispatch procedure table (a_drv_callback) by entries of this module;
- * Create MDBX data files on the specified path, open MDBX context area;
- * Load from the MDBX$MASTER table names of groups - create DB context
- *
- * This is a first routine before any other calls of action routines in this module !!!
- *
- *
- * INPUTS:
- * a_mdbx_path: A root directory for the MDBX database files
- * a_drv_callback
- *
- * IMPLICITE OUTPUTS:
- * s_mdbx_env
- *
- * RETURNS:
- * 0 - SUCCESS
- * 0> - <errno>
- */
-
-int dap_db_driver_mdbx_init(const char *a_mdbx_path, dap_db_driver_callbacks_t *a_drv_dpt)
-{
-int l_rc;
-MDBX_txn *l_txn;
-MDBX_cursor *l_cursor;
-MDBX_val l_key_iov, l_data_iov;
-dap_slist_t l_slist = {0};
-char *l_cp;
-size_t l_upper_limit_of_db_size = 16;
-
- /*
- * [resources]
- * mdbx_upper_limit_of_db_size=32
- */
- l_upper_limit_of_db_size = dap_config_get_item_uint32_default ( g_config, "resources", "mdbx_upper_limit_of_db_size", l_upper_limit_of_db_size);
- l_upper_limit_of_db_size *= 1024*1024*1024ULL;
- log_it(L_INFO, "Set MDBX Upper Limit of DB Size to %zu octets", l_upper_limit_of_db_size);
-
- snprintf(s_db_path, sizeof(s_db_path), "%s/%s", a_mdbx_path, s_subdir );/* Make a path to MDBX root */
- dap_mkdir_with_parents(s_db_path); /* Create directory for the MDBX storage */
-
- log_it(L_NOTICE, "Directory '%s' will be used as an location for MDBX database files", s_db_path);
- s_mdbx_env = NULL;
- if ( MDBX_SUCCESS != (l_rc = mdbx_env_create(&s_mdbx_env)) )
- return log_it(L_CRITICAL, "mdbx_env_create: (%d) %s", l_rc, mdbx_strerror(l_rc)), -ENOENT;
-
-#if 0
- if ( g_dap_global_db_debug_more )
- mdbx_setup_debug ( MDBX_LOG_VERBOSE, 0, 0);
-#endif
-
-
- log_it(L_NOTICE, "Set maximum number of local groups: %lu", DAP_GLOBAL_DB_GROUPS_COUNT_MAX);
- dap_assert ( !(l_rc = mdbx_env_set_maxdbs (s_mdbx_env, DAP_GLOBAL_DB_GROUPS_COUNT_MAX)) );/* Set maximum number of the file-tables (MDBX subDB)
- according to number of supported groups */
-
-
- /* We set "unlim" for all MDBX characteristics at the moment */
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_env_set_geometry(s_mdbx_env, -1, -1, l_upper_limit_of_db_size, -1, -1, -1)) )
- return log_it (L_CRITICAL, "mdbx_env_set_geometry (%s): (%d) %s", s_db_path, l_rc, mdbx_strerror(l_rc)), -EINVAL;
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_env_open(s_mdbx_env, s_db_path, MDBX_CREATE | MDBX_COALESCE | MDBX_LIFORECLAIM, 0664)) )
- return log_it (L_CRITICAL, "mdbx_env_open (%s): (%d) %s", s_db_path, l_rc, mdbx_strerror(l_rc)), -EINVAL;
-
- /*
- * Since MDBX don't maintain a list of subDB with public API, we will use a "MASTER Table",
- * be advised that this MASTER teble is not maintained accurately!!!
- *
- * So, Create (If)/Open a master DB (table) to keep list of subDBs (group/table/subDB name)
- */
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, 0, &l_txn)) )
- return log_it(L_CRITICAL, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), -EIO;
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_dbi_open(l_txn, s_db_master_tbl, MDBX_CREATE, &s_db_master_dbi)) )
- return log_it(L_CRITICAL, "mdbx_dbi_open: (%d) %s", l_rc, mdbx_strerror(l_rc)), -EIO;
-
- dap_assert ( MDBX_SUCCESS == (l_rc = mdbx_txn_commit (l_txn)) );
-
- /*
- * Run over records in the MASTER table to get subDB names
- */
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, MDBX_TXN_RDONLY, &l_txn)) )
- log_it(L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc));
- else if ( MDBX_SUCCESS != (l_rc = mdbx_cursor_open(l_txn, s_db_master_dbi, &l_cursor)) )
- log_it(L_ERROR, "mdbx_cursor_open: (%d) %s", l_rc, mdbx_strerror(l_rc));
- else{
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "--- List of stored groups ---");
-
- for ( int i = 0; !(l_rc = mdbx_cursor_get (l_cursor, &l_key_iov, &l_data_iov, MDBX_NEXT )); i++ )
- {
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "MDBX SubDB #%03d [0:%zu]: '%.*s' = [0:%zu]: '%.*s'", i,
- l_key_iov.iov_len, (int) l_key_iov.iov_len, (char *) l_key_iov.iov_base,
- l_data_iov.iov_len, (int) l_data_iov.iov_len, (char *) l_data_iov.iov_base);
-
- /* Form a simple list of the group/table name to be used after */
- l_cp = dap_strdup(l_data_iov.iov_base); /* We expect an ASCIZ string as the table name */
- l_data_iov.iov_len = strlen(l_cp);
- dap_slist_add2tail(&l_slist, l_cp, l_data_iov.iov_len);
- }
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "--- End-Of-List ---");
- }
-
- dap_assert ( MDBX_SUCCESS == mdbx_txn_commit (l_txn) );
-
-
- /* Run over the list and create/open group/tables and DB context ... */
- while ( !dap_slist_get4head (&l_slist, &l_data_iov.iov_base, &l_data_iov.iov_len) )
- {
- s_cre_db_ctx_for_group(l_data_iov.iov_base, MDBX_CREATE);
- DAP_DELETE(l_data_iov.iov_base);
- }
-
- /*
- ** Fill the Driver Interface Table
- */
- a_drv_dpt->apply_store_obj = s_db_mdbx_apply_store_obj;
- a_drv_dpt->read_last_store_obj = s_db_mdbx_read_last_store_obj;
-
- a_drv_dpt->read_store_obj = s_db_mdbx_read_store_obj;
- a_drv_dpt->read_cond_store_obj = s_db_mdbx_read_cond_store_obj;
- a_drv_dpt->read_count_store = s_db_mdbx_read_count_store;
- a_drv_dpt->get_groups_by_mask = s_db_mdbx_get_groups_by_mask;
- a_drv_dpt->is_obj = s_db_mdbx_is_obj;
- a_drv_dpt->deinit = s_db_mdbx_deinit;
- a_drv_dpt->flush = s_db_mdbx_flush;
-
- /*
- * MDBX support transactions but on the current circuimstance we will not get
- * advantages of using DB Driver level BEGIN/END transactions
- */
- a_drv_dpt->transaction_start = NULL;
- a_drv_dpt->transaction_end = NULL;
-
- return MDBX_SUCCESS;
-}
-
-/*
- * DESCRIPTION: Get a DB context for the specified group/table name
- * from the DB context hash table. This context is just pointer to the DB Context
- * structure, so don't modify it.
- *
- * INPUTS:
- * a_group: Group/table name to be looked for DB context
- *
- * OUTPUTS:
- * NONE
- *
- * RETURNS
- * address of DB Context
- * NULL - no DB context has been craeted for the group
- *
- */
-static dap_db_ctx_t *s_get_db_ctx_for_group(const char *a_group)
-{
-dap_db_ctx_t *l_db_ctx = NULL;
-
- dap_assert ( !pthread_rwlock_rdlock(&s_db_ctxs_rwlock) );
- HASH_FIND_STR(s_db_ctxs, a_group, l_db_ctx);
- dap_assert ( !pthread_rwlock_unlock(&s_db_ctxs_rwlock) );
-
- if ( !l_db_ctx )
- debug_if(g_dap_global_db_debug_more, L_WARNING, "No DB context for the group '%s'", a_group);
-
- return l_db_ctx;
-}
-
-/*
- * DESCRIPTION: Action routine - perform flushing action. Actualy MDBX internaly maintain processes of the flushing
- * and other things related to data integrity.
- *
- * INPUTS:
- * NONE
- *
- * OUTPUTS:
- * NONE
- *
- * RETURNS:
- * 0 - SUCCESS
- */
-static int s_db_mdbx_flush(void)
-{
- return log_it(L_DEBUG, "Flushing resident part of the MDBX to disk"), 0;
-}
-
-/*
- * DESCRIPTION: Action routine - lookup in the group/table a last stored record (with the bigest Id).
- * We mainatain internaly <id> of record (it's just sequence),
- * so actualy we need to performs a full scan of the table to reach a record with the bigest <id>.
- * In case of success create and return <store_object> for the has been found records.
- *
- * INPUTS:
- * a_group: A group/table name to be scanned
- *
- * OUTPUTS:
- * NONE
- *
- * RETURNS:
- * An address to the <store object> with the record
- * NULL - table is empty
- */
-dap_store_obj_t *s_db_mdbx_read_last_store_obj(const char* a_group)
-{
-int l_rc;
-dap_db_ctx_t *l_db_ctx;
-MDBX_val l_key={0}, l_data={0}, l_last_data={0}, l_last_key={0};
-MDBX_cursor *l_cursor = NULL;
-struct __record_suffix__ *l_suff;
-uint64_t l_id;
-dap_store_obj_t *l_obj;
-
- if (!a_group) /* Sanity check */
- return NULL;
-
- if ( !(l_db_ctx = s_get_db_ctx_for_group(a_group)) ) /* Get DB Context for group/table */
- return NULL;
-
- dap_assert ( !pthread_mutex_lock(&l_db_ctx->dbi_mutex) );
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, MDBX_TXN_RDONLY, &l_db_ctx->txn)) )
- {
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return log_it (L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), NULL;
- }
-
- do {
- l_cursor = NULL;
- l_id = 0;
- l_last_key = l_last_data = (MDBX_val) {0, 0};
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_cursor_open(l_db_ctx->txn, l_db_ctx->dbi, &l_cursor)) ) {
- log_it (L_ERROR, "mdbx_cursor_open: (%d) %s", l_rc, mdbx_strerror(l_rc));
- break;
- }
-
- /* Iterate cursor to retrieve records from DB - select a <key> and <data> pair
- ** with maximal <id>
- */
- while ( MDBX_SUCCESS == (l_rc = mdbx_cursor_get(l_cursor, &l_key, &l_data, MDBX_NEXT)) )
- {
- l_suff = (struct __record_suffix__ *) (l_data.iov_base + l_data.iov_len - sizeof(struct __record_suffix__));
- if ( l_id < l_suff->id )
- {
- l_id = l_suff->id;
- l_last_key = l_key; /* <l_last_key> point to real key area in the MDBX DB */
- l_last_data = l_data; /* <l_last_data> point to real data area in the MDBX DB */
- }
- }
-
- } while (0);
-
- if (l_cursor) /* Release uncesessary MDBX cursor area,
- but keep transaction !!! */
- mdbx_cursor_close(l_cursor);
-
-
-
-
-
- if ( !(l_last_key.iov_len || l_data.iov_len) ) /* Not found anything - return NULL */
- {
- mdbx_txn_commit(l_db_ctx->txn);
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return NULL;
- }
-
- /* Found ! Allocate memory for <store object> and <value> */
- if ( (l_obj = DAP_CALLOC(1, sizeof( dap_store_obj_t ))) )
- {
-
- if ( !(l_obj->key = DAP_CALLOC(1, l_last_key.iov_len + 1)) )
- l_rc = MDBX_PROBLEM, log_it (L_ERROR, "Cannot allocate a memory for store object key, errno=%d", errno);
- else if ( (l_obj->value = DAP_CALLOC(1, (l_data.iov_len + 1) - sizeof(struct __record_suffix__))) )
- {
-
- /* Fill the <store obj> by data from the retrieved record */
- l_obj->key_len = l_key.iov_len;
- memcpy((char *) l_obj->key, l_key.iov_base, l_obj->key_len);
-
- l_obj->value_len = l_data.iov_len - sizeof(struct __record_suffix__);
- if(l_last_data.iov_base)
- memcpy(l_obj->value, l_last_data.iov_base, l_obj->value_len);
-
- l_suff = (struct __record_suffix__ *) (l_data.iov_base + l_obj->value_len);
- if(l_suff) {
- l_obj->id = l_suff->id;
- l_obj->timestamp = l_suff->ts;
- l_obj->flags = l_suff->flags;
- }
- dap_assert ( (l_obj->group = dap_strdup(a_group)) );
- }
- else l_rc = MDBX_PROBLEM, log_it (L_ERROR, "Cannot allocate a memory for store object value, errno=%d", errno);
- }
- else l_rc = MDBX_PROBLEM, log_it (L_ERROR, "Cannot allocate a memory for store object, errno=%d", errno);
-
- mdbx_txn_commit(l_db_ctx->txn);
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return l_obj;
-}
-
-/*
- * DESCRIPTION: An action routine to check a presence specified key in the group/table
- *
- * INPUTS:
- * a_group: A group/table to looking in
- * a_key: A key of record to looked for
- *
- * OUTPUTS:
- * NONE
- *
- * RETURNS
- * 1 - SUCCESS, record is exist
- * 0 - Record-No-Found
- */
-bool s_db_mdbx_is_obj(const char *a_group, const char *a_key)
-{
-int l_rc, l_rc2;
-dap_db_ctx_t *l_db_ctx;
-MDBX_val l_key, l_data;
-
- if (!a_group || !a_key) /* Sanity check */
- return 0;
-
- if ( !(l_db_ctx = s_get_db_ctx_for_group(a_group)) ) /* Get DB Context for group/table */
- return 0;
-
- dap_assert ( !pthread_mutex_lock(&l_db_ctx->dbi_mutex) );
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, MDBX_TXN_RDONLY, &l_db_ctx->txn)) )
- {
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return log_it (L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), 0;
- }
-
- l_key.iov_base = (void *) a_key; /* Fill IOV for MDBX key */
- l_key.iov_len = strlen(a_key);
-
- l_rc = mdbx_get(l_db_ctx->txn, l_db_ctx->dbi, &l_key, &l_data);
-
- if ( MDBX_SUCCESS != (l_rc2 = mdbx_txn_commit(l_db_ctx->txn)) )
- log_it (L_ERROR, "mdbx_txn_commit: (%d) %s", l_rc2, mdbx_strerror(l_rc2));
-
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return ( l_rc == MDBX_SUCCESS ); /*0 - RNF, 1 - SUCCESS */
-}
-
-/*
- * DESCRIPTION: Action routine to read record with a give <id > from the table
- *
- * INPUTS:
- * a_group: A group/table name to be looked in
- * a_id: An id of record to be looked for
- *
- * OUTPUTS:
- * NONE
- *
- * RETURNS:
- * An address to the <store object> with the record
- * NULL - table is empty
- */
-
-
-int s_make_store_obj (const char *a_group,
- MDBX_val *a_key,
- MDBX_val *a_data,
- dap_store_obj_t **a_obj
- )
-{
-dap_store_obj_t *l_obj;
-size_t len;
-int rc;
-struct __record_suffix__ *l_suff;
-
- len = dap_strlen(a_group);
- rc = -1; /* ERROR */
-
- if ( (l_obj = DAP_CALLOC(1, sizeof(dap_store_obj_t))) )
- if ( (l_obj->value = DAP_CALLOC(1, (a_data->iov_len + 1) - sizeof(struct __record_suffix__))) )
- if ( (l_obj->key = DAP_CALLOC(1, a_key->iov_len + 1)) )
- if ( (l_obj->group = DAP_CALLOC(1, len + 1)) )
- {
- /* Fill the <store obj> by data from the retrieved record */
- l_obj->key_len = a_key->iov_len;
- memcpy((char *) l_obj->key, a_key->iov_base, l_obj->key_len);
-
- l_obj->value_len = a_data->iov_len - sizeof(struct __record_suffix__);
- memcpy(l_obj->value, a_data->iov_base, l_obj->value_len);
-
- l_obj->group_len = len;
- memcpy(l_obj->group, a_group, l_obj->group_len);
-
- l_suff = (struct __record_suffix__ *) (a_data->iov_base + l_obj->value_len);
- l_obj->id = l_suff->id;
- l_obj->timestamp = l_suff->ts;
- l_obj->flags = l_suff->flags;
-
- rc = 0; /* SUCCESS */
- }
- else log_it (L_ERROR, "Cannot allocate a memory for store object group, errno=%d", errno);
- else log_it (L_ERROR, "Cannot allocate a memory for store object key, errno=%d", errno);
- else log_it (L_ERROR, "Cannot allocate a memory for store object value, errno=%d", errno);
- else log_it (L_ERROR, "Cannot allocate a memory for store object, errno=%d", errno);
-
-
- if ( rc && l_obj ) /* Check final status and do cleanup on error */
- {
- DAP_DEL_Z(l_obj->group);
- DAP_DEL_Z(l_obj->key);
- DAP_DEL_Z(l_obj->value);
- DAP_DEL_Z(l_obj);
- }
-
- *a_obj = l_obj;
-
- return rc;
-}
-
-
-
-static dap_store_obj_t *s_db_mdbx_read_cond_store_obj(const char *a_group, uint64_t a_id, size_t *a_count_out)
-{
-int l_rc = 0;
-dap_db_ctx_t *l_db_ctx = NULL;
-MDBX_val l_key = {0}, l_data = {0};
-MDBX_cursor *l_cursor;
-struct __record_suffix__ *l_suff = NULL;
-dap_store_obj_t *l_obj = NULL, *l_obj_arr = NULL;
-size_t l_cnt = 0, l_count_out = 0;
-
- if (!a_group) /* Sanity check */
- return NULL;
-
- if ( !(l_db_ctx = s_get_db_ctx_for_group(a_group)) ) /* Get DB Context for group/table */
- return NULL;
-
- dap_assert ( !pthread_mutex_lock(&l_db_ctx->dbi_mutex));
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, MDBX_TXN_RDONLY, &l_db_ctx->txn)) )
- {
- dap_assert( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return log_it (L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), NULL;
- }
-
-
- /* Limit a number of objects to be returned */
- l_count_out = (a_count_out && *a_count_out) ? *a_count_out : DAP_GLOBAL_DB_MAX_OBJS;
- l_count_out = MIN(l_count_out, DAP_GLOBAL_DB_MAX_OBJS);
-
- l_cursor = NULL;
- l_obj = l_obj_arr = NULL;
-
-
- do {
- /* Initialize MDBX cursor context area */
- if ( MDBX_SUCCESS != (l_rc = mdbx_cursor_open(l_db_ctx->txn, l_db_ctx->dbi, &l_cursor)) ) {
- log_it (L_ERROR, "mdbx_cursor_open: (%d) %s", l_rc, mdbx_strerror(l_rc));
- break;
- }
-
-
- /* Iterate cursor to retrieve records from DB */
- l_cnt = 0;
- if(a_count_out)
- *a_count_out = l_cnt;
- for (int i = l_count_out; i && (MDBX_SUCCESS == (l_rc = mdbx_cursor_get(l_cursor, &l_key, &l_data, MDBX_NEXT))); i--)
- {
- l_suff = (struct __record_suffix__ *) (l_data.iov_base + l_data.iov_len - sizeof(struct __record_suffix__));
- if ( l_suff->id < a_id )
- continue;
-
- /*
- * Expand a memory for new <store object> structure
- */
- if ( !(l_obj = DAP_REALLOC(l_obj_arr, (1 + l_cnt) * sizeof(dap_store_obj_t))) )
- {
- log_it(L_ERROR, "Cannot expand area to keep %zu <store objects>", l_cnt);
- break;
- }
-
- l_cnt++;
-
- l_obj_arr = l_obj;
- l_obj = l_obj_arr + (l_cnt - 1); /* Point <l_obj> to last array's element */
- memset(l_obj, 0, sizeof(dap_store_obj_t));
-
- if ( !(l_obj->key = DAP_CALLOC(1, l_key.iov_len + 1)) )
- l_rc = MDBX_PROBLEM, log_it (L_ERROR, "Cannot allocate a memory for store object key, errno=%d", errno);
-
- else if ( (l_obj->value = DAP_CALLOC(1, (l_data.iov_len + 1) - sizeof(struct __record_suffix__))) )
- {
- /* Fill the <store obj> by data from the retreived record */
- l_obj->key_len = l_key.iov_len;
- memcpy((char *) l_obj->key, l_key.iov_base, l_obj->key_len);
-
- l_obj->value_len = l_data.iov_len - sizeof(struct __record_suffix__);
- memcpy(l_obj->value, l_data.iov_base, l_obj->value_len);
-
- l_obj->id = l_suff->id;
- l_obj->timestamp = l_suff->ts;
- l_obj->flags = l_suff->flags;
-
- dap_assert ( (l_obj->group = dap_strdup(a_group)) );
- l_obj->group_len = strlen(l_obj->group);
- }
- else l_rc = MDBX_PROBLEM, log_it (L_ERROR, "Cannot allocate a memory for store object value, errno=%d", errno);
- }
-
-
- if ( (MDBX_SUCCESS != l_rc) && (l_rc != MDBX_NOTFOUND) ) {
- log_it (L_ERROR, "mdbx_cursor_get: (%d) %s", l_rc, mdbx_strerror(l_rc)), l_rc = MDBX_SUCCESS;
- break;
- }
-
- } while (0);
-
- if (l_cursor)
- mdbx_cursor_close(l_cursor);
-
- mdbx_txn_commit(l_db_ctx->txn);
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- if(a_count_out)
- *a_count_out = l_cnt;
- return l_obj_arr;
-}
-
-/*
- * DESCRIPTION: Action routine to retrieve a number of records for specified record's id.
- *
- * INPUTS:
- * a_group: A table/group name to be scanned
- * a_id: An id of record to be looked for
- *
- * RETURNS:
- * count of has been found record
- */
-size_t s_db_mdbx_read_count_store(const char *a_group, uint64_t a_id)
-{
-int l_rc, l_count_out;
-dap_db_ctx_t *l_db_ctx;
-MDBX_val l_key, l_data;
-MDBX_cursor *l_cursor;
-struct __record_suffix__ *l_suff;
-MDBX_stat l_stat;
-
- if (!a_group) /* Sanity check */
- return 0;
-
- if ( !(l_db_ctx = s_get_db_ctx_for_group(a_group)) ) /* Get DB Context for group/table */
- return 0;
-
- dap_assert ( !pthread_mutex_lock(&l_db_ctx->dbi_mutex) );
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, MDBX_TXN_RDONLY, &l_db_ctx->txn)) )
- {
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return log_it (L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), 0;
- }
-
- if ( a_id <= 1 ) /* Retrieve a total number of records in the table */
- {
- if ( MDBX_SUCCESS != (l_rc = mdbx_dbi_stat (l_db_ctx->txn, l_db_ctx->dbi, &l_stat, sizeof(MDBX_stat))) )
- log_it (L_ERROR, "mdbx_dbi_stat: (%d) %s", l_rc, mdbx_strerror(l_rc));
-
- mdbx_txn_commit(l_db_ctx->txn);
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return ( l_rc == MDBX_SUCCESS ) ? l_stat.ms_entries : 0;
- }
-
-
-
-
- /*
- * Count a number of records with id = a_id, a_id+1 ...
- */
- l_cursor = NULL;
- l_count_out = 0;
-
- do {
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_cursor_open(l_db_ctx->txn, l_db_ctx->dbi, &l_cursor)) ) {
- log_it (L_ERROR, "mdbx_cursor_open: (%d) %s", l_rc, mdbx_strerror(l_rc));
- break;
- }
-
- /* Iterate cursor to retrieve records from DB */
- while ( MDBX_SUCCESS == (l_rc = mdbx_cursor_get(l_cursor, &l_key, &l_data, MDBX_NEXT)) ) {
- l_suff = (struct __record_suffix__ *) (l_data.iov_base + l_data.iov_len - sizeof(struct __record_suffix__));
- l_count_out += (l_suff->id >= a_id );
- }
-
- } while (0);
-
- if (l_cursor)
- mdbx_cursor_close(l_cursor);
-
- mdbx_txn_commit(l_db_ctx->txn);
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return l_count_out;
-}
-
-/*
- * DESCRIPTION: Action routine - returns a list of group/table names in DB contexts hash table is matched
- * to specified pattern.
- *
- * INPUTS:
- * a_group_mask: A pattern string
- *
- * OUTPUTS:
- * NONE
- *
- * RETURNS:
- * list of has been found groups
- */
-
-static dap_list_t *s_db_mdbx_get_groups_by_mask(const char *a_group_mask)
-{
-dap_list_t *l_ret_list;
-dap_db_ctx_t *l_db_ctx, *l_db_ctx2;
-
- if(!a_group_mask)
- return NULL;
-
- l_ret_list = NULL;
-
- dap_assert ( !pthread_rwlock_rdlock(&s_db_ctxs_rwlock) );
-
- HASH_ITER(hh, s_db_ctxs, l_db_ctx, l_db_ctx2) {
- if (!dap_fnmatch(a_group_mask, l_db_ctx->name, 0) ) /* Name match a pattern/mask ? */
- l_ret_list = dap_list_prepend(l_ret_list, dap_strdup(l_db_ctx->name)); /* Add group name to output list */
- }
-
- dap_assert ( !pthread_rwlock_unlock(&s_db_ctxs_rwlock) );
-
- return l_ret_list;
-}
-
-
-/*
- * DESCRIPTION: Action routine - insert/delete a record with data from the <store_object> to/from database.
- * Be advised that we performs a transaction processing to ensure DB consistency
- *
- * INPUTS:
- * a_store_obj: An object with data to be stored
- *
- * OUTPUTS:
- * NONE:
- *
- * RETURNS:
- * 0 - SUCCESS
- * 0> - <errno>
- */
-static int s_db_mdbx_apply_store_obj (dap_store_obj_t *a_store_obj)
-{
-int l_rc = 0, l_rc2;
-dap_db_ctx_t *l_db_ctx;
-MDBX_val l_key, l_data;
-char *l_val;
-struct __record_suffix__ *l_suff;
-
- if ( !a_store_obj || !a_store_obj->group) /* Sanity checks ... */
- return -EINVAL;
-
-
-
- if ( !(l_db_ctx = s_get_db_ctx_for_group(a_store_obj->group)) ) { /* Get a DB context for the group */
- /* Group is not found ? Try to create table for new group */
- if ( !(l_db_ctx = s_cre_db_ctx_for_group(a_store_obj->group, MDBX_CREATE)) )
- return log_it(L_WARNING, "Cannot create DB context for the group '%s'", a_store_obj->group), -EIO;
-
- log_it(L_NOTICE, "DB context for the group '%s' has been created", a_store_obj->group);
-
- if ( a_store_obj->type == DAP_DB$K_OPTYPE_DEL ) /* Nothing to do anymore */
- return 0;
- }
-
-
- /* At this point we have got the DB Context for the table/group
- * so we are can performs a main work
- */
-
- dap_assert ( !pthread_mutex_lock(&l_db_ctx->dbi_mutex) );
-
-
- if (a_store_obj->type == DAP_DB$K_OPTYPE_ADD ) {
- if( !a_store_obj->key )
- {
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return -ENOENT;
- }
-
- l_key.iov_base = (void *) a_store_obj->key; /* Fill IOV for MDBX key */
- l_key.iov_len = a_store_obj->key_len ? a_store_obj->key_len : strnlen(a_store_obj->key, DAP_GLOBAL_DB_KEY_MAX);
-
- /*
- * Now we are ready to form a record in next format:
- * <value> + <suffix>
- */
- l_rc = a_store_obj->value_len + sizeof(struct __record_suffix__); /* Compute a length of the area to keep value+suffix */
-
- if ( !(l_val = DAP_NEW_Z_SIZE(char, l_rc)) )
- {
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return log_it(L_ERROR, "Cannot allocate memory for new records, %d octets, errno=%d", l_rc, errno), -errno;
- }
-
- l_data.iov_base = l_val; /* Fill IOV for MDBX data */
- l_data.iov_len = l_rc;
-
- /*
- * Fill suffix's fields
- */
- l_suff = (struct __record_suffix__ *) (l_val + a_store_obj->value_len);
- l_suff->flags = a_store_obj->flags;
- l_suff->ts = a_store_obj->timestamp;
-
- memcpy(l_val, a_store_obj->value, a_store_obj->value_len); /* Put <value> into the record */
-
- /* So, finaly: BEGIN transaction, do INSERT, COMMIT or ABORT ... */
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, 0, &l_db_ctx->txn)) )
- {
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return DAP_FREE(l_val), log_it (L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), -EIO;
- }
-
-
- /* Generate <sequence number> for new record */
- if ( MDBX_SUCCESS != mdbx_dbi_sequence (l_db_ctx->txn, l_db_ctx->dbi, &l_suff->id, 1) )
- {
- log_it (L_CRITICAL, "mdbx_dbi_sequence: (%d) %s", l_rc, mdbx_strerror(l_rc));
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_abort(l_db_ctx->txn)) )
- log_it (L_ERROR, "mdbx_txn_abort: (%d) %s", l_rc, mdbx_strerror(l_rc));
-
- mdbx_txn_abort(l_db_ctx->txn);
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return DAP_FREE(l_val), -EIO;
- }
-
-
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_put(l_db_ctx->txn, l_db_ctx->dbi, &l_key, &l_data, 0)) )
- {
- log_it (L_ERROR, "mdbx_put: (%d) %s", l_rc, mdbx_strerror(l_rc));
-
- if ( MDBX_SUCCESS != (l_rc2 = mdbx_txn_abort(l_db_ctx->txn)) )
- log_it (L_ERROR, "mdbx_txn_abort: (%d) %s", l_rc2, mdbx_strerror(l_rc2));
- }
- else if ( MDBX_SUCCESS != (l_rc = mdbx_txn_commit(l_db_ctx->txn)) )
- log_it (L_ERROR, "mdbx_txn_commit: (%d) %s", l_rc, mdbx_strerror(l_rc));
-
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return DAP_FREE(l_val), (( l_rc == MDBX_SUCCESS ) ? 0 : -EIO);
- } /* DAP_DB$K_OPTYPE_ADD */
-
-
-
- if (a_store_obj->type == DAP_DB$K_OPTYPE_DEL) {
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, 0, &l_db_ctx->txn)) )
- {
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return log_it (L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), -ENOENT;
- }
-
- l_rc2 = 0;
-
- if ( a_store_obj->key ) { /* Delete record */
- l_key.iov_base = (void *) a_store_obj->key;
- l_key.iov_len = a_store_obj->key_len ? a_store_obj->key_len : strnlen(a_store_obj->key, DAP_GLOBAL_DB_KEY_MAX);
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_del(l_db_ctx->txn, l_db_ctx->dbi, &l_key, NULL))
- && ( l_rc != MDBX_NOTFOUND) )
- l_rc2 = -EIO, log_it (L_ERROR, "mdbx_del: (%d) %s", l_rc, mdbx_strerror(l_rc));
- }
- else { /* Truncate only table */
- if ( MDBX_SUCCESS != (l_rc = mdbx_drop(l_db_ctx->txn, l_db_ctx->dbi, 0))
- && ( l_rc != MDBX_NOTFOUND) )
- l_rc2 = -EIO, log_it (L_ERROR, "mdbx_drop: (%d) %s", l_rc, mdbx_strerror(l_rc));
- }
-
-
- l_rc = (l_rc == MDBX_NOTFOUND) ? MDBX_SUCCESS : l_rc; /* Not found ?! It's Okay !!! */
-
-
-
- if ( l_rc != MDBX_SUCCESS ) { /* Check result of mdbx_drop/del */
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_abort(l_db_ctx->txn)) )
- l_rc2 = -EIO, log_it (L_ERROR, "mdbx_txn_abort: (%d) %s", l_rc, mdbx_strerror(l_rc));
- }
- else if ( MDBX_SUCCESS != (l_rc = mdbx_txn_commit(l_db_ctx->txn)) )
- l_rc2 = -EIO, log_it (L_ERROR, "mdbx_txn_commit: (%d) %s", l_rc, mdbx_strerror(l_rc));
-
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return ( l_rc2 == MDBX_SUCCESS ) ? 0 : -EIO;
- } /* DAP_DB$K_OPTYPE_DEL */
-
-
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- log_it (L_ERROR, "Unhandle/unknown DB opcode (%d/%#x)", a_store_obj->type, a_store_obj->type);
-
- return -EIO;
-}
-
-
-
-
-
-
-
-
-/*
- * DESCRIPTION: Action routine - retrieve from specified group/table a record with the given key,
- * theoreticaly we can return a set of records - but actualy we don't allow dupplicates in the DB,
- * so count of records is always == 1.
- *
- * INPUTS:
- * a_group: A group/table name to lokkind in
- * a_key: A key's record to looked for
- *
- * OUTPUTS:
- * a_count_out
- *
- * RETURNS
- * Array of store objects
- */
-
-static dap_store_obj_t *s_db_mdbx_read_store_obj(const char *a_group, const char *a_key, size_t *a_count_out)
-{
-int l_rc, l_rc2;
-uint64_t l_count_out;
-dap_db_ctx_t *l_db_ctx;
-dap_store_obj_t *l_obj, *l_obj_arr;
-MDBX_val l_key, l_data;
-MDBX_cursor *l_cursor;
-MDBX_stat l_stat;
-struct __record_suffix__ *l_suff;
-
-
- if (!a_group) /* Sanity check */
- return NULL;
-
- if ( !(l_db_ctx = s_get_db_ctx_for_group(a_group)) ) /* Get DB Context for group/table */
- return NULL;
-
-
- dap_assert ( !pthread_mutex_lock(&l_db_ctx->dbi_mutex) );
-
- if ( MDBX_SUCCESS != (l_rc = mdbx_txn_begin(s_mdbx_env, NULL, MDBX_TXN_RDONLY, &l_db_ctx->txn)) )
- {
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
- return log_it (L_ERROR, "mdbx_txn_begin: (%d) %s", l_rc, mdbx_strerror(l_rc)), NULL;
- }
-
-
- if ( a_count_out )
- *a_count_out = 0;
-
- /*
- * Perfroms a find/get a record with the given key
- */
- if ( a_key )
- {
- l_key.iov_base = (void *) a_key; /* Fill IOV for MDBX key */
- l_key.iov_len = strlen(a_key);
- l_obj = NULL;
-
- if ( MDBX_SUCCESS == (l_rc = mdbx_get(l_db_ctx->txn, l_db_ctx->dbi, &l_key, &l_data)) )
- {
- /* Found ! Make new <store_obj> */
- if ( !s_make_store_obj (a_group, &l_key, &l_data, &l_obj) )
- {
- if ( a_count_out )
- *a_count_out = 1;
- }
- } else if ( l_rc != MDBX_NOTFOUND )
- log_it (L_ERROR, "mdbx_get: (%d) %s", l_rc, mdbx_strerror(l_rc));
-
-
- mdbx_txn_commit(l_db_ctx->txn);
- dap_assert ( !pthread_mutex_unlock(&l_db_ctx->dbi_mutex) );
-
- return ( l_rc == MDBX_SUCCESS ) ? l_obj : NULL;
- }
-
-
- /*
- ** If a_key is NULL - retrieve a requested number of records from the table
- */
- do {
- l_count_out = (a_count_out && *a_count_out)? *a_count_out : DAP_GLOBAL_DB_MAX_OBJS;/* Limit a number of objects to be returned */
- l_cursor = NULL;
- l_obj = l_obj_arr = NULL;
-
- /*
- * Retrieve statistic for group/table, we need to compute a number of records can be retreived
- */
- l_rc2 = 0;
- if ( MDBX_SUCCESS != (l_rc = mdbx_dbi_stat (l_db_ctx->txn, l_db_ctx->dbi, &l_stat, sizeof(MDBX_stat))) ) {
- log_it (L_ERROR, "mdbx_dbi_stat: (%d) %s", l_rc2, mdbx_strerror(l_rc2));
- break;
- }
- else if ( !l_stat.ms_entries ) /* Nothing to retrieve , table contains no record */
- break;
-
- if ( !( l_count_out = min(l_stat.ms_entries, l_count_out)) ) {
- debug_if(g_dap_global_db_debug_more, L_WARNING, "No object (-s) to be retrieved from the group '%s'", a_group);
- break;
- }
-
- /*
- * Allocate memory for array[l_count_out] of returned objects
- */
- l_rc2 = l_count_out * sizeof(dap_store_obj_t);
- if ( !(l_obj_arr = (dap_store_obj_t *) DAP_NEW_Z_SIZE(char, l_rc2)) ) {
- log_it(L_ERROR, "Cannot allocate %zu octets for %"DAP_UINT64_FORMAT_U" store objects", l_count_out * sizeof(dap_store_obj_t), l_count_out);
- break;
- }
-
- /* Initialize MDBX cursor context area */
- if ( MDBX_SUCCESS != (l_rc = mdbx_cursor_open(l_db_ctx->txn, l_db_ctx->dbi, &l_cursor)) ) {
- log_it (L_ERROR, "mdbx_cursor_open: (%d) %s", l_rc, mdbx_strerror(l_rc));
- break;
- }
-
-
- /* Iterate cursor to retrieve records from DB */
- l_obj = l_obj_arr;
- for (int i = l_count_out;
- i && (MDBX_SUCCESS == (l_rc = mdbx_cursor_get(l_cursor, &l_key, &l_data, MDBX_NEXT))); i--, l_obj++)
- {
- if ( !(l_obj->key = DAP_CALLOC(1, l_key.iov_len + 1)) )
- l_rc = MDBX_PROBLEM, log_it (L_ERROR, "Cannot allocate a memory for store object key, errno=%d", errno);
-
- else if ( (l_obj->value = DAP_CALLOC(1, (l_data.iov_len + 1) - sizeof(struct __record_suffix__))) )
- {
- /* Fill the <store obj> by data from the retreived record */
- l_obj->key_len = l_key.iov_len + 1;
- memcpy((char *) l_obj->key, l_key.iov_base, l_obj->key_len);
-
- l_obj->value_len = l_data.iov_len - sizeof(struct __record_suffix__);
- memcpy(l_obj->value, l_data.iov_base, l_obj->value_len);
-
- l_suff = (struct __record_suffix__ *) (l_data.iov_base + l_obj->value_len);
- l_obj->id = l_suff->id;
- l_obj->timestamp = l_suff->ts;
- l_obj->flags = l_suff->flags;
-
- dap_assert ( (l_obj->group = dap_strdup(a_group)) );
- l_obj->group_len = strlen(l_obj->group);
-
- if ( a_count_out )
- *a_count_out += 1;
- }
- else l_rc = MDBX_PROBLEM, log_it (L_ERROR, "Cannot allocate a memory for store object value, errno=%d", errno);
- }
-
-
- if ( (MDBX_SUCCESS != l_rc) && (l_rc != MDBX_NOTFOUND) ) {
- log_it (L_ERROR, "mdbx_cursor_get: (%d) %s", l_rc, mdbx_strerror(l_rc)), l_rc = MDBX_SUCCESS;
- break;
- }
-
- } while (0);
-
-
- if (l_cursor)
- mdbx_cursor_close(l_cursor);
-
- mdbx_txn_commit(l_db_ctx->txn);
- pthread_mutex_unlock(&l_db_ctx->dbi_mutex);
-
- return l_obj_arr;
-}
diff --git a/modules/global-db/dap_chain_global_db_driver_pgsql.c b/modules/global-db/dap_chain_global_db_driver_pgsql.c
deleted file mode 100644
index b3e8dbfde60d67f9648ae5dce4c52ff24ec28ea2..0000000000000000000000000000000000000000
--- a/modules/global-db/dap_chain_global_db_driver_pgsql.c
+++ /dev/null
@@ -1,644 +0,0 @@
-/*
- * Authors:
- * Roman Khlopkov <roman.khlopkov@demlabs.net>
- * DeM Labs Inc. https://demlabs.net
- * CellFrame https://cellframe.net
- * Sources https://gitlab.demlabs.net/cellframe
- * Copyright (c) 2017-2021
- * All rights reserved.
-
- This file is part of CellFrame SDK the open source project
-
- CellFrame SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- CellFrame SDK is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any CellFrame SDK based project. If not, see <http://www.gnu.org/licenses/>.
-*/
-
-#include <stddef.h>
-#include <string.h>
-#include <pthread.h>
-#include <errno.h>
-#include <fcntl.h>
-#include <unistd.h>
-#include <sys/stat.h>
-
-#include "dap_chain_global_db_driver_pgsql.h"
-#include "dap_common.h"
-#include "dap_hash.h"
-#include "dap_file_utils.h"
-#include "dap_strfuncs.h"
-#include "dap_file_utils.h"
-
-#ifdef DAP_CHAIN_GDB_ENGINE_PGSQL
-#include <pwd.h>
-
-#define LOG_TAG "db_pgsql"
-
-struct dap_pgsql_conn_pool_item {
- PGconn *conn;
- int busy;
-};
-
-static PGconn *s_trans_conn = NULL;
-static struct dap_pgsql_conn_pool_item s_conn_pool[DAP_PGSQL_POOL_COUNT];
-static pthread_rwlock_t s_db_rwlock = PTHREAD_RWLOCK_INITIALIZER;
-
-static PGconn *s_pgsql_get_connection(void)
-{
- if (pthread_rwlock_wrlock(&s_db_rwlock) == EDEADLK) {
- return s_trans_conn;
- }
- PGconn *l_ret = NULL;
- for (int i = 0; i < DAP_PGSQL_POOL_COUNT; i++) {
- if (!s_conn_pool[i].busy) {
- l_ret = s_conn_pool[i].conn;
- s_conn_pool[i].busy = 1;
- break;
- }
- }
- pthread_rwlock_unlock(&s_db_rwlock);
- return l_ret;
-}
-
-static void s_pgsql_free_connection(PGconn *a_conn)
-{
- if (pthread_rwlock_wrlock(&s_db_rwlock) == EDEADLK) {
- return;
- }
- for (int i = 0; i < DAP_PGSQL_POOL_COUNT; i++) {
- if (s_conn_pool[i].conn == a_conn) {
- s_conn_pool[i].busy = 0;
- break;
- }
- }
- pthread_rwlock_unlock(&s_db_rwlock);
-}
-
-/**
- * @brief Initializes a PostgreSQL database.
- * @param a_filename_dir a path to the database file
- * @param a_drv_callback a pointer to a structure of callback functions
- * @return If successful returns 0, else a error code <0.
- */
-int dap_db_driver_pgsql_init(const char *a_filename_dir, dap_db_driver_callbacks_t *a_drv_callback)
-{
- dap_hash_fast_t l_dir_hash;
- dap_hash_fast(a_filename_dir, strlen(a_filename_dir), &l_dir_hash);
- char l_db_name[DAP_PGSQL_DBHASHNAME_LEN + 1];
- dap_htoa64(l_db_name, l_dir_hash.raw, DAP_PGSQL_DBHASHNAME_LEN);
- l_db_name[DAP_PGSQL_DBHASHNAME_LEN] = '\0';
- if (!dap_dir_test(a_filename_dir) || !readdir(opendir(a_filename_dir))) {
- // Create PostgreSQL database
- const char *l_base_conn_str = "dbname = postgres";
- PGconn *l_base_conn = PQconnectdb(l_base_conn_str);
- if (PQstatus(l_base_conn) != CONNECTION_OK) {
- log_it(L_ERROR, "Can't init PostgreSQL database: \"%s\"", PQerrorMessage(l_base_conn));
- PQfinish(l_base_conn);
- return -2;
- }
- char *l_query_str = dap_strdup_printf("DROP DATABASE IF EXISTS \"%s\"", l_db_name);
- PGresult *l_res = PQexec(l_base_conn, l_query_str);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Drop database failed: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- PQfinish(l_base_conn);
- return -3;
- }
- PQclear(l_res);
- l_query_str = dap_strdup_printf("DROP TABLESPACE IF EXISTS \"%s\"", l_db_name);
- l_res = PQexec(l_base_conn, l_query_str);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Drop tablespace failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- PQfinish(l_base_conn);
- return -4;
- }
- PQclear(l_res);
- // Check paths and create them if nessesary
- if (!dap_dir_test(a_filename_dir)) {
- log_it(L_NOTICE, "No directory %s, trying to create...", a_filename_dir);
- dap_mkdir_with_parents(a_filename_dir);
- if (!dap_dir_test(a_filename_dir)) {
- char l_errbuf[255];
- l_errbuf[0] = '\0';
- strerror_r(errno, l_errbuf, sizeof(l_errbuf));
- log_it(L_ERROR, "Can't create directory, error code %d, error string \"%s\"", errno, l_errbuf);
- return -1;
- }
- log_it(L_NOTICE,"Directory created");
- chown(a_filename_dir, getpwnam("postgres")->pw_uid, -1);
- }
- l_query_str = dap_strdup_printf("CREATE TABLESPACE \"%s\" LOCATION '%s'", l_db_name, a_filename_dir);
- l_res = PQexec(l_base_conn, l_query_str);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Create tablespace failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- PQfinish(l_base_conn);
- return -5;
- }
- chmod(a_filename_dir, S_IRWXU | S_IRWXG | S_IRWXO);
- PQclear(l_res);
- l_query_str = dap_strdup_printf("CREATE DATABASE \"%s\" WITH TABLESPACE \"%s\"", l_db_name, l_db_name);
- l_res = PQexec(l_base_conn, l_query_str);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Create database failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- PQfinish(l_base_conn);
- return -6;
- }
- PQclear(l_res);
- PQfinish(l_base_conn);
- }
- // Create connection pool for the DAP database
- char *l_conn_str = dap_strdup_printf("dbname = %s", l_db_name);
- for (int i = 0; i < DAP_PGSQL_POOL_COUNT; i++) {
- s_conn_pool[i].conn = PQconnectdb(l_conn_str);
- s_conn_pool[i].busy = 0;
- if (PQstatus(s_conn_pool[i].conn) != CONNECTION_OK) {
- log_it(L_ERROR, "Can't connect PostgreSQL database: \"%s\"", PQerrorMessage(s_conn_pool[i].conn));
- DAP_DELETE(l_conn_str);
- for (int j = 0; j <= i; j++)
- PQfinish(s_conn_pool[j].conn);
- return -7;
- }
- }
- DAP_DELETE(l_conn_str);
- pthread_rwlock_init(&s_db_rwlock, 0);
- a_drv_callback->transaction_start = dap_db_driver_pgsql_start_transaction;
- a_drv_callback->transaction_end = dap_db_driver_pgsql_end_transaction;
- a_drv_callback->apply_store_obj = dap_db_driver_pgsql_apply_store_obj;
- a_drv_callback->read_store_obj = dap_db_driver_pgsql_read_store_obj;
- a_drv_callback->read_cond_store_obj = dap_db_driver_pgsql_read_cond_store_obj;
- a_drv_callback->read_last_store_obj = dap_db_driver_pgsql_read_last_store_obj;
- a_drv_callback->get_groups_by_mask = dap_db_driver_pgsql_get_groups_by_mask;
- a_drv_callback->read_count_store = dap_db_driver_pgsql_read_count_store;
- a_drv_callback->is_obj = dap_db_driver_pgsql_is_obj;
- a_drv_callback->deinit = dap_db_driver_pgsql_deinit;
- a_drv_callback->flush = dap_db_driver_pgsql_flush;
- return 0;
-}
-
-/**
- * @brief Deinitializes a PostgreSQL database.
- *
- * @return Returns 0 if successful.
- */
-int dap_db_driver_pgsql_deinit(void)
-{
- pthread_rwlock_wrlock(&s_db_rwlock);
- for (int j = 0; j < DAP_PGSQL_POOL_COUNT; j++)
- PQfinish(s_conn_pool[j].conn);
- pthread_rwlock_unlock(&s_db_rwlock);
- pthread_rwlock_destroy(&s_db_rwlock);
- return 0;
-}
-
-/**
- * @brief Starts a transaction in a PostgreSQL database.
- *
- * @return Returns 0 if successful, otherwise -1.
- */
-int dap_db_driver_pgsql_start_transaction(void)
-{
- s_trans_conn = s_pgsql_get_connection();
- if (!s_trans_conn)
- return -1;
- pthread_rwlock_wrlock(&s_db_rwlock);
- PGresult *l_res = PQexec(s_trans_conn, "BEGIN");
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Begin transaction failed with message: \"%s\"", PQresultErrorMessage(l_res));
- pthread_rwlock_unlock(&s_db_rwlock);
- s_pgsql_free_connection(s_trans_conn);
- s_trans_conn = NULL;
- }
- return 0;
-}
-
-/**
- * @brief Ends a transaction in a PostgreSQL database.
- *
- * @return Returns 0 if successful, otherwise -1.
- */
-int dap_db_driver_pgsql_end_transaction(void)
-{
- if (!s_trans_conn)
- return -1;
- PGresult *l_res = PQexec(s_trans_conn, "COMMIT");
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "End transaction failed with message: \"%s\"", PQresultErrorMessage(l_res));
- }
- pthread_rwlock_unlock(&s_db_rwlock);
- s_pgsql_free_connection(s_trans_conn);
- s_trans_conn = NULL;
- return 0;
-}
-
-/**
- * @brief Creates a table in a PostgreSQL database.
- *
- * @param a_table_name a table name string
- * @param a_conn a pointer to the connection object
- * @return Returns 0 if successful, otherwise -1.
- */
-static int s_pgsql_create_group_table(const char *a_table_name, PGconn *a_conn)
-{
- if (!a_table_name)
- return -1;
- int l_ret = 0;
- char *l_query_str = dap_strdup_printf("CREATE TABLE \"%s\""
- "(obj_id BIGSERIAL PRIMARY KEY, obj_ts BIGINT, "
- "obj_key TEXT UNIQUE, obj_val BYTEA)",
- a_table_name);
- PGresult *l_res = PQexec(a_conn, l_query_str);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Create table failed with message: \"%s\"", PQresultErrorMessage(l_res));
- l_ret = -3;
- }
- PQclear(l_res);
- return l_ret;
-}
-
-/**
- * @brief Applies an object to a PostgreSQL database.
- *
- * @param a_store_obj a pointer to the object structure
- * @return Returns 0 if successful, else a error code less than zero.
- */
-int dap_db_driver_pgsql_apply_store_obj(dap_store_obj_t *a_store_obj)
-{
- if (!a_store_obj || !a_store_obj->group)
- return -1;
- char *l_query_str = NULL;
- int l_ret = 0;
- PGresult *l_res = NULL;
- PGconn *l_conn = s_pgsql_get_connection();
- if (!l_conn) {
- log_it(L_ERROR, "Can't pick PostgreSQL connection from pool");
- return -2;
- }
- if (a_store_obj->type == 'a') {
- const char *l_param_vals[2];
- time_t l_ts_to_store = htobe64(a_store_obj->timestamp);
- l_param_vals[0] = (const char *)&l_ts_to_store;
- l_param_vals[1] = (const char *)a_store_obj->value;
- int l_param_lens[2] = {sizeof(time_t), a_store_obj->value_len};
- int l_param_formats[2] = {1, 1};
- l_query_str = dap_strdup_printf("INSERT INTO \"%s\" (obj_ts, obj_key, obj_val) VALUES ($1, '%s', $2) "
- "ON CONFLICT (obj_key) DO UPDATE SET "
- "obj_id = EXCLUDED.obj_id, obj_ts = EXCLUDED.obj_ts, obj_val = EXCLUDED.obj_val;",
- a_store_obj->group, a_store_obj->key);
-
- // execute add request
- l_res = PQexecParams(l_conn, l_query_str, 2, NULL, l_param_vals, l_param_lens, l_param_formats, 0);
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- if (s_trans_conn) { //we shouldn't fail within a transaacion
- dap_db_driver_pgsql_end_transaction();
- dap_db_driver_pgsql_start_transaction();
- l_conn = s_pgsql_get_connection();
- }
- if (s_pgsql_create_group_table(a_store_obj->group, l_conn) == 0) {
- PQclear(l_res);
- l_res = PQexecParams(l_conn, l_query_str, 2, NULL, l_param_vals, l_param_lens, l_param_formats, 0);
- }
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Add object failed with message: \"%s\"", PQresultErrorMessage(l_res));
- l_ret = -3;
- }
- }
- } else if (a_store_obj->type == 'd') {
- // delete one record
- if (a_store_obj->key)
- l_query_str = dap_strdup_printf("DELETE FROM \"%s\" WHERE obj_key = '%s'",
- a_store_obj->group, a_store_obj->key);
- // remove all group
- else
- l_query_str = dap_strdup_printf("DROP TABLE \"%s\"", a_store_obj->group);
- // execute delete request
- l_res = PQexec(l_conn, l_query_str);
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- const char *l_err = PQresultErrorField(l_res, PG_DIAG_SQLSTATE);
- if (!l_err || strcmp(l_err, PGSQL_INVALID_TABLE)) {
- log_it(L_ERROR, "Delete object failed with message: \"%s\"", PQresultErrorMessage(l_res));
- l_ret = -4;
- }
- }
- }
- else {
- log_it(L_ERROR, "Unknown store_obj type '0x%x'", a_store_obj->type);
- s_pgsql_free_connection(l_conn);
- return -5;
- }
- DAP_DELETE(l_query_str);
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return l_ret;
-}
-
-/**
- * @brief Fills a object from a row
- * @param a_group a group name string
- * @param a_obj a pointer to the object
- * @param a_res a pointer to the result structure
- * @param a_row a row number
- * @return (none)
- */
-static void s_pgsql_fill_object(const char *a_group, dap_store_obj_t *a_obj, PGresult *a_res, int a_row)
-{
- a_obj->group = dap_strdup(a_group);
-
- for (int i = 0; i < PQnfields(a_res); i++) {
- if (i == PQfnumber(a_res, "obj_id")) {
- a_obj->id = be64toh(*(uint64_t *)PQgetvalue(a_res, a_row, i));
- } else if (i == PQfnumber(a_res, "obj_ts")) {
- a_obj->timestamp = be64toh(*(time_t *)PQgetvalue(a_res, a_row, i));
- } else if ((i == PQfnumber(a_res, "obj_key"))) {
- a_obj->key = dap_strdup(PQgetvalue(a_res, a_row, i));
- } else if ((i == PQfnumber(a_res, "obj_val"))) {
- a_obj->value_len = PQgetlength(a_res, a_row, i);
- a_obj->value = DAP_DUP_SIZE(PQgetvalue(a_res, a_row, i), a_obj->value_len);
- }
- }
-}
-
-/**
- * @brief Reads some objects from a PostgreSQL database by a_group and a_key.
- * @param a_group a group name string
- * @param a_key an object key string, if equals NULL reads the whole group
- * @param a_count_out[in] a number of objects to be read, if equals 0 reads with no limits
- * @param a_count_out[out] a number of objects that were read
- * @return If successful, a pointer to an objects, otherwise a null pointer.
- */
-dap_store_obj_t *dap_db_driver_pgsql_read_store_obj(const char *a_group, const char *a_key, size_t *a_count_out)
-{
- if (!a_group)
- return NULL;
- PGconn *l_conn = s_pgsql_get_connection();
- if (!l_conn) {
- log_it(L_ERROR, "Can't pick PostgreSQL connection from pool");
- return NULL;
- }
- char *l_query_str;
- if (a_key) {
- l_query_str = dap_strdup_printf("SELECT * FROM \"%s\" WHERE obj_key = '%s'", a_group, a_key);
- } else {
- if (a_count_out && *a_count_out)
- l_query_str = dap_strdup_printf("SELECT * FROM \"%s\" ORDER BY obj_id ASC LIMIT %d", a_group, *a_count_out);
- else
- l_query_str = dap_strdup_printf("SELECT * FROM \"%s\" ORDER BY obj_id ASC", a_group);
- }
-
- PGresult *l_res = PQexecParams(l_conn, l_query_str, 0, NULL, NULL, NULL, NULL, 1);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_TUPLES_OK) {
- const char *l_err = PQresultErrorField(l_res, PG_DIAG_SQLSTATE);
- if (!l_err || strcmp(l_err, PGSQL_INVALID_TABLE))
- log_it(L_ERROR, "Read objects failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return NULL;
- }
-
- // parse reply
- size_t l_count = PQntuples(l_res);
- dap_store_obj_t *l_obj = l_count ? DAP_NEW_Z_SIZE(dap_store_obj_t, sizeof(dap_store_obj_t) * l_count) : NULL;
- for (int i = 0; i < l_count; i++) {
- // fill currrent item
- dap_store_obj_t *l_obj_cur = l_obj + i;
- s_pgsql_fill_object(a_group, l_obj_cur, l_res, i);
- }
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- if (a_count_out)
- *a_count_out = l_count;
- return l_obj;
-}
-
-/**
- * @brief Reads a last object from a PostgreSQL database.
- * @param a_group a group name string
- * @return Returns a pointer to the object if successful, otherwise a null pointer.
- */
-dap_store_obj_t *dap_db_driver_pgsql_read_last_store_obj(const char *a_group)
-{
- if (!a_group)
- return NULL;
- PGconn *l_conn = s_pgsql_get_connection();
- if (!l_conn) {
- log_it(L_ERROR, "Can't pick PostgreSQL connection from pool");
- return NULL;
- }
- char *l_query_str = dap_strdup_printf("SELECT * FROM \"%s\" ORDER BY obj_id DESC LIMIT 1", a_group);
- PGresult *l_res = PQexecParams(l_conn, l_query_str, 0, NULL, NULL, NULL, NULL, 1);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_TUPLES_OK) {
- const char *l_err = PQresultErrorField(l_res, PG_DIAG_SQLSTATE);
- if (!l_err || strcmp(l_err, PGSQL_INVALID_TABLE))
- log_it(L_ERROR, "Read last object failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return NULL;
- }
- dap_store_obj_t *l_obj = NULL;
- if (PQntuples(l_res)) {
- l_obj = DAP_NEW_Z(dap_store_obj_t);
- s_pgsql_fill_object(a_group, l_obj, l_res, 0);
- }
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return l_obj;
-}
-
-/**
- * @brief Reads some objects from a PostgreSQL database by conditions.
- * @param a_group a group name string
- * @param a_id id
- * @param a_count_out[in] a number of objects to be read, if equals 0 reads with no limits
- * @param a_count_out[out] a number of objects that were read
- * @return If successful, a pointer to an objects, otherwise a null pointer.
- */
-dap_store_obj_t *dap_db_driver_pgsql_read_cond_store_obj(const char *a_group, uint64_t a_id, size_t *a_count_out)
-{
- if (!a_group)
- return NULL;
- PGconn *l_conn = s_pgsql_get_connection();
- if (!l_conn) {
- log_it(L_ERROR, "Can't pick PostgreSQL connection from pool");
- return NULL;
- }
- char *l_query_str;
- if (a_count_out && *a_count_out)
- l_query_str = dap_strdup_printf("SELECT * FROM \"%s\" WHERE obj_id >= '%"DAP_UINT64_FORMAT_U"' "
- "ORDER BY obj_id ASC LIMIT %d", a_group, a_id, *a_count_out);
- else
- l_query_str = dap_strdup_printf("SELECT * FROM \"%s\" WHERE obj_id >= '%"DAP_UINT64_FORMAT_U"' "
- "ORDER BY obj_id ASC", a_group, a_id);
- PGresult *l_res = PQexecParams(l_conn, l_query_str, 0, NULL, NULL, NULL, NULL, 1);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_TUPLES_OK) {
- const char *l_err = PQresultErrorField(l_res, PG_DIAG_SQLSTATE);
- if (!l_err || strcmp(l_err, PGSQL_INVALID_TABLE))
- log_it(L_ERROR, "Conditional read objects failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return NULL;
- }
-
- // parse reply
- size_t l_count = PQntuples(l_res);
- dap_store_obj_t *l_obj = l_count ? DAP_NEW_Z_SIZE(dap_store_obj_t, sizeof(dap_store_obj_t) * l_count) : NULL;
- for (int i = 0; i < l_count; i++) {
- // fill currrent item
- dap_store_obj_t *l_obj_cur = l_obj + i;
- s_pgsql_fill_object(a_group, l_obj_cur, l_res, i);
- }
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- if (a_count_out)
- *a_count_out = l_count;
- return l_obj;
-}
-
-/**
- * @brief Gets a list of group names from a PostgreSQL database by a_group_mask.
- * @param a_group_mask a group name mask
- * @return Returns a pointer to a list of group names if successful, otherwise a null pointer.
- */
-dap_list_t *dap_db_driver_pgsql_get_groups_by_mask(const char *a_group_mask)
-{
- if (!a_group_mask)
- return NULL;
- PGconn *l_conn = s_pgsql_get_connection();
- if (!l_conn) {
- log_it(L_ERROR, "Can't pick PostgreSQL connection from pool");
- return NULL;
- }
- const char *l_query_str = "SELECT tablename FROM pg_catalog.pg_tables WHERE "
- "schemaname != 'information_schema' AND schemaname != 'pg_catalog'";
- PGresult *l_res = PQexec(l_conn, l_query_str);
- if (PQresultStatus(l_res) != PGRES_TUPLES_OK) {
- log_it(L_ERROR, "Read tables failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return NULL;
- }
-
- dap_list_t *l_ret_list = NULL;
- for (int i = 0; i < PQntuples(l_res); i++) {
- char *l_table_name = (char *)PQgetvalue(l_res, i, 0);
- if(!dap_fnmatch(a_group_mask, l_table_name, 0))
- l_ret_list = dap_list_prepend(l_ret_list, dap_strdup(l_table_name));
- }
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return l_ret_list;
-}
-
-/**
- * @brief Reads a number of objects from a PostgreSQL database by a_group and a_id.
- * @param a_group a group name string
- * @param a_id id starting from which the quantity is calculated
- * @return Returns a number of objects.
- */
-size_t dap_db_driver_pgsql_read_count_store(const char *a_group, uint64_t a_id)
-{
- if (!a_group)
- return 0;
- PGconn *l_conn = s_pgsql_get_connection();
- if (!l_conn) {
- log_it(L_ERROR, "Can't pick PostgreSQL connection from pool");
- return 0;
- }
- char *l_query_str = dap_strdup_printf("SELECT count(*) FROM \"%s\" WHERE obj_id >= '%"DAP_UINT64_FORMAT_U"'",
- a_group, a_id);
- PGresult *l_res = PQexecParams(l_conn, l_query_str, 0, NULL, NULL, NULL, NULL, 1);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_TUPLES_OK) {
- const char *l_err = PQresultErrorField(l_res, PG_DIAG_SQLSTATE);
- if (!l_err || strcmp(l_err, PGSQL_INVALID_TABLE))
- log_it(L_ERROR, "Count objects failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return 0;
- }
- size_t l_ret = be64toh(*(uint64_t *)PQgetvalue(l_res, 0, 0));
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return l_ret;
-}
-
-/**
- * @brief Checks if an object is in a PostgreSQL database by a_group and a_key.
- * @param a_group a group name string
- * @param a_key a object key string
- * @return Returns true if it is, false it's not.
- */
-bool dap_db_driver_pgsql_is_obj(const char *a_group, const char *a_key)
-{
- if (!a_group)
- return NULL;
- PGconn *l_conn = s_pgsql_get_connection();
- if (!l_conn) {
- log_it(L_ERROR, "Can't pick PostgreSQL connection from pool");
- return NULL;
- }
- char *l_query_str = dap_strdup_printf("SELECT EXISTS(SELECT * FROM \"%s\" WHERE obj_key = '%s')", a_group, a_key);
- PGresult *l_res = PQexecParams(l_conn, l_query_str, 0, NULL, NULL, NULL, NULL, 1);
- DAP_DELETE(l_query_str);
- if (PQresultStatus(l_res) != PGRES_TUPLES_OK) {
- const char *l_err = PQresultErrorField(l_res, PG_DIAG_SQLSTATE);
- if (!l_err || strcmp(l_err, PGSQL_INVALID_TABLE))
- log_it(L_ERROR, "Existance check of object failed with message: \"%s\"", PQresultErrorMessage(l_res));
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return 0;
- }
- int l_ret = *PQgetvalue(l_res, 0, 0);
- PQclear(l_res);
- s_pgsql_free_connection(l_conn);
- return l_ret;
-}
-
-/**
- * @brief Flushes a PostgreSQ database cahce to disk.
- * @return Returns 0 if successful, else a error code less than zero.
- */
-int dap_db_driver_pgsql_flush()
-{
- PGconn *l_conn = s_pgsql_get_connection();
- if (!l_conn) {
- log_it(L_ERROR, "Can't pick PostgreSQL connection from pool");
- return -4;
- }
- int l_ret = 0;
- PGresult *l_res = PQexec(l_conn, "CHECKPOINT");
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Flushing database on disk failed with message: \"%s\"", PQresultErrorMessage(l_res));
- l_ret = -5;
- }
- PQclear(l_res);
- if (!l_ret) {
- PGresult *l_res = PQexec(l_conn, "VACUUM");
- if (PQresultStatus(l_res) != PGRES_COMMAND_OK) {
- log_it(L_ERROR, "Vaccuming database failed with message: \"%s\"", PQresultErrorMessage(l_res));
- l_ret = -6;
- }
- PQclear(l_res);
- }
- s_pgsql_free_connection(l_conn);
- return l_ret;
-}
-#endif
diff --git a/modules/global-db/dap_chain_global_db_driver_sqlite.c b/modules/global-db/dap_chain_global_db_driver_sqlite.c
deleted file mode 100644
index ed2772d08484b30627339b248d6660307a61bd94..0000000000000000000000000000000000000000
--- a/modules/global-db/dap_chain_global_db_driver_sqlite.c
+++ /dev/null
@@ -1,1204 +0,0 @@
-/*
- * Authors:
- * Dmitriy A. Gearasimov <gerasimov.dmitriy@demlabs.net>
- * Alexander Lysikov <alexander.lysikov@demlabs.net>
- * DeM Labs Inc. https://demlabs.net
- * CellFrame https://cellframe.net
- * Sources https://gitlab.demlabs.net/cellframe
- * Copyright (c) 2017-2019
- * All rights reserved.
-
- This file is part of CellFrame SDK the open source project
-
- CellFrame SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- CellFrame SDK is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any CellFrame SDK based project. If not, see <http://www.gnu.org/licenses/>.
-*/
-
-#include <sqlite3.h>
-#include <stddef.h>
-#include <string.h>
-#include <pthread.h>
-#include <errno.h>
-#include <stdatomic.h>
-
-#ifdef DAP_OS_UNIX
-#include <unistd.h>
-#endif
-#include "dap_chain_global_db_driver_sqlite.h"
-#include "dap_common.h"
-#include "dap_hash.h"
-#include "dap_file_utils.h"
-#include "dap_strfuncs.h"
-#include "dap_file_utils.h"
-
-#define LOG_TAG "db_sqlite"
-
-static struct conn_pool_item {
- void *flink; /* Forward link to next element in the simple list */
- sqlite3 *conn; /* SQLITE connection context itself */
- int idx; /* Just index, no more */
- atomic_flag busy; /* "Context is busy" flag */
- atomic_ullong usage; /* Usage counter */
-} s_conn_pool [DAP_SQLITE_POOL_COUNT]; /* Preallocate a storage for the SQLITE connections */
-
-
-static struct conn_pool_item *s_trans = NULL; /* SQL context of outstanding transaction */
-static pthread_mutex_t s_trans_mtx = PTHREAD_MUTEX_INITIALIZER;
-static pthread_cond_t s_trans_cnd = PTHREAD_COND_INITIALIZER;
-
-extern int g_dap_global_db_debug_more; /* Enable extensible debug output */
-
-static char s_filename_db [MAX_PATH];
-
-static pthread_mutex_t s_conn_free_mtx = PTHREAD_MUTEX_INITIALIZER; /* Lock to coordinate access to the free connections pool */
-static pthread_cond_t s_conn_free_cnd = PTHREAD_COND_INITIALIZER; /* To signaling to waites of the free connection */
-
-
-static pthread_mutex_t s_db_mtx = PTHREAD_MUTEX_INITIALIZER;
-
-// Value of one field in the table
-typedef struct _sqlite_value_
-{
- int32_t len,
- type;
- /*
- #define SQLITE_INTEGER 1
- #define SQLITE_FLOAT 2
- #define SQLITE_TEXT 3
- #define SQLITE_BLOB 4
- #define SQLITE_NULL 5
- */
-
- union {
- int val_int;
- long long val_int64;
- double val_float;
- const char *val_str;
- const unsigned char *val_blob;
- } val;
-} SQLITE_VALUE;
-
-// Content of one row in the table
-typedef struct _sqlite_row_value_
-{
- int count; // number of columns in a row
- int reserv;
- SQLITE_VALUE *val; // array of field values
-} SQLITE_ROW_VALUE;
-
-/**
- * @brief Closes a SQLite database.
- *
- * @param l_db a pointer to an instance of SQLite database structure
- * @return (none)
- */
-static inline void s_dap_db_driver_sqlite_close(sqlite3 *l_db)
-{
- if(l_db)
- sqlite3_close(l_db);
-}
-
-/**
- * @brief Releases memory allocated by sqlite3_mprintf()
- *
- * @param memory a pointer to a string
- * @return (none)
- */
-static inline void s_dap_db_driver_sqlite_free(char *memory)
-{
- if(memory)
- sqlite3_free(memory);
-}
-
-#define DAP_DB$K_RTR4CONN 3 /* A number of attempts get DB connection */
-#define DAP_DB$K_TMO4CONN 3 /* A number of seconds to wait for connection */
-
-static inline struct conn_pool_item *s_sqlite_get_connection(void)
-{
-int l_rc;
-struct conn_pool_item *l_conn;
-struct timespec tmo = {0};
-
- if ( (l_rc = pthread_mutex_lock(&s_db_mtx)) == EDEADLK ) /* Get the mutex */
- return s_trans; /* DEADLOCK is detected ? Return pointer to current transaction */
- else if ( l_rc )
- return log_it(L_ERROR, "Cannot get free SQLITE connection, errno=%d", l_rc), NULL;
-
- pthread_mutex_unlock(&s_db_mtx);
-
- for (int i = DAP_DB$K_RTR4CONN; i--; )
- {
- /* Run over connection list, try to get a free connection */
- l_conn = s_conn_pool;
- for (int j = DAP_SQLITE_POOL_COUNT; j--; l_conn++)
- {
- if ( !(l_rc = atomic_flag_test_and_set (&l_conn->busy)) ) /* Test-and-set ... */
- {
- /* l_rc == 0 - so connection was free, */
- /* we got free connection, so get out */
-
- atomic_fetch_add(&l_conn->usage, 1);
- if (g_dap_global_db_debug_more )
- log_it(L_DEBUG, "Alloc l_conn: @%p/%p, usage: %llu", l_conn, l_conn->conn, l_conn->usage);
- return l_conn;
- }
- }
-
- log_it(L_INFO, "No free SQLITE connection, wait %d seconds ...", DAP_DB$K_TMO4CONN);
-
- /* No free connection at the moment, so, prepare to wait a condition ... */
-
- clock_gettime(CLOCK_REALTIME, &tmo);
- tmo.tv_sec += DAP_DB$K_TMO4CONN;
-
- pthread_mutex_lock(&s_conn_free_mtx);
- l_rc = pthread_cond_timedwait (&s_conn_free_cnd, &s_conn_free_mtx, &tmo);
- pthread_mutex_unlock(&s_conn_free_mtx);
-
- log_it(L_DEBUG, "pthread_cond_timedwait()->%d, errno=%d", l_rc, errno);
- }
-
- return log_it(L_ERROR, "No free SQLITE connection"), NULL;
-}
-
-static inline int s_sqlite_free_connection(struct conn_pool_item *a_conn)
-{
-int l_rc;
-
- if (g_dap_global_db_debug_more)
- log_it(L_DEBUG, "Free l_conn: @%p/%p, usage: %llu", a_conn, a_conn->conn, a_conn->usage);
-
- atomic_flag_clear(&a_conn->busy); /* Clear busy flag */
-
- l_rc = pthread_mutex_lock (&s_conn_free_mtx); /* Send a signal to waiters to wake-up */
- l_rc = pthread_cond_signal(&s_conn_free_cnd);
- l_rc = pthread_mutex_unlock(&s_conn_free_mtx);
-
- return 0;
-}
-
-/**
- * @brief Deinitializes a SQLite database.
- *
- * @return Returns 0 if successful.
- */
-int dap_db_driver_sqlite_deinit(void)
-{
- pthread_mutex_lock(&s_db_mtx);
- for (int i = 0; i < DAP_SQLITE_POOL_COUNT; i++) {
- if (s_conn_pool[i].conn) {
- s_dap_db_driver_sqlite_close(s_conn_pool[i].conn);
- atomic_flag_clear (&s_conn_pool[i].busy);
- }
- }
- pthread_mutex_unlock(&s_db_mtx);
- //s_db = NULL;
- return sqlite3_shutdown();
-}
-
-// An additional function for SQLite to convert byte to number
-static void s_byte_to_bin(sqlite3_context *l_context, int a_argc, sqlite3_value **a_argv)
-{
- const unsigned char *l_text;
- if(a_argc != 1)
- sqlite3_result_null(l_context);
- l_text = (const unsigned char *) sqlite3_value_blob(a_argv[0]);
- if(l_text && l_text[0]) {
- int l_result = (int) l_text[0];
- sqlite3_result_int(l_context, l_result);
- return;
- }
- sqlite3_result_null(l_context);
-}
-
-/**
- * @brief Opens a SQLite database and adds byte_to_bin function.
- *
- * @param a_filename_utf8 a SQLite database file name
- * @param a_flags database access flags (SQLITE_OPEN_READONLY, SQLITE_OPEN_READWRITE | SQLITE_OPEN_CREATE)
- * @param a_error_message[out] an error message that's received from the SQLite database
- * @return Returns a pointer to an instance of SQLite database structure.
- */
-sqlite3* dap_db_driver_sqlite_open(const char *a_filename_utf8, int a_flags, char **a_error_message)
-{
- sqlite3 *l_db = NULL;
-
- int l_rc = sqlite3_open_v2(a_filename_utf8, &l_db, a_flags | SQLITE_OPEN_NOMUTEX, NULL);
- // if unable to open the database file
- if(l_rc == SQLITE_CANTOPEN) {
- log_it(L_WARNING,"No database on path %s, creating one from scratch", a_filename_utf8);
- if(l_db)
- sqlite3_close(l_db);
- // try to create database
- l_rc = sqlite3_open_v2(a_filename_utf8, &l_db, a_flags | SQLITE_OPEN_NOMUTEX| SQLITE_OPEN_CREATE, NULL);
- }
-
- if(l_rc != SQLITE_OK) {
- log_it(L_CRITICAL,"Can't open database on path %s (code %d: \"%s\" )", a_filename_utf8, l_rc, sqlite3_errstr(l_rc));
- if(a_error_message)
- *a_error_message = sqlite3_mprintf("Can't open database: %s\n", sqlite3_errmsg(l_db));
- sqlite3_close(l_db);
- return NULL;
- }
- // added user functions
- sqlite3_create_function(l_db, "byte_to_bin", 1, SQLITE_UTF8, NULL, &s_byte_to_bin, NULL, NULL);
- return l_db;
-}
-
-
-/**
- * @brief Executes SQL statements.
- *
- * @param l_db a pointer to an instance of SQLite database structure
- * @param l_query the SQL statement
- * @param l_error_message[out] an error message that's received from the SQLite database
- * @return Returns 0 if successful.
- */
-static int s_dap_db_driver_sqlite_exec(sqlite3 *l_db, const char *l_query, char **l_error_message)
-{
-char *l_errmsg = NULL;
-int l_rc;
-struct timespec tmo = {0, 500 * 1024 * 1024 /* ~0.5 sec */}, delta;
-
- for ( int i = 7; i--; )
- { /* Ok or error (exclude SQL_LOCKED) - just exit from loop? */
- if ( SQLITE_LOCKED != (l_rc = sqlite3_exec(l_db, l_query, NULL, 0, &l_errmsg))
- && (l_rc != SQLITE_BUSY) )
- break;
-
- if (g_dap_global_db_debug_more )
- log_it(L_WARNING, "SQL error: \"%s\"%d, dap_db_driver_sqlite_exec(%p, %s), retry ...", sqlite3_errmsg(l_db),l_rc, l_db, l_query);
-
- for ( delta = tmo; nanosleep(&delta, &delta); ); /* Wait some time ... */
- }
-
-
- if ( l_rc != SQLITE_OK)
- {
- if ( l_rc != SQLITE_CONSTRAINT )
- log_it(L_ERROR, "SQL error: \"%s\"%d, dap_db_driver_sqlite_exec(%p, %s), retry ...", sqlite3_errmsg(l_db),l_rc, l_db, l_query);
-
- if(l_error_message && l_errmsg)
- *l_error_message = sqlite3_mprintf("SQL error %d: %s", l_rc, l_errmsg);
- }
-
- if(l_errmsg)
- sqlite3_free(l_errmsg);
-
- return l_rc;
-}
-
-/**
- * @brief Creates a table and unique index in the s_db database.
- *
- * @param a_table_name a table name string
- * @return Returns 0 if successful, otherwise -1.
- */
-static int s_dap_db_driver_sqlite_create_group_table(const char *a_table_name)
-{
-int l_rc;
-struct conn_pool_item *l_conn;
-char *l_error_message, l_query[512];
-
- if( !a_table_name )
- return -EINVAL;
-
- if ( !(l_conn = s_sqlite_get_connection()) )
- return log_it(L_ERROR, "Error create group table '%s'", a_table_name), -ENOENT;
-
- dap_snprintf(l_query, sizeof(l_query) - 1,
- "CREATE TABLE IF NOT EXISTs '%s'(id INTEGER NOT NULL PRIMARY KEY, key TEXT KEY, hash BLOB, ts INTEGER KEY, value BLOB)",
- a_table_name);
-
- if ( (l_rc = s_dap_db_driver_sqlite_exec(l_conn->conn, (const char*) l_query, &l_error_message)) != SQLITE_OK) {
- log_it(L_ERROR, "SQL error: \"%s\"%d, dap_db_driver_sqlite_exec(%p, %s), retry ...", sqlite3_errmsg(l_conn->conn),l_rc, l_conn->conn, l_query);
- s_dap_db_driver_sqlite_free(l_error_message);
- s_sqlite_free_connection(l_conn);
- return -1;
- }
-
- // create unique index - key
- dap_snprintf(l_query, sizeof(l_query) - 1,
- "CREATE UNIQUE INDEX IF NOT EXISTS 'idx_key_%s' ON '%s' (key)", a_table_name,
- a_table_name);
-
- if ( (l_rc = s_dap_db_driver_sqlite_exec(l_conn->conn, (const char*) l_query, &l_error_message)) != SQLITE_OK) {
- log_it(L_ERROR, "SQL error: \"%s\"%d, dap_db_driver_sqlite_exec(%p, %s), retry ...", sqlite3_errmsg(l_conn->conn),l_rc, l_conn->conn, l_query);
- s_dap_db_driver_sqlite_free(l_error_message);
- s_sqlite_free_connection(l_conn);
- return -1;
- }
-
- s_sqlite_free_connection(l_conn);
- return 0;
-}
-
-/**
- * @brief Prepares a SQL query for a database
- * @param db a pointer to an instance of SQLite database structure.
- * @param query the query
- * @param l_res[out] a pointer to a pointer to a structure with result
- * @param l_error_message[out] an error message that's received from the SQLite database
- * @return Returns 0 if successful,
- */
-static int s_dap_db_driver_sqlite_query(sqlite3 *db, char *query, sqlite3_stmt **l_res, char **l_error_message)
-{
- const char *pzTail; // OUT: Pointer to unused portion of zSql
- int l_rc = sqlite3_prepare_v2(db, query, -1, l_res, &pzTail);
- if(l_rc != SQLITE_OK)
- {
- if(l_error_message)
- {
- const char *zErrMsg = sqlite3_errmsg(db);
- if(zErrMsg)
- *l_error_message = sqlite3_mprintf("SQL Query error: %s\n", zErrMsg);
- }
- return l_rc;
- }
- return l_rc;
-}
-
-/**
- * @brief Releases memory allocated for a row.
- *
- * @param row a database row
- * @return (none)
- */
-static void s_dap_db_driver_sqlite_row_free(SQLITE_ROW_VALUE *row)
-{
- if(row) {
- // delete the whole string
- sqlite3_free(row->val);
- // delete structure
- sqlite3_free(row);
- }
-}
-
-
-/**
- * @brief Fetches a result values from a query to l_row_out
- *
- * @param l_res a pointer to a prepared statement structure
- * @param l_row_out a pointer to a pointer to a row structure
- * @return Returns SQLITE_ROW(100) or SQLITE_DONE(101) or SQLITE_BUSY(5)
- */
-static int s_dap_db_driver_sqlite_fetch_array(sqlite3_stmt *l_res, SQLITE_ROW_VALUE **l_row_out)
-{
- SQLITE_ROW_VALUE *l_row = NULL;
- // go to next the string
- int l_rc = sqlite3_step(l_res);
- if(l_rc == SQLITE_ROW) // SQLITE_ROW(100) or SQLITE_DONE(101) or SQLITE_BUSY(5)
- {
- int l_iCol; // number of the column in the row
- // allocate memory for a row with data
- l_row = (SQLITE_ROW_VALUE*) sqlite3_malloc(sizeof(SQLITE_ROW_VALUE));
- int l_count = sqlite3_column_count(l_res); // get the number of columns
- // allocate memory for all columns
- l_row->val = (SQLITE_VALUE*) sqlite3_malloc(l_count * (int)sizeof(SQLITE_VALUE));
- if(l_row->val)
- {
- l_row->count = l_count; // number of columns
- for(l_iCol = 0; l_iCol < l_row->count; l_iCol++)
- {
- SQLITE_VALUE *cur_val = l_row->val + l_iCol;
- cur_val->len = sqlite3_column_bytes(l_res, l_iCol); // how many bytes will be needed
- cur_val->type = sqlite3_column_type(l_res, l_iCol); // field type
- if(cur_val->type == SQLITE_INTEGER)
- cur_val->val.val_int64 = sqlite3_column_int64(l_res, l_iCol);
- else if(cur_val->type == SQLITE_FLOAT)
- cur_val->val.val_float = sqlite3_column_double(l_res, l_iCol);
- else if(cur_val->type == SQLITE_BLOB)
- cur_val->val.val_blob = (const unsigned char*) sqlite3_column_blob(l_res, l_iCol);
- else if(cur_val->type == SQLITE_TEXT)
- cur_val->val.val_str = (const char*) sqlite3_column_text(l_res, l_iCol); //sqlite3_mprintf("%s",sqlite3_column_text(l_res,iCol));
- else
- cur_val->val.val_str = NULL;
- }
- }
- else
- l_row->count = 0; // number of columns
- }
- if(l_row_out)
- *l_row_out = l_row;
- else
- s_dap_db_driver_sqlite_row_free(l_row);
-
-
- return l_rc;
-}
-
-
-/**
- * @brief Destroys a prepared statement structure
- *
- * @param l_res a pointer to the statement structure
- * @return Returnes true if successful, otherwise false.
- */
-static int s_dap_db_driver_sqlite_query_free(sqlite3_stmt *l_res)
-{
-
- if(!l_res)
- return -EINVAL;
-
- return -sqlite3_finalize(l_res);
-
-}
-
-/**
- * @brief Convers a byte array into a hexadecimal string
- *
- * @param blob a byte array
- * @param len a length of byte array
- * @return Returns a hexadecimal string
- */
-static char* s_dap_db_driver_get_string_from_blob(const uint8_t *blob, int len)
-{
- char *str_out;
-
- if(!blob)
- return NULL;
-
- if ( !(str_out = (char*) sqlite3_malloc(len * 2 + 1)) )
- return NULL;
-
- dap_bin2hex(str_out, (const void*)blob, (size_t)len);
- str_out[len * 2] = 0;
- return str_out;
-}
-
-
-/**
- * @brief Executes a VACUUM statement in a database.
- *
- * @param l_db a a pointer to an instance of SQLite database structure
- * @return Returns 0 if successful.
- */
-int s_dap_db_driver_sqlite_vacuum(sqlite3 *l_db)
-{
- if(!l_db)
- return -1;
-
- return s_dap_db_driver_sqlite_exec(l_db, "VACUUM", NULL);
-}
-
-/**
- * @brief Starts a transaction in s_db database.
- *
- * @return Returns 0 if successful, otherwise -1.
- */
-static int s_dap_db_driver_sqlite_start_transaction(void)
-{
-int l_rc;
-
- /* Try to lock */
- if ( EDEADLK == (l_rc = pthread_mutex_lock(&s_trans_mtx)) ) {
- /* DEADLOCK ?! - so transaction is already active ... */
- log_it(L_DEBUG, "Active TX l_conn: @%p/%p", s_trans, s_trans->conn);
- return 0;
- }
-
- if ( ! (s_trans = s_sqlite_get_connection()) )
- return -666;
-
- if ( g_dap_global_db_debug_more )
- log_it(L_DEBUG, "Start TX l_conn: @%p/%p", s_trans, s_trans->conn);
-
- pthread_mutex_lock(&s_db_mtx);
- l_rc = s_dap_db_driver_sqlite_exec(s_trans->conn, "BEGIN", NULL);
- pthread_mutex_unlock(&s_db_mtx);
-
- if ( l_rc != SQLITE_OK ) {
- pthread_mutex_unlock(&s_trans_mtx);
- s_sqlite_free_connection(s_trans);
- s_trans = NULL;
- }
-
- return ( l_rc == SQLITE_OK ) ? 0 : -l_rc;
-}
-
-/**
- * @brief Ends a transaction in s_db database.
- *
- * @return Returns 0 if successful, otherwise -1.
- */
-static int s_dap_db_driver_sqlite_end_transaction(void)
-{
-int l_rc;
-struct conn_pool_item *l_conn;
-
- if ( !s_trans)
- return log_it(L_ERROR, "No active TX!"), -666;
-
- l_conn = s_trans;
- s_trans = NULL; /* Zeroing current TX's context until
- it's protected by the mutex ! */
-
- if ( g_dap_global_db_debug_more )
- log_it(L_DEBUG, "End TX l_conn: @%p/%p", l_conn, l_conn->conn);
-
- pthread_mutex_unlock(&s_trans_mtx); /* Free TX context to other ... */
-
- pthread_mutex_lock(&s_db_mtx);
- l_rc = s_dap_db_driver_sqlite_exec(l_conn->conn, "COMMIT", NULL);
- pthread_mutex_unlock(&s_db_mtx);
-
- s_sqlite_free_connection(l_conn);
-
- return ( l_rc == SQLITE_OK ) ? 0 : -l_rc;
-}
-
-
-/**
- * @brief Replaces '_' char with '.' char in a_table_name.
- *
- * @param a_table_name a table name string
- * @return Returns a group name string with the replaced character
- */
-static inline char *s_sqlite_make_group_name(const char *a_table_name)
-{
- char *l_table_name = dap_strdup(a_table_name), *l_str;
-
- for ( l_str = l_table_name; (l_str = strchr(l_str, '_')); l_str++)
- *l_str = '.';
-
- return l_table_name;
-}
-
-/**
- * @brief Replaces '.' char with '_' char in a_group_name.
- *
- * @param a_group_name a group name string
- * @return Returns a table name string with the replaced character
- */
-static inline char *s_sqlite_make_table_name(const char *a_group_name)
-{
- char *l_group_name = dap_strdup(a_group_name), *l_str;
-
- for ( l_str = l_group_name; (l_str = strchr(l_str, '.')); l_str++)
- *l_str = '_';
-
- return l_group_name;
-}
-
-/**
- * @brief Applies an object to a database.
- *
- * @param a_store_obj a pointer to the object structure
- * @return Returns 0 if successful.
- */
-int dap_db_driver_sqlite_apply_store_obj(dap_store_obj_t *a_store_obj)
-{
- if(!a_store_obj || !a_store_obj->group )
- return -1;
-
- char *l_query = NULL;
- char *l_error_message = NULL;
-
- char *l_table_name = s_sqlite_make_table_name(a_store_obj->group);
-
- if(a_store_obj->type == DAP_DB$K_OPTYPE_ADD) {
- if(!a_store_obj->key)
- return -1;
- char *l_blob_value = s_dap_db_driver_get_string_from_blob(a_store_obj->value, (int)a_store_obj->value_len);
- //add one record
- l_query = sqlite3_mprintf("INSERT INTO '%s' values(NULL, '%s', x'', '%lld', x'%s')",
- l_table_name, a_store_obj->key, a_store_obj->timestamp, l_blob_value);
- s_dap_db_driver_sqlite_free(l_blob_value);
- }
- else if (a_store_obj->type == DAP_DB$K_OPTYPE_DEL) {
- //delete one record
- if (a_store_obj->key) {
- l_query = sqlite3_mprintf("DELETE FROM '%s' where key = '%s'",
- l_table_name, a_store_obj->key);
- } else {
- // remove all group
- l_query = sqlite3_mprintf("DROP TABLE IF EXISTS '%s'", l_table_name);
- }
- }
- else {
- log_it(L_ERROR, "Unknown store_obj type '0x%x'", a_store_obj->type);
- return -1;
- }
- // execute request
- struct conn_pool_item *l_conn = s_sqlite_get_connection();
-
- if( !l_conn )
- return -666;
-
- int l_ret = s_dap_db_driver_sqlite_exec(l_conn->conn, l_query, &l_error_message);
-
- if(l_ret == SQLITE_ERROR) {
- s_dap_db_driver_sqlite_free(l_error_message);
- l_error_message = NULL;
- // create table
- s_dap_db_driver_sqlite_create_group_table(l_table_name);
- // repeat request
- l_ret = s_dap_db_driver_sqlite_exec(l_conn->conn, l_query, &l_error_message);
-
- }
- // entry with the same hash is already present
- if(l_ret == SQLITE_CONSTRAINT) {
- s_dap_db_driver_sqlite_free(l_error_message);
- l_error_message = NULL;
- //replace one record
- char *l_blob_value = s_dap_db_driver_get_string_from_blob(a_store_obj->value, (int)a_store_obj->value_len);
- char *l_query_replace = sqlite3_mprintf("REPLACE INTO '%s' values(NULL, '%s', x'', '%lld', x'%s')",
- l_table_name, a_store_obj->key, a_store_obj->timestamp, l_blob_value);
- s_dap_db_driver_sqlite_free(l_blob_value);
- l_ret = s_dap_db_driver_sqlite_exec(l_conn->conn, l_query_replace, &l_error_message);
- s_dap_db_driver_sqlite_free(l_query_replace);
-
- }
- // missing database
- if(l_ret != SQLITE_OK) {
- log_it(L_ERROR, "sqlite apply error: %s", l_error_message);
- s_dap_db_driver_sqlite_free(l_error_message);
- l_ret = -1;
- }
- s_sqlite_free_connection(l_conn);
- s_dap_db_driver_sqlite_free(l_query);
- DAP_DELETE(l_table_name);
- return l_ret;
-}
-
-/**
- * @brief Fills a object from a row
- *
- * @param a_group a group name string
- * @param a_obj a pointer to the object
- * @param a_row a ponter to the row structure
- */
-static void fill_one_item(const char *a_group, dap_store_obj_t *a_obj, SQLITE_ROW_VALUE *a_row)
-{
- a_obj->group = dap_strdup(a_group);
-
- for(int l_iCol = 0; l_iCol < a_row->count; l_iCol++) {
- SQLITE_VALUE *l_cur_val = a_row->val + l_iCol;
- switch (l_iCol) {
- case 0:
- if(l_cur_val->type == SQLITE_INTEGER)
- a_obj->id = (uint64_t)l_cur_val->val.val_int64;
- break; // id
- case 1:
- if(l_cur_val->type == SQLITE_INTEGER)
- a_obj->timestamp = l_cur_val->val.val_int64;
- break; // ts
- case 2:
- if(l_cur_val->type == SQLITE_TEXT)
- a_obj->key = dap_strdup(l_cur_val->val.val_str);
- break; // key
- case 3:
- if(l_cur_val->type == SQLITE_BLOB)
- {
- a_obj->value_len = (size_t) l_cur_val->len;
- a_obj->value = DAP_NEW_SIZE(uint8_t, a_obj->value_len);
- memcpy((byte_t *)a_obj->value, l_cur_val->val.val_blob, a_obj->value_len);
- }
- break; // value
- }
- }
-
-}
-
-/**
- * @brief Reads a last object from the s_db database.
- *
- * @param a_group a group name string
- * @return Returns a pointer to the object.
- */
-dap_store_obj_t* dap_db_driver_sqlite_read_last_store_obj(const char *a_group)
-{
-dap_store_obj_t *l_obj = NULL;
-char *l_error_message = NULL;
-sqlite3_stmt *l_res;
-struct conn_pool_item *l_conn;
-
- if(!a_group)
- return NULL;
-
- if ( !(l_conn = s_sqlite_get_connection()) )
- return NULL;
-
- char * l_table_name = s_sqlite_make_table_name(a_group);
- char *l_str_query = sqlite3_mprintf("SELECT id,ts,key,value FROM '%s' ORDER BY id DESC LIMIT 1", l_table_name);
- int l_ret = s_dap_db_driver_sqlite_query(l_conn->conn, l_str_query, &l_res, &l_error_message);
-
- sqlite3_free(l_str_query);
- DAP_DEL_Z(l_table_name);
-
- if(l_ret != SQLITE_OK) {
- //log_it(L_ERROR, "read last l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- s_dap_db_driver_sqlite_free(l_error_message);
- s_sqlite_free_connection(l_conn);
- return NULL;
- }
-
- SQLITE_ROW_VALUE *l_row = NULL;
- l_ret = s_dap_db_driver_sqlite_fetch_array(l_res, &l_row);
- if(l_ret != SQLITE_ROW && l_ret != SQLITE_DONE)
- {
- //log_it(L_ERROR, "read l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- }
- if(l_ret == SQLITE_ROW && l_row) {
- l_obj = DAP_NEW_Z(dap_store_obj_t);
- fill_one_item(a_group, l_obj, l_row);
- }
- s_dap_db_driver_sqlite_row_free(l_row);
- s_dap_db_driver_sqlite_query_free(l_res);
-
- s_sqlite_free_connection(l_conn);
-
- return l_obj;
-}
-
-/**
- * @brief Reads some objects from a database by conditions
- *
- * @param a_group a group name string
- * @param a_id id
- * @param a_count_out[in] a number of objects to be read, if equals 0 reads with no limits
- * @param a_count_out[out] a number of objects that were read
- * @return If successful, a pointer to an objects, otherwise NULL.
- */
-dap_store_obj_t* dap_db_driver_sqlite_read_cond_store_obj(const char *a_group, uint64_t a_id, size_t *a_count_out)
-{
- dap_store_obj_t *l_obj = NULL;
- char *l_error_message = NULL;
- sqlite3_stmt *l_res;
- if(!a_group)
- return NULL;
-
- char * l_table_name = s_sqlite_make_table_name(a_group);
- // no limit
- int l_count_out = 0;
- if(a_count_out)
- l_count_out = (int)*a_count_out;
- char *l_str_query = NULL;
- if (l_count_out) {
- l_str_query = sqlite3_mprintf("SELECT id,ts,key,value FROM '%s' WHERE id>='%lld' ORDER BY id ASC LIMIT %d",
- l_table_name, a_id, l_count_out);
- } else {
- l_str_query = sqlite3_mprintf("SELECT id,ts,key,value FROM '%s' WHERE id>='%lld' ORDER BY id ASC",
- l_table_name, a_id);
- }
- struct conn_pool_item *l_conn = s_sqlite_get_connection();
- if(!l_conn) {
- if (l_str_query) sqlite3_free(l_str_query);
- return NULL;
- }
-
- int l_ret = s_dap_db_driver_sqlite_query(l_conn->conn, l_str_query, &l_res, &l_error_message);
- sqlite3_free(l_str_query);
- DAP_DEL_Z(l_table_name);
-
- if(l_ret != SQLITE_OK) {
- //log_it(L_ERROR, "read l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- s_dap_db_driver_sqlite_free(l_error_message);
- s_sqlite_free_connection(l_conn);
- return NULL;
- }
-
- //int b = qlite3_column_count(s_db);
- SQLITE_ROW_VALUE *l_row = NULL;
- l_count_out = 0;
- int l_count_sized = 0;
- do {
- l_ret = s_dap_db_driver_sqlite_fetch_array(l_res, &l_row);
- if(l_ret != SQLITE_ROW && l_ret != SQLITE_DONE)
- {
- // log_it(L_ERROR, "read l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- }
- if(l_ret == SQLITE_ROW && l_row) {
- // realloc memory
- if(l_count_out >= l_count_sized) {
- l_count_sized += 10;
- l_obj = DAP_REALLOC(l_obj, sizeof(dap_store_obj_t) * (uint64_t)l_count_sized);
- memset(l_obj + l_count_out, 0, sizeof(dap_store_obj_t) * (uint64_t)(l_count_sized - l_count_out));
- }
- // fill current item
- dap_store_obj_t *l_obj_cur = l_obj + l_count_out;
- fill_one_item(a_group, l_obj_cur, l_row);
- l_count_out++;
- }
- s_dap_db_driver_sqlite_row_free(l_row);
- } while(l_row);
-
- s_dap_db_driver_sqlite_query_free(l_res);
- s_sqlite_free_connection(l_conn);
-
- if(a_count_out)
- *a_count_out = (size_t)l_count_out;
-
- return l_obj;
-}
-
-/**
- * @brief Reads some objects from a SQLite database by a_group, a_key.
- * @param a_group a group name string
- * @param a_key an object key string, if equals NULL reads the whole group
- * @param a_count_out[in] a number of objects to be read, if equals 0 reads with no limits
- * @param a_count_out[out] a number of objects that were read
- * @return If successful, a pointer to an objects, otherwise NULL.
- */
-dap_store_obj_t* dap_db_driver_sqlite_read_store_obj(const char *a_group, const char *a_key, size_t *a_count_out)
-{
- struct conn_pool_item *l_conn = s_sqlite_get_connection();
-
- if(!a_group || !l_conn)
- return NULL;
-
- dap_store_obj_t *l_obj = NULL;
- sqlite3_stmt *l_res;
- char * l_table_name = s_sqlite_make_table_name(a_group);
- // no limit
- uint64_t l_count_out = 0;
- if(a_count_out)
- l_count_out = *a_count_out;
- char *l_str_query;
- if (a_key) {
- if (l_count_out) {
- l_str_query = sqlite3_mprintf("SELECT id,ts,key,value FROM '%s' WHERE key='%s' ORDER BY id ASC LIMIT %d",
- l_table_name, a_key, l_count_out);
- } else {
- l_str_query = sqlite3_mprintf("SELECT id,ts,key,value FROM '%s' WHERE key='%s' ORDER BY id ASC",
- l_table_name, a_key);
- }
- } else {
- if (l_count_out) {
- l_str_query = sqlite3_mprintf("SELECT id,ts,key,value FROM '%s' ORDER BY id ASC LIMIT %d",
- l_table_name, l_count_out);
- } else {
- l_str_query = sqlite3_mprintf("SELECT id,ts,key,value FROM '%s' ORDER BY id ASC", l_table_name);
- }
- }
- int l_ret = s_dap_db_driver_sqlite_query(l_conn->conn, l_str_query, &l_res, NULL);
-
- sqlite3_free(l_str_query);
- DAP_DEL_Z(l_table_name);
- if(l_ret != SQLITE_OK) {
- //log_it(L_ERROR, "read l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- s_sqlite_free_connection(l_conn);
- return NULL;
- }
-
- //int b = qlite3_column_count(s_db);
- SQLITE_ROW_VALUE *l_row = NULL;
- l_count_out = 0;
- uint64_t l_count_sized = 0;
- do {
- l_ret = s_dap_db_driver_sqlite_fetch_array(l_res, &l_row);
- if(l_ret != SQLITE_ROW && l_ret != SQLITE_DONE)
- {
- // log_it(L_ERROR, "read l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- }
- if(l_ret == SQLITE_ROW && l_row) {
- // realloc memory
- if(l_count_out >= l_count_sized) {
- l_count_sized += 10;
- l_obj = DAP_REALLOC(l_obj, sizeof(dap_store_obj_t) * l_count_sized);
- memset(l_obj + l_count_out, 0, sizeof(dap_store_obj_t) * (l_count_sized - l_count_out));
- }
- // fill currrent item
- dap_store_obj_t *l_obj_cur = l_obj + l_count_out;
- fill_one_item(a_group, l_obj_cur, l_row);
- l_count_out++;
- }
- s_dap_db_driver_sqlite_row_free(l_row);
- } while(l_row);
-
- s_dap_db_driver_sqlite_query_free(l_res);
- s_sqlite_free_connection(l_conn);
-
- if(a_count_out)
- *a_count_out = l_count_out;
-
- return l_obj;
-}
-
-/**
- * @brief Gets a list of group names from a s_db database by a_group_mask.
- *
- * @param a_group_mask a group name mask
- * @return Returns a pointer to a list of group names.
- */
-dap_list_t* dap_db_driver_sqlite_get_groups_by_mask(const char *a_group_mask)
-{
- struct conn_pool_item *l_conn = s_sqlite_get_connection();
-
- if(!a_group_mask || !l_conn)
- return NULL;
-
- sqlite3_stmt *l_res;
- const char *l_str_query = "SELECT name FROM sqlite_master WHERE type ='table' AND name NOT LIKE 'sqlite_%'";
- dap_list_t *l_ret_list = NULL;
- int l_ret = s_dap_db_driver_sqlite_query(l_conn->conn, (char *)l_str_query, &l_res, NULL);
- if(l_ret != SQLITE_OK) {
- //log_it(L_ERROR, "Get tables l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- s_sqlite_free_connection(l_conn);
- return NULL;
- }
- char * l_mask = s_sqlite_make_table_name(a_group_mask);
- SQLITE_ROW_VALUE *l_row = NULL;
- while (s_dap_db_driver_sqlite_fetch_array(l_res, &l_row) == SQLITE_ROW && l_row) {
- char *l_table_name = (char *)l_row->val->val.val_str;
- if(!dap_fnmatch(l_mask, l_table_name, 0))
- l_ret_list = dap_list_prepend(l_ret_list, s_sqlite_make_group_name(l_table_name));
- s_dap_db_driver_sqlite_row_free(l_row);
- }
- s_dap_db_driver_sqlite_query_free(l_res);
-
- s_sqlite_free_connection(l_conn);
-
- return l_ret_list;
-}
-
-/**
- * @brief Reads a number of objects from a s_db database by a_group and a_id
- *
- * @param a_group a group name string
- * @param a_id id starting from which the quantity is calculated
- * @return Returns a number of objects.
- */
-size_t dap_db_driver_sqlite_read_count_store(const char *a_group, uint64_t a_id)
-{
- struct conn_pool_item *l_conn = s_sqlite_get_connection();
-
- if(!a_group || !l_conn)
- return 0;
-
- sqlite3_stmt *l_res;
-
- char * l_table_name = s_sqlite_make_table_name(a_group);
- char *l_str_query = sqlite3_mprintf("SELECT COUNT(*) FROM '%s' WHERE id>='%lld'", l_table_name, a_id);
- int l_ret = s_dap_db_driver_sqlite_query(l_conn->conn, l_str_query, &l_res, NULL);
- sqlite3_free(l_str_query);
- DAP_DEL_Z(l_table_name);
-
- if(l_ret != SQLITE_OK) {
- //log_it(L_ERROR, "Count l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- s_sqlite_free_connection(l_conn);
- return 0;
- }
- size_t l_ret_val = 0;
- SQLITE_ROW_VALUE *l_row = NULL;
- if (s_dap_db_driver_sqlite_fetch_array(l_res, &l_row) == SQLITE_ROW && l_row) {
- l_ret_val = (size_t)l_row->val->val.val_int64;
- s_dap_db_driver_sqlite_row_free(l_row);
- }
- s_dap_db_driver_sqlite_query_free(l_res);
-
- s_sqlite_free_connection(l_conn);
-
- return l_ret_val;
-}
-
-/**
- * @brief Checks if an object is in a s_db database by a_group and a_key.
- *
- * @param a_group a group name string
- * @param a_key a object key string
- * @return Returns true if it is, false it's not.
- */
-bool dap_db_driver_sqlite_is_obj(const char *a_group, const char *a_key)
-{
- struct conn_pool_item *l_conn = s_sqlite_get_connection();
-
- if(!a_group || !l_conn)
- return 0;
-
- sqlite3_stmt *l_res;
-
- char * l_table_name = s_sqlite_make_table_name(a_group);
- char *l_str_query = sqlite3_mprintf("SELECT EXISTS(SELECT * FROM '%s' WHERE key='%s')", l_table_name, a_key);
- int l_ret = s_dap_db_driver_sqlite_query(l_conn->conn, l_str_query, &l_res, NULL);
- sqlite3_free(l_str_query);
- DAP_DEL_Z(l_table_name);
-
- if(l_ret != SQLITE_OK) {
- //log_it(L_ERROR, "Exists l_ret=%d, %s\n", sqlite3_errcode(s_db), sqlite3_errmsg(s_db));
- s_sqlite_free_connection(l_conn);
- return false;
- }
- bool l_ret_val = false;
- SQLITE_ROW_VALUE *l_row = NULL;
- if (s_dap_db_driver_sqlite_fetch_array(l_res, &l_row) == SQLITE_ROW && l_row) {
- l_ret_val = (size_t)l_row->val->val.val_int64;
- s_dap_db_driver_sqlite_row_free(l_row);
- }
- s_dap_db_driver_sqlite_query_free(l_res);
-
- s_sqlite_free_connection(l_conn);
-
- return l_ret_val;
-}
-
-
-
-/**
- * @brief Executes a PRAGMA statement.
- *
- * @param a_db a pointer to an instance of SQLite database structure
- * @param a_param a PRAGMA name
- * @param a_mode a PRAGMA value
- * @return Returns true if successful, otherwise false.
- */
-static int s_dap_db_driver_sqlite_set_pragma(sqlite3 *a_db, char *a_param, char *a_mode)
-{
-char l_query [512];
-int l_rc;
-
- if(!a_param || !a_mode)
- return log_it(L_ERROR, "%s - no param or mode\n", __PRETTY_FUNCTION__), false;
-
- snprintf(l_query, sizeof(l_query) - 1, "PRAGMA %s = %s", a_param, a_mode);
- l_rc = s_dap_db_driver_sqlite_exec(a_db, l_query, NULL); // default synchronous=FULL
-
- return (l_rc == SQLITE_OK);
-}
-
-
-
-/**
- * @brief Flushes a SQLite database cahce to disk.
- * @note The function closes and opens the database
- *
- * @return Returns 0 if successful.
- */
-static int s_dap_db_driver_sqlite_flush()
-{
- struct conn_pool_item *l_conn = s_sqlite_get_connection();
-
- char *l_error_message = NULL;
-
- log_it(L_DEBUG, "Start flush sqlite data base.");
-
- if(!(l_conn = s_sqlite_get_connection()) )
- return -666;
-
- s_dap_db_driver_sqlite_close(l_conn->conn);
-
- if ( !(l_conn->conn = dap_db_driver_sqlite_open(s_filename_db, SQLITE_OPEN_READWRITE, &l_error_message)) ) {
- log_it(L_ERROR, "Can't init sqlite err: \"%s\"", l_error_message? l_error_message: "UNKNOWN");
- s_dap_db_driver_sqlite_free(l_error_message);
- return -3;
- }
-
-#ifndef _WIN32
- sync();
-#endif
-
- if(!s_dap_db_driver_sqlite_set_pragma(l_conn->conn, "synchronous", "NORMAL")) // 0 | OFF | 1 | NORMAL | 2 | FULL
- log_it(L_WARNING, "Can't set new synchronous mode\n");
- if(!s_dap_db_driver_sqlite_set_pragma(l_conn->conn, "journal_mode", "OFF")) // DELETE | TRUNCATE | PERSIST | MEMORY | WAL | OFF
- log_it(L_WARNING, "Can't set new journal mode\n");
-
- if(!s_dap_db_driver_sqlite_set_pragma(l_conn->conn, "page_size", "1024")) // DELETE | TRUNCATE | PERSIST | MEMORY | WAL | OFF
- log_it(L_WARNING, "Can't set page_size\n");
-
- return 0;
-}
-
-
-
-
-/**
- * @brief Initializes a SQLite database.
- * @note no thread safe
- * @param a_filename_db a path to the database file
- * @param a_drv_callback a pointer to a structure of callback functions
- * @return If successful returns 0, else a code < 0.
- */
-int dap_db_driver_sqlite_init(const char *a_filename_db, dap_db_driver_callbacks_t *a_drv_callback)
-{
-int l_ret = -1, l_errno = errno;
-sqlite3 *l_conn;
-char l_errbuf[255] = {0}, *l_error_message = NULL;
-
-
- if ( sqlite3_threadsafe() && !sqlite3_config(SQLITE_CONFIG_SERIALIZED) )
- l_ret = sqlite3_initialize();
-
- if ( l_ret != SQLITE_OK ) {
- log_it(L_ERROR, "Can't init sqlite err=%d (%s)", l_ret, sqlite3_errstr(l_ret));
- return -2;
- }
-
- // Check paths and create them if nessesary
- char * l_filename_dir = dap_path_get_dirname(a_filename_db);
- strncpy(s_filename_db, a_filename_db, sizeof(s_filename_db) );
-
- if ( !dap_dir_test(l_filename_dir) ){
- log_it(L_NOTICE, "No directory %s, trying to create...",l_filename_dir);
-
- int l_mkdir_ret = dap_mkdir_with_parents(l_filename_dir);
- l_errno = errno;
-
- if(!dap_dir_test(l_filename_dir)){
- strerror_r(l_errno,l_errbuf,sizeof(l_errbuf));
- log_it(L_ERROR, "Can't create directory, error code %d, error string \"%s\"", l_mkdir_ret, l_errbuf);
- DAP_DELETE(l_filename_dir);
- return -l_errno;
- }else
- log_it(L_NOTICE, "Directory created");
- }
-
- DAP_DEL_Z(l_filename_dir);
-
- /* Create a pool of connection */
- for (int i = 0; i < DAP_SQLITE_POOL_COUNT; i++)
- {
- if ( !(l_conn = dap_db_driver_sqlite_open(a_filename_db, SQLITE_OPEN_READWRITE|SQLITE_OPEN_CREATE, &l_error_message)) )
- {
- log_it(L_ERROR, "Can't init SQL connection context #%d err: \"%s\"", i, l_error_message);
-
- s_dap_db_driver_sqlite_free(l_error_message);
- l_ret = -3;
- for(int ii = i - 1; ii >= 0; ii--) {
- s_dap_db_driver_sqlite_close(s_conn_pool[ii].conn);
- }
- goto end;
- }
-
- s_conn_pool[i].conn = l_conn;
- s_conn_pool[i].idx = i;
-
- atomic_store(&s_conn_pool[i].usage, 0);
-
- log_it(L_DEBUG, "SQL connection context #%d is created @%p", i, l_conn);
-
- if(!s_dap_db_driver_sqlite_set_pragma(l_conn, "synchronous", "NORMAL"))
- log_it(L_ERROR, "can't set new synchronous mode\n");
- if(!s_dap_db_driver_sqlite_set_pragma(l_conn, "journal_mode", "OFF"))
- log_it(L_ERROR, "can't set new journal mode\n");
- if(!s_dap_db_driver_sqlite_set_pragma(l_conn, "page_size", "4096"))
- log_it(L_ERROR, "can't set page_size\n");
- }
-
- // *PRAGMA page_size = bytes; // page size DB; it is reasonable to make it equal to the size of the disk cluster 4096
- // *PRAGMA cache_size = -kibibytes; // by default it is equal to 2000 pages of database
- //
- a_drv_callback->apply_store_obj = dap_db_driver_sqlite_apply_store_obj;
- a_drv_callback->read_store_obj = dap_db_driver_sqlite_read_store_obj;
- a_drv_callback->read_cond_store_obj = dap_db_driver_sqlite_read_cond_store_obj;
- a_drv_callback->read_last_store_obj = dap_db_driver_sqlite_read_last_store_obj;
- a_drv_callback->transaction_start = s_dap_db_driver_sqlite_start_transaction;
- a_drv_callback->transaction_end = s_dap_db_driver_sqlite_end_transaction;
- a_drv_callback->get_groups_by_mask = dap_db_driver_sqlite_get_groups_by_mask;
- a_drv_callback->read_count_store = dap_db_driver_sqlite_read_count_store;
- a_drv_callback->is_obj = dap_db_driver_sqlite_is_obj;
- a_drv_callback->deinit = dap_db_driver_sqlite_deinit;
- a_drv_callback->flush = s_dap_db_driver_sqlite_flush;
-
-end:
- return l_ret;
-}
diff --git a/modules/global-db/dap_chain_global_db_remote.c b/modules/global-db/dap_chain_global_db_remote.c
deleted file mode 100644
index fae3fe9e0376659ed2406fe3e729f7ab276fd486..0000000000000000000000000000000000000000
--- a/modules/global-db/dap_chain_global_db_remote.c
+++ /dev/null
@@ -1,797 +0,0 @@
-#include <string.h>
-#include <stdlib.h>
-
-#include "dap_global_db.h"
-#include "dap_chain_global_db_remote.h"
-#include "dap_common.h"
-#include "dap_strfuncs.h"
-#include "dap_string.h"
-#include "dap_chain.h"
-#include "dap_time.h"
-
-#define LOG_TAG "dap_chain_global_db_remote"
-
-// Default time of a node address expired in hours
-#define NODE_TIME_EXPIRED_DEFAULT 720
-
-/**
- * @brief A function for a thread for reading a log list
- *
- * @param arg a pointer to the log list structure
- * @return Returns NULL.
- */
-static void *s_list_thread_proc(void *arg)
-{
- dap_db_log_list_t *l_dap_db_log_list = (dap_db_log_list_t *)arg;
- uint32_t l_time_store_lim_hours = dap_config_get_item_uint32_default(g_config, "resources", "dap_global_db_time_store_limit", 72);
- uint64_t l_limit_time = l_time_store_lim_hours ? dap_nanotime_now() - dap_nanotime_from_sec(l_time_store_lim_hours * 3600) : 0;
- for (dap_list_t *l_groups = l_dap_db_log_list->groups; l_groups; l_groups = dap_list_next(l_groups)) {
- dap_db_log_list_group_t *l_group_cur = (dap_db_log_list_group_t *)l_groups->data;
- char *l_del_group_name_replace = NULL;
- char l_obj_type;
- if (!dap_fnmatch("*.del", l_group_cur->name, 0)) {
- l_obj_type = DAP_DB$K_OPTYPE_DEL;
- size_t l_del_name_len = strlen(l_group_cur->name) - 4; //strlen(".del");
- l_del_group_name_replace = DAP_NEW_SIZE(char, l_del_name_len + 1);
- memcpy(l_del_group_name_replace, l_group_cur->name, l_del_name_len);
- l_del_group_name_replace[l_del_name_len] = '\0';
- } else {
- l_obj_type = DAP_DB$K_OPTYPE_ADD;
- }
- uint64_t l_item_start = l_group_cur->last_id_synced + 1;
- dap_nanotime_t l_time_now = dap_nanotime_now();
- while (l_group_cur->count && l_dap_db_log_list->is_process) { // Number of records to be synchronized
- size_t l_item_count = min(64, l_group_cur->count);
- dap_store_obj_t *l_objs = dap_global_db_get_all_raw_sync(l_group_cur->name, l_item_start, &l_item_count);
- pthread_mutex_lock(&l_dap_db_log_list->list_mutex);
- if (!l_dap_db_log_list->is_process){
- pthread_mutex_unlock(&l_dap_db_log_list->list_mutex);
- if(l_objs)
- dap_store_obj_free(l_objs, l_item_count);
- return NULL;
- }
- pthread_mutex_unlock(&l_dap_db_log_list->list_mutex);
-
- // go to next group
- if (!l_objs)
- break;
- // set new start pos = lastitem pos + 1
- l_item_start = l_objs[l_item_count - 1].id + 1;
- l_group_cur->count -= l_item_count;
- dap_list_t *l_list = NULL;
- for (size_t i = 0; i < l_item_count; i++) {
- dap_store_obj_t *l_obj_cur = l_objs + i;
- if (!l_obj_cur)
- continue;
- l_obj_cur->type = l_obj_type;
- if (l_obj_cur->timestamp >> 32 == 0 ||
- l_obj_cur->timestamp > l_time_now ||
- l_obj_cur->group == NULL) {
- dap_chain_global_db_driver_delete(l_obj_cur, 1);
- continue; // the object is broken
- }
- if (l_obj_type == DAP_DB$K_OPTYPE_DEL) {
- if (l_limit_time && l_obj_cur->timestamp < l_limit_time) {
- dap_chain_global_db_driver_delete(l_obj_cur, 1);
- continue;
- }
- DAP_DELETE((char *)l_obj_cur->group);
- l_obj_cur->group = dap_strdup(l_del_group_name_replace);
- }
- dap_db_log_list_obj_t *l_list_obj = DAP_NEW_Z(dap_db_log_list_obj_t);
- uint64_t l_cur_id = l_obj_cur->id;
- l_obj_cur->id = 0;
- dap_global_db_pkt_t *l_pkt = dap_store_packet_single(l_obj_cur);
- dap_hash_fast(l_pkt->data, l_pkt->data_size, &l_list_obj->hash);
- dap_store_packet_change_id(l_pkt, l_cur_id);
- l_list_obj->pkt = l_pkt;
- l_list = dap_list_append(l_list, l_list_obj);
- }
- dap_store_obj_free(l_objs, l_item_count);
- pthread_mutex_lock(&l_dap_db_log_list->list_mutex);
- // add l_list to list_write
- l_dap_db_log_list->list_write = dap_list_concat(l_dap_db_log_list->list_write, l_list);
- // init read list if it ended already
- if(!l_dap_db_log_list->list_read)
- l_dap_db_log_list->list_read = l_list;
- pthread_mutex_unlock(&l_dap_db_log_list->list_mutex);
- }
- DAP_DEL_Z(l_del_group_name_replace);
- if (!l_dap_db_log_list->is_process)
- return NULL;
- }
-
- pthread_mutex_lock(&l_dap_db_log_list->list_mutex);
- l_dap_db_log_list->is_process = false;
- pthread_mutex_unlock(&l_dap_db_log_list->list_mutex);
- return NULL;
-}
-
-/**
- * @brief Starts a thread that readding a log list
- * @note instead dap_db_log_get_list()
- *
- * @param l_net net for sync
- * @param a_addr a pointer to the structure
- * @param a_flags flags
- * @return Returns a pointer to the log list structure if successful, otherwise NULL pointer.
- */
-dap_db_log_list_t* dap_db_log_list_start(dap_chain_net_t *l_net, dap_chain_node_addr_t a_addr, int a_flags)
-{
-#ifdef GDB_SYNC_ALWAYS_FROM_ZERO
- a_flags |= F_DB_LOG_SYNC_FROM_ZERO;
-#endif
-
- const char *l_net_name = NULL;
- if(l_net && l_net->pub.name && l_net->pub.name[0]!='\0') {
- l_net_name = l_net->pub.name;
- }
-
- //log_it(L_DEBUG, "Start loading db list_write...");
- dap_db_log_list_t *l_dap_db_log_list = DAP_NEW_Z(dap_db_log_list_t);
- // Add groups for the selected network only
- dap_list_t *l_groups_masks = dap_chain_db_get_sync_groups(l_net_name);
- if (a_flags & F_DB_LOG_ADD_EXTRA_GROUPS) {
- dap_list_t *l_extra_groups_masks = dap_chain_db_get_sync_extra_groups(l_net_name);
- l_groups_masks = dap_list_concat(l_groups_masks, l_extra_groups_masks);
- }
- dap_list_t *l_groups_names = NULL;
- for (dap_list_t *l_cur_mask = l_groups_masks; l_cur_mask; l_cur_mask = dap_list_next(l_cur_mask)) {
- char *l_cur_mask_data = ((dap_sync_group_item_t *)l_cur_mask->data)->group_mask;
- l_groups_names = dap_list_concat(l_groups_names, dap_chain_global_db_driver_get_groups_by_mask(l_cur_mask_data));
- }
- dap_list_free(l_groups_masks);
-
- static int16_t s_size_ban_list = -1;
- static char **s_ban_list = NULL;
-
- static int16_t s_size_white_list = -1;
- static char **s_white_list = NULL;
- static char **s_white_list_del = NULL;
-
- if (s_size_ban_list == -1)
- s_ban_list = dap_config_get_array_str(g_config, "stream_ch_chain", "ban_list_sync_groups", (uint16_t *)&s_size_ban_list);
- if (s_size_white_list == -1) {
- s_white_list = dap_config_get_array_str(g_config, "stream_ch_chain", "white_list_sync_groups", (uint16_t *)&s_size_white_list);
- if (s_size_white_list > 0) {
- s_white_list_del = DAP_NEW_SIZE(char *, s_size_white_list * sizeof(char *));
- for (int i = 0; i < s_size_white_list; i++) {
- s_white_list_del[i] = dap_strdup_printf("%s.del", s_white_list[i]);
- }
- }
- }
-
- /* delete if not condition */
- if (s_size_white_list > 0) {
- for (dap_list_t *l_group = l_groups_names; l_group; ) {
- bool l_found = false;
- for (int i = 0; i < s_size_white_list; i++) {
- if (!dap_fnmatch(s_white_list[i], l_group->data, FNM_NOESCAPE) ||
- !dap_fnmatch(s_white_list_del[i], l_group->data, FNM_NOESCAPE)) {
- l_found = true;
- break;
- }
- }
- if (!l_found) {
- dap_list_t *l_tmp = l_group->next;
- l_groups_names = dap_list_delete_link(l_groups_names, l_group);
- l_group = l_tmp;
- } else
- l_group = dap_list_next(l_group);
- }
- } else if (s_size_ban_list > 0) {
- for (dap_list_t *l_group = l_groups_names; l_group; ) {
- bool l_found = false;
- for (int i = 0; i < s_size_ban_list; i++) {
- if (!dap_fnmatch(s_ban_list[i], l_group->data, FNM_NOESCAPE)) {
- dap_list_t *l_tmp = l_group->next;
- l_groups_names = dap_list_delete_link(l_groups_names, l_group);
- l_group = l_tmp;
- l_found = true;
- break;
- }
- }
- if (l_found) continue;
- l_group = dap_list_next(l_group);
- }
- }
-
- l_dap_db_log_list->groups = l_groups_names; // repalce name of group with group item
- for (dap_list_t *l_group = l_dap_db_log_list->groups; l_group; l_group = dap_list_next(l_group)) {
- dap_db_log_list_group_t *l_sync_group = DAP_NEW_Z(dap_db_log_list_group_t);
- l_sync_group->name = (char *)l_group->data;
- if (a_flags & F_DB_LOG_SYNC_FROM_ZERO)
- l_sync_group->last_id_synced = 0;
- else
- l_sync_group->last_id_synced = dap_db_get_last_id_remote(a_addr.uint64, l_sync_group->name);
- l_sync_group->count = dap_chain_global_db_driver_count(l_sync_group->name, l_sync_group->last_id_synced + 1);
- l_dap_db_log_list->items_number += l_sync_group->count;
- l_group->data = (void *)l_sync_group;
- }
- l_dap_db_log_list->items_rest = l_dap_db_log_list->items_number;
- if (!l_dap_db_log_list->items_number) {
- DAP_DELETE(l_dap_db_log_list);
- return NULL;
- }
- l_dap_db_log_list->is_process = true;
- pthread_mutex_init(&l_dap_db_log_list->list_mutex, NULL);
- pthread_create(&l_dap_db_log_list->thread, NULL, s_list_thread_proc, l_dap_db_log_list);
- return l_dap_db_log_list;
-}
-
-/**
- * @brief Gets a number of objects from a log list.
- *
- * @param a_db_log_list a pointer to the log list structure
- * @return Returns the number if successful, otherwise 0.
- */
-size_t dap_db_log_list_get_count(dap_db_log_list_t *a_db_log_list)
-{
- if(!a_db_log_list)
- return 0;
- size_t l_items_number;
- pthread_mutex_lock(&a_db_log_list->list_mutex);
- l_items_number = a_db_log_list->items_number;
- pthread_mutex_unlock(&a_db_log_list->list_mutex);
- return l_items_number;
-}
-
-/**
- * @brief Gets a number of rest objects from a log list.
- *
- * @param a_db_log_list a pointer to the log list structure
- * @return Returns the number if successful, otherwise 0.
- */
-size_t dap_db_log_list_get_count_rest(dap_db_log_list_t *a_db_log_list)
-{
- if(!a_db_log_list)
- return 0;
- size_t l_items_rest;
- pthread_mutex_lock(&a_db_log_list->list_mutex);
- l_items_rest = a_db_log_list->items_rest;
- pthread_mutex_unlock(&a_db_log_list->list_mutex);
- return l_items_rest;
-}
-
-/**
- * @brief Gets an object from a list.
- *
- * @param a_db_log_list a pointer to the log list
- * @return Returns a pointer to the object.
- */
-dap_db_log_list_obj_t *dap_db_log_list_get(dap_db_log_list_t *a_db_log_list)
-{
- if (!a_db_log_list)
- return NULL;
- pthread_mutex_lock(&a_db_log_list->list_mutex);
- int l_is_process = a_db_log_list->is_process;
- // check next item
- dap_list_t *l_list = a_db_log_list->list_read;
- if (l_list){
- a_db_log_list->list_read = dap_list_next(a_db_log_list->list_read);
- a_db_log_list->items_rest--;
- }
- pthread_mutex_unlock(&a_db_log_list->list_mutex);
- //log_it(L_DEBUG, "get item n=%d", a_db_log_list->items_number - a_db_log_list->items_rest);
- return l_list ? (dap_db_log_list_obj_t *)l_list->data : DAP_INT_TO_POINTER(l_is_process);
-}
-
-void dap_db_log_list_rewind(dap_db_log_list_t *a_db_log_list)
-{
- if (!a_db_log_list)
- return;
- a_db_log_list->list_read = a_db_log_list->list_write;
- a_db_log_list->items_rest = a_db_log_list->items_number;
-}
-
-/**
- * @brief Deallocates memory of a list item
- *
- * @param a_item a pointer to the list item
- * @returns (none)
- */
-void dap_db_log_list_delete_item(void *a_item)
-{
- dap_db_log_list_obj_t *l_list_item = (dap_db_log_list_obj_t *)a_item;
- DAP_DELETE(l_list_item->pkt);
- DAP_DELETE(l_list_item);
-}
-
-/**
- * @brief Deallocates memory of a log list.
- *
- * @param a_db_log_list a pointer to the log list structure
- * @returns (none)
- */
-void dap_db_log_list_delete(dap_db_log_list_t *a_db_log_list)
-{
- if(!a_db_log_list)
- return;
- // stop thread if it has created
- if(a_db_log_list->thread) {
- pthread_mutex_lock(&a_db_log_list->list_mutex);
- a_db_log_list->is_process = false;
- pthread_mutex_unlock(&a_db_log_list->list_mutex);
- pthread_join(a_db_log_list->thread, NULL);
- }
- dap_list_free_full(a_db_log_list->groups, NULL);
- dap_list_free_full(a_db_log_list->list_write, (dap_callback_destroyed_t)dap_db_log_list_delete_item);
- pthread_mutex_destroy(&a_db_log_list->list_mutex);
- DAP_DELETE(a_db_log_list);
-}
-
-/**
- * @brief Sets a current node adress.
- * @param a_address a current node adress
- * @param a_net_name a net name string
- * @return True if success, otherwise false
- */
-static bool dap_db_set_cur_node_addr_common(uint64_t a_address, char *a_net_name, time_t a_expire_time)
-{
-char l_key [DAP_GLOBAL_DB_KEY_MAX];
-bool l_ret;
-
- if(!a_net_name)
- return false;
-
- dap_snprintf(l_key, sizeof(l_key) - 1, "cur_node_addr_%s", a_net_name);
-
- if ( (l_ret = dap_global_db_set(DAP_GLOBAL_DB_LOCAL_GENERAL, l_key, &a_address, sizeof(a_address),
- true, NULL, NULL)) == 0 ) {
- dap_snprintf(l_key, sizeof(l_key) - 1, "cur_node_addr_%s_time", a_net_name);
- l_ret = dap_global_db_set(DAP_GLOBAL_DB_LOCAL_GENERAL, l_key, &a_expire_time, sizeof(time_t),
- true, NULL, NULL);
- }
-
- return l_ret;
-}
-
-/**
- * @brief Sets an adress of a current node and no expire time.
- *
- * @param a_address an adress of a current node
- * @param a_net_name a net name string
- * @return Returns true if siccessful, otherwise false
- */
-bool dap_db_set_cur_node_addr(uint64_t a_address, char *a_net_name )
-{
- return dap_db_set_cur_node_addr_common(a_address,a_net_name,0);
-}
-
-/**
- * @brief Sets an address of a current node and expire time.
- *
- * @param a_address an address of a current node
- * @param a_net_name a net name string
- * @return Returns true if successful, otherwise false
- */
-bool dap_db_set_cur_node_addr_exp(uint64_t a_address, char *a_net_name )
-{
- return dap_db_set_cur_node_addr_common(a_address,a_net_name, time(NULL));
-}
-
-/**
- * @brief Gets an adress of current node by a net name.
- *
- * @param a_net_name a net name string
- * @return Returns an adress if successful, otherwise 0.
- */
-uint64_t dap_chain_net_get_cur_node_addr_gdb_sync(char *a_net_name)
-{
-char l_key[DAP_GLOBAL_DB_KEY_MAX], l_key_time[DAP_GLOBAL_DB_KEY_MAX];
-uint8_t *l_node_addr_data, *l_node_time_data;
-size_t l_node_addr_len = 0, l_node_time_len = 0;
-uint64_t l_node_addr_ret = 0;
-time_t l_node_time = 0;
-
- if(!a_net_name)
- return 0;
-
- dap_snprintf(l_key, sizeof(l_key) - 1, "cur_node_addr_%s", a_net_name);
- dap_snprintf(l_key_time, sizeof(l_key_time) - 1, "cur_node_addr_%s_time", a_net_name);
-
- l_node_addr_data = dap_global_db_get_sync(DAP_GLOBAL_DB_LOCAL_GENERAL, l_key, &l_node_addr_len, NULL, NULL);
- l_node_time_data = dap_global_db_get_sync(DAP_GLOBAL_DB_LOCAL_GENERAL, l_key_time, &l_node_time_len, NULL, NULL);
-
- if(l_node_addr_data && (l_node_addr_len == sizeof(uint64_t)) )
- l_node_addr_ret = *( (uint64_t *) l_node_addr_data );
-
- if(l_node_time_data && (l_node_time_len == sizeof(time_t)) )
- l_node_time = *( (time_t *) l_node_time_data );
-
- DAP_DELETE(l_node_addr_data);
- DAP_DELETE(l_node_time_data);
-
- // time delta in seconds
- static int64_t addr_time_expired = -1;
- // read time-expired
-
- if(addr_time_expired == -1) {
- dap_string_t *l_cfg_path = dap_string_new("network/");
- dap_string_append(l_cfg_path, a_net_name);
- dap_config_t *l_cfg;
-
- if((l_cfg = dap_config_open(l_cfg_path->str)) == NULL) {
- log_it(L_ERROR, "Can't open default network config");
- addr_time_expired = 0;
- } else {
- addr_time_expired = 3600 *
- dap_config_get_item_int64_default(l_cfg, "general", "node-addr-expired",
- NODE_TIME_EXPIRED_DEFAULT);
- }
- dap_string_free(l_cfg_path, true);
- }
-
- time_t l_dt = time(NULL) - l_node_time;
- //NODE_TIME_EXPIRED
- if(l_node_time && l_dt > addr_time_expired) {
- l_node_addr_ret = 0;
- }
-
- return l_node_addr_ret;
-}
-
-/**
- * @brief Sets last id of a remote node.
- *
- * @param a_node_addr a node adress
- * @param a_id id
- * @param a_group a group name string
- * @return Returns true if successful, otherwise false.
- */
-bool dap_db_set_last_id_remote(uint64_t a_node_addr, uint64_t a_id, char *a_group)
-{
-char l_key[DAP_GLOBAL_DB_KEY_MAX];
-
- dap_snprintf(l_key, sizeof(l_key) - 1, "%"DAP_UINT64_FORMAT_U"%s", a_node_addr, a_group);
- return dap_global_db_set(GROUP_LOCAL_NODE_LAST_ID,l_key, &a_id, sizeof(uint64_t), true, NULL, NULL ) == 0;
-}
-
-/**
- * @brief Gets last id of a remote node.
- *
- * @param a_node_addr a node adress
- * @param a_group a group name string
- * @return Returns id if successful, otherwise 0.
- */
-uint64_t dap_db_get_last_id_remote(uint64_t a_node_addr, char *a_group)
-{
- char *l_node_addr_str = dap_strdup_printf("%"DAP_UINT64_FORMAT_U"%s", a_node_addr, a_group);
- size_t l_id_len = 0;
- byte_t *l_id = dap_global_db_get_sync(GROUP_LOCAL_NODE_LAST_ID, l_node_addr_str, &l_id_len, NULL, NULL);
- uint64_t l_ret_id = 0;
- if (l_id) {
- if (l_id_len == sizeof(uint64_t))
- memcpy(&l_ret_id, l_id, l_id_len);
- DAP_DELETE(l_id);
- }
- DAP_DELETE(l_node_addr_str);
- return l_ret_id;
-}
-
-/**
- * @brief Sets the last hash of a remote node.
- *
- * @param a_node_addr a node adress
- * @param a_chain a pointer to the chain stucture
- * @param a_hash a
- * @return true
- * @return false
- */
-bool dap_db_set_last_hash_remote(uint64_t a_node_addr, dap_chain_t *a_chain, dap_chain_hash_fast_t *a_hash)
-{
-char l_key[DAP_GLOBAL_DB_KEY_MAX];
-
- dap_snprintf(l_key, sizeof(l_key) - 1, "%"DAP_UINT64_FORMAT_U"%s%s", a_node_addr, a_chain->net_name, a_chain->name);
- return dap_global_db_set(GROUP_LOCAL_NODE_LAST_ID, l_key, a_hash, sizeof(dap_chain_hash_fast_t), true, NULL, NULL ) == 0;
-}
-
-/**
- * @brief Gets the last hash of a remote node.
- *
- * @param a_node_addr a node adress
- * @param a_chain a pointer to a chain structure
- * @return Returns a hash if successful.
- */
-dap_chain_hash_fast_t *dap_db_get_last_hash_remote(uint64_t a_node_addr, dap_chain_t *a_chain)
-{
- char *l_node_chain_str = dap_strdup_printf("%ju%s%s", a_node_addr, a_chain->net_name, a_chain->name);
- size_t l_hash_len = 0;
- byte_t *l_hash = dap_global_db_get_sync(GROUP_LOCAL_NODE_LAST_ID,(const char*)l_node_chain_str, &l_hash_len,
- NULL, NULL);
- DAP_DELETE(l_node_chain_str);
- return (dap_chain_hash_fast_t *)l_hash;
-}
-
-/**
- * @brief Gets a size of an object.
- *
- * @param store_obj a pointer to the object
- * @return Returns the size.
- */
-static size_t dap_db_get_size_pdap_store_obj_t(pdap_store_obj_t store_obj)
-{
- size_t size = sizeof(uint32_t) + 2 * sizeof(uint16_t) +
- 3 * sizeof(uint64_t) + dap_strlen(store_obj->group) +
- dap_strlen(store_obj->key) + store_obj->value_len;
- return size;
-}
-
-/**
- * @brief Multiples data into a_old_pkt structure from a_new_pkt structure.
- * @param a_old_pkt a pointer to the old object
- * @param a_new_pkt a pointer to the new object
- * @return Returns a pointer to the multiple object
- */
-dap_global_db_pkt_t *dap_store_packet_multiple(dap_global_db_pkt_t *a_old_pkt, dap_global_db_pkt_t *a_new_pkt)
-{
- if (!a_new_pkt)
- return a_old_pkt;
- if (a_old_pkt)
- a_old_pkt = (dap_global_db_pkt_t *)DAP_REALLOC(a_old_pkt,
- a_old_pkt->data_size + a_new_pkt->data_size + sizeof(dap_global_db_pkt_t));
- else
- a_old_pkt = DAP_NEW_Z_SIZE(dap_global_db_pkt_t, a_new_pkt->data_size + sizeof(dap_global_db_pkt_t));
- memcpy(a_old_pkt->data + a_old_pkt->data_size, a_new_pkt->data, a_new_pkt->data_size);
- a_old_pkt->data_size += a_new_pkt->data_size;
- a_old_pkt->obj_count++;
- return a_old_pkt;
-}
-
-/**
- * @brief Changes id in a packed structure.
- *
- * @param a_pkt a pointer to the packed structure
- * @param a_id id
- * @return (none)
- */
-void dap_store_packet_change_id(dap_global_db_pkt_t *a_pkt, uint64_t a_id)
-{
- uint16_t l_gr_len = *(uint16_t*)(a_pkt->data + sizeof(uint32_t));
- size_t l_id_offset = sizeof(uint32_t) + sizeof(uint16_t) + l_gr_len;
- memcpy(a_pkt->data + l_id_offset, &a_id, sizeof(uint64_t));
-}
-
-/**
- * @brief Serializes an object into a packed structure.
- * @param a_store_obj a pointer to the object to be serialized
- * @return Returns a pointer to the packed sructure if successful, otherwise NULL.
- */
-dap_global_db_pkt_t *dap_store_packet_single(dap_store_obj_t *a_store_obj)
-{
-int len;
-unsigned char *pdata;
-
- if (!a_store_obj)
- return NULL;
-
- uint32_t l_data_size_out = dap_db_get_size_pdap_store_obj_t(a_store_obj);
- dap_global_db_pkt_t *l_pkt = DAP_NEW_SIZE(dap_global_db_pkt_t, l_data_size_out + sizeof(dap_global_db_pkt_t));
-
- /* Fill packet header */
- l_pkt->data_size = l_data_size_out;
- l_pkt->obj_count = 1;
- l_pkt->timestamp = 0;
-
- /* Put serialized data into the payload part of the packet */
- pdata = l_pkt->data;
- *( (uint32_t *) pdata) = a_store_obj->type; pdata += sizeof(uint32_t);
-
- len = dap_strlen(a_store_obj->group);
- *( (uint16_t *) pdata) = (uint16_t) len; pdata += sizeof(uint16_t);
- memcpy(pdata, a_store_obj->group, len); pdata += len;
-
- *( (uint64_t *) pdata) = a_store_obj->id; pdata += sizeof(uint64_t);
- *( (uint64_t *) pdata) = a_store_obj->timestamp; pdata += sizeof(uint64_t);
-
- len = dap_strlen(a_store_obj->key);
- *( (uint16_t *) pdata) = (uint16_t) len; pdata += sizeof(uint16_t);
- memcpy(pdata, a_store_obj->key, len); pdata += len;
-
- *( (uint64_t *) pdata) = a_store_obj->value_len; pdata += sizeof(uint64_t);
- memcpy(pdata, a_store_obj->value, a_store_obj->value_len); pdata += a_store_obj->value_len;
-
- assert( (uint32_t)(pdata - l_pkt->data) == l_data_size_out);
- return l_pkt;
-}
-
-/**
- * @brief Deserializes some objects from a packed structure into an array of objects.
- * @param pkt a pointer to the serialized packed structure
- * @param store_obj_count[out] a number of deserialized objects in the array
- * @return Returns a pointer to the first object in the array, if successful; otherwise NULL.
- */
-#if 0
-dap_store_obj_t *dap_global_db_pkt_deserialize(const dap_global_db_pkt_t *a_pkt, size_t *a_store_obj_count)
-{
- if(!a_pkt || a_pkt->data_size < sizeof(dap_global_db_pkt_t))
- return NULL;
- uint64_t l_offset = 0;
- uint32_t l_count = a_pkt->obj_count, l_cur_count;
- uint64_t l_size = l_count <= UINT16_MAX ? l_count * sizeof(struct dap_store_obj) : 0;
- dap_store_obj_t *l_store_obj = l_size? DAP_NEW_Z_SIZE(dap_store_obj_t, l_size) : NULL;
- if (!l_store_obj || !l_size) {
- log_it(L_ERROR, "Invalid size: can't allocate %"DAP_UINT64_FORMAT_U" bytes", l_size);
- DAP_DEL_Z(l_store_obj)
- return NULL;
- }
- for(l_cur_count = 0; l_cur_count < l_count; ++l_cur_count) {
- dap_store_obj_t *l_obj = l_store_obj + l_cur_count;
- uint16_t l_str_length;
-
- uint32_t l_type;
- if (l_offset+sizeof (uint32_t)> a_pkt->data_size) {log_it(L_ERROR, "Broken GDB element: can't read 'type' field"); break;} // Check for buffer boundries
- memcpy(&l_type, a_pkt->data + l_offset, sizeof(uint32_t));
- l_obj->type = l_type;
- l_offset += sizeof(uint32_t);
-
- if (l_offset+sizeof (uint16_t)> a_pkt->data_size) {log_it(L_ERROR, "Broken GDB element: can't read 'group_length' field"); break;} // Check for buffer boundries
- memcpy(&l_str_length, a_pkt->data + l_offset, sizeof(uint16_t));
- l_offset += sizeof(uint16_t);
-
- if (l_offset + l_str_length > a_pkt->data_size || !l_str_length) {log_it(L_ERROR, "Broken GDB element: can't read 'group' field"); break;} // Check for buffer boundries
- l_obj->group = DAP_NEW_Z_SIZE(char, l_str_length + 1);
- memcpy(l_obj->group, a_pkt->data + l_offset, l_str_length);
- l_offset += l_str_length;
-
- if (l_offset+sizeof (uint64_t)> a_pkt->data_size) {log_it(L_ERROR, "Broken GDB element: can't read 'id' field");
- DAP_DEL_Z(l_obj->group); break;} // Check for buffer boundries
- memcpy(&l_obj->id, a_pkt->data + l_offset, sizeof(uint64_t));
- l_offset += sizeof(uint64_t);
-
- if (l_offset+sizeof (uint64_t)> a_pkt->data_size) {log_it(L_ERROR, "Broken GDB element: can't read 'timestamp' field");
- DAP_DEL_Z(l_obj->group); break;} // Check for buffer boundries
- memcpy(&l_obj->timestamp, a_pkt->data + l_offset, sizeof(uint64_t));
- l_offset += sizeof(uint64_t);
-
- if (l_offset+sizeof (uint16_t)> a_pkt->data_size) {log_it(L_ERROR, "Broken GDB element: can't read 'key_length' field");
- DAP_DEL_Z(l_obj->group); break;} // Check for buffer boundries
- memcpy(&l_str_length, a_pkt->data + l_offset, sizeof(uint16_t));
- l_offset += sizeof(uint16_t);
-
- if (l_offset + l_str_length > a_pkt->data_size || !l_str_length) {log_it(L_ERROR, "Broken GDB element: can't read 'key' field: len %s",
- l_str_length ? "OVER" : "NULL");
- DAP_DEL_Z(l_obj->group); break;} // Check for buffer boundries
- l_obj->key = DAP_NEW_Z_SIZE(char, l_str_length + 1);
- memcpy((char *)l_obj->key, a_pkt->data + l_offset, l_str_length);
- l_offset += l_str_length;
-
- if (l_offset+sizeof (uint64_t)> a_pkt->data_size) {log_it(L_ERROR, "Broken GDB element: can't read 'value_length' field");
- DAP_DEL_Z(l_obj->group); DAP_DEL_Z(l_obj->key); break;} // Check for buffer boundries
- memcpy(&l_obj->value_len, a_pkt->data + l_offset, sizeof(uint64_t));
- l_offset += sizeof(uint64_t);
-
- if (l_offset + l_obj->value_len > a_pkt->data_size || !l_obj->value_len) {log_it(L_ERROR, "Broken GDB element: can't read 'value' field");
- DAP_DEL_Z(l_obj->group); DAP_DEL_Z(l_obj->key);break;} // Check for buffer boundries
- l_obj->value = DAP_NEW_Z_SIZE(uint8_t, l_obj->value_len);
- memcpy((char*)l_obj->value, a_pkt->data + l_offset, l_obj->value_len);
- l_offset += l_obj->value_len;
- }
- if (a_pkt->data_size != l_offset) {
- if (l_store_obj)
- dap_store_obj_free(l_store_obj, l_cur_count);
- return NULL;
- }
- // Return the number of completely filled dap_store_obj_t structures
- // because l_cur_count may be less than l_count due to too little memory
- if(a_store_obj_count)
- *a_store_obj_count = l_cur_count;
- return l_store_obj;
-}
-
-#endif
-
-
-
-
-
-
-dap_store_obj_t *dap_global_db_pkt_deserialize(const dap_global_db_pkt_t *a_pkt, size_t *a_store_obj_count)
-{
-uint32_t l_count, l_cur_count;
-uint64_t l_offset, l_size ;
-unsigned char *pdata, *pdata_end;
-dap_store_obj_t *l_store_obj_arr, *l_obj;
-
- if(!a_pkt || a_pkt->data_size < sizeof(dap_global_db_pkt_t))
- return NULL;
-
- l_count = a_pkt->obj_count;
- l_offset = 0;
- l_size = l_count <= UINT16_MAX ? l_count * sizeof(struct dap_store_obj) : 0;
-
- l_store_obj_arr = DAP_NEW_Z_SIZE(dap_store_obj_t, l_size);
-
- if (!l_store_obj_arr || !l_size)
- {
- log_it(L_ERROR, "Invalid size: can't allocate %"DAP_UINT64_FORMAT_U" bytes, errno=%d", l_size, errno);
- DAP_DEL_Z(l_store_obj_arr)
- return NULL;
- }
-
-
-
- pdata = (unsigned char *) a_pkt->data; /* Set <pdata> to begin of payload */
- pdata_end = pdata + a_pkt->data_size; /* Set <pdata_end> to end of payload area
- will be used to prevent out-of-buffer case */
- l_obj = l_store_obj_arr;
-
- for ( l_cur_count = l_count ; l_cur_count; l_cur_count--, l_obj++ )
- {
- if ( (pdata + sizeof (uint32_t)) > pdata_end ) /* Check for buffer boundaries */
- {log_it(L_ERROR, "Broken GDB element: can't read 'type' field"); break;}
- l_obj->type = *((uint32_t *) pdata);
- pdata += sizeof(uint32_t);
-
-
- if ( (pdata + sizeof (uint16_t)) > pdata_end )
- {log_it(L_ERROR, "Broken GDB element: can't read 'group_length' field"); break;}
- l_obj->group_len = *((uint16_t *) pdata);
- pdata += sizeof(uint16_t);
-
- if ( !l_obj->group_len )
- {log_it(L_ERROR, "Broken GDB element: 'group_len' field is zero"); break;}
-
-
- if ( (pdata + l_obj->group_len) > pdata_end )
- {log_it(L_ERROR, "Broken GDB element: can't read 'group' field"); break;}
- l_obj->group = DAP_NEW_Z_SIZE(char, l_obj->group_len + 1);
- memcpy(l_obj->group, pdata, l_obj->group_len);
- pdata += l_obj->group_len;
-
-
-
- if ( (pdata + sizeof (uint64_t)) > pdata_end )
- {log_it(L_ERROR, "Broken GDB element: can't read 'id' field"); break;}
- l_obj->id = *((uint64_t *) pdata);
- pdata += sizeof(uint64_t);
-
-
-
- if ( (pdata + sizeof (uint64_t)) > pdata_end )
- {log_it(L_ERROR, "Broken GDB element: can't read 'timestamp' field"); break;}
- l_obj->timestamp = *((uint64_t *) pdata);
- pdata += sizeof(uint64_t);
-
-
- if ( (pdata + sizeof (uint16_t)) > pdata_end)
- {log_it(L_ERROR, "Broken GDB element: can't read 'key_length' field"); break;}
- l_obj->key_len = *((uint16_t *) pdata);
- pdata += sizeof(uint16_t);
-
- if ( !l_obj->key_len )
- {log_it(L_ERROR, "Broken GDB element: 'key_length' field is zero"); break;}
-
- if ((pdata + l_obj->key_len) > pdata_end)
- {log_it(L_ERROR, "Broken GDB element: 'key_length' field is out from allocated memory"); break;}
-
- l_obj->key_byte = DAP_NEW_SIZE(byte_t, l_obj->key_len + 1);
- memcpy( l_obj->key_byte, pdata, l_obj->key_len);
- l_obj->key_byte[l_obj->key_len] = '\0';
- pdata += l_obj->key_len;
-
-
- if ( (pdata + sizeof (uint64_t)) > pdata_end )
- {log_it(L_ERROR, "Broken GDB element: can't read 'value_length' field"); break;}
- l_obj->value_len = *((uint64_t *) pdata);
- pdata += sizeof(uint64_t);
-
- if (l_obj->value_len) {
- if ( (pdata + l_obj->value_len) > pdata_end )
- {log_it(L_ERROR, "Broken GDB element: can't read 'value' field"); break;}
- l_obj->value = DAP_NEW_SIZE(uint8_t, l_obj->value_len);
- memcpy(l_obj->value, pdata, l_obj->value_len);
- pdata += l_obj->value_len;
- }
- }
-
- assert(pdata <= pdata_end);
-
- // Return the number of completely filled dap_store_obj_t structures
- // because l_cur_count may be less than l_count due to too little memory
- if (a_store_obj_count)
- *a_store_obj_count = l_count;
-
- return l_store_obj_arr;
-}
-
diff --git a/modules/global-db/dap_global_db.c b/modules/global-db/dap_global_db.c
deleted file mode 100644
index 61f5a5baa6453c8890aa85fe53150e57768b2f76..0000000000000000000000000000000000000000
--- a/modules/global-db/dap_global_db.c
+++ /dev/null
@@ -1,2053 +0,0 @@
-/*
- * Authors:
- * Dmitriy A. Gearasimov <gerasimov.dmitriy@demlabs.net>
- * Demlabs Ltd. https://demlabs.net
- * Copyright (c) 2022
- * All rights reserved.
-
- This file is part of DAP SDK the open source project
-
- DAP SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP SDK is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP SDK based project. If not, see <http://www.gnu.org/licenses/>.
-*/
-#include "uthash.h"
-#include "dap_common.h"
-#include "dap_config.h"
-#include "dap_strfuncs.h"
-#include "dap_file_utils.h"
-#include "dap_chain_common.h"
-#include "dap_time.h"
-#include "dap_context.h"
-#include "dap_worker.h"
-#include "dap_stream_worker.h"
-#include "dap_proc_thread.h"
-#define LOG_TAG "dap_global_db"
-
-#include "dap_global_db.h"
-#include "dap_global_db_sync.h"
-#include "dap_chain_global_db_driver.h"
-
-int g_dap_global_db_debug_more = false; /* Enable extensible debug output */
-
-
-// Queue I/O message op code
-enum queue_io_msg_opcode{
- MSG_OPCODE_UNDEFINED = 0,
- MSG_OPCODE_GET,
- MSG_OPCODE_GET_RAW,
- MSG_OPCODE_GET_DEL_TS,
- MSG_OPCODE_GET_LAST,
- MSG_OPCODE_GET_LAST_RAW,
- MSG_OPCODE_GET_ALL,
- MSG_OPCODE_GET_ALL_RAW,
- MSG_OPCODE_SET,
- MSG_OPCODE_SET_RAW,
- MSG_OPCODE_SET_MULTIPLE,
- MSG_OPCODE_PIN,
- MSG_OPCODE_UNPIN,
- MSG_OPCODE_DELETE,
- MSG_OPCODE_FLUSH,
- MSG_OPCODE_CONTEXT_EXEC
-};
-
-
-// Queue i/o message
-struct queue_io_msg{
- enum queue_io_msg_opcode opcode; // Opcode
-
- // For each message opcode we have only one callback
- union{
- dap_global_db_callback_t callback;
- dap_global_db_callback_result_t callback_result;
- dap_global_db_callback_result_raw_t callback_result_raw;
- dap_global_db_callback_results_t callback_results;
- dap_global_db_callback_results_raw_t callback_results_raw;
- };
- // Custom argument passed to the callback
- void * callback_arg;
- union{
- struct{ // Raw request
- dap_store_obj_t * values_raw;
- uint64_t values_raw_total;
- uint64_t values_raw_shift;
- };
- struct{ //deserialized requests
- // Different variant of message params
- union{
- // values for multile set
- struct{
- dap_global_db_obj_t * values;
- size_t values_count;
- };
-
- // Values for get multiple request
- struct{
- uint64_t values_shift; // For multiple records request here stores next request id
- uint64_t values_total; // Total values
- size_t values_page_size; // Maximum size of results page request. 0 means unlimited
- // TODO implement processing of this value
- };
-
- // Value for singe request
- struct{
- void * value;
- size_t value_length;
- bool value_is_pinned;
- };
-
- };
- char * group; // Group
- char * key; // Key
- };
- };
-
-};
-
-
-
-static uint32_t s_global_db_version = 0; // Current GlobalDB version
-static pthread_cond_t s_check_db_cond = PTHREAD_COND_INITIALIZER; // Check version condition
-static pthread_mutex_t s_check_db_mutex = PTHREAD_MUTEX_INITIALIZER; // Check version condition mutex
-static int s_check_db_ret = 0; // Check version return value
-
-static const char * s_storage_path = NULL; // GlobalDB storage path
-static const char * s_driver_name = NULL; // GlobalDB driver name
-
-static dap_context_t * s_context = NULL; // GlobalDB own context
-static struct dap_global_db_context * s_context_global_db = NULL; // GlobalDB own context custom extension
-
-// Version check& update functiosn
-static int s_check_db_version();
-static void s_check_db_version_callback_get (dap_global_db_context_t * a_global_db_context, int a_errno, const char * a_group, const char * a_key,
- const void * a_value, const size_t a_value_len,
- dap_nanotime_t value_ts,bool a_is_pinned, void * a_arg);
-static void s_check_db_version_callback_set (dap_global_db_context_t * a_global_db_context, int a_errno, const char * a_group, const char * a_key,
- const void * a_value, const size_t a_value_len,
- dap_nanotime_t value_ts,bool a_is_pinned, void * a_arg);
-
-// GlobalDB context start/stop callbacks
-static void s_context_callback_started( dap_context_t * a_context, void *a_arg);
-static void s_context_callback_stopped( dap_context_t * a_context, void *a_arg);
-
-
-// Opcode to string
-static const char *s_msg_opcode_to_str(enum queue_io_msg_opcode a_opcode);
-
-// Queue i/o processing callback
-static void s_queue_io_callback( dap_events_socket_t * a_es, void * a_arg);
-
-// Queue i/o message processing functions
-static bool s_msg_opcode_get(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_get_raw(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_get_del_ts(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_get_last(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_get_last_raw(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_get_all(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_get_all_raw(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_set(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_set_raw(struct queue_io_msg * a_msg);
-
-static bool s_msg_opcode_set_multiple_zc(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_pin(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_unpin(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_delete(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_flush(struct queue_io_msg * a_msg);
-static bool s_msg_opcode_context_exec(struct queue_io_msg * a_msg);
-
-// Free memor for queue i/o message
-static void s_queue_io_msg_delete( struct queue_io_msg * a_msg);
-
-// Delete history add and del
-static int s_record_del_history_add( char *a_key, char *a_group, uint64_t a_timestamp);
-static int s_record_del_history_del( const char *a_key, const char *a_group);
-
-// Call notificators
-static void s_change_notify(dap_store_obj_t * a_store_obj, char a_opcode);
-
-
-
-
-
-/**
- * @brief dap_global_db_init
- * @param a_path
- * @param a_driver
- * @return
- */
-int dap_global_db_init(const char * a_storage_path, const char * a_driver_name)
-{
- int l_rc = 0;
- static bool s_is_check_version = false;
-
- dap_global_db_sync_init();
-
- if ( a_storage_path == NULL && s_storage_path == NULL ){
- log_it(L_CRITICAL, "Can't initialize GlobalDB without storage path");
- }
-
- if ( a_driver_name == NULL && s_driver_name == NULL ){
- log_it(L_CRITICAL, "Can't initialize GlobalDB without driver name");
- }
-
- // For reinitialization it could be NULL but s_storage_path and s_driver_name have to be defined before
-
- if(a_storage_path)
- s_storage_path = dap_strdup(a_storage_path);
-
- if(a_driver_name)
- s_driver_name = dap_strdup(a_driver_name);
-
- // Debug config
- g_dap_global_db_debug_more = dap_config_get_item_bool_default(g_config, "resources", "debug_more", false);
-
-
- // Driver initalization
- if( (l_rc = dap_db_driver_init(s_driver_name, s_storage_path, true)) )
- return log_it(L_CRITICAL, "Hadn't initialized DB driver \"%s\" on path \"%s\", code: %d",
- s_driver_name, s_storage_path, l_rc), l_rc;
-
- // Create and run its own context
- if(s_context == NULL){
- s_context = dap_context_new(DAP_CONTEXT_TYPE_GLOBAL_DB);
- s_context->_inheritor = s_context_global_db = DAP_NEW_Z(struct dap_global_db_context);
- s_context_global_db->context = s_context;
- if (dap_context_run(s_context, -1, DAP_CONTEXT_POLICY_DEFAULT, 0, DAP_CONTEXT_FLAG_WAIT_FOR_STARTED,
- s_context_callback_started, s_context_callback_stopped, NULL) != 0 ){
- l_rc = -2;
- goto lb_return;
- }
- }
-
- // Check version and update if need it
- if(!s_is_check_version){
-
- s_is_check_version = true;
-
- if ( (l_rc = s_check_db_version()) )
- return log_it(L_ERROR, "GlobalDB version changed, please export or remove old version!"), l_rc;
- }
-
-lb_return:
- if (l_rc == 0 )
- log_it(L_NOTICE, "GlobalDB initialized");
- else
- log_it(L_CRITICAL, "GlobalDB wasn't initialized, code %d", l_rc);
-
- return l_rc;
-}
-
-/**
- * @brief dap_global_db_deinit
- */
-void dap_global_db_deinit()
-{
- dap_db_driver_deinit();
- dap_global_db_sync_deinit();
-}
-
-/**
- * @brief dap_global_db_get
- * @details Get record value from GlobalDB group by key
- * @param a_group
- * @param a_key
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_get(const char * a_group, const char *a_key, dap_global_db_callback_result_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_get");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_GET;
- l_msg->group = dap_strdup(a_group);
- l_msg->key = dap_strdup(a_key);
- l_msg->callback_result = a_callback;
- l_msg->callback_arg = a_arg;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec get request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent get request for %s:%s", a_group, a_key);
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_get
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_get(struct queue_io_msg * a_msg)
-{
- size_t l_count_records = 0;
- dap_store_obj_t *l_store_obj = dap_chain_global_db_driver_read( a_msg->group,
- a_msg->key,
- &l_count_records);
- if(l_store_obj != NULL && l_count_records>=1){
- if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_SUCCESS, l_store_obj->group, l_store_obj->key,
- l_store_obj->value, l_store_obj->value_len, l_store_obj->timestamp,
- l_store_obj->flags & RECORD_PINNED, a_msg->callback_arg );
- dap_store_obj_free(l_store_obj,l_count_records);
- }else if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_NO_RESULTS, a_msg->group, a_msg->key,
- NULL, 0, 0,0, a_msg->callback_arg );
- return true;
-}
-
-/**
- * @brief dap_global_db_get_raw
- * @param a_group
- * @param a_key
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_get_raw(const char * a_group, const char *a_key,dap_global_db_callback_result_raw_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_get");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_GET_RAW;
- l_msg->group = dap_strdup(a_group);
- l_msg->key = dap_strdup(a_key);
- l_msg->callback_result_raw = a_callback;
- l_msg->callback_arg = a_arg;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec get request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent get_raw request for %s:%s", a_group, a_key);
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_get_raw
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_get_raw(struct queue_io_msg * a_msg)
-{
- bool l_store_obj_delete = true;
- size_t l_count_records = 0;
- dap_store_obj_t *l_store_obj = dap_chain_global_db_driver_read( a_msg->group,
- a_msg->key,
- &l_count_records);
-
-
- if(a_msg->callback_result_raw)
- l_store_obj_delete = a_msg->callback_result_raw (s_context_global_db,
- l_store_obj ? DAP_GLOBAL_DB_RC_SUCCESS:
- DAP_GLOBAL_DB_RC_NO_RESULTS, l_store_obj,
- a_msg->callback_arg );
- if (l_store_obj && l_store_obj_delete)
- dap_store_obj_free(l_store_obj,l_count_records);
- return true;
-}
-
-/**
- * @brief dap_global_db_get_del_ts
- * @param a_group
- * @param a_key
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_get_del_ts(const char * a_group, const char *a_key,dap_global_db_callback_result_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_get");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_GET_DEL_TS;
- l_msg->group = dap_strdup(a_group);
- l_msg->key = dap_strdup(a_key);
- l_msg->callback_result = a_callback;
- l_msg->callback_arg = a_arg;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec get_del_ts request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent get_del_ts request for \"%s\" group \"%s\" key" , a_group, a_key);
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_get_del_ts
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_get_del_ts(struct queue_io_msg * a_msg)
-{
- uint64_t l_timestamp = 0;
- dap_store_obj_t l_store_obj_del = { 0 };
- char l_group[DAP_GLOBAL_DB_GROUP_NAME_SIZE_MAX];
- size_t l_count_out = 0;
- dap_store_obj_t *l_obj;
-
- if(a_msg->key && a_msg->group){
- l_store_obj_del.key = a_msg->key;
- dap_snprintf(l_group, sizeof(l_group) - 1, "%s.del", a_msg->group);
- l_store_obj_del.group = l_group;
-
- if (dap_chain_global_db_driver_is(l_store_obj_del.group, l_store_obj_del.key))
- {
- if ( (l_obj = dap_chain_global_db_driver_read(l_store_obj_del.group, l_store_obj_del.key, &l_count_out)) )
- {
- if ( (l_count_out > 1) )
- log_it(L_WARNING, "Got more then 1 records (%zu) for group '%s'", l_count_out, l_group);
-
- l_timestamp = l_obj->timestamp;
- dap_store_obj_free(l_obj, l_count_out);
- }
- }
- }
-
- if(l_timestamp){
- if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_SUCCESS, a_msg->group, a_msg->key,
- NULL, 0, l_timestamp,
- false, a_msg->callback_arg );
- }else if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_NO_RESULTS, a_msg->group, a_msg->key,
- NULL, 0, 0,0, a_msg->callback_arg );
- return true;
-}
-
-/**
- * @brief dap_global_db_get_last
- * @details Get the last value in GlobalDB group
- * @param a_group
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_get_last(const char * a_group, dap_global_db_callback_result_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_get_last");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_GET_LAST;
- l_msg->group = dap_strdup(a_group);
- l_msg->callback_arg = a_arg;
- l_msg->callback_result = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec get_last request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent get_last request for \"%s\" group", a_group);
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_get_last
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_get_last(struct queue_io_msg * a_msg)
-{
- dap_store_obj_t *l_store_obj = dap_chain_global_db_driver_read_last(a_msg->group);
- if(l_store_obj){
- if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_SUCCESS, l_store_obj->group, l_store_obj->key,
- l_store_obj->value, l_store_obj->value_len, l_store_obj->timestamp,
- l_store_obj->flags & RECORD_PINNED, a_msg->callback_arg );
- dap_store_obj_free(l_store_obj,1);
- }else if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_NO_RESULTS, a_msg->group, a_msg->key,
- NULL, 0, 0,0, a_msg->callback_arg );
- return true;
-}
-
-/**
- * @brief dap_global_db_get_last_raw
- * @param a_group
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_get_last_raw(const char * a_group, dap_global_db_callback_result_raw_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_get_last");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_GET_LAST_RAW;
- l_msg->group = dap_strdup(a_group);
- l_msg->callback_arg = a_arg;
- l_msg->callback_result_raw = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec get_last request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent get_last request for \"%s\" group", a_group);
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_get_last_raw
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_get_last_raw(struct queue_io_msg * a_msg)
-{
- dap_store_obj_t *l_store_obj = dap_chain_global_db_driver_read_last(a_msg->group);
- bool l_store_obj_delete = true;
- if(l_store_obj){
- if(a_msg->callback_result)
- l_store_obj_delete = a_msg->callback_result_raw(s_context_global_db, DAP_GLOBAL_DB_RC_SUCCESS, l_store_obj, a_msg->callback_arg );
- if(l_store_obj_delete)
- dap_store_obj_free(l_store_obj,1);
- }else if(a_msg->callback_result)
- a_msg->callback_result_raw(s_context_global_db, DAP_GLOBAL_DB_RC_NO_RESULTS, NULL, a_msg->callback_arg );
- return true;
-}
-
-
-
-/**
- * @brief dap_global_db_get_all Get all records from the group
- * @param a_group
- * @param a_results_page_size
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_get_all(const char * a_group,size_t a_results_page_size, dap_global_db_callback_results_t a_callback, void * a_arg )
-{
- // TODO make usable a_results_page_size
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_get_all");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_GET_ALL;
- l_msg->group = dap_strdup(a_group);
- l_msg->callback_arg = a_arg;
- l_msg->callback_results = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec get_all request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent get_all request for \"%s\" group", a_group);
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_get_all
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_get_all(struct queue_io_msg * a_msg)
-{
- size_t l_values_count = 0;
- if(! a_msg->values_total){ // First msg process
- a_msg->values_total = dap_chain_global_db_driver_count(a_msg->group,0);
- }
- dap_store_obj_t *l_store_objs = dap_chain_global_db_driver_cond_read(a_msg->group, a_msg->values_shift , &l_values_count);
- dap_global_db_obj_t *l_objs = NULL;
- debug_if(g_dap_global_db_debug_more, L_DEBUG,"Get all request from group %s recieved %zu values from total %zu",a_msg->group,
- l_values_count, a_msg->value_length );
- // Form objs from store_objs
- if(l_store_objs){
- l_objs = DAP_NEW_Z_SIZE(dap_global_db_obj_t,sizeof(dap_global_db_obj_t)*l_values_count);
- for(size_t i = 0; i < l_values_count; i++){
- l_objs[i].id = l_store_objs[i].id;
- l_objs[i].is_pinned = l_store_objs[i].flags & RECORD_PINNED;
- l_objs[i].key = dap_strdup(l_store_objs[i].key);
- l_objs[i].value = DAP_DUP_SIZE(l_store_objs[i].value, l_store_objs[i].value_len);
- l_objs[i].value_len = l_store_objs[i].value_len;
- l_objs[i].timestamp = l_store_objs[i].timestamp;
- }
- }
-
- bool l_delete_objs = true;
- // Call callback if present
- if(a_msg->callback_results)
- l_delete_objs = a_msg->callback_results(s_context_global_db, l_objs? DAP_GLOBAL_DB_RC_SUCCESS:DAP_GLOBAL_DB_RC_NO_RESULTS
- , a_msg->group, a_msg->key,
- a_msg->values_total,a_msg->values_shift, l_values_count,
- l_objs, a_msg->callback_arg );
- // Clean memory
- if(l_store_objs)
- dap_store_obj_free(l_store_objs,l_values_count);
- if(l_objs && l_delete_objs)
- dap_global_db_objs_delete(l_objs,l_values_count);
-
- // Here we also check if the reply was with zero values. To prevent endless loop we don't resend query request in such cases
- if(a_msg->values_total && l_values_count){
- // Check for values_shift overflow and update it
- if( a_msg->values_shift < (UINT64_MAX - l_values_count) &&
- l_values_count + a_msg->values_shift <= a_msg->values_total ){
- a_msg->values_shift += l_values_count;
-
- }else{
- log_it(L_WARNING, "Values overflow, can't grow up, values_shift:%"DAP_UINT64_FORMAT_U" values_total:%"DAP_UINT64_FORMAT_U" values_current:%zu", a_msg->values_shift, a_msg->values_total, l_values_count );
- a_msg->values_shift += a_msg->values_total;
- }
-
- if(a_msg->values_shift < a_msg->values_total ){ // Have to process callback again
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,a_msg);
- debug_if(g_dap_global_db_debug_more, L_NOTICE, "Resending get all request values_shift:%"DAP_UINT64_FORMAT_U" values_total:%"DAP_UINT64_FORMAT_U" values_current:%zu", a_msg->values_shift, a_msg->values_total, l_values_count );
- if ( l_ret ){
- log_it(L_ERROR,"Can't resend i/o message for opcode GET_ALL after value shift %"
- DAP_UINT64_FORMAT_U" (total values %" DAP_UINT64_FORMAT_U") error code %d", a_msg->values_shift, a_msg->values_total, l_ret);
- }else
- return false; // Don't delete it because it just sent again to the queue
- }
- }
- return true; // All values are sent
-}
-
-/**
- * @brief dap_global_db_get_all_raw
- * @param a_group
- * @param a_first_id
- * @param a_results_page_size
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_get_all_raw(const char * a_group, uint64_t a_first_id,size_t a_results_page_size, dap_global_db_callback_results_raw_t a_callback, void * a_arg )
-{
- // TODO make usable a_results_page_size
-
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_get_all");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_GET_ALL_RAW ;
- l_msg->group = dap_strdup(a_group);
- l_msg->values_raw_shift = a_first_id;
- l_msg->callback_arg = a_arg;
- l_msg->callback_results_raw = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec get_all_raw request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent get_all request for \"%s\" group", a_group);
- return l_ret;
-}
-
-/**
- * @brief Get all records in raw format inside GlobalDB context
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_get_all_raw(struct queue_io_msg * a_msg)
-{
- size_t l_values_count = 0;
- if(! a_msg->values_total){ // First msg process
- a_msg->values_raw_total = dap_chain_global_db_driver_count(a_msg->group,0);
- }
- uint64_t l_values_raw_shift = a_msg->values_raw_shift;
- dap_store_obj_t *l_store_objs = dap_chain_global_db_driver_cond_read(a_msg->group, l_values_raw_shift , &l_values_count);
- bool l_store_objs_delete = true;
- debug_if(g_dap_global_db_debug_more, L_DEBUG,"Get all raw request from group %s recieved %zu values from total %"DAP_UINT64_FORMAT_U,a_msg->group,
- l_values_count, a_msg->values_raw_total );
- // Call callback if present
- if(a_msg->callback_results_raw)
- l_store_objs_delete = a_msg->callback_results_raw(s_context_global_db, l_store_objs? DAP_GLOBAL_DB_RC_SUCCESS:DAP_GLOBAL_DB_RC_NO_RESULTS
- , a_msg->group, a_msg->key,l_values_count, a_msg->values_raw_shift,
- a_msg->values_raw_total,
- l_store_objs, a_msg->callback_arg );
- // Clean memory
- if(l_store_objs && l_store_objs_delete)
- dap_store_obj_free(l_store_objs,l_values_count);
-
- // Here we also check if the reply was with zero values. To prevent endless loop we don't resend query request in such cases
-
- if( a_msg->values_raw_total && l_values_count){
- // Check for values_shift overflow and update it
- if( a_msg->values_raw_total && l_values_count && a_msg->values_raw_shift< UINT64_MAX - l_values_count &&
- l_values_count + a_msg->values_raw_shift <= a_msg->values_raw_total ){
- l_values_raw_shift = (a_msg->values_raw_shift += l_values_count);
-
- } else{
- log_it(L_WARNING, "Values overflow, can't grow up, values_raw_shift:%"DAP_UINT64_FORMAT_U" values_raw_total:%"DAP_UINT64_FORMAT_U" values_raw_current:%zu",
- l_values_raw_shift, a_msg->values_raw_total, l_values_count );
- l_values_raw_shift = (a_msg->values_raw_shift = a_msg->values_raw_total);
- }
-
- if( a_msg->values_raw_total && l_values_count && a_msg->values_raw_shift < a_msg->values_raw_total){ // Have to process callback again
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,a_msg);
- debug_if(g_dap_global_db_debug_more, L_NOTICE, "Resending get all request values_raw_shift:%"DAP_UINT64_FORMAT_U" values_raw_total:%"DAP_UINT64_FORMAT_U" values_raw_current:%zu", a_msg->values_raw_shift, a_msg->values_raw_total, l_values_count );
-
- if ( l_ret ){
- log_it(L_ERROR,"Can't resend i/o message for opcode GET_ALL_RAW after value shift %"
- DAP_UINT64_FORMAT_U" error code %d", a_msg->values_shift,l_ret);
- }else
- return false; // Don't delete it because it just sent again to the queue
- }
- }
- return true; // All values are sent
-
-}
-
-
-/**
- * @brief Set GlobalDB record, identified with group and key
- * @param a_group Group name
- * @param a_key Key string
- * @param a_value Value data's pointer
- * @param a_value_length Value data's length
- * @param a_pin_value Pin value or not
- * @param a_callback Callback executed after request processing
- * @param a_arg Argument passed to the callback
- * @return 0 if success, error code if not
- */
-int dap_global_db_set(const char * a_group, const char *a_key, const void * a_value, const size_t a_value_length, bool a_pin_value, dap_global_db_callback_result_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_set");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_SET;
- l_msg->group = dap_strdup(a_group);
- l_msg->key = dap_strdup(a_key);
- l_msg->value = DAP_DUP_SIZE(a_value, a_value_length);
- l_msg->value_length = a_value_length;
- l_msg->value_is_pinned = a_pin_value;
- l_msg->callback_arg = a_arg;
- l_msg->callback_result = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec set request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent sent request for \"%s\" group \"%s\" key" , a_group, a_key);
-
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_set
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_set(struct queue_io_msg * a_msg)
-{
- dap_store_obj_t l_store_data = { 0 };
- dap_nanotime_t l_ts_now = dap_nanotime_now();
- l_store_data.key = a_msg->key ;
- l_store_data.flags =a_msg->value_is_pinned ? RECORD_PINNED : 0 ;
- l_store_data.value_len = a_msg->value_length;
- l_store_data.value = a_msg->value;
- l_store_data.group = (char*) a_msg->group;
- l_store_data.timestamp = l_ts_now;
-
- int l_res = dap_chain_global_db_driver_add(&l_store_data, 1);
-
- if (l_res == 0){
- s_record_del_history_del( a_msg->group, a_msg->key);
- if(a_msg->callback_result){
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_SUCCESS, a_msg->group, a_msg->key,
- a_msg->value, a_msg->value_length, l_ts_now,
- a_msg->value_is_pinned , a_msg->callback_arg );
- }
- s_change_notify(&l_store_data, DAP_DB$K_OPTYPE_ADD);
-
- }else{
- log_it(L_ERROR, "Save error for %s:%s code %d", a_msg->group,a_msg->key, l_res);
- if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_ERROR , a_msg->group, a_msg->key,
- a_msg->value, a_msg->value_length, l_ts_now,
- a_msg->value_is_pinned , a_msg->callback_arg );
- }
- if(a_msg->value)
- DAP_DELETE(a_msg->value);
- return true;
-}
-
-/**
- * @brief dap_global_db_set_unsafe
- * @param a_global_db_context
- * @param a_group
- * @param a_key
- * @param a_value
- * @param a_value_length
- * @param a_pin_value
- * @return
- */
-int dap_global_db_set_unsafe(dap_global_db_context_t * a_global_db_context, const char * a_group, const char *a_key, const void * a_value, const size_t a_value_length, bool a_pin_value )
-{
- dap_store_obj_t l_store_data = { 0 };
- dap_nanotime_t l_ts_now = dap_nanotime_now();
- l_store_data.key = (char *)a_key ;
- l_store_data.flags = a_pin_value ? RECORD_PINNED : 0 ;
- l_store_data.value_len = a_value_length;
- l_store_data.value = (uint8_t *)a_value;
- l_store_data.group = (char*) a_group;
- l_store_data.timestamp = l_ts_now;
-
- int l_res = dap_chain_global_db_driver_add(&l_store_data, 1);
- if(l_res == 0 ){
- s_record_del_history_del( a_group, a_key);
- s_change_notify(&l_store_data, DAP_DB$K_OPTYPE_ADD);
- }
- return l_res;
-}
-
-
-/**
- * @brief dap_global_db_set_raw
- * @param a_store_objs
- * @param a_store_objs_count
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_set_raw(dap_store_obj_t * a_store_objs, size_t a_store_objs_count, dap_global_db_callback_results_raw_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_set");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_SET_RAW;
- l_msg->callback_arg = a_arg;
- l_msg->callback_results_raw = a_callback;
-
- l_msg->values_raw = dap_store_obj_copy(a_store_objs,a_store_objs_count) ;
- l_msg->values_raw_total = a_store_objs_count;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec set_raw request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent set_raw request for %zu objects" , a_store_objs_count);
- return l_ret;
-
-}
-
-/**
- * @brief s_msg_opcode_set_raw
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_set_raw(struct queue_io_msg * a_msg)
-{
- int l_ret = -1;
- size_t i=0;
- if(a_msg->values_raw_total>0){
- l_ret = dap_chain_global_db_driver_add(a_msg->values_raw,a_msg->values_raw_total);
- int l_res_del = 0;
- if(l_ret == 0) {
- for(; i < a_msg->values_raw_total ; i++ ) {
- if (a_msg->values_raw[i].type == DAP_DB$K_OPTYPE_ADD)
- l_res_del = s_record_del_history_del(a_msg->values_raw[i].key , a_msg->values_raw[i].group);
- else if (a_msg->values_raw[i].type == DAP_DB$K_OPTYPE_DEL)
- l_res_del = s_record_del_history_add((char*)a_msg->values_raw[i].key, a_msg->values_raw[i].group,
- a_msg->values_raw[i].timestamp);
- if (!l_res_del) {
- s_change_notify(&a_msg->values_raw[i] , a_msg->values_raw[i].type);
- }
- }
- }else
- log_it(L_ERROR,"Can't save raw gdb data, code %d ", l_ret);
- }
- if(a_msg->callback_results_raw){
- a_msg->callback_results_raw (s_context_global_db, l_ret==0 ? DAP_GLOBAL_DB_RC_SUCCESS:
- DAP_GLOBAL_DB_RC_ERROR, a_msg->group, a_msg->key,
- a_msg->values_raw_total, 0, a_msg->values_raw_total, a_msg->values_raw ,
- a_msg->callback_arg );
- }
- dap_store_obj_free(a_msg->values_raw, a_msg->values_raw_total);
- return true;
-}
-
-/**
- * @brief dap_global_db_set_multiple_zc Set multiple values, without duplication (zero copy, values are freed after set callback execution )
- * @param a_group
- * @param a_values
- * @param a_values_count
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_set_multiple_zc(const char * a_group, dap_global_db_obj_t * a_values, size_t a_values_count, dap_global_db_callback_results_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_set");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_SET_MULTIPLE;
- l_msg->group = dap_strdup(a_group);
- l_msg->values = a_values;
- l_msg->values_count = a_values_count;
- l_msg->callback_arg = a_arg;
- l_msg->callback_results = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec set_multiple request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent set_multiple request for \"%s\" group with %zu values" , a_group, a_values_count);
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_set_multiple
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_set_multiple_zc(struct queue_io_msg * a_msg)
-{
- int l_ret = -1;
- size_t i=0;
- if(a_msg->values_count>0){
- dap_store_obj_t l_store_obj;
- l_ret = 0;
- for(; i < a_msg->values_count && l_ret == 0 ; i++ ) {
- memset(&l_store_obj,0,sizeof(l_store_obj));
- l_store_obj.type = DAP_DB$K_OPTYPE_ADD;
- l_store_obj.flags = a_msg->values[i].is_pinned;
- l_store_obj.key = a_msg->values[i].key;
- l_store_obj.group = a_msg->group;
- l_store_obj.value = a_msg->values[i].value;
- l_store_obj.value_len = a_msg->values[i].value_len;
- l_store_obj.timestamp = a_msg->values[i].timestamp;
- s_record_del_history_del(a_msg->values[i].key, a_msg->group);
- l_ret = dap_chain_global_db_driver_add(&l_store_obj,1);
- s_change_notify(&l_store_obj , DAP_DB$K_OPTYPE_ADD);
-
- }
- }
- bool l_delete_values = true;
- if(a_msg->callback_results){
- l_delete_values = a_msg->callback_results(s_context_global_db, l_ret==0 ? DAP_GLOBAL_DB_RC_SUCCESS:
- DAP_GLOBAL_DB_RC_ERROR, a_msg->group, a_msg->key,
- i, 0, a_msg->values_count, a_msg->values ,
- a_msg->callback_arg );
- }
- dap_global_db_objs_delete( a_msg->values, a_msg->values_count);
-
- return true;
-}
-
-/**
- * @brief dap_global_db_pin
- * @param a_group
- * @param a_key
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_pin(const char * a_group, const char *a_key, dap_global_db_callback_result_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_pin");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_PIN;
- l_msg->group = dap_strdup(a_group);
- l_msg->key = dap_strdup(a_key);
- l_msg->callback_arg = a_arg;
- l_msg->callback_result = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec pin request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent pin request for \"%s\" group \"%s\" key" , a_group, a_key);
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_pin
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_pin(struct queue_io_msg * a_msg)
-{
- size_t l_count_records = 0;
- dap_store_obj_t *l_store_obj = dap_chain_global_db_driver_read( a_msg->group,
- a_msg->key,
- &l_count_records);
- if(l_store_obj != NULL && l_count_records>=1){
- l_store_obj->flags |= RECORD_PINNED;
- int l_res = dap_chain_global_db_driver_add(l_store_obj,1);
- if(l_res == 0){
- s_record_del_history_del(a_msg->key, a_msg->group);
- s_change_notify(l_store_obj , l_store_obj->type );
- }else
- log_it(L_ERROR,"Can't save pinned gdb data, code %d ", l_res);
-
- if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, l_res == 0? DAP_GLOBAL_DB_RC_SUCCESS : DAP_GLOBAL_DB_RC_ERROR, l_store_obj->group, l_store_obj->key,
- l_store_obj->value, l_store_obj->value_len, l_store_obj->timestamp,
- l_store_obj->flags & RECORD_PINNED, a_msg->callback_arg );
- dap_store_obj_free(l_store_obj,l_count_records);
- }else if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_NO_RESULTS, a_msg->group, a_msg->key,
- NULL, 0, 0,0, a_msg->callback_arg );
- return true;
-}
-
-/**
- * @brief dap_global_db_unpin
- * @param a_group
- * @param a_key
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_unpin(const char * a_group, const char *a_key, dap_global_db_callback_result_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_unpin");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_UNPIN;
- l_msg->group = dap_strdup(a_group);
- l_msg->key = dap_strdup(a_key);
- l_msg->callback_arg = a_arg;
- l_msg->callback_result = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec unpin request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent unpin request for \"%s\" group \"%s\" key" , a_group, a_key);
-
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_unpin
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_unpin(struct queue_io_msg * a_msg)
-{
- size_t l_count_records = 0;
- dap_store_obj_t *l_store_obj = dap_chain_global_db_driver_read( a_msg->group,
- a_msg->key,
- &l_count_records);
- if(l_store_obj != NULL && l_count_records>=1){
- l_store_obj->flags ^= RECORD_PINNED;
- int l_res = dap_chain_global_db_driver_add(l_store_obj,1);
- if(l_res == 0){
- s_record_del_history_del(a_msg->key, a_msg->group);
- s_change_notify(l_store_obj , l_store_obj->type );
- }else
- log_it(L_ERROR,"Can't save pinned gdb data, code %d ", l_res);
-
- if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, l_res == 0? DAP_GLOBAL_DB_RC_SUCCESS : DAP_GLOBAL_DB_RC_ERROR, l_store_obj->group, l_store_obj->key,
- l_store_obj->value, l_store_obj->value_len, l_store_obj->timestamp,
- l_store_obj->flags & RECORD_PINNED, a_msg->callback_arg );
- dap_store_obj_free(l_store_obj,l_count_records);
- }else if(a_msg->callback_result)
- a_msg->callback_result(s_context_global_db, DAP_GLOBAL_DB_RC_NO_RESULTS, a_msg->group, a_msg->key,
- NULL, 0, 0,0, a_msg->callback_arg );
- return true;
-}
-
-/**
- * @brief dap_global_db_delete
- * @param a_group
- * @param a_key
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_del(const char * a_group, const char *a_key, dap_global_db_callback_result_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_delete");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_DELETE;
- l_msg->group = dap_strdup(a_group);
- l_msg->key = dap_strdup(a_key);
- l_msg->callback_arg = a_arg;
- l_msg->callback_result = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec del request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }else
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "Have sent del request for \"%s\" group \"%s\" key" , a_group, a_key);
-
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_delete
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_delete(struct queue_io_msg * a_msg)
-{
-
- int l_res = dap_global_db_del_unsafe(s_context_global_db, a_msg->group, a_msg->key);
-
- if(a_msg->callback_result){
- a_msg->callback_result(s_context_global_db, l_res==0 ? DAP_GLOBAL_DB_RC_SUCCESS:
- DAP_GLOBAL_DB_RC_ERROR,
- a_msg->group, a_msg->key,
- NULL, 0, 0 , false, a_msg->callback_arg );
- }
-
- return true;
-}
-
-/**
- * @brief Deallocates memory of an objs array.
- * @param objs a pointer to the first object of the array
- * @param a_count a number of objects in the array
- * @return (none)
- */
-void dap_global_db_objs_delete(dap_global_db_obj_t *a_objs, size_t a_count)
-{
-dap_global_db_obj_t *l_obj;
-
- if ( !a_objs || !a_count ) /* Sanity checks */
- return;
-
- for(l_obj = a_objs; a_count--; l_obj++) /* Run over array's elements */
- {
- DAP_DELETE(l_obj->key);
- DAP_DELETE(l_obj->value);
- }
-
- DAP_DELETE(a_objs); /* Finaly kill the the array */
-}
-
-/**
- * @brief The objs_get struct
- */
-struct objs_get{
- pthread_mutex_t mutex;
- pthread_cond_t cond;
- dap_global_db_obj_t * objs;
- size_t objs_count;
-};
-
-/**
- * @brief s_objs_get_callback
- * @param a_global_db_context
- * @param a_rc
- * @param a_group
- * @param a_key
- * @param a_values_total
- * @param a_values_shift
- * @param a_values_count
- * @param a_values
- * @param a_arg
- */
-static bool s_objs_get_callback (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg)
-{
- struct objs_get * l_args = (struct objs_get *) a_arg;
- l_args->objs = a_values;
- l_args->objs_count = a_values_count;
- pthread_mutex_lock(&l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_cond_broadcast(&l_args->cond);
- return false;
-}
-
-/**
- * @brief Sync (blocking) function for retrieving of list of GDB content
- * @param a_group
- * @param a_fist_id
- * @param a_objs_count
- * @return Group's objects
- */
-dap_global_db_obj_t* dap_global_db_get_all_sync(const char *a_group, size_t *a_objs_count)
-{
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "get_all sync call executes for group \"%s\"", a_group);
- struct objs_get * l_args = DAP_NEW_Z(struct objs_get);
- pthread_mutex_init(&l_args->mutex,NULL);
- pthread_cond_init(&l_args->cond,NULL);
- pthread_mutex_lock(&l_args->mutex);
- dap_global_db_get_all(a_group, 0,s_objs_get_callback, l_args);
- pthread_cond_wait(&l_args->cond, &l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_mutex_destroy(&l_args->mutex);
- pthread_cond_destroy(&l_args->cond);
-
- dap_global_db_obj_t * l_ret = l_args->objs;
- if(l_args->objs_count)
- *a_objs_count = l_args->objs_count;
- DAP_DELETE(l_args);
- return l_ret;
-}
-
-
-/**
- * @brief The objs_set struct
- */
-struct sync_op_result{
- pthread_mutex_t mutex;
- pthread_cond_t cond;
- int result;
-};
-
-/**
- * @brief s_objs_set_callback
- * @param a_global_db_context
- * @param a_rc
- * @param a_group
- * @param a_key
- * @param a_value
- * @param a_value_len
- * @param a_value_ts
- * @param a_is_pinned
- * @param a_arg
- */
-static void s_sync_op_result_callback (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const void * a_value, const size_t a_value_len, dap_nanotime_t a_value_ts, bool a_is_pinned, void * a_arg)
-{
- struct sync_op_result * l_args = (struct sync_op_result *) a_arg;
- l_args->result = a_rc;
- pthread_mutex_lock(&l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_cond_broadcast(&l_args->cond);
-}
-
-/**
- * @brief Put value into the GlobalDB and waits for result
- * @param a_group
- * @param a_key
- * @param a_value
- * @param a_value_length
- * @param a_pin_value
- * @return 0 if success others if not
- */
-int dap_global_db_set_sync(const char * a_group, const char *a_key, const void * a_value, const size_t a_value_length, bool a_pin_value )
-{
- struct sync_op_result * l_args = DAP_NEW_Z(struct sync_op_result);
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "set sync call executes for group \"%s\" key \"%s\"", a_group, a_key);
-
- pthread_mutex_init(&l_args->mutex,NULL);
- pthread_cond_init(&l_args->cond,NULL);
- pthread_mutex_lock(&l_args->mutex);
- dap_global_db_set(a_group, a_key,a_value,a_value_length, a_pin_value, s_sync_op_result_callback, l_args);
- pthread_cond_wait(&l_args->cond, &l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_mutex_destroy(&l_args->mutex);
- pthread_cond_destroy(&l_args->cond);
-
- int l_ret = l_args->result ;
- DAP_DELETE(l_args);
- return l_ret;
-}
-
-/**
- * @brief dap_global_db_pin_sync
- * @param a_group
- * @param a_key
- * @return
- */
-int dap_global_db_pin_sync(const char * a_group, const char *a_key)
-{
- struct sync_op_result * l_args = DAP_NEW_Z(struct sync_op_result);
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "pin sync call executes for group \"%s\" key \"%s\"", a_group, a_key);
-
- pthread_mutex_init(&l_args->mutex,NULL);
- pthread_cond_init(&l_args->cond,NULL);
- pthread_mutex_lock(&l_args->mutex);
- dap_global_db_pin(a_group, a_key, s_sync_op_result_callback, l_args);
- pthread_cond_wait(&l_args->cond, &l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_mutex_destroy(&l_args->mutex);
- pthread_cond_destroy(&l_args->cond);
-
- int l_ret = l_args->result ;
- DAP_DELETE(l_args);
- return l_ret;
-}
-
-/**
- * @brief dap_global_db_unpin_sync
- * @param a_group
- * @param a_key
- * @return
- */
-int dap_global_db_unpin_sync(const char * a_group, const char *a_key)
-{
- struct sync_op_result * l_args = DAP_NEW_Z(struct sync_op_result);
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "pin sync call executes for group \"%s\" key \"%s\"", a_group, a_key);
-
- pthread_mutex_init(&l_args->mutex,NULL);
- pthread_cond_init(&l_args->cond,NULL);
- pthread_mutex_lock(&l_args->mutex);
- dap_global_db_unpin(a_group, a_key, s_sync_op_result_callback, l_args);
- pthread_cond_wait(&l_args->cond, &l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_mutex_destroy(&l_args->mutex);
- pthread_cond_destroy(&l_args->cond);
-
- int l_ret = l_args->result ;
- DAP_DELETE(l_args);
- return l_ret;
-}
-
-/**
- * @brief dap_global_db_del_sync
- * @param a_group
- * @param a_key
- * @return
- */
-int dap_global_db_del_sync(const char * a_group, const char *a_key )
-{
- struct sync_op_result * l_args = DAP_NEW_Z(struct sync_op_result);
- pthread_mutex_init(&l_args->mutex,NULL);
- pthread_cond_init(&l_args->cond,NULL);
- pthread_mutex_lock(&l_args->mutex);
- if (dap_global_db_del(a_group, a_key, s_sync_op_result_callback, l_args) == 0)
- pthread_cond_wait(&l_args->cond, &l_args->mutex);
- else
- l_args->result = -777;
- pthread_mutex_unlock(&l_args->mutex);
- pthread_mutex_destroy(&l_args->mutex);
- pthread_cond_destroy(&l_args->cond);
-
- int l_ret = l_args->result ;
- DAP_DELETE(l_args);
- return l_ret;
-}
-
-/**
- * @brief The store_obj_get struct
- */
-struct store_obj_get{
- pthread_mutex_t mutex;
- pthread_cond_t cond;
- byte_t * data;
- size_t data_size;
- dap_nanotime_t ts;
- bool is_pinned;
-};
-
-/**
- * @brief s_store_obj_get_callback
- * @param a_global_db_context
- * @param a_rc
- * @param a_group
- * @param a_key
- * @param a_value
- * @param a_value_size
- * @param a_value_ts
- * @param a_is_pinned
- * @param a_arg
- */
-static void s_store_obj_get_callback (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key,
- const void * a_value, const size_t a_value_size, dap_nanotime_t a_value_ts,
- bool a_is_pinned, void * a_arg)
-{
- struct store_obj_get * l_args = (struct store_obj_get *) a_arg;
- assert(l_args);
-
- if( a_value && a_value_size){
- l_args->data = DAP_DUP_SIZE(a_value, a_value_size);
- l_args->data_size = a_value_size;
- l_args->ts = a_value_ts;
- l_args->is_pinned = a_is_pinned;
- }
-
- pthread_mutex_lock(&l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_cond_broadcast(&l_args->cond);
-}
-
-/**
- * @brief dap_global_db_gr_get_sync
- * @param a_group
- * @param a_key
- * @param a_data_size
- * @param a_is_pinned
- * @param a_ts
- * @return
- */
-byte_t* dap_global_db_get_sync(const char * a_group,const char *a_key, size_t *a_data_size, bool *a_is_pinned, dap_nanotime_t * a_ts)
-{
- struct store_obj_get * l_args = DAP_NEW_Z(struct store_obj_get);
- pthread_mutex_init(&l_args->mutex,NULL);
- pthread_cond_init(&l_args->cond,NULL);
- pthread_mutex_lock(&l_args->mutex);
-
-
-
- dap_global_db_get(a_group,a_key, s_store_obj_get_callback, l_args);
- pthread_cond_wait(&l_args->cond, &l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_mutex_destroy(&l_args->mutex);
- pthread_cond_destroy(&l_args->cond);
-
- byte_t * l_ret = l_args->data;
- if( l_ret ){
- if( a_data_size)
- *a_data_size = l_args->data_size;
- if( a_is_pinned)
- *a_is_pinned = l_args->is_pinned;
- if( a_ts)
- *a_ts = l_args->ts;
- }
- DAP_DELETE(l_args);
- return l_ret;
-}
-
-
-/**
- * @brief The store_objs_get struct
- */
-struct store_objs_get{
- pthread_mutex_t mutex;
- pthread_cond_t cond;
- dap_store_obj_t * objs;
- size_t objs_count;
-};
-
-static bool s_get_all_raw_sync_callback (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_store_obj_t * a_values, void * a_arg)
-{
- struct store_objs_get * l_args = (struct store_objs_get *) a_arg;
- l_args->objs = a_values;
- l_args->objs_count = a_values_count;
- pthread_mutex_lock(&l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_cond_broadcast(&l_args->cond);
- return false;
-}
-
-dap_store_obj_t* dap_global_db_get_all_raw_sync(const char *a_group, uint64_t a_first_id, size_t *a_objs_count)
-{
- struct store_objs_get * l_args = DAP_NEW_Z(struct store_objs_get);
- debug_if(g_dap_global_db_debug_more, L_DEBUG, "get_all_raw sync call executes for group %s", a_group);
-
- pthread_mutex_init(&l_args->mutex,NULL);
- pthread_cond_init(&l_args->cond,NULL);
- pthread_mutex_lock(&l_args->mutex);
- dap_global_db_get_all_raw(a_group, a_first_id,0,s_get_all_raw_sync_callback, l_args);
- pthread_cond_wait(&l_args->cond, &l_args->mutex);
- pthread_mutex_unlock(&l_args->mutex);
- pthread_mutex_destroy(&l_args->mutex);
- pthread_cond_destroy(&l_args->cond);
-
- dap_store_obj_t * l_ret = l_args->objs;
- if(l_args->objs_count)
- *a_objs_count = l_args->objs_count;
- DAP_DELETE(l_args);
- return l_ret;
-
-}
-
-/**
- * @brief dap_global_db_context_current
- * @return
- */
-dap_global_db_context_t * dap_global_db_context_current()
-{
- dap_context_t * l_context = dap_context_current();
- if(l_context->type == DAP_CONTEXT_TYPE_GLOBAL_DB)
- return (dap_global_db_context_t *) l_context->_inheritor;
- else
- return NULL;
-
-}
-/**
- * @brief dap_global_db_del_unsafe
- * @param a_group
- * @param a_key
- * @return
- */
-int dap_global_db_del_unsafe(dap_global_db_context_t * a_global_db_context, const char * a_group, const char *a_key)
-{
- dap_store_obj_t l_store_obj = {0};
-
- l_store_obj.key = dap_strdup(a_key);
- l_store_obj.group = dap_strdup(a_group);
-
- int l_res = dap_chain_global_db_driver_delete(&l_store_obj, 1);
-
- if (a_key) {
- if (l_res >= 0) {
- // add to Del group
- l_res = s_record_del_history_add(l_store_obj.group, (char *)l_store_obj.key, dap_nanotime_now() );
- }
- // do not add to history if l_res=1 (already deleted)
- if (!l_res) {
- l_store_obj.key = l_store_obj.key;
- s_change_notify(&l_store_obj, DAP_DB$K_OPTYPE_DEL);
- }
- }
- DAP_DELETE(l_store_obj.key);
- DAP_DELETE(l_store_obj.group);
- return l_res;
-}
-
-/**
- * @brief dap_global_db_flush_sync
- * @return
- */
-int dap_global_db_flush_sync()
-{
- return dap_db_driver_flush();
-}
-
-/**
- * @brief dap_global_db_flush
- * @param a_callback
- * @param a_arg
- * @return
- */
-int dap_global_db_flush( dap_global_db_callback_result_t a_callback, void * a_arg )
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_delete");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_FLUSH;
- l_msg->callback_arg = a_arg;
- l_msg->callback_result = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec flush request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }
- return l_ret;
-}
-
-/**
- * @brief s_msg_opcode_flush
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_flush(struct queue_io_msg * a_msg)
-{
- int l_res = dap_db_driver_flush();
- if(a_msg->callback_result){
- a_msg->callback_result(s_context_global_db, l_res==0 ? DAP_GLOBAL_DB_RC_SUCCESS:
- DAP_GLOBAL_DB_RC_ERROR,
- NULL,NULL,NULL, 0, 0 , false, a_msg->callback_arg );
- }
- return true;
-}
-
-/**
- * @brief Execute callback in GlobalDB context
- * @param a_callback Callback function
- * @param arg Custom argument
- * @return 0 if success, others if not
- */
-int dap_global_db_context_exec (dap_global_db_callback_t a_callback, void * a_arg)
-{
- if(s_context_global_db == NULL){
- log_it(L_ERROR, "GlobalDB context is not initialized, can't call dap_global_db_context_exec");
- return -666;
- }
- struct queue_io_msg * l_msg = DAP_NEW_Z(struct queue_io_msg);
- l_msg->opcode = MSG_OPCODE_CONTEXT_EXEC;
- l_msg->callback_arg = a_arg;
- l_msg->callback = a_callback;
-
- int l_ret = dap_events_socket_queue_ptr_send(s_context_global_db->queue_io,l_msg);
- if (l_ret != 0){
- log_it(L_ERROR, "Can't exec context_exec request, code %d", l_ret);
- s_queue_io_msg_delete(l_msg);
- }
- return l_ret;
-
-}
-
-/**
- * @brief Execute callback in GlobalDB context
- * @param a_msg
- * @return
- */
-static bool s_msg_opcode_context_exec(struct queue_io_msg * a_msg)
-{
- if(a_msg->callback)
- a_msg->callback(s_context_global_db, a_msg->callback_arg );
- return true;
-}
-
-/**
- * @brief s_msg_opcode_to_str
- * @param a_opcode
- * @return
- */
-static const char *s_msg_opcode_to_str(enum queue_io_msg_opcode a_opcode)
-{
- switch(a_opcode){
- case MSG_OPCODE_GET: return "GET";
- case MSG_OPCODE_GET_RAW: return "GET_RAW";
- case MSG_OPCODE_GET_LAST: return "GET_LAST";
- case MSG_OPCODE_GET_DEL_TS: return "GET_DEL_TS";
- case MSG_OPCODE_GET_LAST_RAW: return "GET_LAST_RAW";
- case MSG_OPCODE_GET_ALL: return "GET_ALL";
- case MSG_OPCODE_GET_ALL_RAW: return "GET_ALL_RAW";
- case MSG_OPCODE_SET: return "SET";
- case MSG_OPCODE_SET_MULTIPLE: return "SET_MULTIPLE";
- case MSG_OPCODE_SET_RAW: return "SET_RAW";
- case MSG_OPCODE_PIN: return "PIN";
- case MSG_OPCODE_UNPIN: return "UNPIN";
- case MSG_OPCODE_DELETE: return "DELETE";
- case MSG_OPCODE_FLUSH: return "FLUSH";
- case MSG_OPCODE_CONTEXT_EXEC: return "CONTEXT_EXEC";
- default: return "UNKNOWN";
- }
-}
-
-/**
- * @brief s_queue_io_callback
- * @details Queue I/O process callback
- * @param a_es
- * @param a_arg
- */
-static void s_queue_io_callback( dap_events_socket_t * a_es, void * a_arg)
-{
- (void) a_es;
- struct queue_io_msg * l_msg = (struct queue_io_msg *) a_arg;
- bool l_msg_delete = false; // if msg resent again it shouldn't be deleted in the end of callback
- assert(l_msg);
-
- debug_if(g_dap_global_db_debug_more, L_NOTICE, "Received GlobalDB I/O message with opcode %s", s_msg_opcode_to_str(l_msg->opcode) );
-
- switch(l_msg->opcode){
- case MSG_OPCODE_GET: l_msg_delete = s_msg_opcode_get(l_msg); break;
- case MSG_OPCODE_GET_RAW: l_msg_delete = s_msg_opcode_get_raw(l_msg); break;
- case MSG_OPCODE_GET_LAST: l_msg_delete = s_msg_opcode_get_last(l_msg); break;
- case MSG_OPCODE_GET_LAST_RAW: l_msg_delete = s_msg_opcode_get_last_raw(l_msg); break;
- case MSG_OPCODE_GET_DEL_TS: l_msg_delete = s_msg_opcode_get_del_ts(l_msg); break;
- case MSG_OPCODE_GET_ALL: l_msg_delete = s_msg_opcode_get_all(l_msg); break;
- case MSG_OPCODE_GET_ALL_RAW: l_msg_delete = s_msg_opcode_get_all_raw(l_msg); break;
- case MSG_OPCODE_SET: l_msg_delete = s_msg_opcode_set(l_msg); break;
- case MSG_OPCODE_SET_MULTIPLE: l_msg_delete = s_msg_opcode_set_multiple_zc(l_msg); break;
- case MSG_OPCODE_SET_RAW: l_msg_delete = s_msg_opcode_set_raw(l_msg); break;
- case MSG_OPCODE_PIN: l_msg_delete = s_msg_opcode_pin(l_msg); break;
- case MSG_OPCODE_UNPIN: l_msg_delete = s_msg_opcode_unpin(l_msg); break;
- case MSG_OPCODE_DELETE: l_msg_delete = s_msg_opcode_delete(l_msg); break;
- case MSG_OPCODE_FLUSH: l_msg_delete = s_msg_opcode_flush(l_msg); break;
- case MSG_OPCODE_CONTEXT_EXEC: l_msg_delete = s_msg_opcode_context_exec(l_msg); break;
- default:{
- log_it(L_WARNING, "Message with undefined opcode %d received in queue_io",
- l_msg->opcode);
- }
- }
- if( l_msg_delete )
- s_queue_io_msg_delete(l_msg);
-}
-
-
-/**
- * @brief Adds data to the history log
- *
- * @param a_store_data a pointer to an object
- * @return (none)
- */
-static void s_change_notify(dap_store_obj_t * a_store_obj, char a_opcode)
-{
-dap_list_t *l_items_list = dap_global_db_get_sync_groups_all();
- while (l_items_list) {
- for (dap_list_t *it = dap_global_db_get_sync_groups_all(); it; it = it->next) {
- dap_sync_group_item_t *l_sync_group_item = (dap_sync_group_item_t *)it->data;
- if (dap_fnmatch(l_sync_group_item->group_mask, a_store_obj->group, 0))
- continue;
- if(l_sync_group_item->callback_notify) {
- l_sync_group_item->callback_notify(l_sync_group_item->callback_arg,
- a_opcode,
- a_store_obj->group, a_store_obj->key,
- a_store_obj->value, a_store_obj->value_len);
- }
- return;
- }
- l_items_list = (l_items_list == dap_global_db_get_sync_groups_all()) ?
- dap_global_db_get_sync_groups_extra_all() : NULL;
- }
-}
-
-
-/*
-* @brief s_record_del_history_del Deletes info about the deleted object from the database
-* @param a_key an object key string, looked like "0x8FAFBD00B..."
-* @param a_group a group name string, for example "kelvin-testnet.nodes"
-* @return If successful, returns true; otherwise, false.
-*/
-static int s_record_del_history_del( const char *a_group, const char *a_key)
-{
-dap_store_obj_t store_data = {0};
-char l_group[DAP_GLOBAL_DB_GROUP_NAME_SIZE_MAX];
-int l_res = 0;
-
- if(!a_key)
- return -1;
-
- store_data.key =( char*) a_key;
- dap_snprintf(l_group, sizeof(l_group) - 1, "%s.del", a_group);
- store_data.group = l_group;
-
- if ( dap_chain_global_db_driver_is(store_data.group, store_data.key) )
- l_res = dap_chain_global_db_driver_delete(&store_data, 1);
-
- return l_res;
-}
-
-/**
- * @brief s_record_del_history_add Adds info about the deleted entry to the database.
- * @param a_key an object key string
- * @param a_group a group name string
- * @param a_timestamp an object time stamp
- * @return True if successful, false otherwise.
- */
-static int s_record_del_history_add( char *a_key, char *a_group, uint64_t a_timestamp)
-{
-dap_store_obj_t store_data = {0};
-char l_group[DAP_GLOBAL_DB_GROUP_NAME_SIZE_MAX];
-int l_res = -1;
-
- store_data.key = a_key;
- // group = parent group + '.del'
- dap_snprintf(l_group, sizeof(l_group) - 1, "%s.del", a_group);
- store_data.group = l_group;
- store_data.timestamp = a_timestamp;
-
- l_res = dap_chain_global_db_driver_add(&store_data, 1);
-
- return l_res;
-}
-
-/**
- * @brief s_queue_io_msg_delete
- * @param a_msg
- */
-static void s_queue_io_msg_delete( struct queue_io_msg * a_msg)
-{
- switch(a_msg->opcode){
- case MSG_OPCODE_GET:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- if (a_msg->key)
- DAP_DELETE(a_msg->key);
- break;
- case MSG_OPCODE_GET_RAW:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- if (a_msg->key)
- DAP_DELETE(a_msg->key);
- break;
- case MSG_OPCODE_GET_DEL_TS:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- if (a_msg->key)
- DAP_DELETE(a_msg->key);
- break;
- case MSG_OPCODE_GET_LAST:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- break;
- case MSG_OPCODE_GET_LAST_RAW:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- break;
- case MSG_OPCODE_GET_ALL:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- break;
- case MSG_OPCODE_GET_ALL_RAW:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- break;
- case MSG_OPCODE_SET:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- if (a_msg->key)
- DAP_DELETE(a_msg->key);
- break;
- case MSG_OPCODE_SET_RAW:
- break;
- case MSG_OPCODE_SET_MULTIPLE:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- break;
- case MSG_OPCODE_PIN:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- if (a_msg->key)
- DAP_DELETE(a_msg->key);
- break;
- case MSG_OPCODE_UNPIN:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- if (a_msg->key)
- DAP_DELETE(a_msg->key);
- break;
- case MSG_OPCODE_DELETE:
- if (a_msg->group)
- DAP_DELETE(a_msg->group);
- if (a_msg->key)
- DAP_DELETE(a_msg->key);
- break;
- default:;
- }
-
- DAP_DELETE(a_msg);
-}
-
-
-/**
- * @brief s_context_callback_started
- * @details GlobalDB context started
- * @param a_context
- * @param a_arg
- */
-static void s_context_callback_started( dap_context_t * a_context, void *a_arg)
-{
- // Init its own queue
- s_context_global_db->queue_io = dap_context_create_queue(a_context, s_queue_io_callback);
-
- // Create arrays of inputs for connection with workers and proc threads
- s_context_global_db->queue_worker_io_input = DAP_NEW_Z_SIZE(dap_events_socket_t*, sizeof(dap_events_socket_t*) *
- dap_events_thread_get_count() );
- s_context_global_db->queue_worker_ch_io_input = DAP_NEW_Z_SIZE(dap_events_socket_t*, sizeof(dap_events_socket_t*) *
- dap_events_thread_get_count() );
- s_context_global_db->queue_worker_callback_input = DAP_NEW_Z_SIZE(dap_events_socket_t*, sizeof(dap_events_socket_t*) *
- dap_events_thread_get_count() );
-
- s_context_global_db->queue_proc_thread_callback_input = DAP_NEW_Z_SIZE(dap_events_socket_t*, sizeof(dap_events_socket_t*) *
- dap_events_thread_get_count() );
-
- // Fullful arrays with queue inputs
- for (uint32_t i = 0; i < dap_events_thread_get_count(); i++){
- dap_worker_t * l_worker = dap_events_worker_get(i);
- dap_stream_worker_t * l_stream_worker = DAP_STREAM_WORKER(l_worker);
- dap_proc_thread_t *l_proc_thread = dap_proc_thread_get(i);
- assert(l_worker);
- assert(l_proc_thread);
-
- // Input for I/O operations on worker
- s_context_global_db->queue_worker_io_input[i] =
- dap_events_socket_queue_ptr_create_input( l_worker->queue_es_io);
- dap_context_add( a_context, s_context_global_db->queue_worker_io_input[i]);
-
- // Input for ch I/O operations on worker
- s_context_global_db->queue_worker_ch_io_input[i] =
- dap_events_socket_queue_ptr_create_input( l_stream_worker->queue_ch_io );
- dap_context_add( a_context, s_context_global_db->queue_worker_ch_io_input[i]);
-
- // Input for callback queue on worker
- s_context_global_db->queue_worker_callback_input[i] =
- dap_events_socket_queue_ptr_create_input( l_worker->queue_callback);
- dap_context_add( a_context, s_context_global_db->queue_worker_callback_input[i]);
-
- // Input for callback queue on proc thread
- s_context_global_db->queue_proc_thread_callback_input[i] =
- dap_events_socket_queue_ptr_create_input( l_proc_thread->proc_queue->esocket );
- dap_context_add( a_context, s_context_global_db->queue_proc_thread_callback_input[i]);
- }
-}
-
-
-/**
- * @brief s_context_callback_stopped
- * @details Stop and destroy callback for GlobalDB context
- * @param a_context
- * @param a_arg
- */
-static void s_context_callback_stopped( dap_context_t * a_context, void *a_arg)
-{
- // Fullful arrays with queue inputs
- for (uint32_t i = 0; i < dap_events_thread_get_count(); i++){
- dap_events_socket_remove_and_delete_unsafe(s_context_global_db->queue_worker_io_input[i], true);
- dap_events_socket_remove_and_delete_unsafe(s_context_global_db->queue_worker_callback_input[i], true);
- dap_events_socket_remove_and_delete_unsafe(s_context_global_db->queue_proc_thread_callback_input[i], true);
- }
- DAP_DELETE(s_context_global_db->queue_worker_io_input);
- DAP_DELETE(s_context_global_db->queue_worker_callback_input);
- DAP_DELETE(s_context_global_db->queue_proc_thread_callback_input);
- dap_events_socket_remove_and_delete_unsafe(s_context_global_db->queue_io, false);
-}
-
-
-
-
-/**
- * @brief s_check_db_version
- * @return
- */
-static int s_check_db_version()
-{
- int l_ret;
- pthread_mutex_lock(&s_check_db_mutex);
- l_ret = dap_global_db_get(DAP_GLOBAL_DB_LOCAL_GENERAL, "gdb_version",s_check_db_version_callback_get, NULL);
- if (l_ret == 0){
- pthread_cond_wait(&s_check_db_cond, &s_check_db_mutex);
- l_ret = s_check_db_ret;
- }else{
- log_it(L_CRITICAL, "Can't process get gdb_version request, code %d", l_ret);
- }
- pthread_mutex_unlock(&s_check_db_mutex);
- return l_ret;
-}
-
-/**
- * @brief s_check_db_version_callback_get
- * @details Notify callback on reading GlobalDB version
- * @param a_errno
- * @param a_group
- * @param a_key
- * @param a_value
- * @param a_value_len
- * @param a_arg
- */
-static void s_check_db_version_callback_get (dap_global_db_context_t * a_global_db_context, int a_errno, const char * a_group, const char * a_key,
- const void * a_value, const size_t a_value_len,
- dap_nanotime_t value_ts, bool a_is_pinned, void * a_arg)
-{
- int res = 0;
-
-
- if(a_errno != 0){ // No DB at all
- log_it(L_NOTICE, "No GlobalDB version at all, creating the new GlobalDB from scratch");
- s_global_db_version = DAP_GLOBAL_DB_VERSION;
- if ( (res = dap_global_db_set(DAP_GLOBAL_DB_LOCAL_GENERAL, "gdb_version", &s_global_db_version, sizeof(uint16_t),false,
- s_check_db_version_callback_set, NULL) ) != 0){
- log_it(L_NOTICE, "Can't set GlobalDB version, code %d", res);
- goto lb_exit;
- }
- return; // In this case the condition broadcast should happens in s_check_db_version_callback_set()
-
- }
-
- if (a_value_len == sizeof(uint16_t))
- s_global_db_version = *(uint16_t *)a_value;
-
- if( s_global_db_version < DAP_GLOBAL_DB_VERSION) {
- log_it(L_NOTICE, "GlobalDB version %u, but %u required. The current database will be recreated",
- s_global_db_version, DAP_GLOBAL_DB_VERSION);
- dap_global_db_deinit();
- // Database path
- const char *l_storage_path = dap_config_get_item_str(g_config, "resources", "dap_global_db_path");
- // Delete database
- if(dap_file_test(l_storage_path) || dap_dir_test(l_storage_path)) {
- // Backup filename: backup_global_db_ver.X_DATE_TIME.zip
- char now[255];
- time_t t = time(NULL);
- strftime(now, 200, "%y.%m.%d-%H_%M_%S", localtime(&t));
-#ifdef DAP_BUILD_WITH_ZIP
- char *l_output_file_name = dap_strdup_printf("backup_%s_ver.%d_%s.zip", dap_path_get_basename(l_storage_path), l_gdb_version, now);
- char *l_output_file_path = dap_build_filename(l_storage_path, "../", l_output_file_name, NULL);
- // Create backup as ZIP file
- if(dap_zip_directory(l_storage_path, l_output_file_path)) {
-#else
- char *l_output_file_name = dap_strdup_printf("backup_%s_ver.%d_%s.tar", dap_path_get_basename(s_storage_path), s_global_db_version, now);
- char *l_output_file_path = dap_build_filename(l_storage_path, "../", l_output_file_name, NULL);
- // Create backup as TAR file
- if(dap_tar_directory(l_storage_path, l_output_file_path)) {
-#endif
- // Delete database file or directory
- dap_rm_rf(l_storage_path);
- }
- else {
- log_it(L_ERROR, "Can't backup GlobalDB version %d", s_global_db_version);
- res = -2;
- goto lb_exit;
- }
- DAP_DELETE(l_output_file_name);
- DAP_DELETE(l_output_file_path);
- }
- // Reinitialize database
- res = dap_global_db_init(NULL, NULL);
- // Save current db version
- if(!res) {
- s_global_db_version = DAP_GLOBAL_DB_VERSION;
- if ( (res = dap_global_db_set(DAP_GLOBAL_DB_LOCAL_GENERAL, "gdb_version", &s_global_db_version, sizeof(uint16_t),false,
- s_check_db_version_callback_set, NULL) ) != 0){
- log_it(L_NOTICE, "Can't set GlobalDB version, code %d", res);
- goto lb_exit;
- }
- return; // In this case the condition broadcast should happens in s_check_db_version_callback_set()
- }
- } else if(s_global_db_version > DAP_GLOBAL_DB_VERSION) {
- log_it(L_ERROR, "GlobalDB version %d is newer than supported version %d", s_global_db_version, DAP_GLOBAL_DB_VERSION);
- res = -1;
- goto lb_exit;
- }
- else {
- log_it(L_NOTICE, "GlobalDB version %d", s_global_db_version);
- }
-lb_exit:
- s_check_db_ret = res;
- pthread_mutex_lock(&s_check_db_mutex); // To be sure thats we're on pthread_cond_wait() line
- pthread_cond_broadcast(&s_check_db_cond);
- pthread_mutex_unlock(&s_check_db_mutex); // in calling thread
-}
-
-/**
- * @brief s_check_db_version_callback_set
- * @details GlobalDB version update callback
- * @param a_errno
- * @param a_group
- * @param a_key
- * @param a_value
- * @param a_value_len
- * @param a_arg
- */
-static void s_check_db_version_callback_set (dap_global_db_context_t * a_global_db_context,int a_errno, const char * a_group, const char * a_key,
- const void * a_value, const size_t a_value_len,
- dap_nanotime_t value_ts, bool a_is_pinned, void * a_arg)
-{
- int l_res = 0;
- if(a_errno != 0){
- log_it(L_ERROR, "Can't process request for DB version, error code %d", a_errno);
- l_res = a_errno;
- goto lb_exit;
- }
-
- log_it(L_NOTICE, "GlobalDB version updated to %d", s_global_db_version);
-
-lb_exit:
- s_check_db_ret = l_res;
- pthread_mutex_lock(&s_check_db_mutex); // in calling thread
- pthread_cond_broadcast(&s_check_db_cond);
- pthread_mutex_unlock(&s_check_db_mutex); // in calling thread
-}
diff --git a/modules/global-db/dap_global_db_sync.c b/modules/global-db/dap_global_db_sync.c
deleted file mode 100644
index edea5ea2bdbff5cf289411b08fff62de152f96ce..0000000000000000000000000000000000000000
--- a/modules/global-db/dap_global_db_sync.c
+++ /dev/null
@@ -1,180 +0,0 @@
-/*
- * Authors:
- * Dmitriy A. Gearasimov <gerasimov.dmitriy@demlabs.net>
- * Demlabs Ltd. https://demlabs.net
- * Copyright (c) 2022
- * All rights reserved.
-
- This file is part of DAP SDK the open source project
-
- DAP SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP SDK is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP SDK based project. If not, see <http://www.gnu.org/licenses/>.
-*/
-#include "dap_common.h"
-#include "dap_list.h"
-#include "dap_strfuncs.h"
-#include "dap_fnmatch.h"
-#include "dap_global_db.h"
-#include "dap_global_db_sync.h"
-
-#define LOG_TAG "dap_global_db_sync"
-
-static dap_list_t *s_sync_group_items = NULL;
-static dap_list_t *s_sync_group_extra_items = NULL;
-
-static void s_clear_sync_grp(void *a_elm);
-static int s_db_add_sync_group(dap_list_t **a_grp_list, dap_sync_group_item_t *a_item);
-
-void dap_global_db_sync_init()
-{
-
-}
-
-/**
- * @brief Deinitialize a database.
- * @note You should call this function at the end.
- * @return (none)
- */
-void dap_global_db_sync_deinit()
-{
- dap_list_free_full(s_sync_group_items, s_clear_sync_grp);
- dap_list_free_full(s_sync_group_extra_items, s_clear_sync_grp);
- s_sync_group_extra_items = s_sync_group_items = NULL;
-}
-
-/**
- * @brief Adds a group name for synchronization.
- * @param a_net_name a net name string, for all net a_net_name=null
- * @param a_group_prefix a prefix of the group name
- * @param a_callback a callback function
- * @param a_arg a pointer to an argument
- * @return (none)
- */
-void dap_chain_global_db_add_sync_group(const char *a_net_name, const char *a_group_prefix, dap_global_db_obj_callback_notify_t a_callback, void *a_arg)
-{
- dap_sync_group_item_t *l_item = DAP_NEW_Z(dap_sync_group_item_t);
- l_item->net_name = dap_strdup(a_net_name);
- l_item->group_mask = dap_strdup_printf("%s.*", a_group_prefix);
- l_item->callback_notify = a_callback;
- l_item->callback_arg = a_arg;
- s_db_add_sync_group(&s_sync_group_items, l_item);
-}
-
-/**
- * @brief Adds a group name for synchronization with especially node addresses.
- * @param a_net_name a net name string, for all net a_net_name=null
- * @param a_group_mask a group mask string
- * @param a_callback a callabck function
- * @param a_arg a pointer to an argument
- * @return (none)
- */
-void dap_chain_global_db_add_sync_extra_group(const char *a_net_name, const char *a_group_mask, dap_global_db_obj_callback_notify_t a_callback, void *a_arg)
-{
- dap_sync_group_item_t* l_item = DAP_NEW_Z(dap_sync_group_item_t);
- l_item->net_name = dap_strdup(a_net_name);
- l_item->group_mask = dap_strdup(a_group_mask);
- l_item->callback_notify = a_callback;
- l_item->callback_arg = a_arg;
- s_db_add_sync_group(&s_sync_group_extra_items, l_item);
-}
-
-/**
- * @brief Gets a list of a group mask for s_sync_group_items.
- * @param a_net_name a net name string, for all net a_net_name=null
- * @return Returns a pointer to a list of a group mask.
- */
-dap_list_t* dap_chain_db_get_sync_groups(const char *a_net_name)
-{
- if(!a_net_name)
- return dap_list_copy(s_sync_group_items);
-
- dap_list_t *l_list_out = NULL;
- dap_list_t *l_list_group = s_sync_group_items;
- while(l_list_group) {
- if(!dap_strcmp(a_net_name, ((dap_sync_group_item_t*) l_list_group->data)->net_name)) {
- l_list_out = dap_list_append(l_list_out, l_list_group->data);
- }
- l_list_group = dap_list_next(l_list_group);
- }
- return l_list_out;
-}
-
-/**
- * @brief Gets a list of a group mask for s_sync_group_items.
- * @param a_net_name a net name string, for all net a_net_name=null
- * @return Returns a pointer to a list of a group mask.
- */
-dap_list_t* dap_chain_db_get_sync_extra_groups(const char *a_net_name)
-{
- if(!a_net_name)
- return dap_list_copy(s_sync_group_extra_items);
-
- dap_list_t *l_list_out = NULL;
- dap_list_t *l_list_group = s_sync_group_extra_items;
- while(l_list_group) {
- if(!dap_strcmp(a_net_name, ((dap_sync_group_item_t*) l_list_group->data)->net_name)) {
- l_list_out = dap_list_append(l_list_out, l_list_group->data);
- }
- l_list_group = dap_list_next(l_list_group);
- }
- return l_list_out;
-}
-
-/**
- * @brief dap_global_db_get_sync_groups_all
- * @return
- */
-dap_list_t * dap_global_db_get_sync_groups_all()
-{
- return s_sync_group_items;
-}
-
-/**
- * @brief dap_global_db_get_sync_groups_extra_all
- * @return
- */
-dap_list_t * dap_global_db_get_sync_groups_extra_all()
-{
- return s_sync_group_extra_items;
-}
-
-/**
- * @brief s_clear_sync_grp
- * @param a_elm
- */
-static void s_clear_sync_grp(void *a_elm)
-{
- dap_sync_group_item_t *l_item = (dap_sync_group_item_t *)a_elm;
- DAP_DELETE(l_item->group_mask);
- DAP_DELETE(l_item);
-}
-
-/**
- * @brief s_db_add_sync_group
- * @param a_grp_list
- * @param a_item
- * @return
- */
-static int s_db_add_sync_group(dap_list_t **a_grp_list, dap_sync_group_item_t *a_item)
-{
- for (dap_list_t *it = *a_grp_list; it; it = it->next) {
- dap_sync_group_item_t *l_item = (dap_sync_group_item_t *)it->data;
- if (!dap_strcmp(l_item->group_mask, a_item->group_mask) && !dap_strcmp(l_item->net_name, a_item->net_name)) {
- log_it(L_WARNING, "Group mask '%s' already present in the list, ignore it", a_item->group_mask);
- s_clear_sync_grp(a_item);
- return -1;
- }
- }
- *a_grp_list = dap_list_append(*a_grp_list, a_item);
- return 0;
-}
diff --git a/modules/global-db/include/dap_chain_global_db.h b/modules/global-db/include/dap_chain_global_db.h
deleted file mode 100644
index b3235a6ccbcd525bfc490807750f3e1b5eca4385..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_chain_global_db.h
+++ /dev/null
@@ -1,16 +0,0 @@
-#pragma once
-
-#include <stdint.h>
-#include <stdbool.h>
-#include <time.h>
-
-#include "dap_chain_global_db_driver.h"
-#include "dap_common.h"
-#include "dap_config.h"
-#include "dap_list.h"
-#include "dap_chain_common.h"
-#include "dap_global_db.h"
-#include "dap_global_db_sync.h"
-
-
-
diff --git a/modules/global-db/include/dap_chain_global_db_driver.h b/modules/global-db/include/dap_chain_global_db_driver.h
deleted file mode 100644
index d17c29da500695cbfcceac323b83fff8e064cbb0..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_chain_global_db_driver.h
+++ /dev/null
@@ -1,98 +0,0 @@
-/*
- * Authors:
- * Alexander Lysikov <alexander.lysikov@demlabs.net>
- * DeM Labs Inc. https://demlabs.net
- * Kelvin Project https://github.com/kelvinblockchain
- * Copyright (c) 2019
- * All rights reserved.
-
- This file is part of DAP (Deus Applications Prototypes) the open source project
-
- DAP (Deus Applicaions Prototypes) is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP based project. If not, see <http://www.gnu.org/licenses/>.
-
- MODIFICATION HISTORY:
- 08-MAY-2022 RRL Added <ctx> field to the DB Driver interface table is called as <dap_db_driver_callbacks_t>;
- a set of limits - see DAP$K/SZ constant definitions;
- added lengths for the character fields.
- */
-
-#pragma once
-
-#include "dap_time.h"
-#include "dap_list.h"
-#include "dap_global_db.h"
-#include <stddef.h>
-#include <stdint.h>
-
-#define DAP_GLOBAL_DB_GROUP_NAME_SIZE_MAX 128UL /* A maximum size of group name */
-#define DAP_GLOBAL_DB_GROUPS_COUNT_MAX 1024UL /* A maximum number of groups */
-#define DAP_GLOBAL_DB_KEY_MAX 512UL /* A limit for the key's length in DB */
-#define DAP_GLOBAL_DB_MAX_OBJS 8192UL /* A maximum number of objects to be returned by
- read_srore_obj() */
-
-enum RECORD_FLAGS {
- RECORD_COMMON = 0, // 0000
- RECORD_PINNED = 1, // 0001
-};
-
-
-
-
-typedef int (*dap_db_driver_write_callback_t)(dap_store_obj_t*);
-typedef dap_store_obj_t* (*dap_db_driver_read_callback_t)(const char *,const char *, size_t *);
-typedef dap_store_obj_t* (*dap_db_driver_read_cond_callback_t)(const char *,uint64_t , size_t *);
-typedef dap_store_obj_t* (*dap_db_driver_read_last_callback_t)(const char *);
-typedef size_t (*dap_db_driver_read_count_callback_t)(const char *,uint64_t);
-typedef dap_list_t* (*dap_db_driver_get_groups_callback_t)(const char *);
-typedef bool (*dap_db_driver_is_obj_callback_t)(const char *, const char *);
-typedef int (*dap_db_driver_callback_t)(void);
-
-typedef struct dap_db_driver_callbacks {
- dap_db_driver_write_callback_t apply_store_obj; /* Performs an DB's action like: INSERT/DELETE/UPDATE for the given
- 'store object' */
- dap_db_driver_read_callback_t read_store_obj; /* Retreive 'store object' from DB */
- dap_db_driver_read_last_callback_t read_last_store_obj;
- dap_db_driver_read_cond_callback_t read_cond_store_obj;
- dap_db_driver_read_count_callback_t read_count_store;
-
- dap_db_driver_get_groups_callback_t get_groups_by_mask; /* Return a list of tables/groups has been matched to pattern */
-
- dap_db_driver_is_obj_callback_t is_obj; /* Check for existence of a record in the table/group for
- a given <key> */
-
- dap_db_driver_callback_t transaction_start; /* Allocate DB context for consequtive operations */
- dap_db_driver_callback_t transaction_end; /* Release DB context at end of DB consequtive operations */
-
- dap_db_driver_callback_t deinit;
- dap_db_driver_callback_t flush;
-} dap_db_driver_callbacks_t;
-
-
-int dap_db_driver_init(const char *driver_name, const char *a_filename_db, int a_mode_async);
-void dap_db_driver_deinit(void);
-
-dap_store_obj_t* dap_store_obj_copy(dap_store_obj_t *a_store_obj, size_t a_store_count);
-void dap_store_obj_free(dap_store_obj_t *a_store_obj, size_t a_store_count);
-DAP_STATIC_INLINE void dap_store_obj_free_one(dap_store_obj_t *a_store_obj) { return dap_store_obj_free(a_store_obj, 1); }
-int dap_db_driver_flush(void);
-
-int dap_chain_global_db_driver_apply(dap_store_obj_t *a_store_obj, size_t a_store_count);
-int dap_chain_global_db_driver_add(pdap_store_obj_t a_store_obj, size_t a_store_count);
-int dap_chain_global_db_driver_delete(pdap_store_obj_t a_store_obj, size_t a_store_count);
-dap_store_obj_t* dap_chain_global_db_driver_read_last(const char *a_group);
-dap_store_obj_t* dap_chain_global_db_driver_cond_read(const char *a_group, uint64_t id, size_t *a_count_out);
-dap_store_obj_t* dap_chain_global_db_driver_read(const char *a_group, const char *a_key, size_t *count_out);
-bool dap_chain_global_db_driver_is(const char *a_group, const char *a_key);
-size_t dap_chain_global_db_driver_count(const char *a_group, uint64_t id);
-dap_list_t* dap_chain_global_db_driver_get_groups_by_mask(const char *a_group_mask);
diff --git a/modules/global-db/include/dap_chain_global_db_driver_cdb.h b/modules/global-db/include/dap_chain_global_db_driver_cdb.h
deleted file mode 100644
index 948d0ada5f1473735006fb61a178ea4dc1a48388..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_chain_global_db_driver_cdb.h
+++ /dev/null
@@ -1,52 +0,0 @@
-/*
- * Authors:
- * Konstantin Papizh <konstantin.papizh@demlabs.net>
- * DeM Labs Inc. https://demlabs.net
- * Kelvin Project https://github.com/kelvinblockchain
- * Copyright (c) 2019
- * All rights reserved.
-
- This file is part of DAP (Deus Applications Prototypes) the open source project
-
- DAP (Deus Applicaions Prototypes) is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP based project. If not, see <http://www.gnu.org/licenses/>.
- */
-
-//#include "cuttdb.h"
-#include "cdb_core.h"
-#include "dap_chain_global_db_driver.h"
-#define UNUSED(x) (void)(x)
-
-typedef struct _cdb_options {
- int hsize; // Main hash table size, 1%-10% of total records, immutable
- int rcacheMB; // Record cache in MBytes
- int pcacheMB; // Index page cache in MBytes
-} cdb_options, *pcdb_options;
-
-typedef struct _cdb_record {
- char *key;
- char *val;
-} cdb_record, *pcdb_record;
-
-int dap_db_driver_cdb_init(const char *, dap_db_driver_callbacks_t *);
-int dap_db_driver_cdb_deinit();
-int dap_db_driver_cdb_flush(void);
-
-int dap_db_driver_cdb_apply_store_obj(pdap_store_obj_t);
-
-dap_store_obj_t *dap_db_driver_cdb_read_last_store_obj(const char*);
-dap_store_obj_t *dap_db_driver_cdb_read_store_obj(const char*, const char*, size_t*);
-size_t dap_db_driver_cdb_read_count_store(const char *a_group, uint64_t a_id);
-dap_list_t* dap_db_driver_cdb_get_groups_by_mask(const char *a_group_mask);
-dap_store_obj_t* dap_db_driver_cdb_read_cond_store_obj(const char*, uint64_t, size_t*);
-bool dap_db_driver_cdb_is_obj(const char *a_group, const char *a_key);
diff --git a/modules/global-db/include/dap_chain_global_db_driver_mdbx.h b/modules/global-db/include/dap_chain_global_db_driver_mdbx.h
deleted file mode 100644
index d35ea6e78d92d1c065ab26eea886b2f02c16325d..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_chain_global_db_driver_mdbx.h
+++ /dev/null
@@ -1,26 +0,0 @@
-/*
- * Authors:
- * Gerasimov Dmitriy <gerasimov.dmitriy@demlabs.net>
- * DeM Labs Ltd. https://demlabs.net
- * Copyright (c) 2021
- * All rights reserved.
-
- This file is part of DAP SDK the open source project
-
- DAP SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP based project. If not, see <http://www.gnu.org/licenses/>.
- */
-
-#include "dap_chain_global_db_driver.h"
-
-int dap_db_driver_mdbx_init(const char*, dap_db_driver_callbacks_t *a_drv_dpt);
diff --git a/modules/global-db/include/dap_chain_global_db_driver_pgsql.h b/modules/global-db/include/dap_chain_global_db_driver_pgsql.h
deleted file mode 100644
index 78fb58e3dabc39d33ca5e4ddabd42f8dc513adef..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_chain_global_db_driver_pgsql.h
+++ /dev/null
@@ -1,23 +0,0 @@
-#pragma once
-
-#include "dap_chain_global_db_driver.h"
-#ifdef DAP_CHAIN_GDB_ENGINE_PGSQL
-#include "/usr/include/postgresql/libpq-fe.h"
-#endif
-
-#define DAP_PGSQL_DBHASHNAME_LEN 8
-#define DAP_PGSQL_POOL_COUNT 16
-#define PGSQL_INVALID_TABLE "42P01"
-
-int dap_db_driver_pgsql_init(const char *a_filename_dir, dap_db_driver_callbacks_t *a_drv_callback);
-int dap_db_driver_pgsql_deinit();
-int dap_db_driver_pgsql_start_transaction(void);
-int dap_db_driver_pgsql_end_transaction(void);
-int dap_db_driver_pgsql_apply_store_obj(dap_store_obj_t *a_store_obj);
-dap_store_obj_t *dap_db_driver_pgsql_read_store_obj(const char *a_group, const char *a_key, size_t *a_count_out);
-dap_store_obj_t *dap_db_driver_pgsql_read_last_store_obj(const char *a_group);
-dap_store_obj_t *dap_db_driver_pgsql_read_cond_store_obj(const char *a_group, uint64_t a_id, size_t *a_count_out);
-dap_list_t *dap_db_driver_pgsql_get_groups_by_mask(const char *a_group_mask);
-size_t dap_db_driver_pgsql_read_count_store(const char *a_group, uint64_t a_id);
-bool dap_db_driver_pgsql_is_obj(const char *a_group, const char *a_key);
-int dap_db_driver_pgsql_flush();
diff --git a/modules/global-db/include/dap_chain_global_db_driver_sqlite.h b/modules/global-db/include/dap_chain_global_db_driver_sqlite.h
deleted file mode 100644
index 500d67ed0f123948cd016acc71728d21c40ed6c7..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_chain_global_db_driver_sqlite.h
+++ /dev/null
@@ -1,33 +0,0 @@
-/*
- * Authors:
- * Dmitriy A. Gearasimov <gerasimov.dmitriy@demlabs.net>
- * Alexander Lysikov <alexander.lysikov@demlabs.net>
- * DeM Labs Inc. https://demlabs.net
- * CellFrame https://cellframe.net
- * Sources https://gitlab.demlabs.net/cellframe
- * Copyright (c) 2017-2019
- * All rights reserved.
-
- This file is part of CellFrame SDK the open source project
-
- CellFrame SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- CellFrame SDK is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any CellFrame SDK based project. If not, see <http://www.gnu.org/licenses/>.
-*/
-
-#include <sqlite3.h>
-#include "dap_chain_global_db_driver.h"
-
-#define DAP_SQLITE_POOL_COUNT 16
-
-int dap_db_driver_sqlite_init(const char *a_filename_db, dap_db_driver_callbacks_t *a_drv_callback);
-int dap_db_driver_sqlite_deinit(void);
diff --git a/modules/global-db/include/dap_chain_global_db_remote.h b/modules/global-db/include/dap_chain_global_db_remote.h
deleted file mode 100644
index d6dfc094bae752a521fc517e277f73a43a698bb7..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_chain_global_db_remote.h
+++ /dev/null
@@ -1,77 +0,0 @@
-#pragma once
-
-#include <stdbool.h>
-#include <time.h>
-#include "dap_chain.h"
-#include "dap_chain_common.h"
-#include "dap_chain_net.h"
-#include "dap_chain_global_db_driver.h"
-#include "dap_global_db.h"
-
-#define F_DB_LOG_ADD_EXTRA_GROUPS 1
-#define F_DB_LOG_SYNC_FROM_ZERO 2
-
-#define GROUP_LOCAL_NODE_LAST_ID "local.node.last_id"
-#define GROUP_LOCAL_NODE_ADDR "local.node-addr"
-
-
-#define GDB_SYNC_ALWAYS_FROM_ZERO // For debug purposes
-// for dap_db_log_list_xxx()
-
-typedef struct dap_global_db_pkt {
- dap_nanotime_t timestamp;
- uint64_t data_size;
- uint32_t obj_count;
- uint8_t data[];
-}__attribute__((packed)) dap_global_db_pkt_t;
-
-typedef struct dap_db_log_list_group {
- char *name;
- uint64_t last_id_synced;
- uint64_t count;
-} dap_db_log_list_group_t;
-
-typedef struct dap_db_log_list_obj {
- dap_global_db_pkt_t *pkt;
- dap_hash_fast_t hash;
-} dap_db_log_list_obj_t;
-
-typedef struct dap_db_log_list {
- dap_list_t *list_write; // writed list
- dap_list_t *list_read; // readed list (inside list_write)
- bool is_process;
- size_t items_rest; // rest items to read from list_read
- size_t items_number; // total items in list_write after reading from db
- dap_list_t *groups;
- pthread_t thread;
- pthread_mutex_t list_mutex;
-} dap_db_log_list_t;
-
-// Set addr for current node
-bool dap_db_set_cur_node_addr(uint64_t a_address, char *a_net_name);
-bool dap_db_set_cur_node_addr_exp(uint64_t a_address, char *a_net_name );
-uint64_t dap_chain_net_get_cur_node_addr_gdb_sync(char *a_net_name);
-
-// Set last id for remote node
-bool dap_db_set_last_id_remote(uint64_t a_node_addr, uint64_t a_id, char *a_group);
-// Get last id for remote node
-uint64_t dap_db_get_last_id_remote(uint64_t a_node_addr, char *a_group);
-// Set last hash for chain for remote node
-bool dap_db_set_last_hash_remote(uint64_t a_node_addr, dap_chain_t *a_chain, dap_chain_hash_fast_t *a_hash);
-// Get last hash for chain for remote node
-dap_chain_hash_fast_t *dap_db_get_last_hash_remote(uint64_t a_node_addr, dap_chain_t *a_chain);
-
-dap_global_db_pkt_t *dap_store_packet_single(dap_store_obj_t *a_store_obj);
-dap_global_db_pkt_t *dap_store_packet_multiple(dap_global_db_pkt_t *a_old_pkt, dap_global_db_pkt_t *a_new_pkt);
-dap_store_obj_t *dap_global_db_pkt_deserialize(const dap_global_db_pkt_t *a_pkt, size_t *a_objs_count);
-
-char *dap_store_packet_get_group(dap_global_db_pkt_t *a_pkt);
-uint64_t dap_store_packet_get_id(dap_global_db_pkt_t *a_pkt);
-void dap_store_packet_change_id(dap_global_db_pkt_t *a_pkt, uint64_t a_id);
-
-dap_db_log_list_t* dap_db_log_list_start(dap_chain_net_t *l_net, dap_chain_node_addr_t a_addr, int a_flags);
-size_t dap_db_log_list_get_count(dap_db_log_list_t *a_db_log_list);
-size_t dap_db_log_list_get_count_rest(dap_db_log_list_t *a_db_log_list);
-dap_db_log_list_obj_t *dap_db_log_list_get(dap_db_log_list_t *a_db_log_list);
-void dap_db_log_list_delete(dap_db_log_list_t *a_db_log_list);
-void dap_db_log_list_rewind(dap_db_log_list_t *a_db_log_list);
diff --git a/modules/global-db/include/dap_global_db.h b/modules/global-db/include/dap_global_db.h
deleted file mode 100644
index e83c3d28ab5d5a5ae046ac4f426611f225ba6940..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_global_db.h
+++ /dev/null
@@ -1,184 +0,0 @@
-/*
- * Authors:
- * Dmitriy A. Gearasimov <gerasimov.dmitriy@demlabs.net>
- * Demlabs Ltd. https://demlabs.net
- * Copyright (c) 2022
- * All rights reserved.
-
- This file is part of DAP SDK the open source project
-
- DAP SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP SDK is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP SDK based project. If not, see <http://www.gnu.org/licenses/>.
-*/
-#pragma once
-#include <stddef.h>
-#include <stdbool.h>
-#include "dap_common.h"
-#include "dap_time.h"
-#include "dap_context.h"
-#include "dap_proc_queue.h"
-
-#define DAP_GLOBAL_DB_VERSION 2
-#define DAP_GLOBAL_DB_LOCAL_GENERAL "local.general"
-
-// GlobalDB own context custom extension
-typedef struct dap_global_db_context
-{
- dap_events_socket_t * queue_io; // I/O queue for GlobalDB i/o requests
-
- dap_events_socket_t ** queue_worker_callback_input; // Worker callback queue input
- dap_events_socket_t ** queue_worker_io_input; // Worker io queue input
- dap_events_socket_t ** queue_worker_ch_io_input; // Worker ch io queue input
- dap_events_socket_t ** queue_proc_thread_callback_input; // Worker callback queue input
-
- dap_context_t * context; // parent pointer
-} dap_global_db_context_t;
-
-#define DAP_CONTEXT_TYPE_GLOBAL_DB 100
-
-enum {
- DAP_DB$K_OPTYPE_ADD = 0x61, /* 'a', */ /* Operation Type = INSERT/ADD */
- DAP_DB$K_OPTYPE_DEL = 0x64, /* 'd', */ /* -- // -- DELETE */
- DAP_DB$K_OPTYPE_RETR = 0x72, /* 'r', */ /* -- // -- RETRIEVE/GET */
-};
-
-typedef struct dap_store_obj {
- uint64_t id;
- dap_nanotime_t timestamp;
- uint32_t type; /* Operation type: ADD/DELETE, see DAP_DB$K_OPTYPE_* constants */
- uint8_t flags; /* RECORD_FLAGS */
-
- char *group;
- uint64_t group_len;
-
- union {
- const char *key;
- byte_t * key_byte;
- };
- uint64_t key_len;
-
- uint8_t *value;
- uint64_t value_len;
-
- dap_proc_queue_callback_t callback_proc_thread; /* (Async mode only!) A call back to be called on request completion */
- void *callback_proc_thread_arg; /* (Async mode only!) An argument of the callback rotine */
-} dap_store_obj_t, *pdap_store_obj_t;
-
-typedef struct dap_global_db_obj {
- uint64_t id;
- char *key;
- dap_nanotime_t timestamp;
- uint8_t *value;
- size_t value_len;
- bool is_pinned;
-
-} DAP_ALIGN_PACKED dap_global_db_obj_t;
-
-typedef void (*dap_global_db_callback_t) (dap_global_db_context_t * a_global_db_context, void * a_arg);
-
-/**
- * @brief callback for single result
- * @arg a_rc DAP_GLOBAL_DB_RC_SUCCESS if success others if not
- */
-typedef void (*dap_global_db_callback_result_t) (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const void * a_value,
- const size_t a_value_size, dap_nanotime_t a_value_ts, bool a_is_pinned, void * a_arg);
-
-/**
- * @brief callback for single raw result
- * @arg a_rc DAP_GLOBAL_DB_RC_SUCCESS if success others if not
- * @return true if we need to free a_store_obj, false otherwise.
- */
-typedef bool (*dap_global_db_callback_result_raw_t) (dap_global_db_context_t * a_global_db_context,int a_rc, dap_store_obj_t * a_store_obj, void * a_arg);
-
-
-/**
- * @brief callback for multiple result, with pagination
- * @arg a_rc DAP_GLOBAL_DB_RC_SUCCESS if success others if not
- * @arg a_values_total Total values number
- * @arg a_values_shift Current shift from beginning of values set
- * @arg a_values_count Current number of items in a_values
- * @arg a_values Current items (page of items)
- * @arg a_arg Custom argument
- * @return true if we need to free a_store_obj, false otherwise.
- */
-typedef bool (*dap_global_db_callback_results_t) (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key,
- const size_t a_values_total, const size_t a_values_shift,
- const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg);
-/**
- * @brief callback for multiple raw result, with pagination
- * @arg a_rc DAP_GLOBAL_DB_RC_SUCCESS if success other sif not
- * @arg a_values_total Total values number
- * @arg a_values_shift Current shift from beginning of values set
- * @arg a_values_count Current number of items in a_values
- * @arg a_values Current items (page of items)
- * @return true if we need to free a_store_obj, false otherwise.
- */
-typedef bool (*dap_global_db_callback_results_raw_t) (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key,
- const size_t a_values_current, const size_t a_values_shift,
- const size_t a_values_count, dap_store_obj_t * a_values, void * a_arg);
-// Return codes
-#define DAP_GLOBAL_DB_RC_SUCCESS 0
-#define DAP_GLOBAL_DB_RC_NO_RESULTS -1
-#define DAP_GLOBAL_DB_RC_ERROR -666
-
-extern int g_dap_global_db_debug_more;
-
-int dap_global_db_init(const char * a_path, const char * a_driver);
-void dap_global_db_deinit();
-
-
-// === Async functions ===
-int dap_global_db_get(const char * a_group, const char *a_key,dap_global_db_callback_result_t a_callback, void * a_arg );
-int dap_global_db_get_raw(const char * a_group, const char *a_key,dap_global_db_callback_result_raw_t a_callback, void * a_arg );
-
-int dap_global_db_get_del_ts(const char * a_group, const char *a_key,dap_global_db_callback_result_t a_callback, void * a_arg );
-int dap_global_db_get_last(const char * a_group, dap_global_db_callback_result_t a_callback, void * a_arg );
-int dap_global_db_get_last_raw(const char * a_group, dap_global_db_callback_result_raw_t a_callback, void * a_arg );
-int dap_global_db_get_all(const char * a_group, size_t l_results_page_size, dap_global_db_callback_results_t a_callback, void * a_arg );
-int dap_global_db_get_all_raw(const char * a_group, uint64_t a_first_id, size_t l_results_page_size, dap_global_db_callback_results_raw_t a_callback, void * a_arg );
-
-int dap_global_db_set(const char * a_group, const char *a_key, const void * a_value, const size_t a_value_length, bool a_pin_value, dap_global_db_callback_result_t a_callback, void * a_arg );
-int dap_global_db_set_raw(dap_store_obj_t * a_store_objs, size_t a_store_objs_count, dap_global_db_callback_results_raw_t a_callback, void * a_arg );
-
-// Set multiple. In callback writes total processed objects to a_values_total and a_values_count to the a_values_count as well
-int dap_global_db_set_multiple_zc(const char * a_group, dap_global_db_obj_t * a_values, size_t a_values_count, dap_global_db_callback_results_t a_callback, void * a_arg );
-int dap_global_db_pin(const char * a_group, const char *a_key, dap_global_db_callback_result_t a_callback, void * a_arg );
-int dap_global_db_unpin(const char * a_group, const char *a_key, dap_global_db_callback_result_t a_callback, void * a_arg );
-int dap_global_db_del(const char * a_group, const char *a_key, dap_global_db_callback_result_t a_callback, void * a_arg );
-int dap_global_db_flush( dap_global_db_callback_result_t a_callback, void * a_arg );
-
-// For context unification sometimes we need to exec inside GlobalDB context
-
-int dap_global_db_context_exec (dap_global_db_callback_t a_callback, void * a_arg);
-
-
-// Clear global_db_obj array
-void dap_global_db_objs_delete(dap_global_db_obj_t *a_objs, size_t a_count);
-
-// === Sync functions ===
-
-dap_global_db_obj_t* dap_global_db_get_all_sync(const char *a_group, size_t *a_objs_count);
-dap_store_obj_t* dap_global_db_get_all_raw_sync(const char *a_group, uint64_t a_first_id, size_t *a_objs_count);
-int dap_global_db_set_sync(const char * a_group, const char *a_key, const void * a_value, const size_t a_value_length, bool a_pin_value );
-int dap_global_db_pin_sync(const char * a_group, const char *a_key);
-int dap_global_db_unpin_sync(const char * a_group, const char *a_key);
-
-byte_t* dap_global_db_get_sync(const char * a_group,const char *a_key, size_t *a_data_size, bool *a_is_pinned, dap_nanotime_t * a_ts);
-int dap_global_db_del_sync(const char * a_group, const char *a_key);
-int dap_global_db_flush_sync();
-
-// ==== Unsafe functions (for own context call only) ===
-dap_global_db_context_t * dap_global_db_context_current();
-int dap_global_db_del_unsafe(dap_global_db_context_t * a_global_db_context, const char * a_group, const char *a_key);
-int dap_global_db_set_unsafe(dap_global_db_context_t * a_global_db_context, const char * a_group, const char *a_key, const void * a_value, const size_t a_value_length, bool a_pin_value );
-
diff --git a/modules/global-db/include/dap_global_db_sync.h b/modules/global-db/include/dap_global_db_sync.h
deleted file mode 100644
index 4b132e3e0ce6e0b48dbccdd1ea990626454c79a4..0000000000000000000000000000000000000000
--- a/modules/global-db/include/dap_global_db_sync.h
+++ /dev/null
@@ -1,50 +0,0 @@
-/*
- * Authors:
- * Dmitriy A. Gearasimov <gerasimov.dmitriy@demlabs.net>
- * Demlabs Ltd. https://demlabs.net
- * Copyright (c) 2022
- * All rights reserved.
-
- This file is part of DAP SDK the open source project
-
- DAP SDK is free software: you can redistribute it and/or modify
- it under the terms of the GNU General Public License as published by
- the Free Software Foundation, either version 3 of the License, or
- (at your option) any later version.
-
- DAP SDK is distributed in the hope that it will be useful,
- but WITHOUT ANY WARRANTY; without even the implied warranty of
- MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- GNU General Public License for more details.
-
- You should have received a copy of the GNU General Public License
- along with any DAP SDK based project. If not, see <http://www.gnu.org/licenses/>.
-*/
-
-#pragma once
-#include <stddef.h>
-#include <dap_list.h>
-typedef void (*dap_global_db_obj_callback_notify_t) (void * a_arg, const char a_op_code, const char * a_group,
- const char * a_key, const void * a_value, const size_t a_value_len);
-
-// Callback table item
-typedef struct dap_sync_group_item {
- char *group_mask;
- char *net_name;
- dap_global_db_obj_callback_notify_t callback_notify;
- void * callback_arg;
-} dap_sync_group_item_t;
-
-void dap_global_db_sync_init();
-void dap_global_db_sync_deinit();
-
-/**
- * Setup callbacks and filters
- */
-// Add group name that will be synchronized
-void dap_chain_global_db_add_sync_group(const char *a_net_name, const char *a_group_prefix, dap_global_db_obj_callback_notify_t a_callback, void *a_arg);
-void dap_chain_global_db_add_sync_extra_group(const char *a_net_name, const char *a_group_mask, dap_global_db_obj_callback_notify_t a_callback, void *a_arg);
-dap_list_t *dap_chain_db_get_sync_groups(const char *a_net_name);
-dap_list_t *dap_chain_db_get_sync_extra_groups(const char *a_net_name);
-dap_list_t * dap_global_db_get_sync_groups_all();
-dap_list_t * dap_global_db_get_sync_groups_extra_all();
diff --git a/modules/mempool/CMakeLists.txt b/modules/mempool/CMakeLists.txt
index 95fc039743247a984eed5068182cc11958ac4d6c..42a266f33071d3f7125f2369c13c5a05803073ab 100644
--- a/modules/mempool/CMakeLists.txt
+++ b/modules/mempool/CMakeLists.txt
@@ -6,6 +6,6 @@ file(GLOB DAP_CHAIN_MEMPOOL_HDR include/*.h)
add_library(${PROJECT_NAME} STATIC ${DAP_CHAIN_MEMPOOL_SRC} ${DAP_CHAIN_MEMPOOL_HDR})
-target_link_libraries(${PROJECT_NAME} dap_http_server dap_client dap_chain_net dap_chain_global_db dap_core)
+target_link_libraries(${PROJECT_NAME} dap_http_server dap_client dap_chain_net dap_global_db dap_core)
target_include_directories(${PROJECT_NAME} INTERFACE .)
target_include_directories(${PROJECT_NAME} PUBLIC include)
diff --git a/modules/mempool/dap_chain_mempool.c b/modules/mempool/dap_chain_mempool.c
index 1fe3d74b3a470c33b1fa7fcf140a7ba4c63ed637..dea1a1816f2bd222c965b0d1ba5862f83694ad4b 100644
--- a/modules/mempool/dap_chain_mempool.c
+++ b/modules/mempool/dap_chain_mempool.c
@@ -67,7 +67,7 @@
#define LOG_TAG "dap_chain_mempool"
-static bool s_tx_create_massive_gdb_save_callback( dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
+static bool s_tx_create_massive_gdb_save_callback( dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total,
const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg);
int dap_datum_mempool_init(void)
@@ -461,7 +461,7 @@ int dap_chain_mempool_tx_create_massive( dap_chain_t * a_chain, dap_enc_key_t *a
* @param a_values
* @param a_arg
*/
-static bool s_tx_create_massive_gdb_save_callback (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
+static bool s_tx_create_massive_gdb_save_callback (dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total,
const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg)
{
diff --git a/modules/net/CMakeLists.txt b/modules/net/CMakeLists.txt
index d54d45da2705c42468cc2e9f73c32be903548cf0..3782bc07f981e18f88a331c8acbac5f5297764cf 100644
--- a/modules/net/CMakeLists.txt
+++ b/modules/net/CMakeLists.txt
@@ -38,16 +38,16 @@ add_library(${PROJECT_NAME} STATIC ${DAP_CHAIN_NET_SRCS} ${DAP_CHAIN_NET_HEADERS
if(WIN32)
target_link_libraries(${PROJECT_NAME} dap_core dap_crypto dap_client dap_io dap_notify_srv dap_cli_server dap_stream_ch_chain dap_stream_ch_chain_net dap_stream_ch_chain_net_srv dap_chain dap_chain_wallet dap_chain_net_srv dap_stream_ch_chain_voting
- dap_chain_mempool dap_chain_global_db dap_chain_cs_none)
+ dap_chain_mempool dap_global_db dap_chain_cs_none)
endif()
if(LINUX)
target_link_libraries(${PROJECT_NAME} dap_core dap_crypto dap_io dap_notify_srv dap_cli_server dap_client dap_stream_ch_chain dap_stream_ch_chain_net dap_stream_ch_chain_net_srv dap_stream_ch_chain_voting dap_chain
- dap_chain_wallet dap_chain_net_srv dap_chain_mempool dap_chain_global_db dap_chain_cs_none
+ dap_chain_wallet dap_chain_net_srv dap_chain_mempool dap_global_db dap_chain_cs_none
resolv )
elseif(BSD)
target_link_libraries(${PROJECT_NAME} dap_core dap_crypto dap_io dap_notify_srv dap_cli_server dap_client dap_stream_ch_chain dap_stream_ch_chain_net dap_stream_ch_chain_net_srv dap_stream_ch_chain_voting dap_chain
- dap_chain_wallet dap_chain_net_srv dap_chain_mempool dap_chain_global_db dap_chain_cs_none )
+ dap_chain_wallet dap_chain_net_srv dap_chain_mempool dap_global_db dap_chain_cs_none )
endif()
diff --git a/modules/net/dap_chain_net.c b/modules/net/dap_chain_net.c
index ec020ed75a6f71f883e5cb13f857fc4c40ae2425..cb37f7a51c499b03d435761c888b70ef13bdb34a 100644
--- a/modules/net/dap_chain_net.c
+++ b/modules/net/dap_chain_net.c
@@ -100,7 +100,7 @@
#include "dap_chain_cs_none.h"
#include "dap_client_http.h"
#include "dap_global_db.h"
-#include "dap_chain_global_db_remote.h"
+#include "dap_global_db_remote.h"
#include "dap_stream_ch_chain_net_pkt.h"
#include "dap_stream_ch_chain_net.h"
@@ -129,7 +129,7 @@ struct balancer_link_request {
dap_chain_net_t *net;
dap_worker_t *worker;
bool from_http;
- bool link_replace;
+ int link_replace_tries;
};
struct net_link {
@@ -270,10 +270,8 @@ static void s_gbd_history_callback_notify (void * a_arg, const char a_op_code, c
static void s_chain_callback_notify(void * a_arg, dap_chain_t *a_chain, dap_chain_cell_id_t a_id, void *a_atom, size_t a_atom_size);
static int s_cli_net(int argc, char ** argv, char **str_reply);
static uint8_t *s_net_set_acl(dap_chain_hash_fast_t *a_pkey_hash);
-static bool s_balancer_start_dns_request(dap_chain_net_t *a_net, dap_chain_node_info_t *a_link_node_info, bool a_link_replace);
-static bool s_balancer_start_http_request(dap_chain_net_t *a_net, dap_chain_node_info_t *a_link_node_info, bool a_link_replace);
static void s_prepare_links_from_balancer(dap_chain_net_t *a_net);
-static bool s_new_balancer_link_request(dap_chain_net_t *a_net);
+static bool s_new_balancer_link_request(dap_chain_net_t *a_net, int a_link_replace_tries);
static bool s_seed_mode = false;
@@ -642,7 +640,7 @@ static void s_chain_callback_notify(void *a_arg, dap_chain_t *a_chain, dap_chain
}
/**
- * @brief added like callback in dap_chain_global_db_add_sync_group
+ * @brief added like callback in dap_global_db_add_sync_group
*
* @param a_arg arguments. Can be network object (dap_chain_net_t)
* @param a_op_code object type (f.e. l_net->type from dap_store_obj)
@@ -868,6 +866,21 @@ static void s_node_link_callback_connected(dap_chain_node_client_t * a_node_clie
}
+static bool s_start_free_link(dap_chain_net_t *a_net)
+{
+ struct net_link *l_link, *l_link_tmp;
+ HASH_ITER(hh, PVT(a_net)->net_links, l_link, l_link_tmp) {
+ if (l_link->link == NULL) { // We have a free prepared link
+ dap_chain_node_client_t *l_client_new = dap_chain_net_client_create_n_connect(
+ a_net, l_link->link_info);
+ l_link->link = l_client_new;
+ l_link->client_uuid = l_client_new->uuid;
+ return true;
+ }
+ }
+ return false;
+}
+
/**
* @brief s_node_link_callback_disconnected
* @param a_node_client
@@ -888,23 +901,16 @@ static void s_node_link_callback_disconnected(dap_chain_node_client_t *a_node_cl
pthread_rwlock_wrlock(&l_net_pvt->uplinks_lock);
s_net_link_remove(l_net_pvt, a_node_client->uuid, l_net_pvt->only_static_links);
a_node_client->keep_connection = false;
- struct net_link *l_link, *l_link_tmp;
- HASH_ITER(hh, l_net_pvt->net_links, l_link, l_link_tmp) {
- if (l_link->link == NULL) { // We have a free prepared link
- dap_chain_node_client_t *l_client_new = dap_chain_net_client_create_n_connect(
- l_net, l_link->link_info);
- l_link->link = l_client_new;
- l_link->client_uuid = l_client_new->uuid;
- pthread_rwlock_unlock(&l_net_pvt->uplinks_lock);
- return;
- }
- }
+ if (s_start_free_link(l_net)) {
+ pthread_rwlock_unlock(&l_net_pvt->uplinks_lock);
+ return;
+ }
if (!l_net_pvt->only_static_links) {
size_t l_current_links_prepared = HASH_COUNT(l_net_pvt->net_links);
for (size_t i = l_current_links_prepared; i < l_net_pvt->max_links_count ; i++) {
dap_chain_node_info_t *l_link_node_info = s_get_balancer_link_from_cfg(l_net);
if (l_link_node_info) {
- if (!s_balancer_start_dns_request(l_net, l_link_node_info, true))
+ if (!s_new_balancer_link_request(l_net, 1))
log_it(L_ERROR, "Can't process node info dns request");
DAP_DELETE(l_link_node_info);
}
@@ -1003,8 +1009,8 @@ static void s_net_links_complete_and_start(dap_chain_net_t *a_net, dap_worker_t
// Try to get links from HTTP balancer
l_net_pvt->balancer_http = true;
s_prepare_links_from_balancer(a_net);
- pthread_rwlock_unlock(&l_net_pvt->balancer_lock);
- return;
+ pthread_rwlock_unlock(&l_net_pvt->balancer_lock);
+ return;
}
if (HASH_COUNT(l_net_pvt->net_links) < l_net_pvt->max_links_count)
s_fill_links_from_root_aliases(a_net); // Comlete the sentence
@@ -1038,9 +1044,9 @@ static void s_net_balancer_link_prepare_success(dap_worker_t * a_worker, dap_cha
int l_res = s_net_link_add(l_net, a_node_info);
if (l_res < 0) { // Can't add this link
debug_if(s_debug_more, L_DEBUG, "Can't add link "NODE_ADDR_FP_STR, NODE_ADDR_FP_ARGS_S(a_node_info->hdr.address));
- if (l_balancer_request->link_replace)
+ if (l_balancer_request->link_replace_tries)
// Just try a new one
- s_new_balancer_link_request(l_net);
+ s_new_balancer_link_request(l_net, l_balancer_request->link_replace_tries);
} else if (l_res == 0) {
struct json_object *l_json = s_net_states_json_collect(l_net);
char l_err_str[128] = { };
@@ -1052,19 +1058,23 @@ static void s_net_balancer_link_prepare_success(dap_worker_t * a_worker, dap_cha
json_object_put(l_json);
debug_if(s_debug_more, L_DEBUG, "Link "NODE_ADDR_FP_STR" successfully added",
NODE_ADDR_FP_ARGS_S(a_node_info->hdr.address));
- if (l_balancer_request->link_replace &&
+ if (l_balancer_request->link_replace_tries &&
s_net_get_active_links_count(l_net) < PVT(l_net)->required_links_count) {
// Auto-start new link
- debug_if(s_debug_more, L_DEBUG, "Link "NODE_ADDR_FP_STR" started",
- NODE_ADDR_FP_ARGS_S(a_node_info->hdr.address));
- dap_chain_node_client_t *l_client = dap_chain_net_client_create_n_connect(l_net, a_node_info);
- struct net_link *l_new_link = s_net_link_find(l_net, a_node_info);
- l_new_link->link = l_client;
- l_new_link->client_uuid = l_client->uuid;
+ pthread_rwlock_rdlock(&PVT(l_net)->states_lock);
+ if (PVT(l_net)->state_target != NET_STATE_OFFLINE) {
+ debug_if(s_debug_more, L_DEBUG, "Link "NODE_ADDR_FP_STR" started",
+ NODE_ADDR_FP_ARGS_S(a_node_info->hdr.address));
+ dap_chain_node_client_t *l_client = dap_chain_net_client_create_n_connect(l_net, a_node_info);
+ struct net_link *l_new_link = s_net_link_find(l_net, a_node_info);
+ l_new_link->link = l_client;
+ l_new_link->client_uuid = l_client->uuid;
+ }
+ pthread_rwlock_unlock(&PVT(l_net)->states_lock);
}
} else
debug_if(s_debug_more, L_DEBUG, "Maximum prepared links reached");
- if (!l_balancer_request->link_replace)
+ if (!l_balancer_request->link_replace_tries)
s_net_links_complete_and_start(l_net, a_worker);
DAP_DELETE(l_balancer_request->link_info);
DAP_DELETE(l_balancer_request);
@@ -1094,10 +1104,10 @@ static void s_net_balancer_link_prepare_error(dap_worker_t * a_worker, void * a_
json_object_object_add(l_json, "errorMessage", json_object_new_string(l_err_str));
dap_notify_server_send_mt(json_object_get_string(l_json));
json_object_put(l_json);
- if (!l_balancer_request->link_replace)
+ if (!l_balancer_request->link_replace_tries)
s_net_links_complete_and_start(l_net, a_worker);
else
- s_new_balancer_link_request(l_net);
+ s_new_balancer_link_request(l_net, l_balancer_request->link_replace_tries);
DAP_DELETE(l_node_info);
DAP_DELETE(l_balancer_request);
}
@@ -1108,7 +1118,8 @@ void s_net_http_link_prepare_success(void *a_response, size_t a_response_size, v
struct balancer_link_request *l_balancer_request = (struct balancer_link_request *)a_arg;
if (a_response_size != sizeof(dap_chain_node_info_t)) {
log_it(L_ERROR, "Invalid balancer response size %zu (expect %zu)", a_response_size, sizeof(dap_chain_node_info_t));
- s_new_balancer_link_request(l_balancer_request->net);
+ s_new_balancer_link_request(l_balancer_request->net, l_balancer_request->link_replace_tries);
+ DAP_DELETE(l_balancer_request);
return;
}
s_net_balancer_link_prepare_success(l_balancer_request->worker, (dap_chain_node_info_t *)a_response, a_arg);
@@ -1126,83 +1137,78 @@ void s_net_http_link_prepare_error(int a_error_code, void *a_arg)
* @param a_link_node_info node parameters
* @return list of dap_chain_node_info_t
*/
-static bool s_balancer_start_dns_request(dap_chain_net_t *a_net, dap_chain_node_info_t *a_link_node_info, bool a_link_replace)
+static bool s_new_balancer_link_request(dap_chain_net_t *a_net, int a_link_replace_tries)
{
- char l_node_addr_str[INET_ADDRSTRLEN] = { };
- inet_ntop(AF_INET, &a_link_node_info->hdr.ext_addr_v4, l_node_addr_str, INET_ADDRSTRLEN);
- log_it(L_DEBUG, "Start balancer DNS request to %s", l_node_addr_str);
dap_chain_net_pvt_t *l_net_pvt = a_net ? PVT(a_net) : NULL;
- if(!l_net_pvt)
+ if (!l_net_pvt)
return false;
- struct balancer_link_request *l_balancer_request = DAP_NEW_Z(struct balancer_link_request);
- l_balancer_request->net = a_net;
- l_balancer_request->link_info = DAP_DUP(a_link_node_info);
- l_balancer_request->link_replace = a_link_replace;
- if (dap_chain_node_info_dns_request(a_link_node_info->hdr.ext_addr_v4,
- a_link_node_info->hdr.ext_port,
- a_net->pub.name,
- s_net_balancer_link_prepare_success,
- s_net_balancer_link_prepare_error,
- l_balancer_request)) {
- log_it(L_ERROR, "Can't process balancer link DNS request");
- DAP_DELETE(l_balancer_request->link_info);
- DAP_DELETE(l_balancer_request);
+ pthread_rwlock_rdlock(&l_net_pvt->states_lock);
+ if (l_net_pvt->state_target == NET_STATE_OFFLINE) {
+ pthread_rwlock_unlock(&l_net_pvt->states_lock);
return false;
}
- l_net_pvt->balancer_link_requests++;
- return true;
-}
-
-static bool s_balancer_start_http_request(dap_chain_net_t *a_net, dap_chain_node_info_t *a_link_node_info, bool a_link_replace)
-{
- char l_node_addr_str[INET_ADDRSTRLEN] = { };
- inet_ntop(AF_INET, &a_link_node_info->hdr.ext_addr_v4, l_node_addr_str, INET_ADDRSTRLEN);
- log_it(L_DEBUG, "Start balancer HTTP request to %s", l_node_addr_str);
- dap_chain_net_pvt_t *l_net_pvt = a_net ? PVT(a_net) : NULL;
- if (!l_net_pvt)
+ pthread_rwlock_unlock(&l_net_pvt->states_lock);
+ dap_chain_node_info_t *l_link_node_info = s_get_balancer_link_from_cfg(a_net);
+ if (a_link_replace_tries >= 5) {
+ // network problems, make static links
+ s_fill_links_from_root_aliases(a_net);
+ pthread_rwlock_wrlock(&l_net_pvt->uplinks_lock);
+ s_start_free_link(a_net);
+ pthread_rwlock_unlock(&l_net_pvt->uplinks_lock);
+ return false;
+ }
+ if (!l_link_node_info)
return false;
+ char l_node_addr_str[INET_ADDRSTRLEN] = {};
+ inet_ntop(AF_INET, &l_link_node_info->hdr.ext_addr_v4, l_node_addr_str, INET_ADDRSTRLEN);
+ log_it(L_DEBUG, "Start balancer %s request to %s", PVT(a_net)->balancer_http ? "HTTP" : "DNS", l_node_addr_str);
struct balancer_link_request *l_balancer_request = DAP_NEW_Z(struct balancer_link_request);
- l_balancer_request->from_http = true;
l_balancer_request->net = a_net;
- l_balancer_request->link_info = DAP_DUP(a_link_node_info);
+ l_balancer_request->link_info = l_link_node_info;
l_balancer_request->worker = dap_events_worker_get_auto();
- l_balancer_request->link_replace = a_link_replace;
- char *l_request = dap_strdup_printf("%s/%s?version=1,method=r,net=%s",
- DAP_UPLINK_PATH_BALANCER,
- DAP_BALANCER_URI_HASH,
- a_net->pub.name);
- if (dap_client_http_request(l_balancer_request->worker, l_node_addr_str, a_link_node_info->hdr.ext_port,
- "GET", "text/text", l_request,
- NULL, 0, NULL,
- s_net_http_link_prepare_success, s_net_http_link_prepare_error,
- l_balancer_request, NULL)) {
- log_it(L_ERROR, "Can't process balancer link HTTP request");
+ if (a_link_replace_tries)
+ l_balancer_request->link_replace_tries = a_link_replace_tries + 1;
+ int ret;
+ if (PVT(a_net)->balancer_http) {
+ l_balancer_request->from_http = true;
+ char *l_request = dap_strdup_printf("%s/%s?version=1,method=r,net=%s",
+ DAP_UPLINK_PATH_BALANCER,
+ DAP_BALANCER_URI_HASH,
+ a_net->pub.name);
+ ret = dap_client_http_request(l_balancer_request->worker,
+ l_node_addr_str,
+ l_link_node_info->hdr.ext_port,
+ "GET",
+ "text/text",
+ l_request,
+ NULL,
+ 0,
+ NULL,
+ s_net_http_link_prepare_success,
+ s_net_http_link_prepare_error,
+ l_balancer_request,
+ NULL);
+ } else {
+ l_link_node_info->hdr.ext_port = DNS_LISTEN_PORT;
+ ret = dap_chain_node_info_dns_request(l_balancer_request->worker,
+ l_link_node_info->hdr.ext_addr_v4,
+ l_link_node_info->hdr.ext_port,
+ a_net->pub.name,
+ s_net_balancer_link_prepare_success,
+ s_net_balancer_link_prepare_error,
+ l_balancer_request);
+ }
+ if (ret) {
+ log_it(L_ERROR, "Can't process balancer link %s request", PVT(a_net)->balancer_http ? "HTTP" : "DNS");
DAP_DELETE(l_balancer_request->link_info);
DAP_DELETE(l_balancer_request);
- DAP_DELETE(l_request);
return false;
}
- DAP_DELETE(l_request);
- l_net_pvt->balancer_link_requests++;
+ if (!a_link_replace_tries)
+ l_net_pvt->balancer_link_requests++;
return true;
}
-static bool s_new_balancer_link_request(dap_chain_net_t *a_net)
-{
- bool ret = false;
- dap_chain_node_info_t *l_link_node_info = s_get_balancer_link_from_cfg(a_net);
- if (l_link_node_info) {
- if (PVT(a_net)->balancer_http)
- ret = s_balancer_start_http_request(a_net, l_link_node_info, true);
- else {
- l_link_node_info->hdr.ext_port = DNS_LISTEN_PORT;
- ret = s_balancer_start_dns_request(a_net, l_link_node_info, true);
- }
- DAP_DELETE(l_link_node_info);
- }
- return ret;
-}
-
static void s_prepare_links_from_balancer(dap_chain_net_t *a_net)
{
// Get list of the unique links for l_net
@@ -1214,7 +1220,7 @@ static void s_prepare_links_from_balancer(dap_chain_net_t *a_net)
if (!l_link_node_info)
continue;
// Start connect to link hubs
- s_new_balancer_link_request(a_net);
+ s_new_balancer_link_request(a_net, 0);
l_cur_links_count++;
}
}
@@ -1325,9 +1331,11 @@ static bool s_net_states_proc(dap_proc_thread_t *a_thread, void *a_arg)
break;
}
// Get DNS request result from root nodes as synchronization links
- if (!l_net_pvt->only_static_links)
+ if (!l_net_pvt->only_static_links) {
+ pthread_rwlock_unlock(&l_net_pvt->states_lock);
s_prepare_links_from_balancer(l_net);
- else {
+ pthread_rwlock_wrlock(&l_net_pvt->states_lock);
+ } else {
log_it(L_ATT, "Not use bootstrap addresses, fill seed nodelist from root aliases");
// Add other root nodes as synchronization links
s_fill_links_from_root_aliases(l_net);
@@ -1384,7 +1392,7 @@ int s_net_list_compare_uuids(const void *a_uuid1, const void *a_uuid2)
bool dap_chain_net_sync_trylock(dap_chain_net_t *a_net, dap_chain_node_client_t *a_client)
{
dap_chain_net_pvt_t *l_net_pvt = PVT(a_net);
- int a_err = pthread_rwlock_wrlock(&l_net_pvt->uplinks_lock);
+ int a_err = pthread_rwlock_rdlock(&l_net_pvt->uplinks_lock);
bool l_found = false;
if (l_net_pvt->active_link) {
struct net_link *l_link, *l_link_tmp;
@@ -2248,8 +2256,8 @@ int s_net_load(const char * a_net_name, uint16_t a_acl_idx)
l_net->pub.gdb_groups_prefix = dap_strdup (
dap_config_get_item_str_default(l_cfg , "general" , "gdb_groups_prefix",
dap_config_get_item_str(l_cfg , "general" , "name" ) ) );
- dap_chain_global_db_add_sync_group(l_net->pub.name, "global", s_gbd_history_callback_notify, l_net);
- dap_chain_global_db_add_sync_group(l_net->pub.name, l_net->pub.gdb_groups_prefix, s_gbd_history_callback_notify, l_net);
+ dap_global_db_add_sync_group(l_net->pub.name, "global", s_gbd_history_callback_notify, l_net);
+ dap_global_db_add_sync_group(l_net->pub.name, l_net->pub.gdb_groups_prefix, s_gbd_history_callback_notify, l_net);
l_net->pub.gdb_nodes = dap_strdup_printf("%s.nodes",l_net->pub.gdb_groups_prefix);
l_net->pub.gdb_nodes_aliases = dap_strdup_printf("%s.nodes.aliases",l_net->pub.gdb_groups_prefix);
@@ -2298,7 +2306,7 @@ int s_net_load(const char * a_net_name, uint16_t a_acl_idx)
if (l_gdb_sync_groups && l_gdb_sync_groups_count > 0) {
for(uint16_t i = 0; i < l_gdb_sync_groups_count; i++) {
// add group to special sync
- dap_chain_global_db_add_sync_extra_group(l_net->pub.name, l_gdb_sync_groups[i], s_gbd_history_callback_notify, l_net);
+ dap_global_db_add_sync_extra_group(l_net->pub.name, l_gdb_sync_groups[i], s_gbd_history_callback_notify, l_net);
}
}
@@ -3115,7 +3123,7 @@ bool dap_chain_net_get_flag_sync_from_zero( dap_chain_net_t * a_net)
}
-bool s_proc_mempool_callback_load(dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total, const size_t a_values_shift,
+bool s_proc_mempool_callback_load(dap_global_db_context_t * a_global_db_context,int a_rc, const char * a_group, const char * a_key, const size_t a_values_total,
const size_t a_values_count, dap_global_db_obj_t * a_values, void * a_arg)
{
dap_chain_t * l_chain = (dap_chain_t*) a_arg;
diff --git a/modules/net/dap_chain_node_cli_cmd.c b/modules/net/dap_chain_node_cli_cmd.c
index a8ce008a4056a1edc585a46803473aaf9252ede3..8b744f73348042dbf367056fc93091f1117b5c7a 100644
--- a/modules/net/dap_chain_node_cli_cmd.c
+++ b/modules/net/dap_chain_node_cli_cmd.c
@@ -69,6 +69,7 @@
#include "dap_chain_wallet.h"
#include "dap_chain_node.h"
#include "dap_global_db.h"
+#include "dap_global_db_driver.h"
#include "dap_chain_node_client.h"
#include "dap_chain_node_cli_cmd.h"
#include "dap_chain_node_cli_cmd_tx.h"
@@ -95,7 +96,7 @@
#include "dap_chain_ledger.h"
#include "dap_chain_mempool.h"
#include "dap_global_db.h"
-#include "dap_chain_global_db_remote.h"
+#include "dap_global_db_remote.h"
#include "dap_stream_ch_chain_net.h"
#include "dap_stream_ch_chain.h"
@@ -257,13 +258,13 @@ static bool node_info_save_and_reply(dap_chain_net_t * a_net, dap_chain_node_inf
//memcpy(l_node_info, a_node_info, l_node_info_size );
//size_t data_len_out = 0;
- //dap_chain_node_info_t *a_node_info1 = dap_chain_global_db_gr_get(a_key, &data_len_out, a_net->pub.gdb_nodes);
+ //dap_chain_node_info_t *a_node_info1 = dap_global_db_gr_get(a_key, &data_len_out, a_net->pub.gdb_nodes);
bool res = dap_global_db_set_sync(a_net->pub.gdb_nodes, a_key, (uint8_t *) a_node_info, l_node_info_size,
true) == 0;
//data_len_out = 0;
- //dap_chain_node_info_t *a_node_info2 = dap_chain_global_db_gr_get(a_key, &data_len_out, a_net->pub.gdb_nodes);
+ //dap_chain_node_info_t *a_node_info2 = dap_global_db_gr_get(a_key, &data_len_out, a_net->pub.gdb_nodes);
//DAP_DELETE(a_key);
//DAP_DELETE(a_value);
return res;
@@ -5295,7 +5296,7 @@ int cmd_gdb_export(int argc, char ** argv, char ** a_str_reply)
struct json_object *l_json = json_object_new_array();
dap_list_t *l_groups_list = l_parsed_groups_list
? l_parsed_groups_list
- : dap_chain_global_db_driver_get_groups_by_mask("*");
+ : dap_global_db_driver_get_groups_by_mask("*");
for (dap_list_t *l_list = l_groups_list; l_list; l_list = dap_list_next(l_list)) {
size_t l_store_obj_count = 0;
char *l_group_name = (char *)l_list->data;
@@ -5410,7 +5411,7 @@ int cmd_gdb_import(int argc, char ** argv, char ** a_str_reply)
dap_enc_base64_decode(l_value_str, strlen(l_value_str), l_val, DAP_ENC_DATA_TYPE_B64);
l_group_store[j].value = (uint8_t*)l_val;
}
- if (dap_chain_global_db_driver_apply(l_group_store, l_records_count)) {
+ if (dap_global_db_driver_apply(l_group_store, l_records_count)) {
log_it(L_CRITICAL, "An error occured on importing group %s...", l_group_name);
} else {
log_it(L_INFO, "Imported %zu records of group %s", l_records_count, l_group_name);
diff --git a/modules/net/dap_chain_node_client.c b/modules/net/dap_chain_node_client.c
index 5e574edbf96933c917e34e0731d6b3fd0534974d..faa6313fbdd53862a693cedcd726e5620974d088 100644
--- a/modules/net/dap_chain_node_client.c
+++ b/modules/net/dap_chain_node_client.c
@@ -56,7 +56,7 @@
#include "dap_uuid.h"
#include "dap_client.h"
#include "dap_client_pvt.h"
-#include "dap_chain_global_db_remote.h"
+#include "dap_global_db_remote.h"
#include "dap_chain.h"
#include "dap_chain_cell.h"
#include "dap_chain_net_srv.h"
diff --git a/modules/net/dap_chain_node_dns_client.c b/modules/net/dap_chain_node_dns_client.c
index 1601f8087bb460b56600bee6a33653268f2f68a5..c86d9e282635e355d5fbb753ed297879cf6e2874 100644
--- a/modules/net/dap_chain_node_dns_client.c
+++ b/modules/net/dap_chain_node_dns_client.c
@@ -186,7 +186,7 @@ static void s_dns_client_esocket_worker_assign_callback(dap_events_socket_t * a_
* @param a_callback_error
* @param a_callbacks_arg
*/
-int dap_chain_node_info_dns_request(struct in_addr a_addr, uint16_t a_port, char *a_name,
+int dap_chain_node_info_dns_request(dap_worker_t *a_worker, struct in_addr a_addr, uint16_t a_port, char *a_name,
dap_dns_client_node_info_request_success_callback_t a_callback_success,
dap_dns_client_node_info_request_error_callback_t a_callback_error,void * a_callbacks_arg)
{
@@ -238,8 +238,7 @@ int dap_chain_node_info_dns_request(struct in_addr a_addr, uint16_t a_port, char
l_esocket->remote_addr.sin_addr = a_addr;
l_esocket->_inheritor = l_dns_client;
- dap_worker_t * l_worker = dap_events_worker_get_auto();
- dap_events_socket_assign_on_worker_mt(l_esocket,l_worker);
+ dap_events_socket_assign_on_worker_mt(l_esocket, a_worker);
return 0;
}
diff --git a/modules/net/dap_chain_node_dns_server.c b/modules/net/dap_chain_node_dns_server.c
index 6b04324e136d5822d88220bbdc1a18a3094f8a5f..7ed8fea2822e41a1262efc404947b15571961a7f 100644
--- a/modules/net/dap_chain_node_dns_server.c
+++ b/modules/net/dap_chain_node_dns_server.c
@@ -32,7 +32,7 @@
#include "dap_chain_node.h"
#include "dap_string.h"
#include "dap_global_db.h"
-#include "dap_chain_global_db_remote.h"
+#include "dap_global_db_remote.h"
#include "dap_chain_net_balancer.h"
#define LOG_TAG "dap_chain_node_dns_server"
diff --git a/modules/net/include/dap_chain_node_dns_client.h b/modules/net/include/dap_chain_node_dns_client.h
index ca5f719b00dc6214324a0f893c2ad220d05a91fd..aee9ec1293cd87a8b0d8ff9f88288d917387df6b 100644
--- a/modules/net/include/dap_chain_node_dns_client.h
+++ b/modules/net/include/dap_chain_node_dns_client.h
@@ -40,9 +40,9 @@ typedef struct _dap_dns_buf_t {
typedef void (*dap_dns_client_node_info_request_success_callback_t) (dap_worker_t *a_worker, dap_chain_node_info_t *a_result, void *a_arg);
typedef void (*dap_dns_client_node_info_request_error_callback_t) (dap_worker_t *a_worker, void *a_arg, int a_errno);
-int dap_chain_node_info_dns_request(struct in_addr a_addr, uint16_t a_port, char *a_name,
+int dap_chain_node_info_dns_request(dap_worker_t *a_worker, struct in_addr a_addr, uint16_t a_port, char *a_name,
dap_dns_client_node_info_request_success_callback_t a_callback_success,
- dap_dns_client_node_info_request_error_callback_t a_callback_error,void * a_callback_arg);
+ dap_dns_client_node_info_request_error_callback_t a_callback_error, void * a_callback_arg);
dap_chain_node_info_t *dap_dns_resolve_hostname(char *str);
diff --git a/modules/net/srv/dap_chain_net_srv_order.c b/modules/net/srv/dap_chain_net_srv_order.c
index 23bec9c32f4b37b6f0fc23c22726dde8ea9ded58..85480272b0ed1bd2b42b0573b70776528046cdab 100644
--- a/modules/net/srv/dap_chain_net_srv_order.c
+++ b/modules/net/srv/dap_chain_net_srv_order.c
@@ -622,7 +622,7 @@ static void s_srv_order_callback_notify(void *a_arg, const char a_op_code, const
}
/*dap_chain_hash_fast_t l_pkey_hash;
if (!dap_sign_get_pkey_hash(l_sign, &l_pkey_hash)) {
- dap_chain_global_db_gr_del(dap_strdup(a_key), a_group);
+ dap_global_db_gr_del(dap_strdup(a_key), a_group);
DAP_DELETE(l_gdb_group_str);
return;
}
@@ -631,7 +631,7 @@ static void s_srv_order_callback_notify(void *a_arg, const char a_op_code, const
uint128_t l_balance = dap_chain_ledger_calc_balance(l_net->pub.ledger, &l_addr, l_order->price_ticker);
uint64_t l_solvency = dap_chain_uint128_to(l_balance);
if (l_solvency < l_order->price && !dap_chain_net_srv_stake_key_delegated(&l_addr)) {
- dap_chain_global_db_gr_del(dap_strdup(a_key), a_group);
+ dap_global_db_gr_del(dap_strdup(a_key), a_group);
}*/
}
}
diff --git a/modules/service/stake_lock/dap_chain_net_srv_stake_lock.c b/modules/service/stake_lock/dap_chain_net_srv_stake_lock.c
index cde50a6232ac6fb3fd002f447e5a0503877cbd78..5fae0fbe77461e1d71ad0a1e3cd053fa2fc93dd7 100644
--- a/modules/service/stake_lock/dap_chain_net_srv_stake_lock.c
+++ b/modules/service/stake_lock/dap_chain_net_srv_stake_lock.c
@@ -24,7 +24,6 @@
#include "dap_chain_net_srv.h"
#include "dap_chain_net_srv_stake_lock.h"
-#include "dap_chain_global_db.h"
#include "dap_chain_node_cli.h"
#include "dap_chain_mempool.h"
#include "dap_chain_wallet.h"
diff --git a/modules/service/vpn/dap_chain_net_vpn_client.c b/modules/service/vpn/dap_chain_net_vpn_client.c
index a6a3a14fb60101f8d6a95e4e41e20b8955a32435..72bc6fe573c676a81ec092ac0154ed434b814578 100644
--- a/modules/service/vpn/dap_chain_net_vpn_client.c
+++ b/modules/service/vpn/dap_chain_net_vpn_client.c
@@ -268,7 +268,7 @@ static dap_chain_hash_fast_t* dap_chain_net_vpn_client_tx_cond_hash(dap_chain_ne
if(l_tx_cond_hash)
return l_tx_cond_hash;
- //l_pkey_b64 = (char*) dap_chain_global_db_gr_get(dap_strdup("client_pkey"), &l_gdb_group_size, l_gdb_group);
+ //l_pkey_b64 = (char*) dap_global_db_gr_get(dap_strdup("client_pkey"), &l_gdb_group_size, l_gdb_group);
dap_enc_key_t *l_enc_key = NULL;
if(a_wallet) {
l_enc_key = dap_chain_wallet_get_key(a_wallet, 0);
@@ -302,7 +302,7 @@ static dap_chain_hash_fast_t* dap_chain_net_vpn_client_tx_cond_hash(dap_chain_ne
uint8_t *l_pub_key_data = dap_enc_key_serealize_pub_key(l_cert->enc_key, &l_pub_key_data_size);
// save pub key
if(l_pub_key_data && l_pub_key_data_size > 0){
- if(dap_chain_global_db_gr_set(dap_strdup("client_pkey"), l_pub_key_data, l_pub_key_data_size,
+ if(dap_global_db_gr_set(dap_strdup("client_pkey"), l_pub_key_data, l_pub_key_data_size,
l_gdb_group)){
l_pkey_b64 = l_pub_key_data;
l_pkey_b64_size = l_pub_key_data_size;
diff --git a/modules/type/dag/CMakeLists.txt b/modules/type/dag/CMakeLists.txt
index f77a32ffac6a681c92b26e62e90c7dd84a6c9a45..9dd3c3f2ef89b9b8e30e9f151704a00fb836b73e 100644
--- a/modules/type/dag/CMakeLists.txt
+++ b/modules/type/dag/CMakeLists.txt
@@ -5,7 +5,7 @@ file(GLOB DAP_CHAIN_DAG_SRCS *.c)
file(GLOB DAP_CHAIN_DAG_HEADERS include/*.h)
add_library(${PROJECT_NAME} STATIC ${DAP_CHAIN_DAG_SRCS} ${DAP_CHAIN_DAG_HEADERS})
-target_link_libraries(${PROJECT_NAME} dap_core dap_crypto dap_chain dap_chain_net dap_chain_global_db)
+target_link_libraries(${PROJECT_NAME} dap_core dap_crypto dap_chain dap_chain_net dap_global_db)
target_include_directories(${PROJECT_NAME} INTERFACE .)
target_include_directories(${PROJECT_NAME} PUBLIC include)
target_include_directories(${PROJECT_NAME} PUBLIC ${CMAKE_CURRENT_SOURCE_DIR}/../../../3rdparty/uthash/src)
diff --git a/modules/type/dag/dap_chain_cs_dag.c b/modules/type/dag/dap_chain_cs_dag.c
index 6f1728dd57801f3be868f92cfedd35c5c5b50276..ab08847c8465fc1224e944edebc1377ea6a25f13 100644
--- a/modules/type/dag/dap_chain_cs_dag.c
+++ b/modules/type/dag/dap_chain_cs_dag.c
@@ -48,7 +48,7 @@
#include "dap_chain_cs.h"
#include "dap_chain_cs_dag.h"
#include "dap_global_db.h"
-#include "dap_chain_global_db_driver.h"
+#include "dap_global_db_driver.h"
#include "dap_chain_node_cli.h"
#include "dap_chain_node_cli_cmd.h"
#include "dap_chain_cell.h"
@@ -279,7 +279,7 @@ int dap_chain_cs_dag_new(dap_chain_t * a_chain, dap_config_t * a_chain_cfg)
l_dag->gdb_group_events_round_new = dap_strdup_printf("%s.%s", l_gdb_group, l_round_new_str);
DAP_DELETE(l_gdb_group);
DAP_DELETE(l_round_new_str);
- dap_chain_global_db_add_sync_extra_group(l_net->pub.name, l_dag->gdb_group_events_round_new, s_history_callback_round_notify, l_dag);
+ dap_global_db_add_sync_extra_group(l_net->pub.name, l_dag->gdb_group_events_round_new, s_history_callback_round_notify, l_dag);
l_dag->broadcast_disable = false;
byte_t *l_current_round = dap_global_db_get_sync(l_gdb_group, DAG_ROUND_CURRENT_KEY, NULL, NULL, NULL);
l_dag->round_current = l_current_round? *(uint64_t *)l_current_round : 0;
diff --git a/modules/wallet/dap_chain_wallet.c b/modules/wallet/dap_chain_wallet.c
index fd7e2aceff5130323258b323293a7cba3223ec00..41978b18e4a617a4702e9d8286f89cff146d41f5 100644
--- a/modules/wallet/dap_chain_wallet.c
+++ b/modules/wallet/dap_chain_wallet.c
@@ -65,8 +65,10 @@
#define LOG_TAG "dap_chain_wallet"
+#ifndef DAP_OS_WINDOWS
/* An argument for open()/create() */
static const mode_t s_fileprot = ( S_IREAD | S_IWRITE) | (S_IREAD >> 3) | (S_IREAD >> 6) ;
+#endif
static char s_wallet_ext [] = ".dwallet";
@@ -113,10 +115,10 @@ char *c_wallets_path;
/* Sanity checks ... */
if ( a_name_len > DAP_WALLET$SZ_NAME )
- return log_it(L_ERROR, "Wallet's name is too long (%d > %d)", a_name_len, DAP_WALLET$SZ_NAME), -EINVAL;
+ return log_it(L_ERROR, "Wallet's name is too long (%zd > %d)", a_name_len, DAP_WALLET$SZ_NAME), -EINVAL;
if ( a_pass_len > DAP_WALLET$SZ_PASS )
- return log_it(L_ERROR, "Wallet's password is too long (%d > %d)", a_pass_len, DAP_WALLET$SZ_PASS), -EINVAL;
+ return log_it(L_ERROR, "Wallet's password is too long (%zd > %d)", a_pass_len, DAP_WALLET$SZ_PASS), -EINVAL;
memcpy(l_rec.name, a_name, l_rec.name_len = a_name_len); /* Prefill local record fields */
@@ -206,10 +208,10 @@ struct timespec l_now;
/* Sanity checks ... */
if ( a_name_len > DAP_WALLET$SZ_NAME )
- return log_it(L_ERROR, "Wallet's name is too long (%d > %d)", a_name_len, DAP_WALLET$SZ_NAME), -EINVAL;
+ return log_it(L_ERROR, "Wallet's name is too long (%zd > %d)", a_name_len, DAP_WALLET$SZ_NAME), -EINVAL;
if ( *a_pass_len < DAP_WALLET$SZ_PASS )
- return log_it(L_ERROR, "Wallet's buffer for password is too small (%d < %d)", *a_pass_len, DAP_WALLET$SZ_PASS), -EINVAL;
+ return log_it(L_ERROR, "Wallet's buffer for password is too small (%zd < %d)", *a_pass_len, DAP_WALLET$SZ_PASS), -EINVAL;
clock_gettime(CLOCK_REALTIME, &l_now);
@@ -268,7 +270,7 @@ int l_rc, l_rc2;
dap_chain_wallet_n_pass_t *l_prec;
if ( a_name_len > DAP_WALLET$SZ_NAME )
- return log_it(L_ERROR, "Wallet's name is too long (%d > %d)", a_name_len, DAP_WALLET$SZ_NAME), -EINVAL;
+ return log_it(L_ERROR, "Wallet's name is too long (%zd > %d)", a_name_len, DAP_WALLET$SZ_NAME), -EINVAL;
if ( (l_rc = pthread_rwlock_wrlock(&s_wallet_n_pass_lock)) ) /* Lock for WR access */
return log_it(L_ERROR, "Error locking Wallet table, errno=%d", l_rc), -l_rc;
@@ -280,7 +282,7 @@ dap_chain_wallet_n_pass_t *l_prec;
if ( l_prec )
{
if ( !l_prec->pass_len ) /* Password is zero - has been reset probably */
- log_it(L_WARNING, "The Wallet %.*s is not active", a_name_len, a_name);
+ log_it(L_WARNING, "The Wallet %.*s is not active", (int)a_name_len, a_name);
else if ( (l_prec->pass_len != a_pass_len) /* Check that passwords is equivalent */
|| memcmp(l_prec->pass, a_pass, l_prec->pass_len) )
@@ -295,15 +297,6 @@ dap_chain_wallet_n_pass_t *l_prec;
return l_rc;
}
-
-
-
-
-
-
-
-
-
/**
* @brief dap_chain_wallet_init
* @return
@@ -334,15 +327,17 @@ size_t l_len;
*/
while( (l_dir_entry = readdir(l_dir)))
{
+#ifndef DAP_OS_WINDOWS
if ( l_dir_entry->d_type != DT_REG ) /* Skip unrelated entries */
continue;
-
+#endif
l_len = strlen(l_dir_entry->d_name); /* Check for *.dwallet */
if ( (l_len > 8) && (strcmp(l_dir_entry->d_name + l_len - (sizeof(s_wallet_ext) - 1), s_wallet_ext) == 0) )
{
- dap_snprintf(l_fspec, sizeof(l_fspec) - 1, "%s/%s", c_wallets_path, l_dir_entry->d_name);
-
+ int ret = dap_snprintf(l_fspec, sizeof(l_fspec) - 1, "%s/%s", c_wallets_path, l_dir_entry->d_name);
+ if (ret < 0)
+ continue;
if ( (l_wallet = dap_chain_wallet_open_file(l_fspec, NULL)) )
dap_chain_wallet_close(l_wallet);
}
@@ -412,12 +407,10 @@ int l_rc, l_wallet_name_len, l_pass_len;
return log_it(L_ERROR, "Wallet's password is too long ( > %d)", DAP_WALLET$SZ_PASS), NULL;
if ( !(l_wallet = DAP_NEW_Z(dap_chain_wallet_t)) )
- return log_it(L_ERROR, "Memory allocation error, errno=&d", errno), NULL;
+ return log_it(L_ERROR, "Memory allocation error, errno=%d", errno), NULL;
if ( !(l_wallet->_internal = l_wallet_internal = DAP_NEW_Z(dap_chain_wallet_internal_t)) )
- return DAP_DELETE(l_wallet), log_it(L_ERROR, "Memory allocation error, errno=&d", errno), NULL;
-
-
+ return DAP_DELETE(l_wallet), log_it(L_ERROR, "Memory allocation error, errno=%d", errno), NULL;
strncpy(l_wallet->name, a_wallet_name, DAP_WALLET$SZ_NAME);
l_wallet_internal->certs_count = 1;
@@ -585,6 +578,7 @@ dap_enc_key_t* dap_chain_wallet_get_key( dap_chain_wallet_t * a_wallet,uint32_t
int dap_chain_wallet_save(dap_chain_wallet_t * a_wallet, const char *a_pass)
{
DAP_CHAIN_WALLET_INTERNAL_LOCAL (a_wallet); /* Declare l_wallet_internal */
+#ifndef DAP_OS_WINDOWS
int l_fd = -1, l_rc = 0, l_len = 0;
dap_chain_wallet_file_hdr_t l_file_hdr = {0};
dap_chain_wallet_cert_hdr_t l_wallet_cert_hdr = {0};
@@ -698,7 +692,7 @@ struct iovec l_iov [ WALLET$SZ_IOV_NR ];
if ( l_enc_key )
dap_enc_key_delete(l_enc_key);
-
+#endif
#ifdef DAP_SYS_DEBUG /* @RRL: For debug purpose only!!! */