ISAR/libmdbx
2
0
Fork 0
mirror of https://github.com/isar/libmdbx.git synced 2026-01-27 19:32:22 +08:00
Code Issues Packages Projects Releases Wiki Activity

mdbx: rework using of mdbx_mmap_t and mdbx_msync().

Browse Source
This commit is contained in:
Leo Yuriev
2017-07-11 14:10:24 +03:00
parent 70d54f6f2c
commit 455de97d36
4 changed files with 64 additions and 101 deletions
Download Patch File Download Diff File Expand all files Collapse all files

105
src/mdbx.c
View File

@@ -1852,14 +1852,8 @@ static int mdbx_page_alloc(MDBX_cursor *mc, unsigned num, MDBX_page **mp,
"), %" PRIuPTR " bytes",
growth_pgno, growth_pgno - txn->mt_end_pgno, growth_bytes);
mdbx_mmap_param_t mmap;
mmap.address = env->me_map;
#ifdef MDBX_OSAL_SECTION
mmap.section = env->me_dxb_section;
#endif
mmap.fd = env->me_fd;
rc =
mdbx_mresize(env->me_flags, &mmap, env->me_dbgeo.now, growth_bytes);
rc = mdbx_mresize(env->me_flags, &env->me_dxb_mmap, env->me_dbgeo.now,
growth_bytes);
if (rc == MDBX_SUCCESS) {
txn->mt_end_pgno = growth_pgno;
env->me_dbgeo.now = growth_bytes;
@@ -2146,7 +2140,8 @@ int mdbx_env_sync(MDBX_env *env, int force) {
/* LY: pre-sync without holding lock to reduce latency for writer(s) */
int rc = (flags & MDBX_WRITEMAP)
? mdbx_msync(env->me_map, used_size, flags & MDBX_MAPASYNC)
? mdbx_msync(&env->me_dxb_mmap, 0, used_size,
flags & MDBX_MAPASYNC)
: mdbx_filesync(env->me_fd, false);
if (unlikely(rc != MDBX_SUCCESS))
return rc;
@@ -3783,7 +3778,7 @@ static int mdbx_sync_locked(MDBX_env *env, unsigned flags,
mdbx_assert(env, ((flags ^ env->me_flags) & MDBX_WRITEMAP) == 0);
MDBX_meta *const steady = mdbx_meta_steady(env);
if (flags & MDBX_WRITEMAP) {
rc = mdbx_msync(env->me_map, usedbytes, flags & MDBX_MAPASYNC);
rc = mdbx_msync(&env->me_dxb_mmap, 0, usedbytes, flags & MDBX_MAPASYNC);
if (unlikely(rc != MDBX_SUCCESS))
goto fail;
if ((flags & MDBX_MAPASYNC) == 0) {
@@ -3885,7 +3880,6 @@ static int mdbx_sync_locked(MDBX_env *env, unsigned flags,
mdbx_assert(env, !mdbx_meta_eq(env, pending, meta1));
mdbx_assert(env, !mdbx_meta_eq(env, pending, meta2));
const size_t offset = (char *)target - env->me_map;
mdbx_assert(env, ((env->me_flags ^ flags) & MDBX_WRITEMAP) == 0);
mdbx_ensure(env,
target == head ||
@@ -3932,16 +3926,18 @@ static int mdbx_sync_locked(MDBX_env *env, unsigned flags,
mdbx_jitter4testing(true);
} else {
pending->mm_magic_and_version = MDBX_DATA_MAGIC;
rc = mdbx_pwrite(env->me_fd, pending, sizeof(MDBX_meta), offset);
rc = mdbx_pwrite(env->me_fd, pending, sizeof(MDBX_meta),
(uint8_t *)target - env->me_map);
if (unlikely(rc != MDBX_SUCCESS)) {
undo:
mdbx_debug("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(env->me_fd, (void *)target, sizeof(MDBX_meta), offset);
mdbx_pwrite(env->me_fd, (void *)target, sizeof(MDBX_meta),
(uint8_t *)target - env->me_map);
goto fail;
}
mdbx_invalidate_cache(env->me_map + offset, sizeof(MDBX_meta));
mdbx_invalidate_cache(target, sizeof(MDBX_meta));
}
/* LY: step#3 - sync meta-pages. */
@@ -3949,8 +3945,13 @@ static int mdbx_sync_locked(MDBX_env *env, unsigned flags,
if ((flags & (MDBX_NOSYNC | MDBX_NOMETASYNC)) == 0) {
mdbx_assert(env, ((flags ^ env->me_flags) & MDBX_WRITEMAP) == 0);
if (flags & MDBX_WRITEMAP) {
char *ptr = env->me_map + (offset & ~(env->me_os_psize - 1));
rc = mdbx_msync(ptr, env->me_os_psize, flags & MDBX_MAPASYNC);
const size_t offset = (uint8_t *)container_of(head, MDBX_page, mp_meta) -
env->me_dxb_mmap.dxb;
const size_t paged_offset = offset & ~(env->me_os_psize - 1);
const size_t paged_length = mdbx_roundup2(
env->me_psize + offset - paged_offset, env->me_os_psize);
rc = mdbx_msync(&env->me_dxb_mmap, paged_offset, paged_length,
flags & MDBX_MAPASYNC);
if (unlikely(rc != MDBX_SUCCESS))
goto fail;
} else {
@@ -3965,13 +3966,8 @@ static int mdbx_sync_locked(MDBX_env *env, unsigned flags,
#else
/* LY: shrink datafile if needed */
if (shrink_pgno_delta) {
mdbx_mmap_param_t mmap;
mmap.address = env->me_map;
#ifdef MDBX_OSAL_SECTION
mmap.section = env->me_dxb_section;
#endif
mmap.fd = env->me_fd;
rc = mdbx_mresize(env->me_flags, &mmap, env->me_dbgeo.now, shrink_bytes);
rc = mdbx_mresize(env->me_flags, &env->me_dxb_mmap, env->me_dbgeo.now,
shrink_bytes);
if (rc == MDBX_SUCCESS)
env->me_dbgeo.now = shrink_bytes;
else if (rc != MDBX_RESULT_TRUE)
@@ -4082,19 +4078,13 @@ bailout:
}
static int __cold mdbx_env_map(MDBX_env *env, size_t usedsize) {
mdbx_mmap_param_t mmap;
mmap.fd = env->me_fd;
int rc = mdbx_mmap(env->me_flags, &mmap, env->me_dbgeo.now, env->me_mapsize);
int rc = mdbx_mmap(env->me_flags, &env->me_dxb_mmap, env->me_dbgeo.now,
env->me_mapsize);
if (unlikely(rc != MDBX_SUCCESS)) {
env->me_map = NULL;
return rc;
}
env->me_map = mmap.address;
#ifdef MDBX_OSAL_SECTION
env->me_dxb_section = mmap.section;
#endif
#ifdef MADV_DONTFORK
if (madvise(env->me_map, env->me_mapsize, MADV_DONTFORK))
return errno;
@@ -4130,7 +4120,7 @@ static int __cold mdbx_env_map(MDBX_env *env, size_t usedsize) {
/* Lock meta pages to avoid unexpected write,
* before the data pages would be synchronized. */
if (env->me_flags & MDBX_WRITEMAP) {
rc = mdbx_mlock(&mmap, pgno2bytes(env, NUM_METAS));
rc = mdbx_mlock(&env->me_dxb_mmap, pgno2bytes(env, NUM_METAS));
if (unlikely(rc != MDBX_SUCCESS))
return rc;
}
@@ -4351,17 +4341,11 @@ LIBMDBX_API int mdbx_env_set_geometry(MDBX_env *env, ssize_t size_lower,
const size_t size =
mdbx_roundup2(pgno2bytes(env, meta.mm_geo.upper), env->me_os_psize);
mdbx_mmap_param_t mmap;
mmap.address = env->me_map;
#ifdef MDBX_OSAL_SECTION
mmap.section = env->me_dxb_section;
#endif
mmap.fd = env->me_fd;
rc = mdbx_mremap(env->me_flags, &mmap, env->me_mapsize, size);
rc = mdbx_mremap(env->me_flags, &env->me_dxb_mmap, env->me_mapsize,
size);
if (unlikely(rc != MDBX_SUCCESS))
goto bailout;
env->me_mapsize = size;
env->me_map = mmap.address;
#ifdef USE_VALGRIND
VALGRIND_DISCARD(env->me_valgrind_handle);
env->me_valgrind_handle =
@@ -4372,13 +4356,7 @@ LIBMDBX_API int mdbx_env_set_geometry(MDBX_env *env, ssize_t size_lower,
const size_t size =
mdbx_roundup2(pgno2bytes(env, meta.mm_geo.now), env->me_os_psize);
mdbx_mmap_param_t mmap;
mmap.address = env->me_map;
#ifdef MDBX_OSAL_SECTION
mmap.section = env->me_dxb_section;
#endif
mmap.fd = env->me_fd;
rc = mdbx_mresize(env->me_flags, &mmap,
rc = mdbx_mresize(env->me_flags, &env->me_dxb_mmap,
pgno2bytes(env, head->mm_geo.now), size);
if (unlikely(rc != MDBX_SUCCESS))
goto bailout;
@@ -4736,15 +4714,9 @@ static int __cold mdbx_setup_lck(MDBX_env *env, char *lck_pathname, int mode) {
}
env->me_maxreaders = (unsigned)maxreaders;
mdbx_mmap_param_t mmap;
mmap.fd = env->me_lfd;
err = mdbx_mmap(MDBX_WRITEMAP, &mmap, (size_t)size, (size_t)size);
err = mdbx_mmap(MDBX_WRITEMAP, &env->me_lck_mmap, (size_t)size, (size_t)size);
if (unlikely(err != MDBX_SUCCESS))
return err;
env->me_lck = mmap.address;
#ifdef MDBX_OSAL_SECTION
env->me_lck_section = mmap.section;
#endif
#ifdef MADV_DODUMP
(void)madvise(env->me_lck, size, MADV_DODUMP);
@@ -5011,13 +4983,7 @@ static void __cold mdbx_env_close0(MDBX_env *env) {
}
if (env->me_map) {
mdbx_mmap_param_t mmap;
mmap.address = env->me_map;
#ifdef MDBX_OSAL_SECTION
mmap.section = env->me_dxb_section;
#endif
mmap.fd = env->me_fd;
mdbx_munmap(&mmap, env->me_mapsize);
mdbx_munmap(&env->me_dxb_mmap, env->me_mapsize);
#ifdef USE_VALGRIND
VALGRIND_DISCARD(env->me_valgrind_handle);
env->me_valgrind_handle = -1;
@@ -5029,14 +4995,9 @@ static void __cold mdbx_env_close0(MDBX_env *env) {
}
if (env->me_lck) {
mdbx_mmap_param_t mmap;
mmap.address = env->me_lck;
#ifdef MDBX_OSAL_SECTION
mmap.section = env->me_lck_section;
#endif
mmap.fd = env->me_lfd;
mdbx_munmap(&mmap, (env->me_maxreaders - 1) * sizeof(MDBX_reader) +
sizeof(MDBX_lockinfo));
mdbx_munmap(&env->me_lck_mmap,
(env->me_maxreaders - 1) * sizeof(MDBX_reader) +
sizeof(MDBX_lockinfo));
env->me_lck = nullptr;
}
env->me_pid = 0;
@@ -11027,9 +10988,9 @@ int mdbx_is_dirty(const MDBX_txn *txn, const void *ptr) {
* Тем не менее, однозначно страница "не грязная" (не будет переписана
* во время транзакции) если адрес находится внутри mmap-диапазона
* и в заголовке страницы нет флажка P_DIRTY. */
if (env->me_map < (char *)page) {
if (env->me_map < (uint8_t *)page) {
const size_t used_size = pgno2bytes(env, txn->mt_next_pgno);
if ((char *)page < env->me_map + used_size) {
if ((uint8_t *)page < env->me_map + used_size) {
/* страница внутри диапазона, смотрим на флажки */
return (page->mp_flags & (P_DIRTY | P_LOOSE | P_KEEP))
? MDBX_RESULT_TRUE
@@ -11040,7 +11001,7 @@ int mdbx_is_dirty(const MDBX_txn *txn, const void *ptr) {
* ошибка, к которой не возможно прийти без каких-то больших нарушений.
* Поэтому не проверяем этот случай кроме как assert-ом, на то что
* страница вне mmap-диаппазона. */
mdbx_tassert(txn, (char *)page >= env->me_map + env->me_mapsize);
mdbx_tassert(txn, (uint8_t *)page >= env->me_map + env->me_mapsize);
}
/* Страница вне используемого mmap-диапазона, т.е. либо в функцию был