*
* relmapper.c
* Catalog-to-filenode mapping
*
* For most tables, the physical file underlying the table is specified by
* pg_class.relfilenode. However, that obviously won't work for pg_class
* itself, nor for the other "nailed" catalogs for which we have to be able
* to set up working Relation entries without access to pg_class. It also
* does not work for shared catalogs, since there is no practical way to
* update other databases' pg_class entries when relocating a shared catalog.
* Therefore, for these special catalogs (henceforth referred to as "mapped
* catalogs") we rely on a separately maintained file that shows the mapping
* from catalog OIDs to filenode numbers. Each database has a map file for
* its local mapped catalogs, and there is a separate map file for shared
* catalogs. Mapped catalogs have zero in their pg_class.relfilenode entries.
*
* Relocation of a normal table is committed (ie, the new physical file becomes
* authoritative) when the pg_class row update commits. For mapped catalogs,
* the act of updating the map file is effectively commit of the relocation.
* We postpone the file update till just before commit of the transaction
* doing the rewrite, but there is necessarily a window between. Therefore
* mapped catalogs can only be relocated by operations such as VACUUM FULL
* and CLUSTER, which make no transactionally-significant changes: it must be
* safe for the new file to replace the old, even if the transaction itself
* aborts. An important factor here is that the indexes and toast table of
* a mapped catalog must also be mapped, so that the rewrites/relocations of
* all these files commit in a single map file update rather than being tied
* to transaction commit.
*
* Portions Copyright (c) 1996-2012, PostgreSQL Global Development Group
* Portions Copyright (c) 1994, Regents of the University of California
*
*
* IDENTIFICATION
* src/backend/utils/cache/relmapper.c
*
* -------------------------------------------------------------------------
*/
#include "postgres.h"
#include "knl/knl_variable.h"
#include <fcntl.h>
#include <sys/stat.h>
#include <unistd.h>
#include "access/xact.h"
#include "access/xlog.h"
#include "access/xloginsert.h"
#include "access/multi_redo_api.h"
#include "catalog/catalog.h"
#include "catalog/pg_tablespace.h"
#include "catalog/storage.h"
#include "miscadmin.h"
#include "storage/smgr/fd.h"
#include "storage/lock/lwlock.h"
#include "storage/file/fio_device.h"
#include "utils/inval.h"
#include "utils/relmapper.h"
static void apply_map_update(RelMapFile* map, Oid relationId, Oid fileNode, bool add_okay);
static void merge_map_updates(RelMapFile* map, const RelMapFile* updates, bool add_okay);
static void write_relmap_file(bool shared, RelMapFile* newmap, bool write_wal, bool send_sinval, bool preserve_files,
Oid dbid, Oid tsid, const char* dbpath, XLogRecPtr redo_lsn = 0);
static void perform_relmap_update(bool shared, const RelMapFile* updates);
static int WriteOldVersionRelmap(RelMapFile* map, int fd);
static int ReadOldVersionRelmap(RelMapFile* map, int fd);
static void InitRelmapVerTag(RelMapVerTag* mapTag, bool isNewMap);
static bool IsNewVersionMap(int fd, char* fileName);
static void RegistRelMapWal(RelMapFile* map);
static pg_crc32 RelmapCrcComp(RelMapFile* map);
static int32 ReadRelMapFile(RelMapFile* map, int fd, bool isNewMap);
static int32 WriteRelMapFile(RelMapFile* map, int fd);
static void recover_relmap_file(bool shared, bool backupfile, RelMapFile* real_map);
* RelationMapOidToFilenode
*
* The raison d' etre ... given a relation OID, look up its filenode.
*
* Although shared and local relation OIDs should never overlap, the caller
* always knows which we need --- so pass that information to avoid useless
* searching.
*
* Returns InvalidOid if the OID is not known (which should never happen,
* but the caller is in a better position to report a meaningful error).
*/
Oid RelationMapOidToFilenode(Oid relationId, bool shared)
{
const RelMapFile* map = NULL;
int32 i;
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
if (shared) {
map = relmap_cxt->active_shared_updates;
for (i = 0; i < map->num_mappings; i++) {
if (relationId == map->mappings[i].mapoid) {
return map->mappings[i].mapfilenode;
}
}
map = relmap_cxt->shared_map;
for (i = 0; i < map->num_mappings; i++) {
if (relationId == map->mappings[i].mapoid) {
return map->mappings[i].mapfilenode;
}
}
} else {
map = relmap_cxt->active_local_updates;
for (i = 0; i < map->num_mappings; i++) {
if (relationId == map->mappings[i].mapoid) {
return map->mappings[i].mapfilenode;
}
}
map = relmap_cxt->local_map;
for (i = 0; i < map->num_mappings; i++) {
if (relationId == map->mappings[i].mapoid) {
return map->mappings[i].mapfilenode;
}
}
}
return InvalidOid;
}
* RelationMapFilenodeToOid
*
* Do the reverse of the normal direction of mapping done in
* RelationMapOidToFilenode.
*
* This is not supposed to be used during normal running but rather for
* information purposes when looking at the filesystem or xlog.
*
* Returns InvalidOid if the OID is not known; this can easily happen if the
* relfilenode doesn't pertain to a mapped relation.
*/
Oid RelationMapFilenodeToOid(Oid filenode, bool shared)
{
const RelMapFile* map = NULL;
int32 i;
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
if (shared) {
map = relmap_cxt->active_shared_updates;
for (i = 0; i < map->num_mappings; i++) {
if (filenode == map->mappings[i].mapfilenode) {
return map->mappings[i].mapoid;
}
}
map = relmap_cxt->shared_map;
for (i = 0; i < map->num_mappings; i++) {
if (filenode == map->mappings[i].mapfilenode) {
return map->mappings[i].mapoid;
}
}
} else {
map = relmap_cxt->active_local_updates;
for (i = 0; i < map->num_mappings; i++) {
if (filenode == map->mappings[i].mapfilenode) {
return map->mappings[i].mapoid;
}
}
map = relmap_cxt->local_map;
for (i = 0; i < map->num_mappings; i++) {
if (filenode == map->mappings[i].mapfilenode) {
return map->mappings[i].mapoid;
}
}
}
return InvalidOid;
}
* RelationMapUpdateMap
*
* Install a new relfilenode mapping for the specified relation.
*
* If immediate is true (or we're bootstrapping), the mapping is activated
* immediately. Otherwise it is made pending until CommandCounterIncrement.
*/
void RelationMapUpdateMap(Oid relationId, Oid fileNode, bool shared, bool immediate)
{
RelMapFile* map = NULL;
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
if (IsBootstrapProcessingMode()) {
* In bootstrap mode, the mapping gets installed in permanent map.
*/
if (shared) {
map = relmap_cxt->shared_map;
} else {
map = relmap_cxt->local_map;
}
} else {
* We don't currently support map changes within subtransactions. This
* could be done with more bookkeeping infrastructure, but it doesn't
* presently seem worth it.
*/
if (GetCurrentTransactionNestLevel() > 1) {
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot change relation mapping within subtransaction")));
}
if (immediate) {
if (shared) {
map = relmap_cxt->active_shared_updates;
} else {
map = relmap_cxt->active_local_updates;
}
} else {
if (shared) {
map = relmap_cxt->pending_shared_updates;
} else {
map = relmap_cxt->pending_local_updates;
}
}
}
apply_map_update(map, relationId, fileNode, true);
}
* apply_map_update
*
* Insert a new mapping into the given map variable, replacing any existing
* mapping for the same relation.
*
* In some cases the caller knows there must be an existing mapping; pass
* add_okay = false to draw an error if not.
*/
static void apply_map_update(RelMapFile* map, Oid relationId, Oid fileNode, bool add_okay)
{
int32 i;
RelMapVerTag mapTag;
InitRelmapVerTag(&mapTag, IS_MAGIC_EXIST(map->magic) ? IS_NEW_RELMAP(map->magic) : true);
for (i = 0; i < map->num_mappings; i++) {
if (relationId == map->mappings[i].mapoid) {
map->mappings[i].mapfilenode = fileNode;
return;
}
}
if (!add_okay) {
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("attempt to apply a mapping to unmapped relation %u", relationId)));
}
if (map->num_mappings >= mapTag.relMaxMappings) {
if (map->magic == RELMAPPER_FILEMAGIC) {
map->magic = RELMAPPER_FILEMAGIC_4K;
ereport(LOG, (errmsg("Switch relmap to new version!")));
} else {
ereport(ERROR, (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED), errmsg("ran out of space in relation map")));
}
}
map->mappings[map->num_mappings].mapoid = relationId;
map->mappings[map->num_mappings].mapfilenode = fileNode;
map->num_mappings++;
}
* merge_map_updates
*
* Merge all the updates in the given pending-update map into the target map.
* This is just a bulk form of apply_map_update.
*/
static void merge_map_updates(RelMapFile* map, const RelMapFile* updates, bool add_okay)
{
int32 i;
for (i = 0; i < updates->num_mappings; i++) {
apply_map_update(map, updates->mappings[i].mapoid, updates->mappings[i].mapfilenode, add_okay);
}
}
* RelationMapRemoveMapping
*
* Remove a relation's entry in the map. This is only allowed for "active"
* (but not committed) local mappings. We need it so we can back out the
* entry for the transient target file when doing VACUUM FULL/CLUSTER on
* a mapped relation.
*/
void RelationMapRemoveMapping(Oid relationId)
{
RelMapFile* map = GetRelMapCxt()->active_local_updates;
int32 i;
for (i = 0; i < map->num_mappings; i++) {
if (relationId == map->mappings[i].mapoid) {
map->mappings[i] = map->mappings[map->num_mappings - 1];
map->num_mappings--;
return;
}
}
ereport(ERROR,
(errcode(ERRCODE_NO_DATA_FOUND), errmsg("could not find temporary mapping for relation %u", relationId)));
}
* RelationMapInvalidate
*
* This routine is invoked for SI cache flush messages. We must re-read
* the indicated map file. However, we might receive a SI message in a
* process that hasn't yet, and might never, load the mapping files;
* for example the autovacuum launcher, which *must not* try to read
* a local map since it is attached to no particular database.
* So, re-read only if the map is valid now.
*/
void RelationMapInvalidate(bool shared)
{
if (EnableLocalSysCache()) {
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
RelMapFile *rel_map = shared ? relmap_cxt->shared_map : relmap_cxt->local_map;
if (IS_MAGIC_EXIST(rel_map->magic)) {
t_thrd.lsc_cxt.lsc->LoadRelMapFromGlobal(shared);
}
return;
}
if (shared) {
if (IS_MAGIC_EXIST(u_sess->relmap_cxt.shared_map->magic)) {
LWLockAcquire(RelationMappingLock, LW_SHARED);
load_relmap_file(true, u_sess->relmap_cxt.shared_map);
LWLockRelease(RelationMappingLock);
}
} else {
if (IS_MAGIC_EXIST(u_sess->relmap_cxt.local_map->magic)) {
LWLockAcquire(RelationMappingLock, LW_SHARED);
load_relmap_file(false, u_sess->relmap_cxt.local_map);
LWLockRelease(RelationMappingLock);
}
}
}
* RelationMapInvalidateAll
*
* Reload all map files. This is used to recover from SI message buffer
* overflow: we can't be sure if we missed an inval message.
* Again, reload only currently-valid maps.
*/
void RelationMapInvalidateAll(void)
{
if (EnableLocalSysCache()) {
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
if (IS_MAGIC_EXIST(relmap_cxt->shared_map->magic)) {
t_thrd.lsc_cxt.lsc->LoadRelMapFromGlobal(true);
}
if (IS_MAGIC_EXIST(relmap_cxt->local_map->magic)) {
t_thrd.lsc_cxt.lsc->LoadRelMapFromGlobal(false);
}
return;
}
LWLockAcquire(RelationMappingLock, LW_SHARED);
if (IS_MAGIC_EXIST(u_sess->relmap_cxt.shared_map->magic)) {
load_relmap_file(true, u_sess->relmap_cxt.shared_map);
}
if (IS_MAGIC_EXIST(u_sess->relmap_cxt.local_map->magic)) {
load_relmap_file(false, u_sess->relmap_cxt.local_map);
}
LWLockRelease(RelationMappingLock);
}
* AtCCI_RelationMap
*
* Activate any "pending" relation map updates at CommandCounterIncrement time.
*/
void AtCCI_RelationMap(void)
{
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
if (relmap_cxt->pending_shared_updates->num_mappings != 0) {
merge_map_updates(relmap_cxt->active_shared_updates, relmap_cxt->pending_shared_updates, true);
relmap_cxt->pending_shared_updates->num_mappings = 0;
}
if (relmap_cxt->pending_local_updates->num_mappings != 0) {
merge_map_updates(relmap_cxt->active_local_updates, relmap_cxt->pending_local_updates, true);
relmap_cxt->pending_local_updates->num_mappings = 0;
}
}
* AtEOXact_RelationMap
*
* Handle relation mapping at main-transaction commit or abort.
*
* During commit, this must be called as late as possible before the actual
* transaction commit, so as to minimize the window where the transaction
* could still roll back after committing map changes. Although nothing
* critically bad happens in such a case, we still would prefer that it
* not happen, since we'd possibly be losing useful updates to the relations'
* pg_class row(s).
*
* During abort, we just have to throw away any pending map changes.
* Normal post-abort cleanup will take care of fixing relcache entries.
*/
void AtEOXact_RelationMap(bool isCommit)
{
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
if (isCommit) {
* We should not get here with any "pending" updates. (We could
* logically choose to treat such as committed, but in the current
* code this should never happen.)
*/
Assert(relmap_cxt->pending_shared_updates->num_mappings == 0);
Assert(relmap_cxt->pending_local_updates->num_mappings == 0);
* Write any active updates to the actual map files, then reset them.
*/
if (relmap_cxt->active_shared_updates->num_mappings != 0) {
perform_relmap_update(true, relmap_cxt->active_shared_updates);
relmap_cxt->active_shared_updates->num_mappings = 0;
}
if (relmap_cxt->active_local_updates->num_mappings != 0) {
perform_relmap_update(false, relmap_cxt->active_local_updates);
relmap_cxt->active_local_updates->num_mappings = 0;
}
} else {
relmap_cxt->active_shared_updates->num_mappings = 0;
relmap_cxt->active_local_updates->num_mappings = 0;
relmap_cxt->pending_shared_updates->num_mappings = 0;
relmap_cxt->pending_local_updates->num_mappings = 0;
}
}
* AtPrepare_RelationMap
*
* Handle relation mapping at PREPARE.
*
* Currently, we don't support preparing any transaction that changes the map.
*/
void AtPrepare_RelationMap(void)
{
if (u_sess->attr.attr_common.IsInplaceUpgrade) {
return;
}
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
if (relmap_cxt->active_shared_updates->num_mappings != 0 ||
relmap_cxt->active_local_updates->num_mappings != 0 ||
relmap_cxt->pending_shared_updates->num_mappings != 0 ||
relmap_cxt->pending_local_updates->num_mappings != 0) {
ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("cannot PREPARE a transaction that modified relation mapping")));
}
}
* CheckPointRelationMap
*
* This is called during a checkpoint. It must ensure that any relation map
* updates that were WAL-logged before the start of the checkpoint are
* securely flushed to disk and will not need to be replayed later. This
* seems unlikely to be a performance-critical issue, so we use a simple
* method: we just take and release the RelationMappingLock. This ensures
* that any already-logged map update is complete, because write_relmap_file
* will fsync the map file before the lock is released.
*/
void CheckPointRelationMap(void)
{
LWLockAcquire(RelationMappingLock, LW_SHARED);
LWLockRelease(RelationMappingLock);
}
* RelationMapFinishBootstrap
*
* Write out the initial relation mapping files at the completion of
* bootstrap. All the mapped files should have been made known to us
* via RelationMapUpdateMap calls.
*/
void RelationMapFinishBootstrap(void)
{
Assert(IsBootstrapProcessingMode());
if (ENABLE_DSS) {
char map_file_name[MAXPGPATH];
int rc = snprintf_s(map_file_name, sizeof(map_file_name), sizeof(map_file_name) - 1, "%s/global/%s",
g_instance.attr.attr_storage.dss_attr.ss_dss_data_vg_name, RELMAPPER_FILENAME);
securec_check_ss_c(rc, "\0", "\0");
struct stat st;
if (stat(map_file_name, &st) == 0 && S_ISREG(st.st_mode)) {
return;
}
}
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
Assert(relmap_cxt->active_shared_updates->num_mappings == 0);
Assert(relmap_cxt->active_local_updates->num_mappings == 0);
Assert(relmap_cxt->pending_shared_updates->num_mappings == 0);
Assert(relmap_cxt->pending_local_updates->num_mappings == 0);
write_relmap_file(true, relmap_cxt->shared_map, false, false, false, InvalidOid, GLOBALTABLESPACE_OID, NULL);
write_relmap_file(false,
relmap_cxt->local_map,
false,
false,
false,
u_sess->proc_cxt.MyDatabaseId,
u_sess->proc_cxt.MyDatabaseTableSpace,
u_sess->proc_cxt.DatabasePath);
}
* RelationMapInitialize
*
* This initializes the mapper module at process startup. We can't access the
* database yet, so just make sure the maps are empty.
*/
void RelationMapInitialize(void)
{
if (EnableLocalSysCache()) {
* when switchdb , we memset them zero, nothing need to do */
return;
}
u_sess->relmap_cxt.shared_map->magic = 0;
u_sess->relmap_cxt.local_map->magic = 0;
u_sess->relmap_cxt.shared_map->num_mappings = 0;
u_sess->relmap_cxt.local_map->num_mappings = 0;
u_sess->relmap_cxt.active_shared_updates->num_mappings = 0;
u_sess->relmap_cxt.active_local_updates->num_mappings = 0;
u_sess->relmap_cxt.pending_shared_updates->num_mappings = 0;
u_sess->relmap_cxt.pending_local_updates->num_mappings = 0;
}
* RelationMapInitializePhase2
*
* This is called to prepare for access to pg_database during startup.
* We should be able to read the shared map file now.
*/
void RelationMapInitializePhase2(void)
{
* In bootstrap mode, the map file isn't there yet, so do nothing.
*/
if (IsBootstrapProcessingMode()) {
return;
}
* Load the shared map file, die on error.
*/
if (EnableLocalSysCache()) {
t_thrd.lsc_cxt.lsc->InitRelMapPhase2();
return;
}
LWLockAcquire(RelationMappingLock, LW_SHARED);
load_relmap_file(true, u_sess->relmap_cxt.shared_map);
LWLockRelease(RelationMappingLock);
}
* RelationMapInitializePhase3
*
* This is called as soon as we have determined u_sess->proc_cxt.MyDatabaseId and set up
* u_sess->proc_cxt.DatabasePath. At this point we should be able to read the local map file.
*/
void RelationMapInitializePhase3(void)
{
* In bootstrap mode, the map file isn't there yet, so do nothing.
*/
if (IsBootstrapProcessingMode()) {
return;
}
* Load the local map file, die on error.
*/
if (EnableLocalSysCache()) {
t_thrd.lsc_cxt.lsc->InitRelMapPhase3();
return;
}
LWLockAcquire(RelationMappingLock, LW_SHARED);
load_relmap_file(false, u_sess->relmap_cxt.local_map);
LWLockRelease(RelationMappingLock);
}
* load_relmap_file -- load data from the shared or local map file
*
* Because the map file is essential for access to core system catalogs,
* failure to read it is a fatal error.
*
* Note that the local case requires u_sess->proc_cxt.DatabasePath to be set up.
*/
void load_relmap_file(bool shared, RelMapFile *map)
{
char map_file_name[2][MAXPGPATH];
char* file_name = NULL;
pg_crc32 crc;
int fd;
bool fix_backup = false;
bool retry = false;
struct stat stat_buf;
errno_t rc;
RelMapVerTag mapTag;
bool isNewMap;
if (shared) {
rc = snprintf_s(map_file_name[0],
sizeof(map_file_name[0]),
sizeof(map_file_name[0]) - 1,
"%s/%s",
GLOTBSDIR,
RELMAPPER_FILENAME);
securec_check_ss(rc, "\0", "\0");
rc = snprintf_s(map_file_name[1],
sizeof(map_file_name[1]),
sizeof(map_file_name[1]) - 1,
"%s/%s",
GLOTBSDIR,
RELMAPPER_FILENAME_BAK);
securec_check_ss(rc, "\0", "\0");
} else {
rc = snprintf_s(map_file_name[0],
sizeof(map_file_name[0]),
sizeof(map_file_name[0]) - 1,
"%s/%s",
u_sess->proc_cxt.DatabasePath,
RELMAPPER_FILENAME);
securec_check_ss(rc, "\0", "\0");
rc = snprintf_s(map_file_name[1],
sizeof(map_file_name[1]),
sizeof(map_file_name[1]) - 1,
"%s/%s",
u_sess->proc_cxt.DatabasePath,
RELMAPPER_FILENAME_BAK);
securec_check_ss(rc, "\0", "\0");
}
if (stat(map_file_name[1], &stat_buf) != 0) {
if (!FILE_POSSIBLY_DELETED(errno)) {
ereport(LOG, (errmsg("can not stat file \"%s\", ignore backup file", map_file_name[1])));
} else {
fix_backup = true;
LWLockRelease(RelationMappingLock);
LWLockAcquire(RelationMappingLock, LW_EXCLUSIVE);
}
}
file_name = map_file_name[0];
loop:
fd = BasicOpenFile(file_name, O_RDONLY | PG_BINARY, S_IRUSR | S_IWUSR);
if (fd < 0) {
ereport(
FATAL, (errcode_for_file_access(), errmsg("could not open relation mapping file \"%s\": %m", file_name)));
}
* Note: we take RelationMappingLock in shared mode here, because it
* seems that read() may not be atomic against any
* concurrent updater's write(). If the file is updated shortly after we
* look, the sinval signaling mechanism will make us re-read it before we
* are able to access any relation that's affected by the change.
*/
isNewMap = IsNewVersionMap(fd, file_name);
(void)lseek(fd, 0, SEEK_SET);
InitRelmapVerTag(&mapTag, isNewMap);
int readBytes = ReadRelMapFile(map, fd, isNewMap);
if (readBytes != mapTag.relSize) {
close(fd);
ereport(
FATAL, (errcode_for_file_access(), errmsg("could not read relation mapping file \"%s\": %m", file_name)));
}
if (close(fd)) {
ereport(PANIC, (errcode_for_file_access(), errmsg("could not close control file: %m")));
}
crc = RelmapCrcComp(map);
if (!EQ_CRC32(crc, (pg_crc32)(map->crc))) {
if (retry == false) {
ereport(WARNING,
(errmsg("relation mapping file \"%s\" contains incorrect checksum, try backup file", file_name)));
file_name = map_file_name[1];
retry = true;
if (fix_backup == false) {
LWLockRelease(RelationMappingLock);
LWLockAcquire(RelationMappingLock, LW_EXCLUSIVE);
}
goto loop;
} else {
ereport(FATAL, (errmsg("relation mapping file \"%s\" contains incorrect checksum", file_name)));
}
}
if (map->magic != mapTag.relMagic || map->num_mappings < 0 || map->num_mappings > mapTag.relMaxMappings) {
ereport(FATAL, (errmsg("relation mapping file \"%s\" contains invalid data", file_name)));
}
if (retry == true) {
recover_relmap_file(shared, false, map);
} else if (fix_backup == true) {
recover_relmap_file(shared, true, map);
}
}
* Write out a new shared or local map file with the given contents.
*
* The magic number and CRC are automatically updated in *newmap. On
* success, we copy the data to the appropriate permanent static variable.
*
* If write_wal is TRUE then an appropriate WAL message is emitted.
* (It will be false for bootstrap and WAL replay cases.)
*
* If send_sinval is TRUE then a SI invalidation message is sent.
* (This should be true except in bootstrap case.)
*
* If preserve_files is TRUE then the storage manager is warned not to
* delete the files listed in the map.
*
* Because this may be called during WAL replay when u_sess->proc_cxt.MyDatabaseId,
* u_sess->proc_cxt.DatabasePath, etc aren't valid, we require the caller to pass in suitable
* values. The caller is also responsible for being sure no concurrent
* map update could be happening.
*/
static void write_relmap_file(bool shared, RelMapFile* newmap, bool write_wal, bool send_sinval, bool preserve_files,
Oid dbid, Oid tsid, const char* dbpath, XLogRecPtr redo_lsn)
{
int fd;
RelMapFile* real_map = NULL;
char map_file_name[MAXPGPATH];
errno_t rc = 0;
char* fname[2];
RelMapVerTag mapTag;
if (!IS_MAGIC_EXIST(newmap->magic)) {
newmap->magic = RELMAPPER_FILEMAGIC_4K;
}
InitRelmapVerTag(&mapTag, IS_NEW_RELMAP(newmap->magic));
* Fill in the overhead fields and update CRC.
*/
if (newmap->num_mappings < 0 || newmap->num_mappings > mapTag.relMaxMappings) {
ereport(
ERROR, (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), errmsg("attempt to write bogus relation mapping")));
}
newmap->crc = RelmapCrcComp(newmap);
* Open the target file. We prefer to do this before entering the
* critical section, so that an open() failure need not force PANIC.
*
* Note: since we use BasicOpenFile, we are nominally responsible for
* ensuring the fd is closed on error. In practice, this isn't important
* because either an error happens inside the critical section, or we are
* in bootstrap or WAL replay; so an error past this point is always fatal
* anyway.
*/
fname[0] = RELMAPPER_FILENAME_BAK;
fname[1] = RELMAPPER_FILENAME;
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
for (int i = 0; i < 2; i++) {
if (shared) {
rc = snprintf_s(map_file_name,
sizeof(map_file_name),
sizeof(map_file_name) - 1,
"%s/%s",
GLOTBSDIR,
fname[i]);
securec_check_ss_c(rc, "\0", "\0");
real_map = relmap_cxt->shared_map;
} else {
rc = snprintf_s(map_file_name,
sizeof(map_file_name),
sizeof(map_file_name) - 1,
"%s/%s",
dbpath,
fname[i]);
securec_check_ss_c(rc, "\0", "\0");
real_map = relmap_cxt->local_map;
}
fd = BasicOpenFile(map_file_name, O_WRONLY | O_CREAT | PG_BINARY, S_IRUSR | S_IWUSR);
if (fd < 0) {
ereport(ERROR,
(errcode_for_file_access(), errmsg("could not open relation mapping file \"%s\": %m", map_file_name)));
}
if (write_wal && (i == 0)) {
xl_relmap_update xlrec;
XLogRecPtr lsn;
START_CRIT_SECTION();
xlrec.dbid = dbid;
xlrec.tsid = tsid;
if (IS_NEW_RELMAP(newmap->magic)) {
xlrec.nbytes = sizeof(RelMapFile);
} else {
xlrec.nbytes = RELMAP_SIZE_OLD;
}
XLogBeginInsert();
XLogRegisterData((char*)(&xlrec), MinSizeOfRelmapUpdate);
RegistRelMapWal(newmap);
lsn = XLogInsert(RM_RELMAP_ID, XLOG_RELMAP_UPDATE);
XLogWaitFlush(lsn);
}
errno = 0;
int writeBytes = WriteRelMapFile(newmap, fd);
if (writeBytes != mapTag.relSize) {
close(fd);
if (errno == 0) {
errno = ENOSPC;
}
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not write to relation mapping file \"%s\": %m", map_file_name)));
}
* We choose to fsync the data to disk before considering the task done.
* It would be possible to relax this if it turns out to be a performance
* issue, but it would complicate checkpointing --- see notes for
* CheckPointRelationMap.
*/
if (pg_fsync(fd) != 0) {
close(fd);
ereport(data_sync_elevel(ERROR),
(errcode_for_file_access(), errmsg("could not fsync relation mapping file \"%s\": %m", map_file_name)));
}
if (close(fd)) {
ereport(ERROR,
(errcode_for_file_access(), errmsg("could not close relation mapping file \"%s\": %m", map_file_name)));
}
}
if (EnableLocalSysCache()) {
t_thrd.lsc_cxt.lsc->InvalidateGlobalRelMap(shared, dbid, newmap);
}
* Now that the file is safely on disk, send sinval message to let other
* backends know to re-read it. We must do this inside the critical
* section: if for some reason we fail to send the message, we have to
* force a database-wide PANIC. Otherwise other backends might continue
* execution with stale mapping information, which would be catastrophic
* as soon as others began to use the now-committed data.
*/
if (send_sinval) {
CacheInvalidateRelmap(dbid, redo_lsn);
}
* Make sure that the files listed in the map are not deleted if the outer
* transaction aborts. This had better be within the critical section
* too: it's not likely to fail, but if it did, we'd arrive at transaction
* abort with the files still vulnerable. PANICing will leave things in a
* good state on-disk.
*
* Note: we're cheating a little bit here by assuming that mapped files
* are either in pg_global or the database's default tablespace.
*/
if (preserve_files) {
int32 i;
for (i = 0; i < newmap->num_mappings; i++) {
RelFileNode rnode;
rnode.spcNode = tsid;
rnode.dbNode = dbid;
rnode.relNode = newmap->mappings[i].mapfilenode;
RelationPreserveStorage(rnode, false);
}
}
rc = memcpy_s(real_map, sizeof(RelMapFile), newmap, sizeof(RelMapFile));
securec_check(rc, "", "");
if (write_wal) {
END_CRIT_SECTION();
}
}
* Merge the specified updates into the appropriate "real" map,
* and write out the changes. This function must be used for committing
* updates during normal multiuser operation.
*/
static void perform_relmap_update(bool shared, const RelMapFile* updates)
{
RelMapFile new_map;
errno_t rc;
* Anyone updating a relation's mapping info should take exclusive lock on
* that rel and hold it until commit. This ensures that there will not be
* concurrent updates on the same mapping value; but there could easily be
* concurrent updates on different values in the same file. We cover that
* by acquiring the RelationMappingLock, re-reading the target file to
* ensure it's up to date, applying the updates, and writing the data
* before releasing RelationMappingLock.
*
* There is only one RelationMappingLock. In principle we could try to
* have one per mapping file, but it seems unlikely to be worth the
* trouble.
*/
LWLockAcquire(RelationMappingLock, LW_EXCLUSIVE);
knl_u_relmap_context *relmap_cxt = GetRelMapCxt();
if (EnableLocalSysCache()) {
t_thrd.lsc_cxt.lsc->LoadRelMapFromGlobal(shared);
} else {
load_relmap_file(shared, shared ? relmap_cxt->shared_map : relmap_cxt->local_map);
}
if (shared) {
rc = memcpy_s(&new_map, sizeof(RelMapFile), relmap_cxt->shared_map, sizeof(RelMapFile));
securec_check(rc, "", "");
} else {
rc = memcpy_s(&new_map, sizeof(RelMapFile), relmap_cxt->local_map, sizeof(RelMapFile));
securec_check(rc, "", "");
}
* Apply the updates to newmap. No new mappings should appear, unless
* somebody is adding indexes to system catalogs.
*/
merge_map_updates(&new_map,
updates,
(g_instance.attr.attr_common.allowSystemTableMods || u_sess->attr.attr_common.IsInplaceUpgrade));
Assert(CheckMyDatabaseMatch());
write_relmap_file(shared,
&new_map,
true,
true,
true,
(shared ? InvalidOid : u_sess->proc_cxt.MyDatabaseId),
(shared ? GLOBALTABLESPACE_OID : u_sess->proc_cxt.MyDatabaseTableSpace),
u_sess->proc_cxt.DatabasePath);
LWLockRelease(RelationMappingLock);
}
* when incorrect checksum is detected in relation map file,
* we should recover the file using the content of backup file or,
* if there is no backup file, we create it immediately.
*/
static void recover_relmap_file(bool shared, bool backupfile, RelMapFile* real_map)
{
int fd;
char map_file_name[MAXPGPATH];
char* file_name = NULL;
int level;
errno_t rc;
int32 relmapSize;
if (backupfile) {
file_name = RELMAPPER_FILENAME_BAK;
level = LOG;
} else {
file_name = RELMAPPER_FILENAME;
level = WARNING;
}
if (shared) {
rc = snprintf_s(map_file_name,
sizeof(map_file_name),
sizeof(map_file_name) - 1,
"%s/%s",
GLOTBSDIR,
file_name);
securec_check_ss(rc, "\0", "\0");
} else {
rc = snprintf_s(map_file_name,
sizeof(map_file_name),
sizeof(map_file_name) - 1,
"%s/%s",
u_sess->proc_cxt.DatabasePath,
file_name);
securec_check_ss(rc, "\0", "\0");
}
ereport(level, (errmsg("recover the relation mapping file %s", map_file_name)));
fd = BasicOpenFile(map_file_name, O_CREAT | O_TRUNC | O_WRONLY | PG_BINARY, S_IRUSR | S_IWUSR);
if (fd < 0) {
ereport(ERROR, (errcode_for_file_access(), errmsg("could not open relation mapping file \"%s\": %m",
map_file_name)));
}
errno = 0;
int writeBytes = 0;
if (IS_NEW_RELMAP(real_map->magic)) {
relmapSize = RELMAP_SIZE_NEW;
writeBytes = write(fd, real_map, sizeof(RelMapFile));
} else {
relmapSize = RELMAP_SIZE_OLD;
writeBytes = WriteOldVersionRelmap(real_map, fd);
}
if (writeBytes != relmapSize) {
if (errno == 0) {
errno = ENOSPC;
}
ereport(PANIC,
(errcode_for_file_access(),
errmsg("recover failed could not write to relation mapping file \"%s\": %m", map_file_name)));
}
if (pg_fsync(fd) != 0) {
ereport(PANIC,
(errcode_for_file_access(),
errmsg("recover failed could not fsync relation mapping file \"%s\": %m", map_file_name)));
}
if (close(fd)) {
ereport(ERROR,
(errcode_for_file_access(),
errmsg("recover failed could not close relation mapping file \"%s\": %m", map_file_name)));
}
if (EnableLocalSysCache()) {
t_thrd.lsc_cxt.lsc->InvalidateGlobalRelMap(shared,
shared ? InvalidOid : u_sess->proc_cxt.MyDatabaseId, real_map);
}
}
* RELMAP resource manager's routines
*/
void relmap_redo(XLogReaderState* record)
{
uint8 info = XLogRecGetInfo(record) & ~XLR_INFO_MASK;
Assert(!XLogRecHasAnyBlockRefs(record));
if (info == XLOG_RELMAP_UPDATE) {
xl_relmap_update* xlrec = (xl_relmap_update*)XLogRecGetData(record);
RelMapFile new_map;
char* dbpath = NULL;
errno_t rc;
RelMapVerTag mapTag;
int32 magicNum;
rc = memcpy_s(&magicNum, sizeof(int32), xlrec->data, sizeof(int32));
securec_check(rc, "\0", "\0");
InitRelmapVerTag(&mapTag, IS_NEW_RELMAP(magicNum));
if (xlrec->nbytes != mapTag.relSize) {
ereport(PANIC,
(errcode(ERRCODE_INVALID_PARAMETER_VALUE),
errmsg("relmap_redo: wrong size %d in relmap update record", xlrec->nbytes)));
}
if (IS_NEW_RELMAP(magicNum)) {
rc = memcpy_s(&new_map, sizeof(new_map), xlrec->data, sizeof(new_map));
securec_check(rc, "\0", "\0");
} else {
rc = memcpy_s(&new_map, sizeof(new_map), xlrec->data, MAPPING_LEN_OLDMAP_HEAD);
securec_check(rc, "\0", "\0");
rc = memcpy_s(&new_map.crc, MAPPING_LEN_TAIL, xlrec->data + MAPPING_LEN_OLDMAP_HEAD, MAPPING_LEN_TAIL);
securec_check(rc, "\0", "\0");
}
dbpath = GetDatabasePath(xlrec->dbid, xlrec->tsid);
* Write out the new map and send sinval, but of course don't write a
* new WAL entry. There's no surrounding transaction to tell to
* preserve files, either.
*
* There shouldn't be anyone else updating relmaps during WAL replay,
* so we don't bother to take the RelationMappingLock. We would need
* to do so if load_relmap_file needed to interlock against writers.
*/
XLogRecPtr lsn = record->EndRecPtr;
UpdateMinRecoveryPoint(lsn, false);
if (!IS_EXRTO_READ) {
lsn = 0;
}
write_relmap_file((xlrec->dbid == InvalidOid), &new_map, false, true, false, xlrec->dbid, xlrec->tsid, dbpath,
lsn);
pfree_ext(dbpath);
} else {
ereport(PANIC, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), errmsg("relmap_redo: unknown op code %u", info)));
}
}
static int WriteOldVersionRelmap(RelMapFile* map, int fd)
{
errno_t rc;
char* mapCache_ori = NULL;
char* mapCache = NULL;
if (ENABLE_DSS) {
mapCache_ori = (char*)palloc0(RELMAP_SIZE_OLD + ALIGNOF_BUFFER);
mapCache = (char *)BUFFERALIGN(mapCache_ori);
} else {
mapCache = (char*)palloc0(RELMAP_SIZE_OLD);
}
rc = memcpy_s(mapCache, RELMAP_SIZE_OLD, map, MAPPING_LEN_OLDMAP_HEAD);
securec_check(rc, "\0", "\0");
rc = memcpy_s(
mapCache + MAPPING_LEN_OLDMAP_HEAD, RELMAP_SIZE_OLD - MAPPING_LEN_OLDMAP_HEAD, &(map->crc), MAPPING_LEN_TAIL);
securec_check(rc, "\0", "\0");
int writeBytes = write(fd, mapCache, RELMAP_SIZE_OLD);
if (ENABLE_DSS) {
pfree_ext(mapCache_ori);
} else {
pfree_ext(mapCache);
}
return writeBytes;
}
static int ReadOldVersionRelmap(RelMapFile* map, int fd)
{
errno_t rc;
char* mapCache = (char*)palloc0(RELMAP_SIZE_OLD);
int readByte = read(fd, mapCache, RELMAP_SIZE_OLD);
rc = memcpy_s(map, sizeof(RelMapFile), mapCache, MAPPING_LEN_OLDMAP_HEAD);
securec_check(rc, "\0", "\0");
rc = memcpy_s(&(map->crc), MAPPING_LEN_TAIL, mapCache + MAPPING_LEN_OLDMAP_HEAD, MAPPING_LEN_TAIL);
securec_check(rc, "\0", "\0");
pfree_ext(mapCache);
return readByte;
}
static void InitRelmapVerTag(RelMapVerTag* mapTag, bool isNewMap)
{
if (isNewMap) {
mapTag->relMagic = RELMAPPER_FILEMAGIC_4K;
mapTag->relMaxMappings = MAX_MAPPINGS_4K;
mapTag->relSize = RELMAP_SIZE_NEW;
} else {
mapTag->relMagic = RELMAPPER_FILEMAGIC;
mapTag->relMaxMappings = MAX_MAPPINGS;
mapTag->relSize = RELMAP_SIZE_OLD;
}
}
static bool IsNewVersionMap(int fd, char* fileName)
{
int32 magicNum;
bool ret = false;
int readByte = read(fd, &magicNum, sizeof(int32));
if (readByte != sizeof(int32)) {
close(fd);
ereport(
FATAL, (errcode_for_file_access(), errmsg("could not read relation mapping file \"%s\": %m", fileName)));
}
switch (magicNum) {
case RELMAPPER_FILEMAGIC:
ret = false;
break;
case RELMAPPER_FILEMAGIC_4K:
ret = true;
break;
default:
close(fd);
ereport(FATAL,
(errmsg("Illegal magic number %d for relation mapping file \"%s\": %m", magicNum, fileName)));
}
return ret;
}
static void RegistRelMapWal(RelMapFile* map)
{
if (IS_NEW_RELMAP(map->magic)) {
XLogRegisterData((char*)map, sizeof(RelMapFile));
} else {
XLogRegisterData((char*)map, MAPPING_LEN_OLDMAP_HEAD);
XLogRegisterData((char*)&map->crc, MAPPING_LEN_TAIL);
}
}
static pg_crc32 RelmapCrcComp(RelMapFile* map)
{
pg_crc32 crc;
INIT_CRC32(crc);
if (IS_NEW_RELMAP(map->magic)) {
COMP_CRC32(crc, (char*)map, offsetof(RelMapFile, crc));
} else {
COMP_CRC32(crc, (char*)map, MAPPING_LEN_OLDMAP_HEAD);
}
FIN_CRC32(crc);
return crc;
}
static int32 ReadRelMapFile(RelMapFile* map, int fd, bool isNewMap)
{
int32 readBytes = 0;
if (isNewMap) {
readBytes = read(fd, map, sizeof(RelMapFile));
} else {
readBytes = ReadOldVersionRelmap(map, fd);
}
return readBytes;
}
static int32 WriteRelMapFile(RelMapFile* map, int fd)
{
int32 writeBytes = 0;
char* unalignRelMap = NULL;
RelMapFile* relMap = NULL;
if (IS_NEW_RELMAP(map->magic)) {
if (ENABLE_DSS) {
unalignRelMap = (char*)palloc0(sizeof(RelMapFile) + ALIGNOF_BUFFER);
relMap = (RelMapFile*)BUFFERALIGN(unalignRelMap);
errno_t err = memcpy_s(relMap, sizeof(RelMapFile), map, sizeof(RelMapFile));
securec_check(err, "\0", "\0");
} else {
relMap = map;
}
writeBytes = write(fd, relMap, sizeof(RelMapFile));
} else {
writeBytes = WriteOldVersionRelmap(map, fd);
}
if (unalignRelMap != NULL) {
pfree(unalignRelMap);
}
return writeBytes;
}