diff options
Diffstat (limited to 'src/backend')
236 files changed, 9709 insertions, 5169 deletions
diff --git a/src/backend/access/brin/brin_minmax_multi.c b/src/backend/access/brin/brin_minmax_multi.c index 0d1507a2a36..b85a70a0db2 100644 --- a/src/backend/access/brin/brin_minmax_multi.c +++ b/src/backend/access/brin/brin_minmax_multi.c @@ -624,7 +624,7 @@ brin_range_serialize(Ranges *range) for (i = 0; i < nvalues; i++) { - len += VARSIZE_ANY(range->values[i]); + len += VARSIZE_ANY(DatumGetPointer(range->values[i])); } } else if (typlen == -2) /* cstring */ diff --git a/src/backend/access/common/heaptuple.c b/src/backend/access/common/heaptuple.c index 969d1028cae..a410b5eb99b 100644 --- a/src/backend/access/common/heaptuple.c +++ b/src/backend/access/common/heaptuple.c @@ -189,7 +189,7 @@ getmissingattr(TupleDesc tupleDesc, if (att->attlen > 0) key.len = att->attlen; else - key.len = VARSIZE_ANY(attrmiss->am_value); + key.len = VARSIZE_ANY(DatumGetPointer(attrmiss->am_value)); key.value = attrmiss->am_value; entry = hash_search(missing_cache, &key, HASH_ENTER, &found); @@ -901,9 +901,9 @@ expand_tuple(HeapTuple *targetHeapTuple, att->attlen, attrmiss[attnum].am_value); - targetDataLen = att_addlength_pointer(targetDataLen, - att->attlen, - attrmiss[attnum].am_value); + targetDataLen = att_addlength_datum(targetDataLen, + att->attlen, + attrmiss[attnum].am_value); } else { diff --git a/src/backend/access/common/printsimple.c b/src/backend/access/common/printsimple.c index f346ab3e812..a09c8fcd332 100644 --- a/src/backend/access/common/printsimple.c +++ b/src/backend/access/common/printsimple.c @@ -123,7 +123,7 @@ printsimple(TupleTableSlot *slot, DestReceiver *self) case OIDOID: { - Oid num = ObjectIdGetDatum(value); + Oid num = DatumGetObjectId(value); char str[10]; /* 10 digits */ int len; diff --git a/src/backend/access/common/reloptions.c b/src/backend/access/common/reloptions.c index 46c1dce222d..594a657ea1a 100644 --- a/src/backend/access/common/reloptions.c +++ b/src/backend/access/common/reloptions.c @@ -1190,8 +1190,8 @@ transformRelOptions(Datum oldOptions, List *defList, const char *namspace, for (i = 0; i < noldoptions; i++) { - char *text_str = VARDATA(oldoptions[i]); - int text_len = VARSIZE(oldoptions[i]) - VARHDRSZ; + char *text_str = VARDATA(DatumGetPointer(oldoptions[i])); + int text_len = VARSIZE(DatumGetPointer(oldoptions[i])) - VARHDRSZ; /* Search for a match in defList */ foreach(cell, defList) @@ -1243,8 +1243,9 @@ transformRelOptions(Datum oldOptions, List *defList, const char *namspace, } else { - text *t; + const char *name; const char *value; + text *t; Size len; /* @@ -1291,11 +1292,19 @@ transformRelOptions(Datum oldOptions, List *defList, const char *namspace, * have just "name", assume "name=true" is meant. Note: the * namespace is not output. */ + name = def->defname; if (def->arg != NULL) value = defGetString(def); else value = "true"; + /* Insist that name not contain "=", else "a=b=c" is ambiguous */ + if (strchr(name, '=') != NULL) + ereport(ERROR, + (errcode(ERRCODE_INVALID_PARAMETER_VALUE), + errmsg("invalid option name \"%s\": must not contain \"=\"", + name))); + /* * This is not a great place for this test, but there's no other * convenient place to filter the option out. As WITH (oids = @@ -1303,7 +1312,7 @@ transformRelOptions(Datum oldOptions, List *defList, const char *namspace, * amount of ugly. */ if (acceptOidsOff && def->defnamespace == NULL && - strcmp(def->defname, "oids") == 0) + strcmp(name, "oids") == 0) { if (defGetBoolean(def)) ereport(ERROR, @@ -1313,11 +1322,11 @@ transformRelOptions(Datum oldOptions, List *defList, const char *namspace, continue; } - len = VARHDRSZ + strlen(def->defname) + 1 + strlen(value); + len = VARHDRSZ + strlen(name) + 1 + strlen(value); /* +1 leaves room for sprintf's trailing null */ t = (text *) palloc(len + 1); SET_VARSIZE(t, len); - sprintf(VARDATA(t), "%s=%s", def->defname, value); + sprintf(VARDATA(t), "%s=%s", name, value); astate = accumArrayResult(astate, PointerGetDatum(t), false, TEXTOID, @@ -1447,8 +1456,8 @@ parseRelOptionsInternal(Datum options, bool validate, for (i = 0; i < noptions; i++) { - char *text_str = VARDATA(optiondatums[i]); - int text_len = VARSIZE(optiondatums[i]) - VARHDRSZ; + char *text_str = VARDATA(DatumGetPointer(optiondatums[i])); + int text_len = VARSIZE(DatumGetPointer(optiondatums[i])) - VARHDRSZ; int j; /* Search for a match in reloptions */ diff --git a/src/backend/access/common/toast_compression.c b/src/backend/access/common/toast_compression.c index 21f2f4af97e..926f1e4008a 100644 --- a/src/backend/access/common/toast_compression.c +++ b/src/backend/access/common/toast_compression.c @@ -25,11 +25,11 @@ /* GUC */ int default_toast_compression = TOAST_PGLZ_COMPRESSION; -#define NO_LZ4_SUPPORT() \ +#define NO_COMPRESSION_SUPPORT(method) \ ereport(ERROR, \ (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), \ - errmsg("compression method lz4 not supported"), \ - errdetail("This functionality requires the server to be built with lz4 support."))) + errmsg("compression method %s not supported", method), \ + errdetail("This functionality requires the server to be built with %s support.", method))) /* * Compress a varlena using PGLZ. @@ -139,7 +139,7 @@ struct varlena * lz4_compress_datum(const struct varlena *value) { #ifndef USE_LZ4 - NO_LZ4_SUPPORT(); + NO_COMPRESSION_SUPPORT("lz4"); return NULL; /* keep compiler quiet */ #else int32 valsize; @@ -182,7 +182,7 @@ struct varlena * lz4_decompress_datum(const struct varlena *value) { #ifndef USE_LZ4 - NO_LZ4_SUPPORT(); + NO_COMPRESSION_SUPPORT("lz4"); return NULL; /* keep compiler quiet */ #else int32 rawsize; @@ -215,7 +215,7 @@ struct varlena * lz4_decompress_datum_slice(const struct varlena *value, int32 slicelength) { #ifndef USE_LZ4 - NO_LZ4_SUPPORT(); + NO_COMPRESSION_SUPPORT("lz4"); return NULL; /* keep compiler quiet */ #else int32 rawsize; @@ -289,7 +289,7 @@ CompressionNameToMethod(const char *compression) else if (strcmp(compression, "lz4") == 0) { #ifndef USE_LZ4 - NO_LZ4_SUPPORT(); + NO_COMPRESSION_SUPPORT("lz4"); #endif return TOAST_LZ4_COMPRESSION; } diff --git a/src/backend/access/common/toast_internals.c b/src/backend/access/common/toast_internals.c index 7d8be8346ce..196e06115e9 100644 --- a/src/backend/access/common/toast_internals.c +++ b/src/backend/access/common/toast_internals.c @@ -144,7 +144,7 @@ toast_save_datum(Relation rel, Datum value, int num_indexes; int validIndex; - Assert(!VARATT_IS_EXTERNAL(value)); + Assert(!VARATT_IS_EXTERNAL(dval)); /* * Open the toast relation and its indexes. We can use the index to check diff --git a/src/backend/access/common/tupdesc.c b/src/backend/access/common/tupdesc.c index ffd0c78f905..be60005ae46 100644 --- a/src/backend/access/common/tupdesc.c +++ b/src/backend/access/common/tupdesc.c @@ -142,11 +142,18 @@ void verify_compact_attribute(TupleDesc tupdesc, int attnum) { #ifdef USE_ASSERT_CHECKING - CompactAttribute *cattr = &tupdesc->compact_attrs[attnum]; + CompactAttribute cattr; Form_pg_attribute attr = TupleDescAttr(tupdesc, attnum); CompactAttribute tmp; /* + * Make a temp copy of the TupleDesc's CompactAttribute. This may be a + * shared TupleDesc and the attcacheoff might get changed by another + * backend. + */ + memcpy(&cattr, &tupdesc->compact_attrs[attnum], sizeof(CompactAttribute)); + + /* * Populate the temporary CompactAttribute from the corresponding * Form_pg_attribute */ @@ -156,11 +163,11 @@ verify_compact_attribute(TupleDesc tupdesc, int attnum) * Make the attcacheoff match since it's been reset to -1 by * populate_compact_attribute_internal. Same with attnullability. */ - tmp.attcacheoff = cattr->attcacheoff; - tmp.attnullability = cattr->attnullability; + tmp.attcacheoff = cattr.attcacheoff; + tmp.attnullability = cattr.attnullability; /* Check the freshly populated CompactAttribute matches the TupleDesc's */ - Assert(memcmp(&tmp, cattr, sizeof(CompactAttribute)) == 0); + Assert(memcmp(&tmp, &cattr, sizeof(CompactAttribute)) == 0); #endif } @@ -808,10 +815,10 @@ hashRowType(TupleDesc desc) uint32 s; int i; - s = hash_combine(0, hash_uint32(desc->natts)); - s = hash_combine(s, hash_uint32(desc->tdtypeid)); + s = hash_combine(0, hash_bytes_uint32(desc->natts)); + s = hash_combine(s, hash_bytes_uint32(desc->tdtypeid)); for (i = 0; i < desc->natts; ++i) - s = hash_combine(s, hash_uint32(TupleDescAttr(desc, i)->atttypid)); + s = hash_combine(s, hash_bytes_uint32(TupleDescAttr(desc, i)->atttypid)); return s; } diff --git a/src/backend/access/gin/gininsert.c b/src/backend/access/gin/gininsert.c index a65acd89104..47b1898a064 100644 --- a/src/backend/access/gin/gininsert.c +++ b/src/backend/access/gin/gininsert.c @@ -2233,7 +2233,7 @@ _gin_build_tuple(OffsetNumber attrnum, unsigned char category, else if (typlen > 0) keylen = typlen; else if (typlen == -1) - keylen = VARSIZE_ANY(key); + keylen = VARSIZE_ANY(DatumGetPointer(key)); else if (typlen == -2) keylen = strlen(DatumGetPointer(key)) + 1; else diff --git a/src/backend/access/gist/gistutil.c b/src/backend/access/gist/gistutil.c index a6b701943d3..c0aa7d0222f 100644 --- a/src/backend/access/gist/gistutil.c +++ b/src/backend/access/gist/gistutil.c @@ -1058,11 +1058,11 @@ gistGetFakeLSN(Relation rel) } /* - * This is a stratnum support function for GiST opclasses that use the - * RT*StrategyNumber constants. + * This is a stratnum translation support function for GiST opclasses that use + * the RT*StrategyNumber constants. */ Datum -gist_stratnum_common(PG_FUNCTION_ARGS) +gist_translate_cmptype_common(PG_FUNCTION_ARGS) { CompareType cmptype = PG_GETARG_INT32(0); @@ -1090,9 +1090,9 @@ gist_stratnum_common(PG_FUNCTION_ARGS) /* * Returns the opclass's private stratnum used for the given compare type. * - * Calls the opclass's GIST_STRATNUM_PROC support function, if any, - * and returns the result. - * Returns InvalidStrategy if the function is not defined. + * Calls the opclass's GIST_TRANSLATE_CMPTYPE_PROC support function, if any, + * and returns the result. Returns InvalidStrategy if the function is not + * defined. */ StrategyNumber gisttranslatecmptype(CompareType cmptype, Oid opfamily) @@ -1101,7 +1101,7 @@ gisttranslatecmptype(CompareType cmptype, Oid opfamily) Datum result; /* Check whether the function is provided. */ - funcid = get_opfamily_proc(opfamily, ANYOID, ANYOID, GIST_STRATNUM_PROC); + funcid = get_opfamily_proc(opfamily, ANYOID, ANYOID, GIST_TRANSLATE_CMPTYPE_PROC); if (!OidIsValid(funcid)) return InvalidStrategy; diff --git a/src/backend/access/gist/gistvalidate.c b/src/backend/access/gist/gistvalidate.c index 2a49e6d20f0..2ed6f74fce9 100644 --- a/src/backend/access/gist/gistvalidate.c +++ b/src/backend/access/gist/gistvalidate.c @@ -138,7 +138,7 @@ gistvalidate(Oid opclassoid) ok = check_amproc_signature(procform->amproc, VOIDOID, true, 1, 1, INTERNALOID); break; - case GIST_STRATNUM_PROC: + case GIST_TRANSLATE_CMPTYPE_PROC: ok = check_amproc_signature(procform->amproc, INT2OID, true, 1, 1, INT4OID) && procform->amproclefttype == ANYOID && @@ -265,7 +265,7 @@ gistvalidate(Oid opclassoid) if (i == GIST_DISTANCE_PROC || i == GIST_FETCH_PROC || i == GIST_COMPRESS_PROC || i == GIST_DECOMPRESS_PROC || i == GIST_OPTIONS_PROC || i == GIST_SORTSUPPORT_PROC || - i == GIST_STRATNUM_PROC) + i == GIST_TRANSLATE_CMPTYPE_PROC) continue; /* optional methods */ ereport(INFO, (errcode(ERRCODE_INVALID_OBJECT_DEFINITION), @@ -336,7 +336,7 @@ gistadjustmembers(Oid opfamilyoid, case GIST_FETCH_PROC: case GIST_OPTIONS_PROC: case GIST_SORTSUPPORT_PROC: - case GIST_STRATNUM_PROC: + case GIST_TRANSLATE_CMPTYPE_PROC: /* Optional, so force it to be a soft family dependency */ op->ref_is_hard = false; op->ref_is_family = true; diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 2be7f817c78..0dcd6ee817e 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -4982,7 +4982,7 @@ l3: case LockWaitError: if (!ConditionalMultiXactIdWait((MultiXactId) xwait, status, infomask, relation, - NULL, log_lock_failure)) + NULL, log_lock_failures)) ereport(ERROR, (errcode(ERRCODE_LOCK_NOT_AVAILABLE), errmsg("could not obtain lock on row in relation \"%s\"", @@ -5020,7 +5020,7 @@ l3: } break; case LockWaitError: - if (!ConditionalXactLockTableWait(xwait, log_lock_failure)) + if (!ConditionalXactLockTableWait(xwait, log_lock_failures)) ereport(ERROR, (errcode(ERRCODE_LOCK_NOT_AVAILABLE), errmsg("could not obtain lock on row in relation \"%s\"", @@ -5285,7 +5285,7 @@ heap_acquire_tuplock(Relation relation, ItemPointer tid, LockTupleMode mode, break; case LockWaitError: - if (!ConditionalLockTupleTuplock(relation, tid, mode, log_lock_failure)) + if (!ConditionalLockTupleTuplock(relation, tid, mode, log_lock_failures)) ereport(ERROR, (errcode(ERRCODE_LOCK_NOT_AVAILABLE), errmsg("could not obtain lock on row in relation \"%s\"", diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c index ac082fefa77..cb4bc35c93e 100644 --- a/src/backend/access/heap/heapam_handler.c +++ b/src/backend/access/heap/heapam_handler.c @@ -464,7 +464,7 @@ tuple_lock_retry: return TM_WouldBlock; break; case LockWaitError: - if (!ConditionalXactLockTableWait(SnapshotDirty.xmax, log_lock_failure)) + if (!ConditionalXactLockTableWait(SnapshotDirty.xmax, log_lock_failures)) ereport(ERROR, (errcode(ERRCODE_LOCK_NOT_AVAILABLE), errmsg("could not obtain lock on row in relation \"%s\"", diff --git a/src/backend/access/heap/heapam_xlog.c b/src/backend/access/heap/heapam_xlog.c index 30f4c2d3c67..eb4bd3d6ae3 100644 --- a/src/backend/access/heap/heapam_xlog.c +++ b/src/backend/access/heap/heapam_xlog.c @@ -438,6 +438,9 @@ heap_xlog_insert(XLogReaderState *record) ItemPointerSetBlockNumber(&target_tid, blkno); ItemPointerSetOffsetNumber(&target_tid, xlrec->offnum); + /* No freezing in the heap_insert() code path */ + Assert(!(xlrec->flags & XLH_INSERT_ALL_FROZEN_SET)); + /* * The visibility map may need to be fixed even if the heap page is * already up-to-date. @@ -508,10 +511,6 @@ heap_xlog_insert(XLogReaderState *record) if (xlrec->flags & XLH_INSERT_ALL_VISIBLE_CLEARED) PageClearAllVisible(page); - /* XLH_INSERT_ALL_FROZEN_SET implies that all tuples are visible */ - if (xlrec->flags & XLH_INSERT_ALL_FROZEN_SET) - PageSetAllVisible(page); - MarkBufferDirty(buffer); } if (BufferIsValid(buffer)) diff --git a/src/backend/access/heap/vacuumlazy.c b/src/backend/access/heap/vacuumlazy.c index 708674d8fcf..14036c27e87 100644 --- a/src/backend/access/heap/vacuumlazy.c +++ b/src/backend/access/heap/vacuumlazy.c @@ -423,7 +423,7 @@ typedef struct LVSavedErrInfo /* non-export function prototypes */ static void lazy_scan_heap(LVRelState *vacrel); static void heap_vacuum_eager_scan_setup(LVRelState *vacrel, - VacuumParams *params); + const VacuumParams params); static BlockNumber heap_vac_scan_next_block(ReadStream *stream, void *callback_private_data, void *per_buffer_data); @@ -431,7 +431,7 @@ static void find_next_unskippable_block(LVRelState *vacrel, bool *skipsallvis); static bool lazy_scan_new_or_empty(LVRelState *vacrel, Buffer buf, BlockNumber blkno, Page page, bool sharelock, Buffer vmbuffer); -static void lazy_scan_prune(LVRelState *vacrel, Buffer buf, +static int lazy_scan_prune(LVRelState *vacrel, Buffer buf, BlockNumber blkno, Page page, Buffer vmbuffer, bool all_visible_according_to_vm, bool *has_lpdead_items, bool *vm_page_frozen); @@ -485,7 +485,7 @@ static void restore_vacuum_error_info(LVRelState *vacrel, * vacuum options or for relfrozenxid/relminmxid advancement. */ static void -heap_vacuum_eager_scan_setup(LVRelState *vacrel, VacuumParams *params) +heap_vacuum_eager_scan_setup(LVRelState *vacrel, const VacuumParams params) { uint32 randseed; BlockNumber allvisible; @@ -504,7 +504,7 @@ heap_vacuum_eager_scan_setup(LVRelState *vacrel, VacuumParams *params) vacrel->eager_scan_remaining_successes = 0; /* If eager scanning is explicitly disabled, just return. */ - if (params->max_eager_freeze_failure_rate == 0) + if (params.max_eager_freeze_failure_rate == 0) return; /* @@ -581,11 +581,11 @@ heap_vacuum_eager_scan_setup(LVRelState *vacrel, VacuumParams *params) vacrel->next_eager_scan_region_start = randseed % EAGER_SCAN_REGION_SIZE; - Assert(params->max_eager_freeze_failure_rate > 0 && - params->max_eager_freeze_failure_rate <= 1); + Assert(params.max_eager_freeze_failure_rate > 0 && + params.max_eager_freeze_failure_rate <= 1); vacrel->eager_scan_max_fails_per_region = - params->max_eager_freeze_failure_rate * + params.max_eager_freeze_failure_rate * EAGER_SCAN_REGION_SIZE; /* @@ -612,7 +612,7 @@ heap_vacuum_eager_scan_setup(LVRelState *vacrel, VacuumParams *params) * and locked the relation. */ void -heap_vacuum_rel(Relation rel, VacuumParams *params, +heap_vacuum_rel(Relation rel, const VacuumParams params, BufferAccessStrategy bstrategy) { LVRelState *vacrel; @@ -634,9 +634,9 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, ErrorContextCallback errcallback; char **indnames = NULL; - verbose = (params->options & VACOPT_VERBOSE) != 0; + verbose = (params.options & VACOPT_VERBOSE) != 0; instrument = (verbose || (AmAutoVacuumWorkerProcess() && - params->log_min_duration >= 0)); + params.log_min_duration >= 0)); if (instrument) { pg_rusage_init(&ru0); @@ -699,9 +699,9 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, * The truncate param allows user to avoid attempting relation truncation, * though it can't force truncation to happen. */ - Assert(params->index_cleanup != VACOPTVALUE_UNSPECIFIED); - Assert(params->truncate != VACOPTVALUE_UNSPECIFIED && - params->truncate != VACOPTVALUE_AUTO); + Assert(params.index_cleanup != VACOPTVALUE_UNSPECIFIED); + Assert(params.truncate != VACOPTVALUE_UNSPECIFIED && + params.truncate != VACOPTVALUE_AUTO); /* * While VacuumFailSafeActive is reset to false before calling this, we @@ -711,14 +711,14 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, vacrel->consider_bypass_optimization = true; vacrel->do_index_vacuuming = true; vacrel->do_index_cleanup = true; - vacrel->do_rel_truncate = (params->truncate != VACOPTVALUE_DISABLED); - if (params->index_cleanup == VACOPTVALUE_DISABLED) + vacrel->do_rel_truncate = (params.truncate != VACOPTVALUE_DISABLED); + if (params.index_cleanup == VACOPTVALUE_DISABLED) { /* Force disable index vacuuming up-front */ vacrel->do_index_vacuuming = false; vacrel->do_index_cleanup = false; } - else if (params->index_cleanup == VACOPTVALUE_ENABLED) + else if (params.index_cleanup == VACOPTVALUE_ENABLED) { /* Force index vacuuming. Note that failsafe can still bypass. */ vacrel->consider_bypass_optimization = false; @@ -726,7 +726,7 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, else { /* Default/auto, make all decisions dynamically */ - Assert(params->index_cleanup == VACOPTVALUE_AUTO); + Assert(params.index_cleanup == VACOPTVALUE_AUTO); } /* Initialize page counters explicitly (be tidy) */ @@ -757,7 +757,6 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, vacrel->vm_new_visible_pages = 0; vacrel->vm_new_visible_frozen_pages = 0; vacrel->vm_new_frozen_pages = 0; - vacrel->rel_pages = orig_rel_pages = RelationGetNumberOfBlocks(rel); /* * Get cutoffs that determine which deleted tuples are considered DEAD, @@ -776,7 +775,9 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, * to increase the number of dead tuples it can prune away.) */ vacrel->aggressive = vacuum_get_cutoffs(rel, params, &vacrel->cutoffs); + vacrel->rel_pages = orig_rel_pages = RelationGetNumberOfBlocks(rel); vacrel->vistest = GlobalVisTestFor(rel); + /* Initialize state used to track oldest extant XID/MXID */ vacrel->NewRelfrozenXid = vacrel->cutoffs.OldestXmin; vacrel->NewRelminMxid = vacrel->cutoffs.OldestMxact; @@ -788,7 +789,7 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, */ vacrel->skippedallvis = false; skipwithvm = true; - if (params->options & VACOPT_DISABLE_PAGE_SKIPPING) + if (params.options & VACOPT_DISABLE_PAGE_SKIPPING) { /* * Force aggressive mode, and disable skipping blocks using the @@ -829,7 +830,7 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, * is already dangerously old.) */ lazy_check_wraparound_failsafe(vacrel); - dead_items_alloc(vacrel, params->nworkers); + dead_items_alloc(vacrel, params.nworkers); /* * Call lazy_scan_heap to perform all required heap pruning, index @@ -946,9 +947,9 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, { TimestampTz endtime = GetCurrentTimestamp(); - if (verbose || params->log_min_duration == 0 || + if (verbose || params.log_min_duration == 0 || TimestampDifferenceExceeds(starttime, endtime, - params->log_min_duration)) + params.log_min_duration)) { long secs_dur; int usecs_dur; @@ -983,10 +984,10 @@ heap_vacuum_rel(Relation rel, VacuumParams *params, * Aggressiveness already reported earlier, in dedicated * VACUUM VERBOSE ereport */ - Assert(!params->is_wraparound); + Assert(!params.is_wraparound); msgfmt = _("finished vacuuming \"%s.%s.%s\": index scans: %d\n"); } - else if (params->is_wraparound) + else if (params.is_wraparound) { /* * While it's possible for a VACUUM to be both is_wraparound @@ -1244,6 +1245,7 @@ lazy_scan_heap(LVRelState *vacrel) Buffer buf; Page page; uint8 blk_info = 0; + int ndeleted = 0; bool has_lpdead_items; void *per_buffer_data = NULL; bool vm_page_frozen = false; @@ -1386,10 +1388,10 @@ lazy_scan_heap(LVRelState *vacrel) * line pointers previously marked LP_DEAD. */ if (got_cleanup_lock) - lazy_scan_prune(vacrel, buf, blkno, page, - vmbuffer, - blk_info & VAC_BLK_ALL_VISIBLE_ACCORDING_TO_VM, - &has_lpdead_items, &vm_page_frozen); + ndeleted = lazy_scan_prune(vacrel, buf, blkno, page, + vmbuffer, + blk_info & VAC_BLK_ALL_VISIBLE_ACCORDING_TO_VM, + &has_lpdead_items, &vm_page_frozen); /* * Count an eagerly scanned page as a failure or a success. @@ -1427,7 +1429,7 @@ lazy_scan_heap(LVRelState *vacrel) */ if (vacrel->eager_scan_max_fails_per_region > 0) ereport(vacrel->verbose ? INFO : DEBUG2, - (errmsg("disabling eager scanning after freezing %u eagerly scanned blocks of \"%s.%s.%s\"", + (errmsg("disabling eager scanning after freezing %u eagerly scanned blocks of relation \"%s.%s.%s\"", orig_eager_scan_success_limit, vacrel->dbname, vacrel->relnamespace, vacrel->relname))); @@ -1480,7 +1482,7 @@ lazy_scan_heap(LVRelState *vacrel) * table has indexes. There will only be newly-freed space if we * held the cleanup lock and lazy_scan_prune() was called. */ - if (got_cleanup_lock && vacrel->nindexes == 0 && has_lpdead_items && + if (got_cleanup_lock && vacrel->nindexes == 0 && ndeleted > 0 && blkno - next_fsm_block_to_vacuum >= VACUUM_FSM_EVERY_PAGES) { FreeSpaceMapVacuumRange(vacrel->rel, next_fsm_block_to_vacuum, @@ -1871,8 +1873,6 @@ lazy_scan_new_or_empty(LVRelState *vacrel, Buffer buf, BlockNumber blkno, */ if (!PageIsAllVisible(page)) { - uint8 old_vmbits; - START_CRIT_SECTION(); /* mark buffer dirty before writing a WAL record */ @@ -1892,24 +1892,16 @@ lazy_scan_new_or_empty(LVRelState *vacrel, Buffer buf, BlockNumber blkno, log_newpage_buffer(buf, true); PageSetAllVisible(page); - old_vmbits = visibilitymap_set(vacrel->rel, blkno, buf, - InvalidXLogRecPtr, - vmbuffer, InvalidTransactionId, - VISIBILITYMAP_ALL_VISIBLE | - VISIBILITYMAP_ALL_FROZEN); + visibilitymap_set(vacrel->rel, blkno, buf, + InvalidXLogRecPtr, + vmbuffer, InvalidTransactionId, + VISIBILITYMAP_ALL_VISIBLE | + VISIBILITYMAP_ALL_FROZEN); END_CRIT_SECTION(); - /* - * If the page wasn't already set all-visible and/or all-frozen in - * the VM, count it as newly set for logging. - */ - if ((old_vmbits & VISIBILITYMAP_ALL_VISIBLE) == 0) - { - vacrel->vm_new_visible_pages++; - vacrel->vm_new_visible_frozen_pages++; - } - else if ((old_vmbits & VISIBILITYMAP_ALL_FROZEN) == 0) - vacrel->vm_new_frozen_pages++; + /* Count the newly all-frozen pages for logging */ + vacrel->vm_new_visible_pages++; + vacrel->vm_new_visible_frozen_pages++; } freespace = PageGetHeapFreeSpace(page); @@ -1945,8 +1937,10 @@ cmpOffsetNumbers(const void *a, const void *b) * *vm_page_frozen is set to true if the page is newly set all-frozen in the * VM. The caller currently only uses this for determining whether an eagerly * scanned page was successfully set all-frozen. + * + * Returns the number of tuples deleted from the page during HOT pruning. */ -static void +static int lazy_scan_prune(LVRelState *vacrel, Buffer buf, BlockNumber blkno, @@ -2217,6 +2211,8 @@ lazy_scan_prune(LVRelState *vacrel, *vm_page_frozen = true; } } + + return presult.ndeleted; } /* @@ -2914,7 +2910,6 @@ lazy_vacuum_heap_page(LVRelState *vacrel, BlockNumber blkno, Buffer buffer, if (heap_page_is_all_visible(vacrel, buffer, &visibility_cutoff_xid, &all_frozen)) { - uint8 old_vmbits; uint8 flags = VISIBILITYMAP_ALL_VISIBLE; if (all_frozen) @@ -2924,25 +2919,15 @@ lazy_vacuum_heap_page(LVRelState *vacrel, BlockNumber blkno, Buffer buffer, } PageSetAllVisible(page); - old_vmbits = visibilitymap_set(vacrel->rel, blkno, buffer, - InvalidXLogRecPtr, - vmbuffer, visibility_cutoff_xid, - flags); - - /* - * If the page wasn't already set all-visible and/or all-frozen in the - * VM, count it as newly set for logging. - */ - if ((old_vmbits & VISIBILITYMAP_ALL_VISIBLE) == 0) - { - vacrel->vm_new_visible_pages++; - if (all_frozen) - vacrel->vm_new_visible_frozen_pages++; - } + visibilitymap_set(vacrel->rel, blkno, buffer, + InvalidXLogRecPtr, + vmbuffer, visibility_cutoff_xid, + flags); - else if ((old_vmbits & VISIBILITYMAP_ALL_FROZEN) == 0 && - all_frozen) - vacrel->vm_new_frozen_pages++; + /* Count the newly set VM page for logging */ + vacrel->vm_new_visible_pages++; + if (all_frozen) + vacrel->vm_new_visible_frozen_pages++; } /* Revert to the previous phase information for error traceback */ diff --git a/src/backend/access/heap/visibilitymap.c b/src/backend/access/heap/visibilitymap.c index 745a04ef26e..8f918e00af7 100644 --- a/src/backend/access/heap/visibilitymap.c +++ b/src/backend/access/heap/visibilitymap.c @@ -364,7 +364,7 @@ visibilitymap_get_status(Relation rel, BlockNumber heapBlk, Buffer *vmbuf) { *vmbuf = vm_readbuf(rel, mapBlock, false); if (!BufferIsValid(*vmbuf)) - return false; + return (uint8) 0; } map = PageGetContents(BufferGetPage(*vmbuf)); diff --git a/src/backend/access/index/amapi.c b/src/backend/access/index/amapi.c index f0f4f974bce..60684c53422 100644 --- a/src/backend/access/index/amapi.c +++ b/src/backend/access/index/amapi.c @@ -42,6 +42,19 @@ GetIndexAmRoutine(Oid amhandler) elog(ERROR, "index access method handler function %u did not return an IndexAmRoutine struct", amhandler); + /* Assert that all required callbacks are present. */ + Assert(routine->ambuild != NULL); + Assert(routine->ambuildempty != NULL); + Assert(routine->aminsert != NULL); + Assert(routine->ambulkdelete != NULL); + Assert(routine->amvacuumcleanup != NULL); + Assert(routine->amcostestimate != NULL); + Assert(routine->amoptions != NULL); + Assert(routine->amvalidate != NULL); + Assert(routine->ambeginscan != NULL); + Assert(routine->amrescan != NULL); + Assert(routine->amendscan != NULL); + return routine; } diff --git a/src/backend/access/nbtree/nbtcompare.c b/src/backend/access/nbtree/nbtcompare.c index 4da5a3c1d16..e1b52acd20d 100644 --- a/src/backend/access/nbtree/nbtcompare.c +++ b/src/backend/access/nbtree/nbtcompare.c @@ -555,7 +555,7 @@ btcharcmp(PG_FUNCTION_ARGS) static Datum char_decrement(Relation rel, Datum existing, bool *underflow) { - uint8 cexisting = UInt8GetDatum(existing); + uint8 cexisting = DatumGetUInt8(existing); if (cexisting == 0) { @@ -571,7 +571,7 @@ char_decrement(Relation rel, Datum existing, bool *underflow) static Datum char_increment(Relation rel, Datum existing, bool *overflow) { - uint8 cexisting = UInt8GetDatum(existing); + uint8 cexisting = DatumGetUInt8(existing); if (cexisting == UCHAR_MAX) { diff --git a/src/backend/access/nbtree/nbtpreprocesskeys.c b/src/backend/access/nbtree/nbtpreprocesskeys.c index a136e4bbfdf..21c519cd108 100644 --- a/src/backend/access/nbtree/nbtpreprocesskeys.c +++ b/src/backend/access/nbtree/nbtpreprocesskeys.c @@ -16,6 +16,7 @@ #include "postgres.h" #include "access/nbtree.h" +#include "common/int.h" #include "lib/qunique.h" #include "utils/array.h" #include "utils/lsyscache.h" @@ -56,6 +57,8 @@ static void _bt_skiparray_strat_decrement(IndexScanDesc scan, ScanKey arraysk, BTArrayKeyInfo *array); static void _bt_skiparray_strat_increment(IndexScanDesc scan, ScanKey arraysk, BTArrayKeyInfo *array); +static void _bt_unmark_keys(IndexScanDesc scan, int *keyDataMap); +static int _bt_reorder_array_cmp(const void *a, const void *b); static ScanKey _bt_preprocess_array_keys(IndexScanDesc scan, int *new_numberOfKeys); static void _bt_preprocess_array_keys_final(IndexScanDesc scan, int *keyDataMap); static int _bt_num_array_keys(IndexScanDesc scan, Oid *skip_eq_ops_out, @@ -96,7 +99,7 @@ static int _bt_compare_array_elements(const void *a, const void *b, void *arg); * incomplete sets of cross-type operators, we may fail to detect redundant * or contradictory keys, but we can survive that.) * - * The output keys must be sorted by index attribute. Presently we expect + * Required output keys are sorted by index attribute. Presently we expect * (but verify) that the input keys are already so sorted --- this is done * by match_clauses_to_index() in indxpath.c. Some reordering of the keys * within each attribute may be done as a byproduct of the processing here. @@ -127,29 +130,36 @@ static int _bt_compare_array_elements(const void *a, const void *b, void *arg); * This has the potential to be much more efficient than a full index scan * (though it behaves like a full scan when there's many distinct "x" values). * - * If possible, redundant keys are eliminated: we keep only the tightest + * Typically, redundant keys are eliminated: we keep only the tightest * >/>= bound and the tightest </<= bound, and if there's an = key then * that's the only one returned. (So, we return either a single = key, * or one or two boundary-condition keys for each attr.) However, if we * cannot compare two keys for lack of a suitable cross-type operator, - * we cannot eliminate either. If there are two such keys of the same - * operator strategy, the second one is just pushed into the output array - * without further processing here. We may also emit both >/>= or both - * </<= keys if we can't compare them. The logic about required keys still - * works if we don't eliminate redundant keys. - * - * Note that one reason we need direction-sensitive required-key flags is - * precisely that we may not be able to eliminate redundant keys. Suppose - * we have "x > 4::int AND x > 10::bigint", and we are unable to determine - * which key is more restrictive for lack of a suitable cross-type operator. - * _bt_first will arbitrarily pick one of the keys to do the initial - * positioning with. If it picks x > 4, then the x > 10 condition will fail - * until we reach index entries > 10; but we can't stop the scan just because - * x > 10 is failing. On the other hand, if we are scanning backwards, then - * failure of either key is indeed enough to stop the scan. (In general, when - * inequality keys are present, the initial-positioning code only promises to - * position before the first possible match, not exactly at the first match, - * for a forward scan; or after the last match for a backward scan.) + * we cannot eliminate either key. + * + * When all redundant keys could not be eliminated, we'll output a key array + * that can more or less be treated as if it had no redundant keys. Suppose + * we have "x > 4::int AND x > 10::bigint AND x < 70", and we are unable to + * determine which > key is more restrictive for lack of a suitable cross-type + * operator. We'll arbitrarily pick one of the > keys; the other > key won't + * be marked required. Obviously, the scan will be less efficient if we + * choose x > 4 over x > 10 -- but it can still largely proceed as if there + * was only a single > condition. "x > 10" will be placed at the end of the + * so->keyData[] output array. It'll always be evaluated last, after the keys + * that could be marked required in the usual way (after "x > 4 AND x < 70"). + * This can sometimes result in so->keyData[] keys that aren't even in index + * attribute order (if the qual involves multiple attributes). The scan's + * required keys will still be in attribute order, though, so it can't matter. + * + * This scheme ensures that _bt_first always uses the same set of keys at the + * start of a forwards scan as those _bt_checkkeys uses to determine when to + * end a similar backwards scan (and vice-versa). _bt_advance_array_keys + * depends on this: it expects to be able to reliably predict what the next + * _bt_first call will do by testing whether _bt_checkkeys' routines report + * that the final tuple on the page is past the end of matches for the scan's + * keys with the scan direction flipped. If it is (if continuescan=false), + * then it follows that calling _bt_first will, at a minimum, relocate the + * scan to the very next leaf page (in the current scan direction). * * As a byproduct of this work, we can detect contradictory quals such * as "x = 1 AND x > 2". If we see that, we return so->qual_ok = false, @@ -188,7 +198,8 @@ _bt_preprocess_keys(IndexScanDesc scan) int numberOfEqualCols; ScanKey inkeys; BTScanKeyPreproc xform[BTMaxStrategyNumber]; - bool test_result; + bool test_result, + redundant_key_kept = false; AttrNumber attno; ScanKey arrayKeyData; int *keyDataMap = NULL; @@ -388,7 +399,8 @@ _bt_preprocess_keys(IndexScanDesc scan) xform[j].inkey = NULL; xform[j].inkeyi = -1; } - /* else, cannot determine redundancy, keep both keys */ + else + redundant_key_kept = true; } /* track number of attrs for which we have "=" keys */ numberOfEqualCols++; @@ -409,6 +421,8 @@ _bt_preprocess_keys(IndexScanDesc scan) else xform[BTLessStrategyNumber - 1].inkey = NULL; } + else + redundant_key_kept = true; } /* try to keep only one of >, >= */ @@ -426,6 +440,8 @@ _bt_preprocess_keys(IndexScanDesc scan) else xform[BTGreaterStrategyNumber - 1].inkey = NULL; } + else + redundant_key_kept = true; } /* @@ -466,25 +482,6 @@ _bt_preprocess_keys(IndexScanDesc scan) /* check strategy this key's operator corresponds to */ j = inkey->sk_strategy - 1; - /* if row comparison, push it directly to the output array */ - if (inkey->sk_flags & SK_ROW_HEADER) - { - ScanKey outkey = &so->keyData[new_numberOfKeys++]; - - memcpy(outkey, inkey, sizeof(ScanKeyData)); - if (arrayKeyData) - keyDataMap[new_numberOfKeys - 1] = i; - if (numberOfEqualCols == attno - 1) - _bt_mark_scankey_required(outkey); - - /* - * We don't support RowCompare using equality; such a qual would - * mess up the numberOfEqualCols tracking. - */ - Assert(j != (BTEqualStrategyNumber - 1)); - continue; - } - if (inkey->sk_strategy == BTEqualStrategyNumber && (inkey->sk_flags & SK_SEARCHARRAY)) { @@ -593,9 +590,8 @@ _bt_preprocess_keys(IndexScanDesc scan) * the new scan key. * * Note: We do things this way around so that our arrays are - * always in the same order as their corresponding scan keys, - * even with incomplete opfamilies. _bt_advance_array_keys - * depends on this. + * always in the same order as their corresponding scan keys. + * _bt_preprocess_array_keys_final expects this. */ ScanKey outkey = &so->keyData[new_numberOfKeys++]; @@ -607,6 +603,7 @@ _bt_preprocess_keys(IndexScanDesc scan) xform[j].inkey = inkey; xform[j].inkeyi = i; xform[j].arrayidx = arrayidx; + redundant_key_kept = true; } } } @@ -622,6 +619,15 @@ _bt_preprocess_keys(IndexScanDesc scan) if (arrayKeyData) _bt_preprocess_array_keys_final(scan, keyDataMap); + /* + * If there are remaining redundant inequality keys, we must make sure + * that each index attribute has no more than one required >/>= key, and + * no more than one required </<= key. Attributes that have one or more + * required = keys now must keep only one required key (the first = key). + */ + if (unlikely(redundant_key_kept) && so->qual_ok) + _bt_unmark_keys(scan, keyDataMap); + /* Could pfree arrayKeyData/keyDataMap now, but not worth the cycles */ } @@ -746,9 +752,12 @@ _bt_fix_scankey_strategy(ScanKey skey, int16 *indoption) * * Depending on the operator type, the key may be required for both scan * directions or just one. Also, if the key is a row comparison header, - * we have to mark its first subsidiary ScanKey as required. (Subsequent - * subsidiary ScanKeys are normally for lower-order columns, and thus - * cannot be required, since they're after the first non-equality scankey.) + * we have to mark the appropriate subsidiary ScanKeys as required. In such + * cases, the first subsidiary key is required, but subsequent ones are + * required only as long as they correspond to successive index columns and + * match the leading column as to sort direction. Otherwise the row + * comparison ordering is different from the index ordering and so we can't + * stop the scan on the basis of those lower-order columns. * * Note: when we set required-key flag bits in a subsidiary scankey, we are * scribbling on a data structure belonging to the index AM's caller, not on @@ -786,12 +795,25 @@ _bt_mark_scankey_required(ScanKey skey) if (skey->sk_flags & SK_ROW_HEADER) { ScanKey subkey = (ScanKey) DatumGetPointer(skey->sk_argument); + AttrNumber attno = skey->sk_attno; /* First subkey should be same column/operator as the header */ - Assert(subkey->sk_flags & SK_ROW_MEMBER); - Assert(subkey->sk_attno == skey->sk_attno); + Assert(subkey->sk_attno == attno); Assert(subkey->sk_strategy == skey->sk_strategy); - subkey->sk_flags |= addflags; + + for (;;) + { + Assert(subkey->sk_flags & SK_ROW_MEMBER); + if (subkey->sk_attno != attno) + break; /* non-adjacent key, so not required */ + if (subkey->sk_strategy != skey->sk_strategy) + break; /* wrong direction, so not required */ + subkey->sk_flags |= addflags; + if (subkey->sk_flags & SK_ROW_END) + break; + subkey++; + attno++; + } } } @@ -847,8 +869,7 @@ _bt_compare_scankey_args(IndexScanDesc scan, ScanKey op, cmp_op; StrategyNumber strat; - Assert(!((leftarg->sk_flags | rightarg->sk_flags) & - (SK_ROW_HEADER | SK_ROW_MEMBER))); + Assert(!((leftarg->sk_flags | rightarg->sk_flags) & SK_ROW_MEMBER)); /* * First, deal with cases where one or both args are NULL. This should @@ -925,6 +946,16 @@ _bt_compare_scankey_args(IndexScanDesc scan, ScanKey op, } /* + * We don't yet know how to determine redundancy when it involves a row + * compare key (barring simple cases involving IS NULL/IS NOT NULL) + */ + if ((leftarg->sk_flags | rightarg->sk_flags) & SK_ROW_HEADER) + { + Assert(!((leftarg->sk_flags | rightarg->sk_flags) & SK_BT_SKIP)); + return false; + } + + /* * If either leftarg or rightarg are equality-type array scankeys, we need * specialized handling (since by now we know that IS NULL wasn't used) */ @@ -1468,6 +1499,283 @@ _bt_skiparray_strat_increment(IndexScanDesc scan, ScanKey arraysk, } /* + * _bt_unmark_keys() -- make superfluous required keys nonrequired after all + * + * When _bt_preprocess_keys fails to eliminate one or more redundant keys, it + * calls here to make sure that no index attribute has more than one > or >= + * key marked required, and no more than one required < or <= key. Attributes + * with = keys will always get one = key as their required key. All other + * keys that were initially marked required get "unmarked" here. That way, + * _bt_first and _bt_checkkeys will reliably agree on which keys to use to + * start and/or to end the scan. + * + * We also relocate keys that become/started out nonrequired to the end of + * so->keyData[]. That way, _bt_first and _bt_checkkeys cannot fail to reach + * a required key due to some earlier nonrequired key getting in the way. + * + * Only call here when _bt_compare_scankey_args returned false at least once + * (otherwise, calling here will just waste cycles). + */ +static void +_bt_unmark_keys(IndexScanDesc scan, int *keyDataMap) +{ + BTScanOpaque so = (BTScanOpaque) scan->opaque; + AttrNumber attno; + bool *unmarkikey; + int nunmark, + nunmarked, + nkept, + firsti; + ScanKey keepKeys, + unmarkKeys; + FmgrInfo *keepOrderProcs = NULL, + *unmarkOrderProcs = NULL; + bool haveReqEquals, + haveReqForward, + haveReqBackward; + + /* + * Do an initial pass over so->keyData[] that determines which keys to + * keep as required. We expect so->keyData[] to still be in attribute + * order when we're called (though we don't expect any particular order + * among each attribute's keys). + * + * When both equality and inequality keys remain on a single attribute, we + * *must* make sure that exactly one of the equalities remains required. + * Any requiredness markings that we might leave on later keys/attributes + * are predicated on there being required = keys on all prior columns. + */ + unmarkikey = palloc0(so->numberOfKeys * sizeof(bool)); + nunmark = 0; + + /* Set things up for first key's attribute */ + attno = so->keyData[0].sk_attno; + firsti = 0; + haveReqEquals = false; + haveReqForward = false; + haveReqBackward = false; + for (int i = 0; i < so->numberOfKeys; i++) + { + ScanKey origkey = &so->keyData[i]; + + if (origkey->sk_attno != attno) + { + /* Reset for next attribute */ + attno = origkey->sk_attno; + firsti = i; + + haveReqEquals = false; + haveReqForward = false; + haveReqBackward = false; + } + + /* Equalities get priority over inequalities */ + if (haveReqEquals) + { + /* + * We already found the first "=" key for this attribute. We've + * already decided that all its other keys will be unmarked. + */ + Assert(!(origkey->sk_flags & SK_SEARCHNULL)); + unmarkikey[i] = true; + nunmark++; + continue; + } + else if ((origkey->sk_flags & SK_BT_REQFWD) && + (origkey->sk_flags & SK_BT_REQBKWD)) + { + /* + * Found the first "=" key for attno. All other attno keys will + * be unmarked. + */ + Assert(origkey->sk_strategy == BTEqualStrategyNumber); + + haveReqEquals = true; + for (int j = firsti; j < i; j++) + { + /* Unmark any prior inequality keys on attno after all */ + if (!unmarkikey[j]) + { + unmarkikey[j] = true; + nunmark++; + } + } + continue; + } + + /* Deal with inequalities next */ + if ((origkey->sk_flags & SK_BT_REQFWD) && !haveReqForward) + { + haveReqForward = true; + continue; + } + else if ((origkey->sk_flags & SK_BT_REQBKWD) && !haveReqBackward) + { + haveReqBackward = true; + continue; + } + + /* + * We have either a redundant inequality key that will be unmarked, or + * we have a key that wasn't marked required in the first place + */ + unmarkikey[i] = true; + nunmark++; + } + + /* Should only be called when _bt_compare_scankey_args reported failure */ + Assert(nunmark > 0); + + /* + * Next, allocate temp arrays: one for required keys that'll remain + * required, the other for all remaining keys + */ + unmarkKeys = palloc(nunmark * sizeof(ScanKeyData)); + keepKeys = palloc((so->numberOfKeys - nunmark) * sizeof(ScanKeyData)); + nunmarked = 0; + nkept = 0; + if (so->numArrayKeys) + { + unmarkOrderProcs = palloc(nunmark * sizeof(FmgrInfo)); + keepOrderProcs = palloc((so->numberOfKeys - nunmark) * sizeof(FmgrInfo)); + } + + /* + * Next, copy the contents of so->keyData[] into the appropriate temp + * array. + * + * Scans with = array keys need us to maintain invariants around the order + * of so->orderProcs[] and so->arrayKeys[] relative to so->keyData[]. See + * _bt_preprocess_array_keys_final for a full explanation. + */ + for (int i = 0; i < so->numberOfKeys; i++) + { + ScanKey origkey = &so->keyData[i]; + ScanKey unmark; + + if (!unmarkikey[i]) + { + /* + * Key gets to keep its original requiredness markings. + * + * Key will stay in its original position, unless we're going to + * unmark an earlier key (in which case this key gets moved back). + */ + memcpy(keepKeys + nkept, origkey, sizeof(ScanKeyData)); + + if (so->numArrayKeys) + { + keyDataMap[i] = nkept; + memcpy(keepOrderProcs + nkept, &so->orderProcs[i], + sizeof(FmgrInfo)); + } + + nkept++; + continue; + } + + /* + * Key will be unmarked as needed, and moved to the end of the array, + * next to other keys that will become (or always were) nonrequired + */ + unmark = unmarkKeys + nunmarked; + memcpy(unmark, origkey, sizeof(ScanKeyData)); + + if (so->numArrayKeys) + { + keyDataMap[i] = (so->numberOfKeys - nunmark) + nunmarked; + memcpy(&unmarkOrderProcs[nunmarked], &so->orderProcs[i], + sizeof(FmgrInfo)); + } + + /* + * Preprocessing only generates skip arrays when it knows that they'll + * be the only required = key on the attr. We'll never unmark them. + */ + Assert(!(unmark->sk_flags & SK_BT_SKIP)); + + /* + * Also shouldn't have to unmark an IS NULL or an IS NOT NULL key. + * They aren't cross-type, so an incomplete opfamily can't matter. + */ + Assert(!(unmark->sk_flags & SK_ISNULL) || + !(unmark->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD))); + + /* Clear requiredness flags on redundant key (and on any subkeys) */ + unmark->sk_flags &= ~(SK_BT_REQFWD | SK_BT_REQBKWD); + if (unmark->sk_flags & SK_ROW_HEADER) + { + ScanKey subkey = (ScanKey) DatumGetPointer(unmark->sk_argument); + + Assert(subkey->sk_strategy == unmark->sk_strategy); + for (;;) + { + Assert(subkey->sk_flags & SK_ROW_MEMBER); + subkey->sk_flags &= ~(SK_BT_REQFWD | SK_BT_REQBKWD); + if (subkey->sk_flags & SK_ROW_END) + break; + subkey++; + } + } + + nunmarked++; + } + + /* Copy both temp arrays back into so->keyData[] to reorder */ + Assert(nkept == so->numberOfKeys - nunmark); + Assert(nunmarked == nunmark); + memcpy(so->keyData, keepKeys, sizeof(ScanKeyData) * nkept); + memcpy(so->keyData + nkept, unmarkKeys, sizeof(ScanKeyData) * nunmarked); + + /* Done with temp arrays */ + pfree(unmarkikey); + pfree(keepKeys); + pfree(unmarkKeys); + + /* + * Now copy so->orderProcs[] temp entries needed by scans with = array + * keys back (just like with the so->keyData[] temp arrays) + */ + if (so->numArrayKeys) + { + memcpy(so->orderProcs, keepOrderProcs, sizeof(FmgrInfo) * nkept); + memcpy(so->orderProcs + nkept, unmarkOrderProcs, + sizeof(FmgrInfo) * nunmarked); + + /* Also fix-up array->scan_key references */ + for (int arridx = 0; arridx < so->numArrayKeys; arridx++) + { + BTArrayKeyInfo *array = &so->arrayKeys[arridx]; + + array->scan_key = keyDataMap[array->scan_key]; + } + + /* + * Sort so->arrayKeys[] based on its new BTArrayKeyInfo.scan_key + * offsets, so that its order matches so->keyData[] order as expected + */ + qsort(so->arrayKeys, so->numArrayKeys, sizeof(BTArrayKeyInfo), + _bt_reorder_array_cmp); + + /* Done with temp arrays */ + pfree(unmarkOrderProcs); + pfree(keepOrderProcs); + } +} + +/* + * qsort comparator for reordering so->arrayKeys[] BTArrayKeyInfo entries + */ +static int +_bt_reorder_array_cmp(const void *a, const void *b) +{ + BTArrayKeyInfo *arraya = (BTArrayKeyInfo *) a; + BTArrayKeyInfo *arrayb = (BTArrayKeyInfo *) b; + + return pg_cmp_s32(arraya->scan_key, arrayb->scan_key); +} + +/* * _bt_preprocess_array_keys() -- Preprocess SK_SEARCHARRAY scan keys * * If there are any SK_SEARCHARRAY scan keys, deconstruct the array(s) and diff --git a/src/backend/access/nbtree/nbtree.c b/src/backend/access/nbtree/nbtree.c index 765659887af..fdff960c130 100644 --- a/src/backend/access/nbtree/nbtree.c +++ b/src/backend/access/nbtree/nbtree.c @@ -228,6 +228,8 @@ btgettuple(IndexScanDesc scan, ScanDirection dir) BTScanOpaque so = (BTScanOpaque) scan->opaque; bool res; + Assert(scan->heapRelation != NULL); + /* btree indexes are never lossy */ scan->xs_recheck = false; @@ -289,6 +291,8 @@ btgetbitmap(IndexScanDesc scan, TIDBitmap *tbm) int64 ntids = 0; ItemPointer heapTid; + Assert(scan->heapRelation == NULL); + /* Each loop iteration performs another primitive index scan */ do { @@ -393,6 +397,34 @@ btrescan(IndexScanDesc scan, ScanKey scankey, int nscankeys, BTScanPosInvalidate(so->currPos); } + /* + * We prefer to eagerly drop leaf page pins before btgettuple returns. + * This avoids making VACUUM wait to acquire a cleanup lock on the page. + * + * We cannot safely drop leaf page pins during index-only scans due to a + * race condition involving VACUUM setting pages all-visible in the VM. + * It's also unsafe for plain index scans that use a non-MVCC snapshot. + * + * When we drop pins eagerly, the mechanism that marks so->killedItems[] + * index tuples LP_DEAD has to deal with concurrent TID recycling races. + * The scheme used to detect unsafe TID recycling won't work when scanning + * unlogged relations (since it involves saving an affected page's LSN). + * Opt out of eager pin dropping during unlogged relation scans for now + * (this is preferable to opting out of kill_prior_tuple LP_DEAD setting). + * + * Also opt out of dropping leaf page pins eagerly during bitmap scans. + * Pins cannot be held for more than an instant during bitmap scans either + * way, so we might as well avoid wasting cycles on acquiring page LSNs. + * + * See nbtree/README section on making concurrent TID recycling safe. + * + * Note: so->dropPin should never change across rescans. + */ + so->dropPin = (!scan->xs_want_itup && + IsMVCCSnapshot(scan->xs_snapshot) && + RelationNeedsWAL(scan->indexRelation) && + scan->heapRelation != NULL); + so->markItemIndex = -1; so->needPrimScan = false; so->scanBehind = false; diff --git a/src/backend/access/nbtree/nbtsearch.c b/src/backend/access/nbtree/nbtsearch.c index fe9a3886913..d69798795b4 100644 --- a/src/backend/access/nbtree/nbtsearch.c +++ b/src/backend/access/nbtree/nbtsearch.c @@ -25,7 +25,7 @@ #include "utils/rel.h" -static void _bt_drop_lock_and_maybe_pin(IndexScanDesc scan, BTScanPos sp); +static inline void _bt_drop_lock_and_maybe_pin(Relation rel, BTScanOpaque so); static Buffer _bt_moveright(Relation rel, Relation heaprel, BTScanInsert key, Buffer buf, bool forupdate, BTStack stack, int access); @@ -57,24 +57,29 @@ static bool _bt_endpoint(IndexScanDesc scan, ScanDirection dir); /* * _bt_drop_lock_and_maybe_pin() * - * Unlock the buffer; and if it is safe to release the pin, do that, too. - * This will prevent vacuum from stalling in a blocked state trying to read a - * page when a cursor is sitting on it. - * - * See nbtree/README section on making concurrent TID recycling safe. + * Unlock so->currPos.buf. If scan is so->dropPin, drop the pin, too. + * Dropping the pin prevents VACUUM from blocking on acquiring a cleanup lock. */ -static void -_bt_drop_lock_and_maybe_pin(IndexScanDesc scan, BTScanPos sp) +static inline void +_bt_drop_lock_and_maybe_pin(Relation rel, BTScanOpaque so) { - _bt_unlockbuf(scan->indexRelation, sp->buf); - - if (IsMVCCSnapshot(scan->xs_snapshot) && - RelationNeedsWAL(scan->indexRelation) && - !scan->xs_want_itup) + if (!so->dropPin) { - ReleaseBuffer(sp->buf); - sp->buf = InvalidBuffer; + /* Just drop the lock (not the pin) */ + _bt_unlockbuf(rel, so->currPos.buf); + return; } + + /* + * Drop both the lock and the pin. + * + * Have to set so->currPos.lsn so that _bt_killitems has a way to detect + * when concurrent heap TID recycling by VACUUM might have taken place. + */ + Assert(RelationNeedsWAL(rel)); + so->currPos.lsn = BufferGetLSNAtomic(so->currPos.buf); + _bt_relbuf(rel, so->currPos.buf); + so->currPos.buf = InvalidBuffer; } /* @@ -866,8 +871,8 @@ _bt_compare(Relation rel, * if backwards scan, the last item) in the tree that satisfies the * qualifications in the scan key. On success exit, data about the * matching tuple(s) on the page has been loaded into so->currPos. We'll - * drop all locks and hold onto a pin on page's buffer, except when - * _bt_drop_lock_and_maybe_pin dropped the pin to avoid blocking VACUUM. + * drop all locks and hold onto a pin on page's buffer, except during + * so->dropPin scans, when we drop both the lock and the pin. * _bt_returnitem sets the next item to return to scan on success exit. * * If there are no matching items in the index, we return false, with no @@ -887,9 +892,9 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) OffsetNumber offnum; BTScanInsertData inskey; ScanKey startKeys[INDEX_MAX_KEYS]; - ScanKeyData notnullkeys[INDEX_MAX_KEYS]; + ScanKeyData notnullkey; int keysz = 0; - StrategyNumber strat_total; + StrategyNumber strat_total = InvalidStrategy; BlockNumber blkno = InvalidBlockNumber, lastcurrblkno; @@ -955,46 +960,51 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) /*---------- * Examine the scan keys to discover where we need to start the scan. + * The selected scan keys (at most one per index column) are remembered by + * storing their addresses into the local startKeys[] array. The final + * startKeys[] entry's strategy is set in strat_total. (Actually, there + * are a couple of cases where we force a less/more restrictive strategy.) * - * We want to identify the keys that can be used as starting boundaries; - * these are =, >, or >= keys for a forward scan or =, <, <= keys for - * a backwards scan. We can use keys for multiple attributes so long as - * the prior attributes had only =, >= (resp. =, <=) keys. Once we accept - * a > or < boundary or find an attribute with no boundary (which can be - * thought of as the same as "> -infinity"), we can't use keys for any - * attributes to its right, because it would break our simplistic notion - * of what initial positioning strategy to use. + * We must use the key that was marked required (in the direction opposite + * our own scan's) during preprocessing. Each index attribute can only + * have one such required key. In general, the keys that we use to find + * an initial position when scanning forwards are the same keys that end + * the scan on the leaf level when scanning backwards (and vice-versa). * * When the scan keys include cross-type operators, _bt_preprocess_keys - * may not be able to eliminate redundant keys; in such cases we will - * arbitrarily pick a usable one for each attribute. This is correct - * but possibly not optimal behavior. (For example, with keys like - * "x >= 4 AND x >= 5" we would elect to scan starting at x=4 when - * x=5 would be more efficient.) Since the situation only arises given - * a poorly-worded query plus an incomplete opfamily, live with it. + * may not be able to eliminate redundant keys; in such cases it will + * arbitrarily pick a usable key for each attribute (and scan direction), + * ensuring that there is no more than one key required in each direction. + * We stop considering further keys once we reach the first nonrequired + * key (which must come after all required keys), so this can't affect us. + * + * The required keys that we use as starting boundaries have to be =, >, + * or >= keys for a forward scan or =, <, <= keys for a backwards scan. + * We can use keys for multiple attributes so long as the prior attributes + * had only =, >= (resp. =, <=) keys. These rules are very similar to the + * rules that preprocessing used to determine which keys to mark required. + * We cannot always use every required key as a positioning key, though. + * Skip arrays necessitate independently applying our own rules here. + * Skip arrays are always generally considered = array keys, but we'll + * nevertheless treat them as inequalities at certain points of the scan. + * When that happens, it _might_ have implications for the number of + * required keys that we can safely use for initial positioning purposes. * - * When both equality and inequality keys appear for a single attribute - * (again, only possible when cross-type operators appear), we *must* - * select one of the equality keys for the starting point, because - * _bt_checkkeys() will stop the scan as soon as an equality qual fails. - * For example, if we have keys like "x >= 4 AND x = 10" and we elect to - * start at x=4, we will fail and stop before reaching x=10. If multiple - * equality quals survive preprocessing, however, it doesn't matter which - * one we use --- by definition, they are either redundant or - * contradictory. + * For example, a forward scan with a skip array on its leading attribute + * (with no low_compare/high_compare) will have at least two required scan + * keys, but we won't use any of them as boundary keys during the scan's + * initial call here. Our positioning key during the first call here can + * be thought of as representing "> -infinity". Similarly, if such a skip + * array's low_compare is "a > 'foo'", then we position using "a > 'foo'" + * during the scan's initial call here; a lower-order key such as "b = 42" + * can't be used until the "a" array advances beyond MINVAL/low_compare. * - * In practice we rarely see any "attribute boundary key gaps" here. - * Preprocessing can usually backfill skip array keys for any attributes - * that were omitted from the original scan->keyData[] input keys. All - * array keys are always considered = keys, but we'll sometimes need to - * treat the current key value as if we were using an inequality strategy. - * This happens with range skip arrays, which store inequality keys in the - * array's low_compare/high_compare fields (used to find the first/last - * set of matches, when = key will lack a usable sk_argument value). - * These are always preferred over any redundant "standard" inequality - * keys on the same column (per the usual rule about preferring = keys). - * Note also that any column with an = skip array key can never have an - * additional, contradictory = key. + * On the other hand, if such a skip array's low_compare was "a >= 'foo'", + * then we _can_ use "a >= 'foo' AND b = 42" during the initial call here. + * A subsequent call here might have us use "a = 'fop' AND b = 42". Note + * that we treat = and >= as equivalent when scanning forwards (just as we + * treat = and <= as equivalent when scanning backwards). We effectively + * do the same thing (though with a distinct "a" element/value) each time. * * All keys (with the exception of SK_SEARCHNULL keys and SK_BT_SKIP * array keys whose array is "null_elem=true") imply a NOT NULL qualifier. @@ -1006,41 +1016,38 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) * traversing a lot of null entries at the start of the scan. * * In this loop, row-comparison keys are treated the same as keys on their - * first (leftmost) columns. We'll add on lower-order columns of the row - * comparison below, if possible. + * first (leftmost) columns. We'll add all lower-order columns of the row + * comparison that were marked required during preprocessing below. * - * The selected scan keys (at most one per index column) are remembered by - * storing their addresses into the local startKeys[] array. - * - * _bt_checkkeys/_bt_advance_array_keys decide whether and when to start - * the next primitive index scan (for scans with array keys) based in part - * on an understanding of how it'll enable us to reposition the scan. - * They're directly aware of how we'll sometimes cons up an explicit - * SK_SEARCHNOTNULL key. They'll even end primitive scans by applying a - * symmetric "deduce NOT NULL" rule of their own. This allows top-level - * scans to skip large groups of NULLs through repeated deductions about - * key strictness (for a required inequality key) and whether NULLs in the - * key's index column are stored last or first (relative to non-NULLs). + * _bt_advance_array_keys needs to know exactly how we'll reposition the + * scan (should it opt to schedule another primitive index scan). It is + * critical that primscans only be scheduled when they'll definitely make + * some useful progress. _bt_advance_array_keys does this by calling + * _bt_checkkeys routines that report whether a tuple is past the end of + * matches for the scan's keys (given the scan's current array elements). + * If the page's final tuple is "after the end of matches" for a scan that + * uses the *opposite* scan direction, then it must follow that it's also + * "before the start of matches" for the actual current scan direction. + * It is therefore essential that all of our initial positioning rules are + * symmetric with _bt_checkkeys's corresponding continuescan=false rule. * If you update anything here, _bt_checkkeys/_bt_advance_array_keys might * need to be kept in sync. *---------- */ - strat_total = BTEqualStrategyNumber; if (so->numberOfKeys > 0) { AttrNumber curattr; - ScanKey chosen; + ScanKey bkey; ScanKey impliesNN; ScanKey cur; /* - * chosen is the so-far-chosen key for the current attribute, if any. - * We don't cast the decision in stone until we reach keys for the - * next attribute. + * bkey will be set to the key that preprocessing left behind as the + * boundary key for this attribute, in this scan direction (if any) */ cur = so->keyData; curattr = 1; - chosen = NULL; + bkey = NULL; /* Also remember any scankey that implies a NOT NULL constraint */ impliesNN = NULL; @@ -1053,23 +1060,29 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) { if (i >= so->numberOfKeys || cur->sk_attno != curattr) { + /* Done looking for the curattr boundary key */ + Assert(bkey == NULL || + (bkey->sk_attno == curattr && + (bkey->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)))); + Assert(impliesNN == NULL || + (impliesNN->sk_attno == curattr && + (impliesNN->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)))); + /* - * Done looking at keys for curattr. - * * If this is a scan key for a skip array whose current * element is MINVAL, choose low_compare (when scanning * backwards it'll be MAXVAL, and we'll choose high_compare). * - * Note: if the array's low_compare key makes 'chosen' NULL, + * Note: if the array's low_compare key makes 'bkey' NULL, * then we behave as if the array's first element is -inf, * except when !array->null_elem implies a usable NOT NULL * constraint. */ - if (chosen != NULL && - (chosen->sk_flags & (SK_BT_MINVAL | SK_BT_MAXVAL))) + if (bkey != NULL && + (bkey->sk_flags & (SK_BT_MINVAL | SK_BT_MAXVAL))) { - int ikey = chosen - so->keyData; - ScanKey skipequalitykey = chosen; + int ikey = bkey - so->keyData; + ScanKey skipequalitykey = bkey; BTArrayKeyInfo *array = NULL; for (int arridx = 0; arridx < so->numArrayKeys; arridx++) @@ -1082,42 +1095,41 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) if (ScanDirectionIsForward(dir)) { Assert(!(skipequalitykey->sk_flags & SK_BT_MAXVAL)); - chosen = array->low_compare; + bkey = array->low_compare; } else { Assert(!(skipequalitykey->sk_flags & SK_BT_MINVAL)); - chosen = array->high_compare; + bkey = array->high_compare; } - Assert(chosen == NULL || - chosen->sk_attno == skipequalitykey->sk_attno); + Assert(bkey == NULL || + bkey->sk_attno == skipequalitykey->sk_attno); if (!array->null_elem) impliesNN = skipequalitykey; else - Assert(chosen == NULL && impliesNN == NULL); + Assert(bkey == NULL && impliesNN == NULL); } /* * If we didn't find a usable boundary key, see if we can * deduce a NOT NULL key */ - if (chosen == NULL && impliesNN != NULL && + if (bkey == NULL && impliesNN != NULL && ((impliesNN->sk_flags & SK_BT_NULLS_FIRST) ? ScanDirectionIsForward(dir) : ScanDirectionIsBackward(dir))) { - /* Yes, so build the key in notnullkeys[keysz] */ - chosen = ¬nullkeys[keysz]; - ScanKeyEntryInitialize(chosen, + /* Final startKeys[] entry will be deduced NOT NULL key */ + bkey = ¬nullkey; + ScanKeyEntryInitialize(bkey, (SK_SEARCHNOTNULL | SK_ISNULL | (impliesNN->sk_flags & (SK_BT_DESC | SK_BT_NULLS_FIRST))), curattr, - ((impliesNN->sk_flags & SK_BT_NULLS_FIRST) ? - BTGreaterStrategyNumber : - BTLessStrategyNumber), + ScanDirectionIsForward(dir) ? + BTGreaterStrategyNumber : BTLessStrategyNumber, InvalidOid, InvalidOid, InvalidOid, @@ -1125,12 +1137,12 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) } /* - * If we still didn't find a usable boundary key, quit; else - * save the boundary key pointer in startKeys. + * If preprocessing didn't leave a usable boundary key, quit; + * else save the boundary key pointer in startKeys[] */ - if (chosen == NULL) + if (bkey == NULL) break; - startKeys[keysz++] = chosen; + startKeys[keysz++] = bkey; /* * We can only consider adding more boundary keys when the one @@ -1138,7 +1150,7 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) * (during backwards scans we can only do so when the key that * we just added to startKeys[] uses the = or <= strategy) */ - strat_total = chosen->sk_strategy; + strat_total = bkey->sk_strategy; if (strat_total == BTGreaterStrategyNumber || strat_total == BTLessStrategyNumber) break; @@ -1149,19 +1161,19 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) * make strat_total > or < (and stop adding boundary keys). * This can only happen with opclasses that lack skip support. */ - if (chosen->sk_flags & (SK_BT_NEXT | SK_BT_PRIOR)) + if (bkey->sk_flags & (SK_BT_NEXT | SK_BT_PRIOR)) { - Assert(chosen->sk_flags & SK_BT_SKIP); + Assert(bkey->sk_flags & SK_BT_SKIP); Assert(strat_total == BTEqualStrategyNumber); if (ScanDirectionIsForward(dir)) { - Assert(!(chosen->sk_flags & SK_BT_PRIOR)); + Assert(!(bkey->sk_flags & SK_BT_PRIOR)); strat_total = BTGreaterStrategyNumber; } else { - Assert(!(chosen->sk_flags & SK_BT_NEXT)); + Assert(!(bkey->sk_flags & SK_BT_NEXT)); strat_total = BTLessStrategyNumber; } @@ -1175,24 +1187,30 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) /* * Done if that was the last scan key output by preprocessing. - * Also done if there is a gap index attribute that lacks a - * usable key (only possible when preprocessing was unable to - * generate a skip array key to "fill in the gap"). + * Also done if we've now examined all keys marked required. */ if (i >= so->numberOfKeys || - cur->sk_attno != curattr + 1) + !(cur->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD))) break; /* * Reset for next attr. */ + Assert(cur->sk_attno == curattr + 1); curattr = cur->sk_attno; - chosen = NULL; + bkey = NULL; impliesNN = NULL; } /* - * Can we use this key as a starting boundary for this attr? + * If we've located the starting boundary key for curattr, we have + * no interest in curattr's other required key + */ + if (bkey != NULL) + continue; + + /* + * Is this key the starting boundary key for curattr? * * If not, does it imply a NOT NULL constraint? (Because * SK_SEARCHNULL keys are always assigned BTEqualStrategyNumber, @@ -1202,27 +1220,20 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) { case BTLessStrategyNumber: case BTLessEqualStrategyNumber: - if (chosen == NULL) - { - if (ScanDirectionIsBackward(dir)) - chosen = cur; - else - impliesNN = cur; - } + if (ScanDirectionIsBackward(dir)) + bkey = cur; + else if (impliesNN == NULL) + impliesNN = cur; break; case BTEqualStrategyNumber: - /* override any non-equality choice */ - chosen = cur; + bkey = cur; break; case BTGreaterEqualStrategyNumber: case BTGreaterStrategyNumber: - if (chosen == NULL) - { - if (ScanDirectionIsForward(dir)) - chosen = cur; - else - impliesNN = cur; - } + if (ScanDirectionIsForward(dir)) + bkey = cur; + else if (impliesNN == NULL) + impliesNN = cur; break; } } @@ -1248,16 +1259,18 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) Assert(keysz <= INDEX_MAX_KEYS); for (int i = 0; i < keysz; i++) { - ScanKey cur = startKeys[i]; + ScanKey bkey = startKeys[i]; - Assert(cur->sk_attno == i + 1); + Assert(bkey->sk_attno == i + 1); - if (cur->sk_flags & SK_ROW_HEADER) + if (bkey->sk_flags & SK_ROW_HEADER) { /* * Row comparison header: look to the first row member instead */ - ScanKey subkey = (ScanKey) DatumGetPointer(cur->sk_argument); + ScanKey subkey = (ScanKey) DatumGetPointer(bkey->sk_argument); + bool loosen_strat = false, + tighten_strat = false; /* * Cannot be a NULL in the first row member: _bt_preprocess_keys @@ -1265,122 +1278,160 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) * ever getting this far */ Assert(subkey->sk_flags & SK_ROW_MEMBER); - Assert(subkey->sk_attno == cur->sk_attno); + Assert(subkey->sk_attno == bkey->sk_attno); Assert(!(subkey->sk_flags & SK_ISNULL)); /* + * This is either a > or >= key (during backwards scans it is + * either < or <=) that was marked required during preprocessing. + * Later so->keyData[] keys can't have been marked required, so + * our row compare header key must be the final startKeys[] entry. + */ + Assert(subkey->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)); + Assert(i == keysz - 1); + + /* * The member scankeys are already in insertion format (ie, they * have sk_func = 3-way-comparison function) */ memcpy(inskey.scankeys + i, subkey, sizeof(ScanKeyData)); /* - * If the row comparison is the last positioning key we accepted, - * try to add additional keys from the lower-order row members. - * (If we accepted independent conditions on additional index - * columns, we use those instead --- doesn't seem worth trying to - * determine which is more restrictive.) Note that this is OK - * even if the row comparison is of ">" or "<" type, because the - * condition applied to all but the last row member is effectively - * ">=" or "<=", and so the extra keys don't break the positioning - * scheme. But, by the same token, if we aren't able to use all - * the row members, then the part of the row comparison that we - * did use has to be treated as just a ">=" or "<=" condition, and - * so we'd better adjust strat_total accordingly. + * Now look to later row compare members. + * + * If there's an "index attribute gap" between two row compare + * members, the second member won't have been marked required, and + * so can't be used as a starting boundary key here. The part of + * the row comparison that we do still use has to be treated as a + * ">=" or "<=" condition. For example, a qual "(a, c) > (1, 42)" + * with an omitted intervening index attribute "b" will use an + * insertion scan key "a >= 1". Even the first "a = 1" tuple on + * the leaf level might satisfy the row compare qual. + * + * We're able to use a _more_ restrictive strategy when we reach a + * NULL row compare member, since they're always unsatisfiable. + * For example, a qual "(a, b, c) >= (1, NULL, 77)" will use an + * insertion scan key "a > 1". All tuples where "a = 1" cannot + * possibly satisfy the row compare qual, so this is safe. */ - if (i == keysz - 1) + Assert(!(subkey->sk_flags & SK_ROW_END)); + for (;;) { - bool used_all_subkeys = false; + subkey++; + Assert(subkey->sk_flags & SK_ROW_MEMBER); - Assert(!(subkey->sk_flags & SK_ROW_END)); - for (;;) + if (subkey->sk_flags & SK_ISNULL) { - subkey++; - Assert(subkey->sk_flags & SK_ROW_MEMBER); - if (subkey->sk_attno != keysz + 1) - break; /* out-of-sequence, can't use it */ - if (subkey->sk_strategy != cur->sk_strategy) - break; /* wrong direction, can't use it */ - if (subkey->sk_flags & SK_ISNULL) - break; /* can't use null keys */ - Assert(keysz < INDEX_MAX_KEYS); - memcpy(inskey.scankeys + keysz, subkey, - sizeof(ScanKeyData)); - keysz++; - if (subkey->sk_flags & SK_ROW_END) - { - used_all_subkeys = true; - break; - } + /* + * NULL member key, can only use earlier keys. + * + * We deliberately avoid checking if this key is marked + * required. All earlier keys are required, and this key + * is unsatisfiable either way, so we can't miss anything. + */ + tighten_strat = true; + break; } - if (!used_all_subkeys) + + if (!(subkey->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD))) { - switch (strat_total) - { - case BTLessStrategyNumber: - strat_total = BTLessEqualStrategyNumber; - break; - case BTGreaterStrategyNumber: - strat_total = BTGreaterEqualStrategyNumber; - break; - } + /* nonrequired member key, can only use earlier keys */ + loosen_strat = true; + break; } - break; /* done with outer loop */ + + Assert(subkey->sk_attno == keysz + 1); + Assert(subkey->sk_strategy == bkey->sk_strategy); + Assert(keysz < INDEX_MAX_KEYS); + + memcpy(inskey.scankeys + keysz, subkey, + sizeof(ScanKeyData)); + keysz++; + if (subkey->sk_flags & SK_ROW_END) + break; } - } - else - { - /* - * Ordinary comparison key. Transform the search-style scan key - * to an insertion scan key by replacing the sk_func with the - * appropriate btree comparison function. - * - * If scankey operator is not a cross-type comparison, we can use - * the cached comparison function; otherwise gotta look it up in - * the catalogs. (That can't lead to infinite recursion, since no - * indexscan initiated by syscache lookup will use cross-data-type - * operators.) - * - * We support the convention that sk_subtype == InvalidOid means - * the opclass input type; this is a hack to simplify life for - * ScanKeyInit(). - */ - if (cur->sk_subtype == rel->rd_opcintype[i] || - cur->sk_subtype == InvalidOid) + Assert(!(loosen_strat && tighten_strat)); + if (loosen_strat) { - FmgrInfo *procinfo; - - procinfo = index_getprocinfo(rel, cur->sk_attno, BTORDER_PROC); - ScanKeyEntryInitializeWithInfo(inskey.scankeys + i, - cur->sk_flags, - cur->sk_attno, - InvalidStrategy, - cur->sk_subtype, - cur->sk_collation, - procinfo, - cur->sk_argument); + /* Use less restrictive strategy (and fewer member keys) */ + switch (strat_total) + { + case BTLessStrategyNumber: + strat_total = BTLessEqualStrategyNumber; + break; + case BTGreaterStrategyNumber: + strat_total = BTGreaterEqualStrategyNumber; + break; + } } - else + if (tighten_strat) { - RegProcedure cmp_proc; - - cmp_proc = get_opfamily_proc(rel->rd_opfamily[i], - rel->rd_opcintype[i], - cur->sk_subtype, - BTORDER_PROC); - if (!RegProcedureIsValid(cmp_proc)) - elog(ERROR, "missing support function %d(%u,%u) for attribute %d of index \"%s\"", - BTORDER_PROC, rel->rd_opcintype[i], cur->sk_subtype, - cur->sk_attno, RelationGetRelationName(rel)); - ScanKeyEntryInitialize(inskey.scankeys + i, - cur->sk_flags, - cur->sk_attno, - InvalidStrategy, - cur->sk_subtype, - cur->sk_collation, - cmp_proc, - cur->sk_argument); + /* Use more restrictive strategy (and fewer member keys) */ + switch (strat_total) + { + case BTLessEqualStrategyNumber: + strat_total = BTLessStrategyNumber; + break; + case BTGreaterEqualStrategyNumber: + strat_total = BTGreaterStrategyNumber; + break; + } } + + /* done adding to inskey (row comparison keys always come last) */ + break; + } + + /* + * Ordinary comparison key/search-style key. + * + * Transform the search-style scan key to an insertion scan key by + * replacing the sk_func with the appropriate btree 3-way-comparison + * function. + * + * If scankey operator is not a cross-type comparison, we can use the + * cached comparison function; otherwise gotta look it up in the + * catalogs. (That can't lead to infinite recursion, since no + * indexscan initiated by syscache lookup will use cross-data-type + * operators.) + * + * We support the convention that sk_subtype == InvalidOid means the + * opclass input type; this hack simplifies life for ScanKeyInit(). + */ + if (bkey->sk_subtype == rel->rd_opcintype[i] || + bkey->sk_subtype == InvalidOid) + { + FmgrInfo *procinfo; + + procinfo = index_getprocinfo(rel, bkey->sk_attno, BTORDER_PROC); + ScanKeyEntryInitializeWithInfo(inskey.scankeys + i, + bkey->sk_flags, + bkey->sk_attno, + InvalidStrategy, + bkey->sk_subtype, + bkey->sk_collation, + procinfo, + bkey->sk_argument); + } + else + { + RegProcedure cmp_proc; + + cmp_proc = get_opfamily_proc(rel->rd_opfamily[i], + rel->rd_opcintype[i], + bkey->sk_subtype, BTORDER_PROC); + if (!RegProcedureIsValid(cmp_proc)) + elog(ERROR, "missing support function %d(%u,%u) for attribute %d of index \"%s\"", + BTORDER_PROC, rel->rd_opcintype[i], bkey->sk_subtype, + bkey->sk_attno, RelationGetRelationName(rel)); + ScanKeyEntryInitialize(inskey.scankeys + i, + bkey->sk_flags, + bkey->sk_attno, + InvalidStrategy, + bkey->sk_subtype, + bkey->sk_collation, + cmp_proc, + bkey->sk_argument); } } @@ -1469,6 +1520,8 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) if (!BufferIsValid(so->currPos.buf)) { + Assert(!so->needPrimScan); + /* * We only get here if the index is completely empty. Lock relation * because nothing finer to lock exists. Without a buffer lock, it's @@ -1487,7 +1540,6 @@ _bt_first(IndexScanDesc scan, ScanDirection dir) if (!BufferIsValid(so->currPos.buf)) { - Assert(!so->needPrimScan); _bt_parallel_done(scan); return false; } @@ -1610,7 +1662,13 @@ _bt_readpage(IndexScanDesc scan, ScanDirection dir, OffsetNumber offnum, so->currPos.currPage = BufferGetBlockNumber(so->currPos.buf); so->currPos.prevPage = opaque->btpo_prev; so->currPos.nextPage = opaque->btpo_next; + /* delay setting so->currPos.lsn until _bt_drop_lock_and_maybe_pin */ + so->currPos.dir = dir; + so->currPos.nextTupleOffset = 0; + /* either moreRight or moreLeft should be set now (may be unset later) */ + Assert(ScanDirectionIsForward(dir) ? so->currPos.moreRight : + so->currPos.moreLeft); Assert(!P_IGNORE(opaque)); Assert(BTScanPosIsPinned(so->currPos)); Assert(!so->needPrimScan); @@ -1626,14 +1684,6 @@ _bt_readpage(IndexScanDesc scan, ScanDirection dir, OffsetNumber offnum, so->currPos.currPage); } - /* initialize remaining currPos fields related to current page */ - so->currPos.lsn = BufferGetLSNAtomic(so->currPos.buf); - so->currPos.dir = dir; - so->currPos.nextTupleOffset = 0; - /* either moreLeft or moreRight should be set now (may be unset later) */ - Assert(ScanDirectionIsForward(dir) ? so->currPos.moreRight : - so->currPos.moreLeft); - PredicateLockPage(rel, so->currPos.currPage, scan->xs_snapshot); /* initialize local variables */ @@ -2107,10 +2157,9 @@ _bt_returnitem(IndexScanDesc scan, BTScanOpaque so) * * Wrapper on _bt_readnextpage that performs final steps for the current page. * - * On entry, if so->currPos.buf is valid the buffer is pinned but not locked. - * If there's no pin held, it's because _bt_drop_lock_and_maybe_pin dropped - * the pin eagerly earlier on. The scan must have so->currPos.currPage set to - * a valid block, in any case. + * On entry, so->currPos must be valid. Its buffer will be pinned, though + * never locked. (Actually, when so->dropPin there won't even be a pin held, + * though so->currPos.currPage must still be set to a valid block number.) */ static bool _bt_steppage(IndexScanDesc scan, ScanDirection dir) @@ -2251,12 +2300,14 @@ _bt_readfirstpage(IndexScanDesc scan, OffsetNumber offnum, ScanDirection dir) */ if (_bt_readpage(scan, dir, offnum, true)) { + Relation rel = scan->indexRelation; + /* * _bt_readpage succeeded. Drop the lock (and maybe the pin) on * so->currPos.buf in preparation for btgettuple returning tuples. */ Assert(BTScanPosIsPinned(so->currPos)); - _bt_drop_lock_and_maybe_pin(scan, &so->currPos); + _bt_drop_lock_and_maybe_pin(rel, so); return true; } @@ -2278,9 +2329,12 @@ _bt_readfirstpage(IndexScanDesc scan, OffsetNumber offnum, ScanDirection dir) * previously-saved right link or left link. lastcurrblkno is the page that * was current at the point where the blkno link was saved, which we use to * reason about concurrent page splits/page deletions during backwards scans. + * In the common case where seized=false, blkno is either so->currPos.nextPage + * or so->currPos.prevPage, and lastcurrblkno is so->currPos.currPage. * - * On entry, caller shouldn't hold any locks or pins on any page (we work - * directly off of blkno and lastcurrblkno instead). Parallel scan callers + * On entry, so->currPos shouldn't be locked by caller. so->currPos.buf must + * be InvalidBuffer/unpinned as needed by caller (note that lastcurrblkno + * won't need to be read again in almost all cases). Parallel scan callers * that seized the scan before calling here should pass seized=true; such a * caller's blkno and lastcurrblkno arguments come from the seized scan. * seized=false callers just pass us the blkno/lastcurrblkno taken from their @@ -2294,11 +2348,11 @@ _bt_readfirstpage(IndexScanDesc scan, OffsetNumber offnum, ScanDirection dir) * * On success exit, so->currPos is updated to contain data from the next * interesting page, and we return true. We hold a pin on the buffer on - * success exit, except when _bt_drop_lock_and_maybe_pin decided it was safe - * to eagerly drop the pin (to avoid blocking VACUUM). + * success exit (except during so->dropPin index scans, when we drop the pin + * eagerly to avoid blocking VACUUM). * - * If there are no more matching records in the given direction, we drop all - * locks and pins, invalidate so->currPos, and return false. + * If there are no more matching records in the given direction, we invalidate + * so->currPos (while ensuring it retains no locks or pins), and return false. * * We always release the scan for a parallel scan caller, regardless of * success or failure; we'll call _bt_parallel_release as soon as possible. @@ -2413,7 +2467,7 @@ _bt_readnextpage(IndexScanDesc scan, BlockNumber blkno, */ Assert(so->currPos.currPage == blkno); Assert(BTScanPosIsPinned(so->currPos)); - _bt_drop_lock_and_maybe_pin(scan, &so->currPos); + _bt_drop_lock_and_maybe_pin(rel, so); return true; } diff --git a/src/backend/access/nbtree/nbtutils.c b/src/backend/access/nbtree/nbtutils.c index 1a15dfcb7d3..9aed207995f 100644 --- a/src/backend/access/nbtree/nbtutils.c +++ b/src/backend/access/nbtree/nbtutils.c @@ -44,7 +44,6 @@ static bool _bt_array_decrement(Relation rel, ScanKey skey, BTArrayKeyInfo *arra static bool _bt_array_increment(Relation rel, ScanKey skey, BTArrayKeyInfo *array); static bool _bt_advance_array_keys_increment(IndexScanDesc scan, ScanDirection dir, bool *skip_array_set); -static void _bt_rewind_nonrequired_arrays(IndexScanDesc scan, ScanDirection dir); static bool _bt_tuple_before_array_skeys(IndexScanDesc scan, ScanDirection dir, IndexTuple tuple, TupleDesc tupdesc, int tupnatts, bool readpagetup, int sktrig, bool *scanBehind); @@ -52,7 +51,6 @@ static bool _bt_advance_array_keys(IndexScanDesc scan, BTReadPageState *pstate, IndexTuple tuple, int tupnatts, TupleDesc tupdesc, int sktrig, bool sktrig_required); #ifdef USE_ASSERT_CHECKING -static bool _bt_verify_arrays_bt_first(IndexScanDesc scan, ScanDirection dir); static bool _bt_verify_keys_with_arraykeys(IndexScanDesc scan); #endif static bool _bt_oppodir_checkkeys(IndexScanDesc scan, ScanDirection dir, @@ -1035,73 +1033,6 @@ _bt_advance_array_keys_increment(IndexScanDesc scan, ScanDirection dir, } /* - * _bt_rewind_nonrequired_arrays() -- Rewind SAOP arrays not marked required - * - * Called when _bt_advance_array_keys decides to start a new primitive index - * scan on the basis of the current scan position being before the position - * that _bt_first is capable of repositioning the scan to by applying an - * inequality operator required in the opposite-to-scan direction only. - * - * Although equality strategy scan keys (for both arrays and non-arrays alike) - * are either marked required in both directions or in neither direction, - * there is a sense in which non-required arrays behave like required arrays. - * With a qual such as "WHERE a IN (100, 200) AND b >= 3 AND c IN (5, 6, 7)", - * the scan key on "c" is non-required, but nevertheless enables positioning - * the scan at the first tuple >= "(100, 3, 5)" on the leaf level during the - * first descent of the tree by _bt_first. Later on, there could also be a - * second descent, that places the scan right before tuples >= "(200, 3, 5)". - * _bt_first must never be allowed to build an insertion scan key whose "c" - * entry is set to a value other than 5, the "c" array's first element/value. - * (Actually, it's the first in the current scan direction. This example uses - * a forward scan.) - * - * Calling here resets the array scan key elements for the scan's non-required - * arrays. This is strictly necessary for correctness in a subset of cases - * involving "required in opposite direction"-triggered primitive index scans. - * Not all callers are at risk of _bt_first using a non-required array like - * this, but advancement always resets the arrays when another primitive scan - * is scheduled, just to keep things simple. Array advancement even makes - * sure to reset non-required arrays during scans that have no inequalities. - * (Advancement still won't call here when there are no inequalities, though - * that's just because it's all handled indirectly instead.) - * - * Note: _bt_verify_arrays_bt_first is called by an assertion to enforce that - * everybody got this right. - * - * Note: In practice almost all SAOP arrays are marked required during - * preprocessing (if necessary by generating skip arrays). It is hardly ever - * truly necessary to call here, but consistently doing so is simpler. - */ -static void -_bt_rewind_nonrequired_arrays(IndexScanDesc scan, ScanDirection dir) -{ - Relation rel = scan->indexRelation; - BTScanOpaque so = (BTScanOpaque) scan->opaque; - int arrayidx = 0; - - for (int ikey = 0; ikey < so->numberOfKeys; ikey++) - { - ScanKey cur = so->keyData + ikey; - BTArrayKeyInfo *array = NULL; - - if (!(cur->sk_flags & SK_SEARCHARRAY) || - cur->sk_strategy != BTEqualStrategyNumber) - continue; - - array = &so->arrayKeys[arrayidx++]; - Assert(array->scan_key == ikey); - - if ((cur->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD))) - continue; - - Assert(array->num_elems != -1); /* No non-required skip arrays */ - - _bt_array_set_low_or_high(rel, cur, array, - ScanDirectionIsForward(dir)); - } -} - -/* * _bt_tuple_before_array_skeys() -- too early to advance required arrays? * * We always compare the tuple using the current array keys (which we assume @@ -1380,8 +1311,6 @@ _bt_start_prim_scan(IndexScanDesc scan, ScanDirection dir) */ if (so->needPrimScan) { - Assert(_bt_verify_arrays_bt_first(scan, dir)); - /* * Flag was set -- must call _bt_first again, which will reset the * scan's needPrimScan flag @@ -2007,14 +1936,7 @@ _bt_advance_array_keys(IndexScanDesc scan, BTReadPageState *pstate, */ else if (has_required_opposite_direction_only && pstate->finaltup && unlikely(!_bt_oppodir_checkkeys(scan, dir, pstate->finaltup))) - { - /* - * Make sure that any SAOP arrays that were not marked required by - * preprocessing are reset to their first element for this direction - */ - _bt_rewind_nonrequired_arrays(scan, dir); goto new_prim_scan; - } continue_scan: @@ -2045,8 +1967,6 @@ continue_scan: */ so->oppositeDirCheck = has_required_opposite_direction_only; - _bt_rewind_nonrequired_arrays(scan, dir); - /* * skip by setting "look ahead" mechanism's offnum for forwards scans * (backwards scans check scanBehind flag directly instead) @@ -2143,48 +2063,6 @@ end_toplevel_scan: #ifdef USE_ASSERT_CHECKING /* - * Verify that the scan's qual state matches what we expect at the point that - * _bt_start_prim_scan is about to start a just-scheduled new primitive scan. - * - * We enforce a rule against non-required array scan keys: they must start out - * with whatever element is the first for the scan's current scan direction. - * See _bt_rewind_nonrequired_arrays comments for an explanation. - */ -static bool -_bt_verify_arrays_bt_first(IndexScanDesc scan, ScanDirection dir) -{ - BTScanOpaque so = (BTScanOpaque) scan->opaque; - int arrayidx = 0; - - for (int ikey = 0; ikey < so->numberOfKeys; ikey++) - { - ScanKey cur = so->keyData + ikey; - BTArrayKeyInfo *array = NULL; - int first_elem_dir; - - if (!(cur->sk_flags & SK_SEARCHARRAY) || - cur->sk_strategy != BTEqualStrategyNumber) - continue; - - array = &so->arrayKeys[arrayidx++]; - - if (((cur->sk_flags & SK_BT_REQFWD) && ScanDirectionIsForward(dir)) || - ((cur->sk_flags & SK_BT_REQBKWD) && ScanDirectionIsBackward(dir))) - continue; - - if (ScanDirectionIsForward(dir)) - first_elem_dir = 0; - else - first_elem_dir = array->num_elems - 1; - - if (array->cur_elem != first_elem_dir) - return false; - } - - return _bt_verify_keys_with_arraykeys(scan); -} - -/* * Verify that the scan's "so->keyData[]" scan keys are in agreement with * its array key state */ @@ -2194,6 +2072,7 @@ _bt_verify_keys_with_arraykeys(IndexScanDesc scan) BTScanOpaque so = (BTScanOpaque) scan->opaque; int last_sk_attno = InvalidAttrNumber, arrayidx = 0; + bool nonrequiredseen = false; if (!so->qual_ok) return false; @@ -2217,8 +2096,16 @@ _bt_verify_keys_with_arraykeys(IndexScanDesc scan) if (array->num_elems != -1 && cur->sk_argument != array->elem_values[array->cur_elem]) return false; - if (last_sk_attno > cur->sk_attno) - return false; + if (cur->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)) + { + if (last_sk_attno > cur->sk_attno) + return false; + if (nonrequiredseen) + return false; + } + else + nonrequiredseen = true; + last_sk_attno = cur->sk_attno; } @@ -2551,37 +2438,12 @@ _bt_set_startikey(IndexScanDesc scan, BTReadPageState *pstate) if (!(key->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD))) { /* Scan key isn't marked required (corner case) */ - Assert(!(key->sk_flags & SK_ROW_HEADER)); break; /* unsafe */ } if (key->sk_flags & SK_ROW_HEADER) { - /* - * RowCompare inequality. - * - * Only the first subkey from a RowCompare can ever be marked - * required (that happens when the row header is marked required). - * There is no simple, general way for us to transitively deduce - * whether or not every tuple on the page satisfies a RowCompare - * key based only on firsttup and lasttup -- so we just give up. - */ - if (!start_past_saop_eq && !so->skipScan) - break; /* unsafe to go further */ - - /* - * We have to be even more careful with RowCompares that come - * after an array: we assume it's unsafe to even bypass the array. - * Calling _bt_start_array_keys to recover the scan's arrays - * following use of forcenonrequired mode isn't compatible with - * _bt_check_rowcompare's continuescan=false behavior with NULL - * row compare members. _bt_advance_array_keys must not make a - * decision on the basis of a key not being satisfied in the - * opposite-to-scan direction until the scan reaches a leaf page - * where the same key begins to be satisfied in scan direction. - * The _bt_first !used_all_subkeys behavior makes this limitation - * hard to work around some other way. - */ - return; /* completely unsafe to set pstate.startikey */ + /* RowCompare inequalities currently aren't supported */ + break; /* "unsafe" */ } if (key->sk_strategy != BTEqualStrategyNumber) { @@ -3078,6 +2940,31 @@ _bt_check_rowcompare(ScanKey skey, IndexTuple tuple, int tupnatts, Assert(subkey->sk_flags & SK_ROW_MEMBER); + /* When a NULL row member is compared, the row never matches */ + if (subkey->sk_flags & SK_ISNULL) + { + /* + * Unlike the simple-scankey case, this isn't a disallowed case + * (except when it's the first row element that has the NULL arg). + * But it can never match. If all the earlier row comparison + * columns are required for the scan direction, we can stop the + * scan, because there can't be another tuple that will succeed. + */ + Assert(subkey != (ScanKey) DatumGetPointer(skey->sk_argument)); + subkey--; + if (forcenonrequired) + { + /* treating scan's keys as non-required */ + } + else if ((subkey->sk_flags & SK_BT_REQFWD) && + ScanDirectionIsForward(dir)) + *continuescan = false; + else if ((subkey->sk_flags & SK_BT_REQBKWD) && + ScanDirectionIsBackward(dir)) + *continuescan = false; + return false; + } + if (subkey->sk_attno > tupnatts) { /* @@ -3087,11 +2974,7 @@ _bt_check_rowcompare(ScanKey skey, IndexTuple tuple, int tupnatts, * attribute passes the qual. */ Assert(BTreeTupleIsPivot(tuple)); - cmpresult = 0; - if (subkey->sk_flags & SK_ROW_END) - break; - subkey++; - continue; + return true; } datum = index_getattr(tuple, @@ -3101,6 +2984,8 @@ _bt_check_rowcompare(ScanKey skey, IndexTuple tuple, int tupnatts, if (isNull) { + int reqflags; + if (forcenonrequired) { /* treating scan's keys as non-required */ @@ -3111,15 +2996,35 @@ _bt_check_rowcompare(ScanKey skey, IndexTuple tuple, int tupnatts, * Since NULLs are sorted before non-NULLs, we know we have * reached the lower limit of the range of values for this * index attr. On a backward scan, we can stop if this qual - * is one of the "must match" subset. We can stop regardless - * of whether the qual is > or <, so long as it's required, - * because it's not possible for any future tuples to pass. On - * a forward scan, however, we must keep going, because we may - * have initially positioned to the start of the index. - * (_bt_advance_array_keys also relies on this behavior during - * forward scans.) + * is one of the "must match" subset. However, on a forwards + * scan, we must keep going, because we may have initially + * positioned to the start of the index. + * + * All required NULLS FIRST > row members can use NULL tuple + * values to end backwards scans, just like with other values. + * A qual "WHERE (a, b, c) > (9, 42, 'foo')" can terminate a + * backwards scan upon reaching the index's rightmost "a = 9" + * tuple whose "b" column contains a NULL (if not sooner). + * Since "b" is NULLS FIRST, we can treat its NULLs as "<" 42. */ - if ((subkey->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)) && + reqflags = SK_BT_REQBKWD; + + /* + * When a most significant required NULLS FIRST < row compare + * member sees NULL tuple values during a backwards scan, it + * signals the end of matches for the whole row compare/scan. + * A qual "WHERE (a, b, c) < (9, 42, 'foo')" will terminate a + * backwards scan upon reaching the rightmost tuple whose "a" + * column has a NULL. The "a" NULL value is "<" 9, and yet + * our < row compare will still end the scan. (This isn't + * safe with later/lower-order row members. Notice that it + * can only happen with an "a" NULL some time after the scan + * completely stops needing to use its "b" and "c" members.) + */ + if (subkey == (ScanKey) DatumGetPointer(skey->sk_argument)) + reqflags |= SK_BT_REQFWD; /* safe, first row member */ + + if ((subkey->sk_flags & reqflags) && ScanDirectionIsBackward(dir)) *continuescan = false; } @@ -3129,15 +3034,35 @@ _bt_check_rowcompare(ScanKey skey, IndexTuple tuple, int tupnatts, * Since NULLs are sorted after non-NULLs, we know we have * reached the upper limit of the range of values for this * index attr. On a forward scan, we can stop if this qual is - * one of the "must match" subset. We can stop regardless of - * whether the qual is > or <, so long as it's required, - * because it's not possible for any future tuples to pass. On - * a backward scan, however, we must keep going, because we - * may have initially positioned to the end of the index. - * (_bt_advance_array_keys also relies on this behavior during - * backward scans.) + * one of the "must match" subset. However, on a backward + * scan, we must keep going, because we may have initially + * positioned to the end of the index. + * + * All required NULLS LAST < row members can use NULL tuple + * values to end forwards scans, just like with other values. + * A qual "WHERE (a, b, c) < (9, 42, 'foo')" can terminate a + * forwards scan upon reaching the index's leftmost "a = 9" + * tuple whose "b" column contains a NULL (if not sooner). + * Since "b" is NULLS LAST, we can treat its NULLs as ">" 42. + */ + reqflags = SK_BT_REQFWD; + + /* + * When a most significant required NULLS LAST > row compare + * member sees NULL tuple values during a forwards scan, it + * signals the end of matches for the whole row compare/scan. + * A qual "WHERE (a, b, c) > (9, 42, 'foo')" will terminate a + * forwards scan upon reaching the leftmost tuple whose "a" + * column has a NULL. The "a" NULL value is ">" 9, and yet + * our > row compare will end the scan. (This isn't safe with + * later/lower-order row members. Notice that it can only + * happen with an "a" NULL some time after the scan completely + * stops needing to use its "b" and "c" members.) */ - if ((subkey->sk_flags & (SK_BT_REQFWD | SK_BT_REQBKWD)) && + if (subkey == (ScanKey) DatumGetPointer(skey->sk_argument)) + reqflags |= SK_BT_REQBKWD; /* safe, first row member */ + + if ((subkey->sk_flags & reqflags) && ScanDirectionIsForward(dir)) *continuescan = false; } @@ -3148,30 +3073,6 @@ _bt_check_rowcompare(ScanKey skey, IndexTuple tuple, int tupnatts, return false; } - if (subkey->sk_flags & SK_ISNULL) - { - /* - * Unlike the simple-scankey case, this isn't a disallowed case - * (except when it's the first row element that has the NULL arg). - * But it can never match. If all the earlier row comparison - * columns are required for the scan direction, we can stop the - * scan, because there can't be another tuple that will succeed. - */ - Assert(subkey != (ScanKey) DatumGetPointer(skey->sk_argument)); - subkey--; - if (forcenonrequired) - { - /* treating scan's keys as non-required */ - } - else if ((subkey->sk_flags & SK_BT_REQFWD) && - ScanDirectionIsForward(dir)) - *continuescan = false; - else if ((subkey->sk_flags & SK_BT_REQBKWD) && - ScanDirectionIsBackward(dir)) - *continuescan = false; - return false; - } - /* Perform the test --- three-way comparison not bool operator */ cmpresult = DatumGetInt32(FunctionCall2Coll(&subkey->sk_func, subkey->sk_collation, @@ -3330,87 +3231,85 @@ _bt_checkkeys_look_ahead(IndexScanDesc scan, BTReadPageState *pstate, * current page and killed tuples thereon (generally, this should only be * called if so->numKilled > 0). * - * The caller does not have a lock on the page and may or may not have the - * page pinned in a buffer. Note that read-lock is sufficient for setting - * LP_DEAD status (which is only a hint). - * - * We match items by heap TID before assuming they are the right ones to - * delete. We cope with cases where items have moved right due to insertions. - * If an item has moved off the current page due to a split, we'll fail to - * find it and do nothing (this is not an error case --- we assume the item - * will eventually get marked in a future indexscan). + * Caller should not have a lock on the so->currPos page, but must hold a + * buffer pin when !so->dropPin. When we return, it still won't be locked. + * It'll continue to hold whatever pins were held before calling here. * - * Note that if we hold a pin on the target page continuously from initially - * reading the items until applying this function, VACUUM cannot have deleted - * any items from the page, and so there is no need to search left from the - * recorded offset. (This observation also guarantees that the item is still - * the right one to delete, which might otherwise be questionable since heap - * TIDs can get recycled.) This holds true even if the page has been modified - * by inserts and page splits, so there is no need to consult the LSN. + * We match items by heap TID before assuming they are the right ones to set + * LP_DEAD. If the scan is one that holds a buffer pin on the target page + * continuously from initially reading the items until applying this function + * (if it is a !so->dropPin scan), VACUUM cannot have deleted any items on the + * page, so the page's TIDs can't have been recycled by now. There's no risk + * that we'll confuse a new index tuple that happens to use a recycled TID + * with a now-removed tuple with the same TID (that used to be on this same + * page). We can't rely on that during scans that drop buffer pins eagerly + * (so->dropPin scans), though, so we must condition setting LP_DEAD bits on + * the page LSN having not changed since back when _bt_readpage saw the page. + * We totally give up on setting LP_DEAD bits when the page LSN changed. * - * If the pin was released after reading the page, then we re-read it. If it - * has been modified since we read it (as determined by the LSN), we dare not - * flag any entries because it is possible that the old entry was vacuumed - * away and the TID was re-used by a completely different heap tuple. + * We give up much less often during !so->dropPin scans, but it still happens. + * We cope with cases where items have moved right due to insertions. If an + * item has moved off the current page due to a split, we'll fail to find it + * and just give up on it. */ void _bt_killitems(IndexScanDesc scan) { + Relation rel = scan->indexRelation; BTScanOpaque so = (BTScanOpaque) scan->opaque; Page page; BTPageOpaque opaque; OffsetNumber minoff; OffsetNumber maxoff; - int i; int numKilled = so->numKilled; bool killedsomething = false; - bool droppedpin PG_USED_FOR_ASSERTS_ONLY; + Buffer buf; + Assert(numKilled > 0); Assert(BTScanPosIsValid(so->currPos)); + Assert(scan->heapRelation != NULL); /* can't be a bitmap index scan */ - /* - * Always reset the scan state, so we don't look for same items on other - * pages. - */ + /* Always invalidate so->killedItems[] before leaving so->currPos */ so->numKilled = 0; - if (BTScanPosIsPinned(so->currPos)) + if (!so->dropPin) { /* * We have held the pin on this page since we read the index tuples, * so all we need to do is lock it. The pin will have prevented - * re-use of any TID on the page, so there is no need to check the - * LSN. + * concurrent VACUUMs from recycling any of the TIDs on the page. */ - droppedpin = false; - _bt_lockbuf(scan->indexRelation, so->currPos.buf, BT_READ); - - page = BufferGetPage(so->currPos.buf); + Assert(BTScanPosIsPinned(so->currPos)); + buf = so->currPos.buf; + _bt_lockbuf(rel, buf, BT_READ); } else { - Buffer buf; + XLogRecPtr latestlsn; - droppedpin = true; - /* Attempt to re-read the buffer, getting pin and lock. */ - buf = _bt_getbuf(scan->indexRelation, so->currPos.currPage, BT_READ); + Assert(!BTScanPosIsPinned(so->currPos)); + Assert(RelationNeedsWAL(rel)); + buf = _bt_getbuf(rel, so->currPos.currPage, BT_READ); - page = BufferGetPage(buf); - if (BufferGetLSNAtomic(buf) == so->currPos.lsn) - so->currPos.buf = buf; - else + latestlsn = BufferGetLSNAtomic(buf); + Assert(!XLogRecPtrIsInvalid(so->currPos.lsn)); + Assert(so->currPos.lsn <= latestlsn); + if (so->currPos.lsn != latestlsn) { - /* Modified while not pinned means hinting is not safe. */ - _bt_relbuf(scan->indexRelation, buf); + /* Modified, give up on hinting */ + _bt_relbuf(rel, buf); return; } + + /* Unmodified, hinting is safe */ } + page = BufferGetPage(buf); opaque = BTPageGetOpaque(page); minoff = P_FIRSTDATAKEY(opaque); maxoff = PageGetMaxOffsetNumber(page); - for (i = 0; i < numKilled; i++) + for (int i = 0; i < numKilled; i++) { int itemIndex = so->killedItems[i]; BTScanPosItem *kitem = &so->currPos.items[itemIndex]; @@ -3442,7 +3341,7 @@ _bt_killitems(IndexScanDesc scan) * correctness. * * Note that the page may have been modified in almost any way - * since we first read it (in the !droppedpin case), so it's + * since we first read it (in the !so->dropPin case), so it's * possible that this posting list tuple wasn't a posting list * tuple when we first encountered its heap TIDs. */ @@ -3458,7 +3357,7 @@ _bt_killitems(IndexScanDesc scan) * though only in the common case where the page can't * have been concurrently modified */ - Assert(kitem->indexOffset == offnum || !droppedpin); + Assert(kitem->indexOffset == offnum || !so->dropPin); /* * Read-ahead to later kitems here. @@ -3522,10 +3421,13 @@ _bt_killitems(IndexScanDesc scan) if (killedsomething) { opaque->btpo_flags |= BTP_HAS_GARBAGE; - MarkBufferDirtyHint(so->currPos.buf, true); + MarkBufferDirtyHint(buf, true); } - _bt_unlockbuf(scan->indexRelation, so->currPos.buf); + if (!so->dropPin) + _bt_unlockbuf(rel, buf); + else + _bt_relbuf(rel, buf); } diff --git a/src/backend/access/rmgrdesc/replorigindesc.c b/src/backend/access/rmgrdesc/replorigindesc.c index 5dd74233996..35e3af2903e 100644 --- a/src/backend/access/rmgrdesc/replorigindesc.c +++ b/src/backend/access/rmgrdesc/replorigindesc.c @@ -29,7 +29,7 @@ replorigin_desc(StringInfo buf, XLogReaderState *record) xlrec = (xl_replorigin_set *) rec; - appendStringInfo(buf, "set %u; lsn %X/%X; force: %d", + appendStringInfo(buf, "set %u; lsn %X/%08X; force: %d", xlrec->node_id, LSN_FORMAT_ARGS(xlrec->remote_lsn), xlrec->force); diff --git a/src/backend/access/rmgrdesc/xactdesc.c b/src/backend/access/rmgrdesc/xactdesc.c index 305598e2865..f0f696855b9 100644 --- a/src/backend/access/rmgrdesc/xactdesc.c +++ b/src/backend/access/rmgrdesc/xactdesc.c @@ -359,7 +359,7 @@ xact_desc_commit(StringInfo buf, uint8 info, xl_xact_commit *xlrec, RepOriginId if (parsed.xinfo & XACT_XINFO_HAS_ORIGIN) { - appendStringInfo(buf, "; origin: node %u, lsn %X/%X, at %s", + appendStringInfo(buf, "; origin: node %u, lsn %X/%08X, at %s", origin_id, LSN_FORMAT_ARGS(parsed.origin_lsn), timestamptz_to_str(parsed.origin_timestamp)); @@ -384,7 +384,7 @@ xact_desc_abort(StringInfo buf, uint8 info, xl_xact_abort *xlrec, RepOriginId or if (parsed.xinfo & XACT_XINFO_HAS_ORIGIN) { - appendStringInfo(buf, "; origin: node %u, lsn %X/%X, at %s", + appendStringInfo(buf, "; origin: node %u, lsn %X/%08X, at %s", origin_id, LSN_FORMAT_ARGS(parsed.origin_lsn), timestamptz_to_str(parsed.origin_timestamp)); @@ -418,7 +418,7 @@ xact_desc_prepare(StringInfo buf, uint8 info, xl_xact_prepare *xlrec, RepOriginI * way as PrepareRedoAdd(). */ if (origin_id != InvalidRepOriginId) - appendStringInfo(buf, "; origin: node %u, lsn %X/%X, at %s", + appendStringInfo(buf, "; origin: node %u, lsn %X/%08X, at %s", origin_id, LSN_FORMAT_ARGS(parsed.origin_lsn), timestamptz_to_str(parsed.origin_timestamp)); diff --git a/src/backend/access/rmgrdesc/xlogdesc.c b/src/backend/access/rmgrdesc/xlogdesc.c index 58040f28656..cd6c2a2f650 100644 --- a/src/backend/access/rmgrdesc/xlogdesc.c +++ b/src/backend/access/rmgrdesc/xlogdesc.c @@ -65,7 +65,7 @@ xlog_desc(StringInfo buf, XLogReaderState *record) { CheckPoint *checkpoint = (CheckPoint *) rec; - appendStringInfo(buf, "redo %X/%X; " + appendStringInfo(buf, "redo %X/%08X; " "tli %u; prev tli %u; fpw %s; wal_level %s; xid %u:%u; oid %u; multi %u; offset %u; " "oldest xid %u in DB %u; oldest multi %u in DB %u; " "oldest/newest commit timestamp xid: %u/%u; " @@ -111,7 +111,7 @@ xlog_desc(StringInfo buf, XLogReaderState *record) XLogRecPtr startpoint; memcpy(&startpoint, rec, sizeof(XLogRecPtr)); - appendStringInfo(buf, "%X/%X", LSN_FORMAT_ARGS(startpoint)); + appendStringInfo(buf, "%X/%08X", LSN_FORMAT_ARGS(startpoint)); } else if (info == XLOG_PARAMETER_CHANGE) { @@ -156,7 +156,7 @@ xlog_desc(StringInfo buf, XLogReaderState *record) xl_overwrite_contrecord xlrec; memcpy(&xlrec, rec, sizeof(xl_overwrite_contrecord)); - appendStringInfo(buf, "lsn %X/%X; time %s", + appendStringInfo(buf, "lsn %X/%08X; time %s", LSN_FORMAT_ARGS(xlrec.overwritten_lsn), timestamptz_to_str(xlrec.overwrite_time)); } diff --git a/src/backend/access/spgist/spgutils.c b/src/backend/access/spgist/spgutils.c index 95fea74e296..9b86c016acb 100644 --- a/src/backend/access/spgist/spgutils.c +++ b/src/backend/access/spgist/spgutils.c @@ -785,7 +785,7 @@ SpGistGetInnerTypeSize(SpGistTypeDesc *att, Datum datum) else if (att->attlen > 0) size = att->attlen; else - size = VARSIZE_ANY(datum); + size = VARSIZE_ANY(DatumGetPointer(datum)); return MAXALIGN(size); } @@ -804,7 +804,7 @@ memcpyInnerDatum(void *target, SpGistTypeDesc *att, Datum datum) } else { - size = (att->attlen > 0) ? att->attlen : VARSIZE_ANY(datum); + size = (att->attlen > 0) ? att->attlen : VARSIZE_ANY(DatumGetPointer(datum)); memcpy(target, DatumGetPointer(datum), size); } } diff --git a/src/backend/access/table/toast_helper.c b/src/backend/access/table/toast_helper.c index b60fab0a4d2..11f97d65367 100644 --- a/src/backend/access/table/toast_helper.c +++ b/src/backend/access/table/toast_helper.c @@ -330,7 +330,7 @@ toast_delete_external(Relation rel, const Datum *values, const bool *isnull, if (isnull[i]) continue; - else if (VARATT_IS_EXTERNAL_ONDISK(value)) + else if (VARATT_IS_EXTERNAL_ONDISK(DatumGetPointer(value))) toast_delete_datum(rel, value, is_speculative); } } diff --git a/src/backend/access/transam/clog.c b/src/backend/access/transam/clog.c index 48f10bec91e..e80fbe109cf 100644 --- a/src/backend/access/transam/clog.c +++ b/src/backend/access/transam/clog.c @@ -110,9 +110,7 @@ static SlruCtlData XactCtlData; #define XactCtl (&XactCtlData) -static int ZeroCLOGPage(int64 pageno, bool writeXlog); static bool CLOGPagePrecedes(int64 page1, int64 page2); -static void WriteZeroPageXlogRec(int64 pageno); static void WriteTruncateXlogRec(int64 pageno, TransactionId oldestXact, Oid oldestXactDb); static void TransactionIdSetPageStatus(TransactionId xid, int nsubxids, @@ -832,41 +830,8 @@ check_transaction_buffers(int *newval, void **extra, GucSource source) void BootStrapCLOG(void) { - int slotno; - LWLock *lock = SimpleLruGetBankLock(XactCtl, 0); - - LWLockAcquire(lock, LW_EXCLUSIVE); - - /* Create and zero the first page of the commit log */ - slotno = ZeroCLOGPage(0, false); - - /* Make sure it's written out */ - SimpleLruWritePage(XactCtl, slotno); - Assert(!XactCtl->shared->page_dirty[slotno]); - - LWLockRelease(lock); -} - -/* - * Initialize (or reinitialize) a page of CLOG to zeroes. - * If writeXlog is true, also emit an XLOG record saying we did this. - * - * The page is not actually written, just set up in shared memory. - * The slot number of the new page is returned. - * - * Control lock must be held at entry, and will be held at exit. - */ -static int -ZeroCLOGPage(int64 pageno, bool writeXlog) -{ - int slotno; - - slotno = SimpleLruZeroPage(XactCtl, pageno); - - if (writeXlog) - WriteZeroPageXlogRec(pageno); - - return slotno; + /* Zero the initial page and flush it to disk */ + SimpleLruZeroAndWritePage(XactCtl, 0); } /* @@ -974,8 +939,9 @@ ExtendCLOG(TransactionId newestXact) LWLockAcquire(lock, LW_EXCLUSIVE); - /* Zero the page and make an XLOG entry about it */ - ZeroCLOGPage(pageno, true); + /* Zero the page and make a WAL entry about it */ + SimpleLruZeroPage(XactCtl, pageno); + XLogSimpleInsertInt64(RM_CLOG_ID, CLOG_ZEROPAGE, pageno); LWLockRelease(lock); } @@ -1068,17 +1034,6 @@ CLOGPagePrecedes(int64 page1, int64 page2) /* - * Write a ZEROPAGE xlog record - */ -static void -WriteZeroPageXlogRec(int64 pageno) -{ - XLogBeginInsert(); - XLogRegisterData(&pageno, sizeof(pageno)); - (void) XLogInsert(RM_CLOG_ID, CLOG_ZEROPAGE); -} - -/* * Write a TRUNCATE xlog record * * We must flush the xlog record to disk before returning --- see notes @@ -1114,19 +1069,9 @@ clog_redo(XLogReaderState *record) if (info == CLOG_ZEROPAGE) { int64 pageno; - int slotno; - LWLock *lock; memcpy(&pageno, XLogRecGetData(record), sizeof(pageno)); - - lock = SimpleLruGetBankLock(XactCtl, pageno); - LWLockAcquire(lock, LW_EXCLUSIVE); - - slotno = ZeroCLOGPage(pageno, false); - SimpleLruWritePage(XactCtl, slotno); - Assert(!XactCtl->shared->page_dirty[slotno]); - - LWLockRelease(lock); + SimpleLruZeroAndWritePage(XactCtl, pageno); } else if (info == CLOG_TRUNCATE) { diff --git a/src/backend/access/transam/commit_ts.c b/src/backend/access/transam/commit_ts.c index 113fae1437a..370b38e048b 100644 --- a/src/backend/access/transam/commit_ts.c +++ b/src/backend/access/transam/commit_ts.c @@ -114,11 +114,9 @@ static void SetXidCommitTsInPage(TransactionId xid, int nsubxids, static void TransactionIdSetCommitTs(TransactionId xid, TimestampTz ts, RepOriginId nodeid, int slotno); static void error_commit_ts_disabled(void); -static int ZeroCommitTsPage(int64 pageno, bool writeXlog); static bool CommitTsPagePrecedes(int64 page1, int64 page2); static void ActivateCommitTs(void); static void DeactivateCommitTs(void); -static void WriteZeroPageXlogRec(int64 pageno); static void WriteTruncateXlogRec(int64 pageno, TransactionId oldestXid); /* @@ -603,28 +601,6 @@ BootStrapCommitTs(void) } /* - * Initialize (or reinitialize) a page of CommitTs to zeroes. - * If writeXlog is true, also emit an XLOG record saying we did this. - * - * The page is not actually written, just set up in shared memory. - * The slot number of the new page is returned. - * - * Control lock must be held at entry, and will be held at exit. - */ -static int -ZeroCommitTsPage(int64 pageno, bool writeXlog) -{ - int slotno; - - slotno = SimpleLruZeroPage(CommitTsCtl, pageno); - - if (writeXlog) - WriteZeroPageXlogRec(pageno); - - return slotno; -} - -/* * This must be called ONCE during postmaster or standalone-backend startup, * after StartupXLOG has initialized TransamVariables->nextXid. */ @@ -707,6 +683,13 @@ ActivateCommitTs(void) TransactionId xid; int64 pageno; + /* + * During bootstrap, we should not register commit timestamps so skip the + * activation in this case. + */ + if (IsBootstrapProcessingMode()) + return; + /* If we've done this already, there's nothing to do */ LWLockAcquire(CommitTsLock, LW_EXCLUSIVE); if (commitTsShared->commitTsActive) @@ -747,16 +730,7 @@ ActivateCommitTs(void) /* Create the current segment file, if necessary */ if (!SimpleLruDoesPhysicalPageExist(CommitTsCtl, pageno)) - { - LWLock *lock = SimpleLruGetBankLock(CommitTsCtl, pageno); - int slotno; - - LWLockAcquire(lock, LW_EXCLUSIVE); - slotno = ZeroCommitTsPage(pageno, false); - SimpleLruWritePage(CommitTsCtl, slotno); - Assert(!CommitTsCtl->shared->page_dirty[slotno]); - LWLockRelease(lock); - } + SimpleLruZeroAndWritePage(CommitTsCtl, pageno); /* Change the activation status in shared memory. */ LWLockAcquire(CommitTsLock, LW_EXCLUSIVE); @@ -867,8 +841,12 @@ ExtendCommitTs(TransactionId newestXact) LWLockAcquire(lock, LW_EXCLUSIVE); - /* Zero the page and make an XLOG entry about it */ - ZeroCommitTsPage(pageno, !InRecovery); + /* Zero the page ... */ + SimpleLruZeroPage(CommitTsCtl, pageno); + + /* and make a WAL entry about that, unless we're in REDO */ + if (!InRecovery) + XLogSimpleInsertInt64(RM_COMMIT_TS_ID, COMMIT_TS_ZEROPAGE, pageno); LWLockRelease(lock); } @@ -983,17 +961,6 @@ CommitTsPagePrecedes(int64 page1, int64 page2) /* - * Write a ZEROPAGE xlog record - */ -static void -WriteZeroPageXlogRec(int64 pageno) -{ - XLogBeginInsert(); - XLogRegisterData(&pageno, sizeof(pageno)); - (void) XLogInsert(RM_COMMIT_TS_ID, COMMIT_TS_ZEROPAGE); -} - -/* * Write a TRUNCATE xlog record */ static void @@ -1023,19 +990,9 @@ commit_ts_redo(XLogReaderState *record) if (info == COMMIT_TS_ZEROPAGE) { int64 pageno; - int slotno; - LWLock *lock; memcpy(&pageno, XLogRecGetData(record), sizeof(pageno)); - - lock = SimpleLruGetBankLock(CommitTsCtl, pageno); - LWLockAcquire(lock, LW_EXCLUSIVE); - - slotno = ZeroCommitTsPage(pageno, false); - SimpleLruWritePage(CommitTsCtl, slotno); - Assert(!CommitTsCtl->shared->page_dirty[slotno]); - - LWLockRelease(lock); + SimpleLruZeroAndWritePage(CommitTsCtl, pageno); } else if (info == COMMIT_TS_TRUNCATE) { diff --git a/src/backend/access/transam/multixact.c b/src/backend/access/transam/multixact.c index 3c06ac45532..3cb09c3d598 100644 --- a/src/backend/access/transam/multixact.c +++ b/src/backend/access/transam/multixact.c @@ -401,8 +401,6 @@ static void mXactCachePut(MultiXactId multi, int nmembers, static char *mxstatus_to_string(MultiXactStatus status); /* management of SLRU infrastructure */ -static int ZeroMultiXactOffsetPage(int64 pageno, bool writeXlog); -static int ZeroMultiXactMemberPage(int64 pageno, bool writeXlog); static bool MultiXactOffsetPagePrecedes(int64 page1, int64 page2); static bool MultiXactMemberPagePrecedes(int64 page1, int64 page2); static bool MultiXactOffsetPrecedes(MultiXactOffset offset1, @@ -413,7 +411,6 @@ static bool MultiXactOffsetWouldWrap(MultiXactOffset boundary, MultiXactOffset start, uint32 distance); static bool SetOffsetVacuumLimit(bool is_startup); static bool find_multixact_start(MultiXactId multi, MultiXactOffset *result); -static void WriteMZeroPageXlogRec(int64 pageno, uint8 info); static void WriteMTruncateXlogRec(Oid oldestMultiDB, MultiXactId startTruncOff, MultiXactId endTruncOff, @@ -1847,7 +1844,7 @@ AtPrepare_MultiXact(void) * Clean up after successful PREPARE TRANSACTION */ void -PostPrepare_MultiXact(TransactionId xid) +PostPrepare_MultiXact(FullTransactionId fxid) { MultiXactId myOldestMember; @@ -1858,7 +1855,7 @@ PostPrepare_MultiXact(TransactionId xid) myOldestMember = OldestMemberMXactId[MyProcNumber]; if (MultiXactIdIsValid(myOldestMember)) { - ProcNumber dummyProcNumber = TwoPhaseGetDummyProcNumber(xid, false); + ProcNumber dummyProcNumber = TwoPhaseGetDummyProcNumber(fxid, false); /* * Even though storing MultiXactId is atomic, acquire lock to make @@ -1896,10 +1893,10 @@ PostPrepare_MultiXact(TransactionId xid) * Recover the state of a prepared transaction at startup */ void -multixact_twophase_recover(TransactionId xid, uint16 info, +multixact_twophase_recover(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { - ProcNumber dummyProcNumber = TwoPhaseGetDummyProcNumber(xid, false); + ProcNumber dummyProcNumber = TwoPhaseGetDummyProcNumber(fxid, false); MultiXactId oldestMember; /* @@ -1917,10 +1914,10 @@ multixact_twophase_recover(TransactionId xid, uint16 info, * Similar to AtEOXact_MultiXact but for COMMIT PREPARED */ void -multixact_twophase_postcommit(TransactionId xid, uint16 info, +multixact_twophase_postcommit(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { - ProcNumber dummyProcNumber = TwoPhaseGetDummyProcNumber(xid, true); + ProcNumber dummyProcNumber = TwoPhaseGetDummyProcNumber(fxid, true); Assert(len == sizeof(MultiXactId)); @@ -1932,10 +1929,10 @@ multixact_twophase_postcommit(TransactionId xid, uint16 info, * This is actually just the same as the COMMIT case. */ void -multixact_twophase_postabort(TransactionId xid, uint16 info, +multixact_twophase_postabort(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { - multixact_twophase_postcommit(xid, info, recdata, len); + multixact_twophase_postcommit(fxid, info, recdata, len); } /* @@ -2033,70 +2030,9 @@ check_multixact_member_buffers(int *newval, void **extra, GucSource source) void BootStrapMultiXact(void) { - int slotno; - LWLock *lock; - - lock = SimpleLruGetBankLock(MultiXactOffsetCtl, 0); - LWLockAcquire(lock, LW_EXCLUSIVE); - - /* Create and zero the first page of the offsets log */ - slotno = ZeroMultiXactOffsetPage(0, false); - - /* Make sure it's written out */ - SimpleLruWritePage(MultiXactOffsetCtl, slotno); - Assert(!MultiXactOffsetCtl->shared->page_dirty[slotno]); - - LWLockRelease(lock); - - lock = SimpleLruGetBankLock(MultiXactMemberCtl, 0); - LWLockAcquire(lock, LW_EXCLUSIVE); - - /* Create and zero the first page of the members log */ - slotno = ZeroMultiXactMemberPage(0, false); - - /* Make sure it's written out */ - SimpleLruWritePage(MultiXactMemberCtl, slotno); - Assert(!MultiXactMemberCtl->shared->page_dirty[slotno]); - - LWLockRelease(lock); -} - -/* - * Initialize (or reinitialize) a page of MultiXactOffset to zeroes. - * If writeXlog is true, also emit an XLOG record saying we did this. - * - * The page is not actually written, just set up in shared memory. - * The slot number of the new page is returned. - * - * Control lock must be held at entry, and will be held at exit. - */ -static int -ZeroMultiXactOffsetPage(int64 pageno, bool writeXlog) -{ - int slotno; - - slotno = SimpleLruZeroPage(MultiXactOffsetCtl, pageno); - - if (writeXlog) - WriteMZeroPageXlogRec(pageno, XLOG_MULTIXACT_ZERO_OFF_PAGE); - - return slotno; -} - -/* - * Ditto, for MultiXactMember - */ -static int -ZeroMultiXactMemberPage(int64 pageno, bool writeXlog) -{ - int slotno; - - slotno = SimpleLruZeroPage(MultiXactMemberCtl, pageno); - - if (writeXlog) - WriteMZeroPageXlogRec(pageno, XLOG_MULTIXACT_ZERO_MEM_PAGE); - - return slotno; + /* Zero the initial pages and flush them to disk */ + SimpleLruZeroAndWritePage(MultiXactOffsetCtl, 0); + SimpleLruZeroAndWritePage(MultiXactMemberCtl, 0); } /* @@ -2134,7 +2070,7 @@ MaybeExtendOffsetSlru(void) * with creating a new segment file even if the page we're writing is * not the first in it, so this is enough. */ - slotno = ZeroMultiXactOffsetPage(pageno, false); + slotno = SimpleLruZeroPage(MultiXactOffsetCtl, pageno); SimpleLruWritePage(MultiXactOffsetCtl, slotno); } @@ -2568,8 +2504,10 @@ ExtendMultiXactOffset(MultiXactId multi) LWLockAcquire(lock, LW_EXCLUSIVE); - /* Zero the page and make an XLOG entry about it */ - ZeroMultiXactOffsetPage(pageno, true); + /* Zero the page and make a WAL entry about it */ + SimpleLruZeroPage(MultiXactOffsetCtl, pageno); + XLogSimpleInsertInt64(RM_MULTIXACT_ID, XLOG_MULTIXACT_ZERO_OFF_PAGE, + pageno); LWLockRelease(lock); } @@ -2611,8 +2549,10 @@ ExtendMultiXactMember(MultiXactOffset offset, int nmembers) LWLockAcquire(lock, LW_EXCLUSIVE); - /* Zero the page and make an XLOG entry about it */ - ZeroMultiXactMemberPage(pageno, true); + /* Zero the page and make a WAL entry about it */ + SimpleLruZeroPage(MultiXactMemberCtl, pageno); + XLogSimpleInsertInt64(RM_MULTIXACT_ID, + XLOG_MULTIXACT_ZERO_MEM_PAGE, pageno); LWLockRelease(lock); } @@ -3348,18 +3288,6 @@ MultiXactOffsetPrecedes(MultiXactOffset offset1, MultiXactOffset offset2) } /* - * Write an xlog record reflecting the zeroing of either a MEMBERs or - * OFFSETs page (info shows which) - */ -static void -WriteMZeroPageXlogRec(int64 pageno, uint8 info) -{ - XLogBeginInsert(); - XLogRegisterData(&pageno, sizeof(pageno)); - (void) XLogInsert(RM_MULTIXACT_ID, info); -} - -/* * Write a TRUNCATE xlog record * * We must flush the xlog record to disk before returning --- see notes in @@ -3401,36 +3329,16 @@ multixact_redo(XLogReaderState *record) if (info == XLOG_MULTIXACT_ZERO_OFF_PAGE) { int64 pageno; - int slotno; - LWLock *lock; memcpy(&pageno, XLogRecGetData(record), sizeof(pageno)); - - lock = SimpleLruGetBankLock(MultiXactOffsetCtl, pageno); - LWLockAcquire(lock, LW_EXCLUSIVE); - - slotno = ZeroMultiXactOffsetPage(pageno, false); - SimpleLruWritePage(MultiXactOffsetCtl, slotno); - Assert(!MultiXactOffsetCtl->shared->page_dirty[slotno]); - - LWLockRelease(lock); + SimpleLruZeroAndWritePage(MultiXactOffsetCtl, pageno); } else if (info == XLOG_MULTIXACT_ZERO_MEM_PAGE) { int64 pageno; - int slotno; - LWLock *lock; memcpy(&pageno, XLogRecGetData(record), sizeof(pageno)); - - lock = SimpleLruGetBankLock(MultiXactMemberCtl, pageno); - LWLockAcquire(lock, LW_EXCLUSIVE); - - slotno = ZeroMultiXactMemberPage(pageno, false); - SimpleLruWritePage(MultiXactMemberCtl, slotno); - Assert(!MultiXactMemberCtl->shared->page_dirty[slotno]); - - LWLockRelease(lock); + SimpleLruZeroAndWritePage(MultiXactMemberCtl, pageno); } else if (info == XLOG_MULTIXACT_CREATE_ID) { diff --git a/src/backend/access/transam/slru.c b/src/backend/access/transam/slru.c index fe56286d9a9..10ec259f382 100644 --- a/src/backend/access/transam/slru.c +++ b/src/backend/access/transam/slru.c @@ -434,6 +434,31 @@ SimpleLruZeroLSNs(SlruCtl ctl, int slotno) } /* + * This is a convenience wrapper for the common case of zeroing a page and + * immediately flushing it to disk. + * + * Control lock is acquired and released here. + */ +void +SimpleLruZeroAndWritePage(SlruCtl ctl, int64 pageno) +{ + int slotno; + LWLock *lock; + + lock = SimpleLruGetBankLock(ctl, pageno); + LWLockAcquire(lock, LW_EXCLUSIVE); + + /* Create and zero the page */ + slotno = SimpleLruZeroPage(ctl, pageno); + + /* Make sure it's written out */ + SimpleLruWritePage(ctl, slotno); + Assert(!ctl->shared->page_dirty[slotno]); + + LWLockRelease(lock); +} + +/* * Wait for any active I/O on a page slot to finish. (This does not * guarantee that new I/O hasn't been started before we return, though. * In fact the slot might not even contain the same page anymore.) diff --git a/src/backend/access/transam/subtrans.c b/src/backend/access/transam/subtrans.c index 15153618fad..09aace9e09f 100644 --- a/src/backend/access/transam/subtrans.c +++ b/src/backend/access/transam/subtrans.c @@ -74,7 +74,6 @@ static SlruCtlData SubTransCtlData; #define SubTransCtl (&SubTransCtlData) -static int ZeroSUBTRANSPage(int64 pageno); static bool SubTransPagePrecedes(int64 page1, int64 page2); @@ -269,33 +268,8 @@ check_subtrans_buffers(int *newval, void **extra, GucSource source) void BootStrapSUBTRANS(void) { - int slotno; - LWLock *lock = SimpleLruGetBankLock(SubTransCtl, 0); - - LWLockAcquire(lock, LW_EXCLUSIVE); - - /* Create and zero the first page of the subtrans log */ - slotno = ZeroSUBTRANSPage(0); - - /* Make sure it's written out */ - SimpleLruWritePage(SubTransCtl, slotno); - Assert(!SubTransCtl->shared->page_dirty[slotno]); - - LWLockRelease(lock); -} - -/* - * Initialize (or reinitialize) a page of SUBTRANS to zeroes. - * - * The page is not actually written, just set up in shared memory. - * The slot number of the new page is returned. - * - * Control lock must be held at entry, and will be held at exit. - */ -static int -ZeroSUBTRANSPage(int64 pageno) -{ - return SimpleLruZeroPage(SubTransCtl, pageno); + /* Zero the initial page and flush it to disk */ + SimpleLruZeroAndWritePage(SubTransCtl, 0); } /* @@ -335,7 +309,7 @@ StartupSUBTRANS(TransactionId oldestActiveXID) prevlock = lock; } - (void) ZeroSUBTRANSPage(startPage); + (void) SimpleLruZeroPage(SubTransCtl, startPage); if (startPage == endPage) break; @@ -395,7 +369,7 @@ ExtendSUBTRANS(TransactionId newestXact) LWLockAcquire(lock, LW_EXCLUSIVE); /* Zero the page */ - ZeroSUBTRANSPage(pageno); + SimpleLruZeroPage(SubTransCtl, pageno); LWLockRelease(lock); } diff --git a/src/backend/access/transam/timeline.c b/src/backend/access/transam/timeline.c index a27f27cc037..186eb91f609 100644 --- a/src/backend/access/transam/timeline.c +++ b/src/backend/access/transam/timeline.c @@ -154,7 +154,7 @@ readTimeLineHistory(TimeLineID targetTLI) if (*ptr == '\0' || *ptr == '#') continue; - nfields = sscanf(fline, "%u\t%X/%X", &tli, &switchpoint_hi, &switchpoint_lo); + nfields = sscanf(fline, "%u\t%X/%08X", &tli, &switchpoint_hi, &switchpoint_lo); if (nfields < 1) { @@ -399,7 +399,7 @@ writeTimeLineHistory(TimeLineID newTLI, TimeLineID parentTLI, * parent file failed to end with one. */ snprintf(buffer, sizeof(buffer), - "%s%u\t%X/%X\t%s\n", + "%s%u\t%X/%08X\t%s\n", (srcfd < 0) ? "" : "\n", parentTLI, LSN_FORMAT_ARGS(switchpoint), diff --git a/src/backend/access/transam/twophase.c b/src/backend/access/transam/twophase.c index 73a80559194..7918176fc58 100644 --- a/src/backend/access/transam/twophase.c +++ b/src/backend/access/transam/twophase.c @@ -159,7 +159,7 @@ typedef struct GlobalTransactionData */ XLogRecPtr prepare_start_lsn; /* XLOG offset of prepare record start */ XLogRecPtr prepare_end_lsn; /* XLOG offset of prepare record end */ - TransactionId xid; /* The GXACT id */ + FullTransactionId fxid; /* The GXACT full xid */ Oid owner; /* ID of user that executed the xact */ ProcNumber locking_backend; /* backend currently working on the xact */ @@ -197,6 +197,7 @@ static GlobalTransaction MyLockedGxact = NULL; static bool twophaseExitRegistered = false; +static void PrepareRedoRemoveFull(FullTransactionId fxid, bool giveWarning); static void RecordTransactionCommitPrepared(TransactionId xid, int nchildren, TransactionId *children, @@ -216,19 +217,19 @@ static void RecordTransactionAbortPrepared(TransactionId xid, int nstats, xl_xact_stats_item *stats, const char *gid); -static void ProcessRecords(char *bufptr, TransactionId xid, +static void ProcessRecords(char *bufptr, FullTransactionId fxid, const TwoPhaseCallback callbacks[]); static void RemoveGXact(GlobalTransaction gxact); static void XlogReadTwoPhaseData(XLogRecPtr lsn, char **buf, int *len); -static char *ProcessTwoPhaseBuffer(TransactionId xid, +static char *ProcessTwoPhaseBuffer(FullTransactionId fxid, XLogRecPtr prepare_start_lsn, bool fromdisk, bool setParent, bool setNextXid); -static void MarkAsPreparingGuts(GlobalTransaction gxact, TransactionId xid, +static void MarkAsPreparingGuts(GlobalTransaction gxact, FullTransactionId fxid, const char *gid, TimestampTz prepared_at, Oid owner, Oid databaseid); -static void RemoveTwoPhaseFile(TransactionId xid, bool giveWarning); -static void RecreateTwoPhaseFile(TransactionId xid, void *content, int len); +static void RemoveTwoPhaseFile(FullTransactionId fxid, bool giveWarning); +static void RecreateTwoPhaseFile(FullTransactionId fxid, void *content, int len); /* * Initialization of shared memory @@ -356,7 +357,7 @@ PostPrepare_Twophase(void) * Reserve the GID for the given transaction. */ GlobalTransaction -MarkAsPreparing(TransactionId xid, const char *gid, +MarkAsPreparing(FullTransactionId fxid, const char *gid, TimestampTz prepared_at, Oid owner, Oid databaseid) { GlobalTransaction gxact; @@ -407,7 +408,7 @@ MarkAsPreparing(TransactionId xid, const char *gid, gxact = TwoPhaseState->freeGXacts; TwoPhaseState->freeGXacts = gxact->next; - MarkAsPreparingGuts(gxact, xid, gid, prepared_at, owner, databaseid); + MarkAsPreparingGuts(gxact, fxid, gid, prepared_at, owner, databaseid); gxact->ondisk = false; @@ -430,11 +431,13 @@ MarkAsPreparing(TransactionId xid, const char *gid, * Note: This function should be called with appropriate locks held. */ static void -MarkAsPreparingGuts(GlobalTransaction gxact, TransactionId xid, const char *gid, - TimestampTz prepared_at, Oid owner, Oid databaseid) +MarkAsPreparingGuts(GlobalTransaction gxact, FullTransactionId fxid, + const char *gid, TimestampTz prepared_at, Oid owner, + Oid databaseid) { PGPROC *proc; int i; + TransactionId xid = XidFromFullTransactionId(fxid); Assert(LWLockHeldByMeInMode(TwoPhaseStateLock, LW_EXCLUSIVE)); @@ -479,7 +482,7 @@ MarkAsPreparingGuts(GlobalTransaction gxact, TransactionId xid, const char *gid, proc->subxidStatus.count = 0; gxact->prepared_at = prepared_at; - gxact->xid = xid; + gxact->fxid = fxid; gxact->owner = owner; gxact->locking_backend = MyProcNumber; gxact->valid = false; @@ -797,12 +800,12 @@ pg_prepared_xact(PG_FUNCTION_ARGS) * caller had better hold it. */ static GlobalTransaction -TwoPhaseGetGXact(TransactionId xid, bool lock_held) +TwoPhaseGetGXact(FullTransactionId fxid, bool lock_held) { GlobalTransaction result = NULL; int i; - static TransactionId cached_xid = InvalidTransactionId; + static FullTransactionId cached_fxid = {InvalidTransactionId}; static GlobalTransaction cached_gxact = NULL; Assert(!lock_held || LWLockHeldByMe(TwoPhaseStateLock)); @@ -811,7 +814,7 @@ TwoPhaseGetGXact(TransactionId xid, bool lock_held) * During a recovery, COMMIT PREPARED, or ABORT PREPARED, we'll be called * repeatedly for the same XID. We can save work with a simple cache. */ - if (xid == cached_xid) + if (FullTransactionIdEquals(fxid, cached_fxid)) return cached_gxact; if (!lock_held) @@ -821,7 +824,7 @@ TwoPhaseGetGXact(TransactionId xid, bool lock_held) { GlobalTransaction gxact = TwoPhaseState->prepXacts[i]; - if (gxact->xid == xid) + if (FullTransactionIdEquals(gxact->fxid, fxid)) { result = gxact; break; @@ -832,9 +835,10 @@ TwoPhaseGetGXact(TransactionId xid, bool lock_held) LWLockRelease(TwoPhaseStateLock); if (result == NULL) /* should not happen */ - elog(ERROR, "failed to find GlobalTransaction for xid %u", xid); + elog(ERROR, "failed to find GlobalTransaction for xid %u", + XidFromFullTransactionId(fxid)); - cached_xid = xid; + cached_fxid = fxid; cached_gxact = result; return result; @@ -881,7 +885,7 @@ TwoPhaseGetXidByVirtualXID(VirtualTransactionId vxid, *have_more = true; break; } - result = gxact->xid; + result = XidFromFullTransactionId(gxact->fxid); } } @@ -892,7 +896,7 @@ TwoPhaseGetXidByVirtualXID(VirtualTransactionId vxid, /* * TwoPhaseGetDummyProcNumber - * Get the dummy proc number for prepared transaction specified by XID + * Get the dummy proc number for prepared transaction * * Dummy proc numbers are similar to proc numbers of real backends. They * start at MaxBackends, and are unique across all currently active real @@ -900,24 +904,24 @@ TwoPhaseGetXidByVirtualXID(VirtualTransactionId vxid, * TwoPhaseStateLock will not be taken, so the caller had better hold it. */ ProcNumber -TwoPhaseGetDummyProcNumber(TransactionId xid, bool lock_held) +TwoPhaseGetDummyProcNumber(FullTransactionId fxid, bool lock_held) { - GlobalTransaction gxact = TwoPhaseGetGXact(xid, lock_held); + GlobalTransaction gxact = TwoPhaseGetGXact(fxid, lock_held); return gxact->pgprocno; } /* * TwoPhaseGetDummyProc - * Get the PGPROC that represents a prepared transaction specified by XID + * Get the PGPROC that represents a prepared transaction * * If lock_held is set to true, TwoPhaseStateLock will not be taken, so the * caller had better hold it. */ PGPROC * -TwoPhaseGetDummyProc(TransactionId xid, bool lock_held) +TwoPhaseGetDummyProc(FullTransactionId fxid, bool lock_held) { - GlobalTransaction gxact = TwoPhaseGetGXact(xid, lock_held); + GlobalTransaction gxact = TwoPhaseGetGXact(fxid, lock_held); return GetPGProcByNumber(gxact->pgprocno); } @@ -942,10 +946,8 @@ AdjustToFullTransactionId(TransactionId xid) } static inline int -TwoPhaseFilePath(char *path, TransactionId xid) +TwoPhaseFilePath(char *path, FullTransactionId fxid) { - FullTransactionId fxid = AdjustToFullTransactionId(xid); - return snprintf(path, MAXPGPATH, TWOPHASE_DIR "/%08X%08X", EpochFromFullTransactionId(fxid), XidFromFullTransactionId(fxid)); @@ -1049,7 +1051,7 @@ void StartPrepare(GlobalTransaction gxact) { PGPROC *proc = GetPGProcByNumber(gxact->pgprocno); - TransactionId xid = gxact->xid; + TransactionId xid = XidFromFullTransactionId(gxact->fxid); TwoPhaseFileHeader hdr; TransactionId *children; RelFileLocator *commitrels; @@ -1181,7 +1183,11 @@ EndPrepare(GlobalTransaction gxact) * starting immediately after the WAL record is inserted could complete * without fsync'ing our state file. (This is essentially the same kind * of race condition as the COMMIT-to-clog-write case that - * RecordTransactionCommit uses DELAY_CHKPT_START for; see notes there.) + * RecordTransactionCommit uses DELAY_CHKPT_IN_COMMIT for; see notes + * there.) Note that DELAY_CHKPT_IN_COMMIT is used to find transactions in + * the critical commit section. We need to know about such transactions + * for conflict detection in logical replication. See + * GetOldestActiveTransactionId(true, false) and its use. * * We save the PREPARE record's location in the gxact for later use by * CheckPointTwoPhase. @@ -1281,10 +1287,11 @@ RegisterTwoPhaseRecord(TwoPhaseRmgrId rmid, uint16 info, * If it looks OK (has a valid magic number and CRC), return the palloc'd * contents of the file, issuing an error when finding corrupted data. If * missing_ok is true, which indicates that missing files can be safely - * ignored, then return NULL. This state can be reached when doing recovery. + * ignored, then return NULL. This state can be reached when doing recovery + * after discarding two-phase files from frozen epochs. */ static char * -ReadTwoPhaseFile(TransactionId xid, bool missing_ok) +ReadTwoPhaseFile(FullTransactionId fxid, bool missing_ok) { char path[MAXPGPATH]; char *buf; @@ -1296,7 +1303,7 @@ ReadTwoPhaseFile(TransactionId xid, bool missing_ok) file_crc; int r; - TwoPhaseFilePath(path, xid); + TwoPhaseFilePath(path, fxid); fd = OpenTransientFile(path, O_RDONLY | PG_BINARY); if (fd < 0) @@ -1426,12 +1433,12 @@ XlogReadTwoPhaseData(XLogRecPtr lsn, char **buf, int *len) if (errormsg) ereport(ERROR, (errcode_for_file_access(), - errmsg("could not read two-phase state from WAL at %X/%X: %s", + errmsg("could not read two-phase state from WAL at %X/%08X: %s", LSN_FORMAT_ARGS(lsn), errormsg))); else ereport(ERROR, (errcode_for_file_access(), - errmsg("could not read two-phase state from WAL at %X/%X", + errmsg("could not read two-phase state from WAL at %X/%08X", LSN_FORMAT_ARGS(lsn)))); } @@ -1439,7 +1446,7 @@ XlogReadTwoPhaseData(XLogRecPtr lsn, char **buf, int *len) (XLogRecGetInfo(xlogreader) & XLOG_XACT_OPMASK) != XLOG_XACT_PREPARE) ereport(ERROR, (errcode_for_file_access(), - errmsg("expected two-phase state data is not present in WAL at %X/%X", + errmsg("expected two-phase state data is not present in WAL at %X/%08X", LSN_FORMAT_ARGS(lsn)))); if (len != NULL) @@ -1461,6 +1468,7 @@ StandbyTransactionIdIsPrepared(TransactionId xid) char *buf; TwoPhaseFileHeader *hdr; bool result; + FullTransactionId fxid; Assert(TransactionIdIsValid(xid)); @@ -1468,7 +1476,8 @@ StandbyTransactionIdIsPrepared(TransactionId xid) return false; /* nothing to do */ /* Read and validate file */ - buf = ReadTwoPhaseFile(xid, true); + fxid = AdjustToFullTransactionId(xid); + buf = ReadTwoPhaseFile(fxid, true); if (buf == NULL) return false; @@ -1488,6 +1497,7 @@ FinishPreparedTransaction(const char *gid, bool isCommit) { GlobalTransaction gxact; PGPROC *proc; + FullTransactionId fxid; TransactionId xid; bool ondisk; char *buf; @@ -1509,7 +1519,8 @@ FinishPreparedTransaction(const char *gid, bool isCommit) */ gxact = LockGXact(gid, GetUserId()); proc = GetPGProcByNumber(gxact->pgprocno); - xid = gxact->xid; + fxid = gxact->fxid; + xid = XidFromFullTransactionId(fxid); /* * Read and validate 2PC state data. State data will typically be stored @@ -1517,7 +1528,7 @@ FinishPreparedTransaction(const char *gid, bool isCommit) * to disk if for some reason they have lived for a long time. */ if (gxact->ondisk) - buf = ReadTwoPhaseFile(xid, false); + buf = ReadTwoPhaseFile(fxid, false); else XlogReadTwoPhaseData(gxact->prepare_start_lsn, &buf, NULL); @@ -1636,11 +1647,11 @@ FinishPreparedTransaction(const char *gid, bool isCommit) /* And now do the callbacks */ if (isCommit) - ProcessRecords(bufptr, xid, twophase_postcommit_callbacks); + ProcessRecords(bufptr, fxid, twophase_postcommit_callbacks); else - ProcessRecords(bufptr, xid, twophase_postabort_callbacks); + ProcessRecords(bufptr, fxid, twophase_postabort_callbacks); - PredicateLockTwoPhaseFinish(xid, isCommit); + PredicateLockTwoPhaseFinish(fxid, isCommit); /* * Read this value while holding the two-phase lock, as the on-disk 2PC @@ -1664,7 +1675,7 @@ FinishPreparedTransaction(const char *gid, bool isCommit) * And now we can clean up any files we may have left. */ if (ondisk) - RemoveTwoPhaseFile(xid, true); + RemoveTwoPhaseFile(fxid, true); MyLockedGxact = NULL; @@ -1677,7 +1688,7 @@ FinishPreparedTransaction(const char *gid, bool isCommit) * Scan 2PC state data in memory and call the indicated callbacks for each 2PC record. */ static void -ProcessRecords(char *bufptr, TransactionId xid, +ProcessRecords(char *bufptr, FullTransactionId fxid, const TwoPhaseCallback callbacks[]) { for (;;) @@ -1691,24 +1702,28 @@ ProcessRecords(char *bufptr, TransactionId xid, bufptr += MAXALIGN(sizeof(TwoPhaseRecordOnDisk)); if (callbacks[record->rmid] != NULL) - callbacks[record->rmid] (xid, record->info, bufptr, record->len); + callbacks[record->rmid] (fxid, record->info, bufptr, record->len); bufptr += MAXALIGN(record->len); } } /* - * Remove the 2PC file for the specified XID. + * Remove the 2PC file. * * If giveWarning is false, do not complain about file-not-present; * this is an expected case during WAL replay. + * + * This routine is used at early stages at recovery where future and + * past orphaned files are checked, hence the FullTransactionId to build + * a complete file name fit for the removal. */ static void -RemoveTwoPhaseFile(TransactionId xid, bool giveWarning) +RemoveTwoPhaseFile(FullTransactionId fxid, bool giveWarning) { char path[MAXPGPATH]; - TwoPhaseFilePath(path, xid); + TwoPhaseFilePath(path, fxid); if (unlink(path)) if (errno != ENOENT || giveWarning) ereport(WARNING, @@ -1723,7 +1738,7 @@ RemoveTwoPhaseFile(TransactionId xid, bool giveWarning) * Note: content and len don't include CRC. */ static void -RecreateTwoPhaseFile(TransactionId xid, void *content, int len) +RecreateTwoPhaseFile(FullTransactionId fxid, void *content, int len) { char path[MAXPGPATH]; pg_crc32c statefile_crc; @@ -1734,7 +1749,7 @@ RecreateTwoPhaseFile(TransactionId xid, void *content, int len) COMP_CRC32C(statefile_crc, content, len); FIN_CRC32C(statefile_crc); - TwoPhaseFilePath(path, xid); + TwoPhaseFilePath(path, fxid); fd = OpenTransientFile(path, O_CREAT | O_TRUNC | O_WRONLY | PG_BINARY); @@ -1846,7 +1861,7 @@ CheckPointTwoPhase(XLogRecPtr redo_horizon) int len; XlogReadTwoPhaseData(gxact->prepare_start_lsn, &buf, &len); - RecreateTwoPhaseFile(gxact->xid, buf, len); + RecreateTwoPhaseFile(gxact->fxid, buf, len); gxact->ondisk = true; gxact->prepare_start_lsn = InvalidXLogRecPtr; gxact->prepare_end_lsn = InvalidXLogRecPtr; @@ -1897,19 +1912,17 @@ restoreTwoPhaseData(void) if (strlen(clde->d_name) == 16 && strspn(clde->d_name, "0123456789ABCDEF") == 16) { - TransactionId xid; FullTransactionId fxid; char *buf; fxid = FullTransactionIdFromU64(strtou64(clde->d_name, NULL, 16)); - xid = XidFromFullTransactionId(fxid); - buf = ProcessTwoPhaseBuffer(xid, InvalidXLogRecPtr, + buf = ProcessTwoPhaseBuffer(fxid, InvalidXLogRecPtr, true, false, false); if (buf == NULL) continue; - PrepareRedoAdd(buf, InvalidXLogRecPtr, + PrepareRedoAdd(fxid, buf, InvalidXLogRecPtr, InvalidXLogRecPtr, InvalidRepOriginId); } } @@ -1968,9 +1981,7 @@ PrescanPreparedTransactions(TransactionId **xids_p, int *nxids_p) Assert(gxact->inredo); - xid = gxact->xid; - - buf = ProcessTwoPhaseBuffer(xid, + buf = ProcessTwoPhaseBuffer(gxact->fxid, gxact->prepare_start_lsn, gxact->ondisk, false, true); @@ -1981,6 +1992,7 @@ PrescanPreparedTransactions(TransactionId **xids_p, int *nxids_p) * OK, we think this file is valid. Incorporate xid into the * running-minimum result. */ + xid = XidFromFullTransactionId(gxact->fxid); if (TransactionIdPrecedes(xid, result)) result = xid; @@ -2036,15 +2048,12 @@ StandbyRecoverPreparedTransactions(void) LWLockAcquire(TwoPhaseStateLock, LW_EXCLUSIVE); for (i = 0; i < TwoPhaseState->numPrepXacts; i++) { - TransactionId xid; char *buf; GlobalTransaction gxact = TwoPhaseState->prepXacts[i]; Assert(gxact->inredo); - xid = gxact->xid; - - buf = ProcessTwoPhaseBuffer(xid, + buf = ProcessTwoPhaseBuffer(gxact->fxid, gxact->prepare_start_lsn, gxact->ondisk, true, false); if (buf != NULL) @@ -2077,16 +2086,14 @@ RecoverPreparedTransactions(void) LWLockAcquire(TwoPhaseStateLock, LW_EXCLUSIVE); for (i = 0; i < TwoPhaseState->numPrepXacts; i++) { - TransactionId xid; char *buf; GlobalTransaction gxact = TwoPhaseState->prepXacts[i]; + FullTransactionId fxid = gxact->fxid; char *bufptr; TwoPhaseFileHeader *hdr; TransactionId *subxids; const char *gid; - xid = gxact->xid; - /* * Reconstruct subtrans state for the transaction --- needed because * pg_subtrans is not preserved over a restart. Note that we are @@ -2096,17 +2103,20 @@ RecoverPreparedTransactions(void) * SubTransSetParent has been set before, if the prepared transaction * generated xid assignment records. */ - buf = ProcessTwoPhaseBuffer(xid, + buf = ProcessTwoPhaseBuffer(gxact->fxid, gxact->prepare_start_lsn, gxact->ondisk, true, false); if (buf == NULL) continue; ereport(LOG, - (errmsg("recovering prepared transaction %u from shared memory", xid))); + (errmsg("recovering prepared transaction %u of epoch %u from shared memory", + XidFromFullTransactionId(gxact->fxid), + EpochFromFullTransactionId(gxact->fxid)))); hdr = (TwoPhaseFileHeader *) buf; - Assert(TransactionIdEquals(hdr->xid, xid)); + Assert(TransactionIdEquals(hdr->xid, + XidFromFullTransactionId(gxact->fxid))); bufptr = buf + MAXALIGN(sizeof(TwoPhaseFileHeader)); gid = (const char *) bufptr; bufptr += MAXALIGN(hdr->gidlen); @@ -2122,7 +2132,7 @@ RecoverPreparedTransactions(void) * Recreate its GXACT and dummy PGPROC. But, check whether it was * added in redo and already has a shmem entry for it. */ - MarkAsPreparingGuts(gxact, xid, gid, + MarkAsPreparingGuts(gxact, gxact->fxid, gid, hdr->prepared_at, hdr->owner, hdr->database); @@ -2137,7 +2147,7 @@ RecoverPreparedTransactions(void) /* * Recover other state (notably locks) using resource managers. */ - ProcessRecords(bufptr, xid, twophase_recover_callbacks); + ProcessRecords(bufptr, fxid, twophase_recover_callbacks); /* * Release locks held by the standby process after we process each @@ -2145,7 +2155,7 @@ RecoverPreparedTransactions(void) * additional locks at any one time. */ if (InHotStandby) - StandbyReleaseLockTree(xid, hdr->nsubxacts, subxids); + StandbyReleaseLockTree(hdr->xid, hdr->nsubxacts, subxids); /* * We're done with recovering this transaction. Clear MyLockedGxact, @@ -2164,7 +2174,7 @@ RecoverPreparedTransactions(void) /* * ProcessTwoPhaseBuffer * - * Given a transaction id, read it either from disk or read it directly + * Given a FullTransactionId, read it either from disk or read it directly * via shmem xlog record pointer using the provided "prepare_start_lsn". * * If setParent is true, set up subtransaction parent linkages. @@ -2173,13 +2183,12 @@ RecoverPreparedTransactions(void) * value scanned. */ static char * -ProcessTwoPhaseBuffer(TransactionId xid, +ProcessTwoPhaseBuffer(FullTransactionId fxid, XLogRecPtr prepare_start_lsn, bool fromdisk, bool setParent, bool setNextXid) { FullTransactionId nextXid = TransamVariables->nextXid; - TransactionId origNextXid = XidFromFullTransactionId(nextXid); TransactionId *subxids; char *buf; TwoPhaseFileHeader *hdr; @@ -2191,41 +2200,46 @@ ProcessTwoPhaseBuffer(TransactionId xid, Assert(prepare_start_lsn != InvalidXLogRecPtr); /* Already processed? */ - if (TransactionIdDidCommit(xid) || TransactionIdDidAbort(xid)) + if (TransactionIdDidCommit(XidFromFullTransactionId(fxid)) || + TransactionIdDidAbort(XidFromFullTransactionId(fxid))) { if (fromdisk) { ereport(WARNING, - (errmsg("removing stale two-phase state file for transaction %u", - xid))); - RemoveTwoPhaseFile(xid, true); + (errmsg("removing stale two-phase state file for transaction %u of epoch %u", + XidFromFullTransactionId(fxid), + EpochFromFullTransactionId(fxid)))); + RemoveTwoPhaseFile(fxid, true); } else { ereport(WARNING, - (errmsg("removing stale two-phase state from memory for transaction %u", - xid))); - PrepareRedoRemove(xid, true); + (errmsg("removing stale two-phase state from memory for transaction %u of epoch %u", + XidFromFullTransactionId(fxid), + EpochFromFullTransactionId(fxid)))); + PrepareRedoRemoveFull(fxid, true); } return NULL; } /* Reject XID if too new */ - if (TransactionIdFollowsOrEquals(xid, origNextXid)) + if (FullTransactionIdFollowsOrEquals(fxid, nextXid)) { if (fromdisk) { ereport(WARNING, - (errmsg("removing future two-phase state file for transaction %u", - xid))); - RemoveTwoPhaseFile(xid, true); + (errmsg("removing future two-phase state file for transaction %u of epoch %u", + XidFromFullTransactionId(fxid), + EpochFromFullTransactionId(fxid)))); + RemoveTwoPhaseFile(fxid, true); } else { ereport(WARNING, - (errmsg("removing future two-phase state from memory for transaction %u", - xid))); - PrepareRedoRemove(xid, true); + (errmsg("removing future two-phase state from memory for transaction %u of epoch %u", + XidFromFullTransactionId(fxid), + EpochFromFullTransactionId(fxid)))); + PrepareRedoRemoveFull(fxid, true); } return NULL; } @@ -2233,7 +2247,7 @@ ProcessTwoPhaseBuffer(TransactionId xid, if (fromdisk) { /* Read and validate file */ - buf = ReadTwoPhaseFile(xid, false); + buf = ReadTwoPhaseFile(fxid, false); } else { @@ -2243,18 +2257,20 @@ ProcessTwoPhaseBuffer(TransactionId xid, /* Deconstruct header */ hdr = (TwoPhaseFileHeader *) buf; - if (!TransactionIdEquals(hdr->xid, xid)) + if (!TransactionIdEquals(hdr->xid, XidFromFullTransactionId(fxid))) { if (fromdisk) ereport(ERROR, (errcode(ERRCODE_DATA_CORRUPTED), - errmsg("corrupted two-phase state file for transaction %u", - xid))); + errmsg("corrupted two-phase state file for transaction %u of epoch %u", + XidFromFullTransactionId(fxid), + EpochFromFullTransactionId(fxid)))); else ereport(ERROR, (errcode(ERRCODE_DATA_CORRUPTED), - errmsg("corrupted two-phase state in memory for transaction %u", - xid))); + errmsg("corrupted two-phase state in memory for transaction %u of epoch %u", + XidFromFullTransactionId(fxid), + EpochFromFullTransactionId(fxid)))); } /* @@ -2268,14 +2284,14 @@ ProcessTwoPhaseBuffer(TransactionId xid, { TransactionId subxid = subxids[i]; - Assert(TransactionIdFollows(subxid, xid)); + Assert(TransactionIdFollows(subxid, XidFromFullTransactionId(fxid))); /* update nextXid if needed */ if (setNextXid) AdvanceNextFullTransactionIdPastXid(subxid); if (setParent) - SubTransSetParent(subxid, xid); + SubTransSetParent(subxid, XidFromFullTransactionId(fxid)); } return buf; @@ -2286,7 +2302,7 @@ ProcessTwoPhaseBuffer(TransactionId xid, * RecordTransactionCommitPrepared * * This is basically the same as RecordTransactionCommit (q.v. if you change - * this function): in particular, we must set DELAY_CHKPT_START to avoid a + * this function): in particular, we must set DELAY_CHKPT_IN_COMMIT to avoid a * race condition. * * We know the transaction made at least one XLOG entry (its PREPARE), @@ -2306,7 +2322,7 @@ RecordTransactionCommitPrepared(TransactionId xid, const char *gid) { XLogRecPtr recptr; - TimestampTz committs = GetCurrentTimestamp(); + TimestampTz committs; bool replorigin; /* @@ -2319,8 +2335,24 @@ RecordTransactionCommitPrepared(TransactionId xid, START_CRIT_SECTION(); /* See notes in RecordTransactionCommit */ - Assert((MyProc->delayChkptFlags & DELAY_CHKPT_START) == 0); - MyProc->delayChkptFlags |= DELAY_CHKPT_START; + Assert((MyProc->delayChkptFlags & DELAY_CHKPT_IN_COMMIT) == 0); + MyProc->delayChkptFlags |= DELAY_CHKPT_IN_COMMIT; + + /* + * Ensures the DELAY_CHKPT_IN_COMMIT flag write is globally visible before + * commit time is written. + */ + pg_write_barrier(); + + /* + * Note it is important to set committs value after marking ourselves as + * in the commit critical section (DELAY_CHKPT_IN_COMMIT). This is because + * we want to ensure all transactions that have acquired commit timestamp + * are finished before we allow the logical replication client to advance + * its xid which is used to hold back dead rows for conflict detection. + * See comments atop worker.c. + */ + committs = GetCurrentTimestamp(); /* * Emit the XLOG commit record. Note that we mark 2PC commits as @@ -2369,7 +2401,7 @@ RecordTransactionCommitPrepared(TransactionId xid, TransactionIdCommitTree(xid, nchildren, children); /* Checkpoint can proceed now */ - MyProc->delayChkptFlags &= ~DELAY_CHKPT_START; + MyProc->delayChkptFlags &= ~DELAY_CHKPT_IN_COMMIT; END_CRIT_SECTION(); @@ -2466,8 +2498,9 @@ RecordTransactionAbortPrepared(TransactionId xid, * data, the entry is marked as located on disk. */ void -PrepareRedoAdd(char *buf, XLogRecPtr start_lsn, - XLogRecPtr end_lsn, RepOriginId origin_id) +PrepareRedoAdd(FullTransactionId fxid, char *buf, + XLogRecPtr start_lsn, XLogRecPtr end_lsn, + RepOriginId origin_id) { TwoPhaseFileHeader *hdr = (TwoPhaseFileHeader *) buf; char *bufptr; @@ -2477,6 +2510,13 @@ PrepareRedoAdd(char *buf, XLogRecPtr start_lsn, Assert(LWLockHeldByMeInMode(TwoPhaseStateLock, LW_EXCLUSIVE)); Assert(RecoveryInProgress()); + if (!FullTransactionIdIsValid(fxid)) + { + Assert(InRecovery); + fxid = FullTransactionIdFromAllowableAt(TransamVariables->nextXid, + hdr->xid); + } + bufptr = buf + MAXALIGN(sizeof(TwoPhaseFileHeader)); gid = (const char *) bufptr; @@ -2505,14 +2545,15 @@ PrepareRedoAdd(char *buf, XLogRecPtr start_lsn, { char path[MAXPGPATH]; - TwoPhaseFilePath(path, hdr->xid); + Assert(InRecovery); + TwoPhaseFilePath(path, fxid); if (access(path, F_OK) == 0) { ereport(reachedConsistency ? ERROR : WARNING, (errmsg("could not recover two-phase state file for transaction %u", hdr->xid), - errdetail("Two-phase state file has been found in WAL record %X/%X, but this transaction has already been restored from disk.", + errdetail("Two-phase state file has been found in WAL record %X/%08X, but this transaction has already been restored from disk.", LSN_FORMAT_ARGS(start_lsn)))); return; } @@ -2536,7 +2577,7 @@ PrepareRedoAdd(char *buf, XLogRecPtr start_lsn, gxact->prepared_at = hdr->prepared_at; gxact->prepare_start_lsn = start_lsn; gxact->prepare_end_lsn = end_lsn; - gxact->xid = hdr->xid; + gxact->fxid = fxid; gxact->owner = hdr->owner; gxact->locking_backend = INVALID_PROC_NUMBER; gxact->valid = false; @@ -2555,11 +2596,13 @@ PrepareRedoAdd(char *buf, XLogRecPtr start_lsn, false /* backward */ , false /* WAL */ ); } - elog(DEBUG2, "added 2PC data in shared memory for transaction %u", gxact->xid); + elog(DEBUG2, "added 2PC data in shared memory for transaction %u of epoch %u", + XidFromFullTransactionId(gxact->fxid), + EpochFromFullTransactionId(gxact->fxid)); } /* - * PrepareRedoRemove + * PrepareRedoRemoveFull * * Remove the corresponding gxact entry from TwoPhaseState. Also remove * the 2PC file if a prepared transaction was saved via an earlier checkpoint. @@ -2567,8 +2610,8 @@ PrepareRedoAdd(char *buf, XLogRecPtr start_lsn, * Caller must hold TwoPhaseStateLock in exclusive mode, because TwoPhaseState * is updated. */ -void -PrepareRedoRemove(TransactionId xid, bool giveWarning) +static void +PrepareRedoRemoveFull(FullTransactionId fxid, bool giveWarning) { GlobalTransaction gxact = NULL; int i; @@ -2581,7 +2624,7 @@ PrepareRedoRemove(TransactionId xid, bool giveWarning) { gxact = TwoPhaseState->prepXacts[i]; - if (gxact->xid == xid) + if (FullTransactionIdEquals(gxact->fxid, fxid)) { Assert(gxact->inredo); found = true; @@ -2598,13 +2641,29 @@ PrepareRedoRemove(TransactionId xid, bool giveWarning) /* * And now we can clean up any files we may have left. */ - elog(DEBUG2, "removing 2PC data for transaction %u", xid); + elog(DEBUG2, "removing 2PC data for transaction %u of epoch %u ", + XidFromFullTransactionId(fxid), + EpochFromFullTransactionId(fxid)); + if (gxact->ondisk) - RemoveTwoPhaseFile(xid, giveWarning); + RemoveTwoPhaseFile(fxid, giveWarning); + RemoveGXact(gxact); } /* + * Wrapper of PrepareRedoRemoveFull(), for TransactionIds. + */ +void +PrepareRedoRemove(TransactionId xid, bool giveWarning) +{ + FullTransactionId fxid = + FullTransactionIdFromAllowableAt(TransamVariables->nextXid, xid); + + PrepareRedoRemoveFull(fxid, giveWarning); +} + +/* * LookupGXact * Check if the prepared transaction with the given GID, lsn and timestamp * exists. @@ -2648,7 +2707,7 @@ LookupGXact(const char *gid, XLogRecPtr prepare_end_lsn, * between publisher and subscriber. */ if (gxact->ondisk) - buf = ReadTwoPhaseFile(gxact->xid, false); + buf = ReadTwoPhaseFile(gxact->fxid, false); else { Assert(gxact->prepare_start_lsn); diff --git a/src/backend/access/transam/xact.c b/src/backend/access/transam/xact.c index b885513f765..b46e7e9c2a6 100644 --- a/src/backend/access/transam/xact.c +++ b/src/backend/access/transam/xact.c @@ -1431,10 +1431,22 @@ RecordTransactionCommit(void) * without holding the ProcArrayLock, since we're the only one * modifying it. This makes checkpoint's determination of which xacts * are delaying the checkpoint a bit fuzzy, but it doesn't matter. + * + * Note, it is important to get the commit timestamp after marking the + * transaction in the commit critical section. See + * RecordTransactionCommitPrepared. */ - Assert((MyProc->delayChkptFlags & DELAY_CHKPT_START) == 0); + Assert((MyProc->delayChkptFlags & DELAY_CHKPT_IN_COMMIT) == 0); START_CRIT_SECTION(); - MyProc->delayChkptFlags |= DELAY_CHKPT_START; + MyProc->delayChkptFlags |= DELAY_CHKPT_IN_COMMIT; + + Assert(xactStopTimestamp == 0); + + /* + * Ensures the DELAY_CHKPT_IN_COMMIT flag write is globally visible + * before commit time is written. + */ + pg_write_barrier(); /* * Insert the commit XLOG record. @@ -1537,7 +1549,7 @@ RecordTransactionCommit(void) */ if (markXidCommitted) { - MyProc->delayChkptFlags &= ~DELAY_CHKPT_START; + MyProc->delayChkptFlags &= ~DELAY_CHKPT_IN_COMMIT; END_CRIT_SECTION(); } @@ -2515,7 +2527,7 @@ static void PrepareTransaction(void) { TransactionState s = CurrentTransactionState; - TransactionId xid = GetCurrentTransactionId(); + FullTransactionId fxid = GetCurrentFullTransactionId(); GlobalTransaction gxact; TimestampTz prepared_at; @@ -2644,7 +2656,7 @@ PrepareTransaction(void) * Reserve the GID for this transaction. This could fail if the requested * GID is invalid or already in use. */ - gxact = MarkAsPreparing(xid, prepareGID, prepared_at, + gxact = MarkAsPreparing(fxid, prepareGID, prepared_at, GetUserId(), MyDatabaseId); prepareGID = NULL; @@ -2694,7 +2706,7 @@ PrepareTransaction(void) * ProcArrayClearTransaction(). Otherwise, a GetLockConflicts() would * conclude "xact already committed or aborted" for our locks. */ - PostPrepare_Locks(xid); + PostPrepare_Locks(fxid); /* * Let others know about no transaction in progress by me. This has to be @@ -2738,9 +2750,9 @@ PrepareTransaction(void) PostPrepare_smgr(); - PostPrepare_MultiXact(xid); + PostPrepare_MultiXact(fxid); - PostPrepare_PredicateLocks(xid); + PostPrepare_PredicateLocks(fxid); ResourceOwnerRelease(TopTransactionResourceOwner, RESOURCE_RELEASE_LOCKS, @@ -6420,7 +6432,8 @@ xact_redo(XLogReaderState *record) * gxact entry. */ LWLockAcquire(TwoPhaseStateLock, LW_EXCLUSIVE); - PrepareRedoAdd(XLogRecGetData(record), + PrepareRedoAdd(InvalidFullTransactionId, + XLogRecGetData(record), record->ReadRecPtr, record->EndRecPtr, XLogRecGetOrigin(record)); diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c index 1914859b2ee..9a4de1616bc 100644 --- a/src/backend/access/transam/xlog.c +++ b/src/backend/access/transam/xlog.c @@ -96,6 +96,7 @@ #include "utils/guc_hooks.h" #include "utils/guc_tables.h" #include "utils/injection_point.h" +#include "utils/pgstat_internal.h" #include "utils/ps_status.h" #include "utils/relmapper.h" #include "utils/snapmgr.h" @@ -449,7 +450,6 @@ typedef struct XLogCtlData /* Protected by info_lck: */ XLogwrtRqst LogwrtRqst; XLogRecPtr RedoRecPtr; /* a recent copy of Insert->RedoRecPtr */ - FullTransactionId ckptFullXid; /* nextXid of latest checkpoint */ XLogRecPtr asyncXactLSN; /* LSN of newest async commit/abort */ XLogRecPtr replicationSlotMinLSN; /* oldest LSN needed by any slot */ @@ -703,7 +703,7 @@ static void InitControlFile(uint64 sysidentifier, uint32 data_checksum_version); static void WriteControlFile(void); static void ReadControlFile(void); static void UpdateControlFile(void); -static char *str_time(pg_time_t tnow); +static char *str_time(pg_time_t tnow, char *buf, size_t bufsize); static int get_sync_bit(int method); @@ -1028,7 +1028,7 @@ XLogInsertRecord(XLogRecData *rdata, oldCxt = MemoryContextSwitchTo(walDebugCxt); initStringInfo(&buf); - appendStringInfo(&buf, "INSERT @ %X/%X: ", LSN_FORMAT_ARGS(EndPos)); + appendStringInfo(&buf, "INSERT @ %X/%08X: ", LSN_FORMAT_ARGS(EndPos)); /* * We have to piece together the WAL record data from the XLogRecData @@ -1092,6 +1092,9 @@ XLogInsertRecord(XLogRecData *rdata, pgWalUsage.wal_bytes += rechdr->xl_tot_len; pgWalUsage.wal_records++; pgWalUsage.wal_fpi += num_fpi; + + /* Required for the flush of pending stats WAL data */ + pgstat_report_fixed = true; } return EndPos; @@ -1549,8 +1552,8 @@ WaitXLogInsertionsToFinish(XLogRecPtr upto) if (upto > reservedUpto) { ereport(LOG, - (errmsg("request to flush past end of generated WAL; request %X/%X, current position %X/%X", - LSN_FORMAT_ARGS(upto), LSN_FORMAT_ARGS(reservedUpto)))); + errmsg("request to flush past end of generated WAL; request %X/%08X, current position %X/%08X", + LSN_FORMAT_ARGS(upto), LSN_FORMAT_ARGS(reservedUpto))); upto = reservedUpto; } @@ -1716,7 +1719,7 @@ GetXLogBuffer(XLogRecPtr ptr, TimeLineID tli) endptr = pg_atomic_read_u64(&XLogCtl->xlblocks[idx]); if (expectedEndPtr != endptr) - elog(PANIC, "could not find WAL buffer for %X/%X", + elog(PANIC, "could not find WAL buffer for %X/%08X", LSN_FORMAT_ARGS(ptr)); } else @@ -1776,7 +1779,7 @@ WALReadFromBuffers(char *dstbuf, XLogRecPtr startptr, Size count, inserted = pg_atomic_read_u64(&XLogCtl->logInsertResult); if (startptr + count > inserted) ereport(ERROR, - errmsg("cannot read past end of generated WAL: requested %X/%X, current position %X/%X", + errmsg("cannot read past end of generated WAL: requested %X/%08X, current position %X/%08X", LSN_FORMAT_ARGS(startptr + count), LSN_FORMAT_ARGS(inserted))); @@ -2109,6 +2112,12 @@ AdvanceXLInsertBuffer(XLogRecPtr upto, TimeLineID tli, bool opportunistic) LWLockRelease(WALWriteLock); pgWalUsage.wal_buffers_full++; TRACE_POSTGRESQL_WAL_BUFFER_WRITE_DIRTY_DONE(); + + /* + * Required for the flush of pending stats WAL data, per + * update of pgWalUsage. + */ + pgstat_report_fixed = true; } } } @@ -2281,7 +2290,7 @@ AdvanceXLInsertBuffer(XLogRecPtr upto, TimeLineID tli, bool opportunistic) #ifdef WAL_DEBUG if (XLOG_DEBUG && npages > 0) { - elog(DEBUG1, "initialized %d pages, up to %X/%X", + elog(DEBUG1, "initialized %d pages, up to %X/%08X", npages, LSN_FORMAT_ARGS(NewPageEndPtr)); } #endif @@ -2347,25 +2356,6 @@ check_wal_segment_size(int *newval, void **extra, GucSource source) } /* - * GUC check_hook for max_slot_wal_keep_size - * - * We don't allow the value of max_slot_wal_keep_size other than -1 during the - * binary upgrade. See start_postmaster() in pg_upgrade for more details. - */ -bool -check_max_slot_wal_keep_size(int *newval, void **extra, GucSource source) -{ - if (IsBinaryUpgrade && *newval != -1) - { - GUC_check_errdetail("\"%s\" must be set to -1 during binary upgrade mode.", - "max_slot_wal_keep_size"); - return false; - } - - return true; -} - -/* * At a checkpoint, how many WAL segments to recycle as preallocated future * XLOG segments? Returns the highest segment that should be preallocated. */ @@ -2492,7 +2482,7 @@ XLogWrite(XLogwrtRqst WriteRqst, TimeLineID tli, bool flexible) XLogRecPtr EndPtr = pg_atomic_read_u64(&XLogCtl->xlblocks[curridx]); if (LogwrtResult.Write >= EndPtr) - elog(PANIC, "xlog write request %X/%X is past end of log %X/%X", + elog(PANIC, "xlog write request %X/%08X is past end of log %X/%08X", LSN_FORMAT_ARGS(LogwrtResult.Write), LSN_FORMAT_ARGS(EndPtr)); @@ -2892,7 +2882,7 @@ UpdateMinRecoveryPoint(XLogRecPtr lsn, bool force) newMinRecoveryPoint = GetCurrentReplayRecPtr(&newMinRecoveryPointTLI); if (!force && newMinRecoveryPoint < lsn) elog(WARNING, - "xlog min recovery request %X/%X is past current point %X/%X", + "xlog min recovery request %X/%08X is past current point %X/%08X", LSN_FORMAT_ARGS(lsn), LSN_FORMAT_ARGS(newMinRecoveryPoint)); /* update control file */ @@ -2905,9 +2895,9 @@ UpdateMinRecoveryPoint(XLogRecPtr lsn, bool force) LocalMinRecoveryPointTLI = newMinRecoveryPointTLI; ereport(DEBUG2, - (errmsg_internal("updated min recovery point to %X/%X on timeline %u", - LSN_FORMAT_ARGS(newMinRecoveryPoint), - newMinRecoveryPointTLI))); + errmsg_internal("updated min recovery point to %X/%08X on timeline %u", + LSN_FORMAT_ARGS(newMinRecoveryPoint), + newMinRecoveryPointTLI)); } } LWLockRelease(ControlFileLock); @@ -2945,7 +2935,7 @@ XLogFlush(XLogRecPtr record) #ifdef WAL_DEBUG if (XLOG_DEBUG) - elog(LOG, "xlog flush request %X/%X; write %X/%X; flush %X/%X", + elog(LOG, "xlog flush request %X/%08X; write %X/%08X; flush %X/%08X", LSN_FORMAT_ARGS(record), LSN_FORMAT_ARGS(LogwrtResult.Write), LSN_FORMAT_ARGS(LogwrtResult.Flush)); @@ -3078,7 +3068,7 @@ XLogFlush(XLogRecPtr record) */ if (LogwrtResult.Flush < record) elog(ERROR, - "xlog flush request %X/%X is not satisfied --- flushed only to %X/%X", + "xlog flush request %X/%08X is not satisfied --- flushed only to %X/%08X", LSN_FORMAT_ARGS(record), LSN_FORMAT_ARGS(LogwrtResult.Flush)); } @@ -3205,7 +3195,7 @@ XLogBackgroundFlush(void) #ifdef WAL_DEBUG if (XLOG_DEBUG) - elog(LOG, "xlog bg flush request write %X/%X; flush: %X/%X, current is write %X/%X; flush %X/%X", + elog(LOG, "xlog bg flush request write %X/%08X; flush: %X/%08X, current is write %X/%08X; flush %X/%08X", LSN_FORMAT_ARGS(WriteRqst.Write), LSN_FORMAT_ARGS(WriteRqst.Flush), LSN_FORMAT_ARGS(LogwrtResult.Write), @@ -5381,11 +5371,9 @@ BootStrapXLOG(uint32 data_checksum_version) } static char * -str_time(pg_time_t tnow) +str_time(pg_time_t tnow, char *buf, size_t bufsize) { - char *buf = palloc(128); - - pg_strftime(buf, 128, + pg_strftime(buf, bufsize, "%Y-%m-%d %H:%M:%S %Z", pg_localtime(&tnow, log_timezone)); @@ -5628,6 +5616,7 @@ StartupXLOG(void) XLogRecPtr missingContrecPtr; TransactionId oldestActiveXID; bool promoted = false; + char timebuf[128]; /* * We should have an aux process resource owner to use, and we should not @@ -5656,25 +5645,29 @@ StartupXLOG(void) */ ereport(IsPostmasterEnvironment ? LOG : NOTICE, (errmsg("database system was shut down at %s", - str_time(ControlFile->time)))); + str_time(ControlFile->time, + timebuf, sizeof(timebuf))))); break; case DB_SHUTDOWNED_IN_RECOVERY: ereport(LOG, (errmsg("database system was shut down in recovery at %s", - str_time(ControlFile->time)))); + str_time(ControlFile->time, + timebuf, sizeof(timebuf))))); break; case DB_SHUTDOWNING: ereport(LOG, (errmsg("database system shutdown was interrupted; last known up at %s", - str_time(ControlFile->time)))); + str_time(ControlFile->time, + timebuf, sizeof(timebuf))))); break; case DB_IN_CRASH_RECOVERY: ereport(LOG, (errmsg("database system was interrupted while in recovery at %s", - str_time(ControlFile->time)), + str_time(ControlFile->time, + timebuf, sizeof(timebuf))), errhint("This probably means that some data is corrupted and" " you will have to use the last backup for recovery."))); break; @@ -5682,7 +5675,8 @@ StartupXLOG(void) case DB_IN_ARCHIVE_RECOVERY: ereport(LOG, (errmsg("database system was interrupted while in recovery at log time %s", - str_time(ControlFile->checkPointCopy.time)), + str_time(ControlFile->checkPointCopy.time, + timebuf, sizeof(timebuf))), errhint("If this has occurred more than once some data might be corrupted" " and you might need to choose an earlier recovery target."))); break; @@ -5690,7 +5684,8 @@ StartupXLOG(void) case DB_IN_PRODUCTION: ereport(LOG, (errmsg("database system was interrupted; last known up at %s", - str_time(ControlFile->time)))); + str_time(ControlFile->time, + timebuf, sizeof(timebuf))))); break; default: @@ -5763,7 +5758,6 @@ StartupXLOG(void) SetMultiXactIdLimit(checkPoint.oldestMulti, checkPoint.oldestMultiDB, true); SetCommitTsLimit(checkPoint.oldestCommitTsXid, checkPoint.newestCommitTsXid); - XLogCtl->ckptFullXid = checkPoint.nextXid; /* * Clear out any old relcache cache files. This is *necessary* if we do @@ -6336,6 +6330,12 @@ StartupXLOG(void) */ CompleteCommitTsInitialization(); + /* Clean up EndOfWalRecoveryInfo data to appease Valgrind leak checking */ + if (endOfRecoveryInfo->lastPage) + pfree(endOfRecoveryInfo->lastPage); + pfree(endOfRecoveryInfo->recoveryStopReason); + pfree(endOfRecoveryInfo); + /* * All done with end-of-recovery actions. * @@ -6505,7 +6505,7 @@ PerformRecoveryXLogAction(void) else { RequestCheckpoint(CHECKPOINT_END_OF_RECOVERY | - CHECKPOINT_IMMEDIATE | + CHECKPOINT_FAST | CHECKPOINT_WAIT); } @@ -6814,7 +6814,7 @@ ShutdownXLOG(int code, Datum arg) WalSndWaitStopping(); if (RecoveryInProgress()) - CreateRestartPoint(CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_IMMEDIATE); + CreateRestartPoint(CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_FAST); else { /* @@ -6826,7 +6826,7 @@ ShutdownXLOG(int code, Datum arg) if (XLogArchivingActive()) RequestXLogSwitch(false); - CreateCheckPoint(CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_IMMEDIATE); + CreateCheckPoint(CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_FAST); } } @@ -6842,24 +6842,24 @@ LogCheckpointStart(int flags, bool restartpoint) (errmsg("restartpoint starting:%s%s%s%s%s%s%s%s", (flags & CHECKPOINT_IS_SHUTDOWN) ? " shutdown" : "", (flags & CHECKPOINT_END_OF_RECOVERY) ? " end-of-recovery" : "", - (flags & CHECKPOINT_IMMEDIATE) ? " immediate" : "", + (flags & CHECKPOINT_FAST) ? " fast" : "", (flags & CHECKPOINT_FORCE) ? " force" : "", (flags & CHECKPOINT_WAIT) ? " wait" : "", (flags & CHECKPOINT_CAUSE_XLOG) ? " wal" : "", (flags & CHECKPOINT_CAUSE_TIME) ? " time" : "", - (flags & CHECKPOINT_FLUSH_ALL) ? " flush-all" : ""))); + (flags & CHECKPOINT_FLUSH_UNLOGGED) ? " flush-unlogged" : ""))); else ereport(LOG, /* translator: the placeholders show checkpoint options */ (errmsg("checkpoint starting:%s%s%s%s%s%s%s%s", (flags & CHECKPOINT_IS_SHUTDOWN) ? " shutdown" : "", (flags & CHECKPOINT_END_OF_RECOVERY) ? " end-of-recovery" : "", - (flags & CHECKPOINT_IMMEDIATE) ? " immediate" : "", + (flags & CHECKPOINT_FAST) ? " fast" : "", (flags & CHECKPOINT_FORCE) ? " force" : "", (flags & CHECKPOINT_WAIT) ? " wait" : "", (flags & CHECKPOINT_CAUSE_XLOG) ? " wal" : "", (flags & CHECKPOINT_CAUSE_TIME) ? " time" : "", - (flags & CHECKPOINT_FLUSH_ALL) ? " flush-all" : ""))); + (flags & CHECKPOINT_FLUSH_UNLOGGED) ? " flush-unlogged" : ""))); } /* @@ -6921,7 +6921,7 @@ LogCheckpointEnd(bool restartpoint) "%d removed, %d recycled; write=%ld.%03d s, " "sync=%ld.%03d s, total=%ld.%03d s; sync files=%d, " "longest=%ld.%03d s, average=%ld.%03d s; distance=%d kB, " - "estimate=%d kB; lsn=%X/%X, redo lsn=%X/%X", + "estimate=%d kB; lsn=%X/%08X, redo lsn=%X/%08X", CheckpointStats.ckpt_bufs_written, (double) CheckpointStats.ckpt_bufs_written * 100 / NBuffers, CheckpointStats.ckpt_slru_written, @@ -6945,7 +6945,7 @@ LogCheckpointEnd(bool restartpoint) "%d removed, %d recycled; write=%ld.%03d s, " "sync=%ld.%03d s, total=%ld.%03d s; sync files=%d, " "longest=%ld.%03d s, average=%ld.%03d s; distance=%d kB, " - "estimate=%d kB; lsn=%X/%X, redo lsn=%X/%X", + "estimate=%d kB; lsn=%X/%08X, redo lsn=%X/%08X", CheckpointStats.ckpt_bufs_written, (double) CheckpointStats.ckpt_bufs_written * 100 / NBuffers, CheckpointStats.ckpt_slru_written, @@ -7042,12 +7042,12 @@ update_checkpoint_display(int flags, bool restartpoint, bool reset) * flags is a bitwise OR of the following: * CHECKPOINT_IS_SHUTDOWN: checkpoint is for database shutdown. * CHECKPOINT_END_OF_RECOVERY: checkpoint is for end of WAL recovery. - * CHECKPOINT_IMMEDIATE: finish the checkpoint ASAP, - * ignoring checkpoint_completion_target parameter. + * CHECKPOINT_FAST: finish the checkpoint ASAP, ignoring + * checkpoint_completion_target parameter. * CHECKPOINT_FORCE: force a checkpoint even if no XLOG activity has occurred * since the last one (implied by CHECKPOINT_IS_SHUTDOWN or * CHECKPOINT_END_OF_RECOVERY). - * CHECKPOINT_FLUSH_ALL: also flush buffers of unlogged tables. + * CHECKPOINT_FLUSH_UNLOGGED: also flush buffers of unlogged tables. * * Note: flags contains other bits, of interest here only for logging purposes. * In particular note that this routine is synchronous and does not pay @@ -7142,7 +7142,7 @@ CreateCheckPoint(int flags) * starting snapshot of locks and transactions. */ if (!shutdown && XLogStandbyInfoActive()) - checkPoint.oldestActiveXid = GetOldestActiveTransactionId(); + checkPoint.oldestActiveXid = GetOldestActiveTransactionId(false, true); else checkPoint.oldestActiveXid = InvalidTransactionId; @@ -7456,11 +7456,6 @@ CreateCheckPoint(int flags) UpdateControlFile(); LWLockRelease(ControlFileLock); - /* Update shared-memory copy of checkpoint XID/epoch */ - SpinLockAcquire(&XLogCtl->info_lck); - XLogCtl->ckptFullXid = checkPoint.nextXid; - SpinLockRelease(&XLogCtl->info_lck); - /* * We are now done with critical updates; no need for system panic if we * have trouble while fooling with old log segments. @@ -7498,6 +7493,10 @@ CreateCheckPoint(int flags) if (PriorRedoPtr != InvalidXLogRecPtr) UpdateCheckPointDistanceEstimate(RedoRecPtr - PriorRedoPtr); +#ifdef USE_INJECTION_POINTS + INJECTION_POINT("checkpoint-before-old-wal-removal", NULL); +#endif + /* * Delete old log files, those no longer needed for last checkpoint to * prevent the disk holding the xlog from growing full. @@ -7637,7 +7636,7 @@ CreateOverwriteContrecordRecord(XLogRecPtr aborted_lsn, XLogRecPtr pagePtr, if (!RecoveryInProgress()) elog(ERROR, "can only be used at end of recovery"); if (pagePtr % XLOG_BLCKSZ != 0) - elog(ERROR, "invalid position for missing continuation record %X/%X", + elog(ERROR, "invalid position for missing continuation record %X/%08X", LSN_FORMAT_ARGS(pagePtr)); /* The current WAL insert position should be right after the page header */ @@ -7648,7 +7647,7 @@ CreateOverwriteContrecordRecord(XLogRecPtr aborted_lsn, XLogRecPtr pagePtr, startPos += SizeOfXLogShortPHD; recptr = GetXLogInsertRecPtr(); if (recptr != startPos) - elog(ERROR, "invalid WAL insert position %X/%X for OVERWRITE_CONTRECORD", + elog(ERROR, "invalid WAL insert position %X/%08X for OVERWRITE_CONTRECORD", LSN_FORMAT_ARGS(recptr)); START_CRIT_SECTION(); @@ -7678,7 +7677,7 @@ CreateOverwriteContrecordRecord(XLogRecPtr aborted_lsn, XLogRecPtr pagePtr, /* check that the record was inserted to the right place */ if (ProcLastRecPtr != startPos) - elog(ERROR, "OVERWRITE_CONTRECORD was inserted to unexpected position %X/%X", + elog(ERROR, "OVERWRITE_CONTRECORD was inserted to unexpected position %X/%08X", LSN_FORMAT_ARGS(ProcLastRecPtr)); XLogFlush(recptr); @@ -7747,8 +7746,7 @@ RecoveryRestartPoint(const CheckPoint *checkPoint, XLogReaderState *record) if (XLogHaveInvalidPages()) { elog(DEBUG2, - "could not record restart point at %X/%X because there " - "are unresolved references to invalid pages", + "could not record restart point at %X/%08X because there are unresolved references to invalid pages", LSN_FORMAT_ARGS(checkPoint->redo)); return; } @@ -7828,8 +7826,8 @@ CreateRestartPoint(int flags) lastCheckPoint.redo <= ControlFile->checkPointCopy.redo) { ereport(DEBUG2, - (errmsg_internal("skipping restartpoint, already performed at %X/%X", - LSN_FORMAT_ARGS(lastCheckPoint.redo)))); + errmsg_internal("skipping restartpoint, already performed at %X/%08X", + LSN_FORMAT_ARGS(lastCheckPoint.redo))); UpdateMinRecoveryPoint(InvalidXLogRecPtr, true); if (flags & CHECKPOINT_IS_SHUTDOWN) @@ -8013,10 +8011,10 @@ CreateRestartPoint(int flags) xtime = GetLatestXTime(); ereport((log_checkpoints ? LOG : DEBUG2), - (errmsg("recovery restart point at %X/%X", - LSN_FORMAT_ARGS(lastCheckPoint.redo)), - xtime ? errdetail("Last completed transaction was at log time %s.", - timestamptz_to_str(xtime)) : 0)); + errmsg("recovery restart point at %X/%08X", + LSN_FORMAT_ARGS(lastCheckPoint.redo)), + xtime ? errdetail("Last completed transaction was at log time %s.", + timestamptz_to_str(xtime)) : 0); /* * Finally, execute archive_cleanup_command, if any. @@ -8147,17 +8145,19 @@ KeepLogSeg(XLogRecPtr recptr, XLogSegNo *logSegNo) XLByteToSeg(recptr, currSegNo, wal_segment_size); segno = currSegNo; - /* - * Calculate how many segments are kept by slots first, adjusting for - * max_slot_wal_keep_size. - */ + /* Calculate how many segments are kept by slots. */ keep = XLogGetReplicationSlotMinimumLSN(); if (keep != InvalidXLogRecPtr && keep < recptr) { XLByteToSeg(keep, segno, wal_segment_size); - /* Cap by max_slot_wal_keep_size ... */ - if (max_slot_wal_keep_size_mb >= 0) + /* + * Account for max_slot_wal_keep_size to avoid keeping more than + * configured. However, don't do that during a binary upgrade: if + * slots were to be invalidated because of this, it would not be + * possible to preserve logical ones during the upgrade. + */ + if (max_slot_wal_keep_size_mb >= 0 && !IsBinaryUpgrade) { uint64 slot_keep_segs; @@ -8277,8 +8277,8 @@ XLogRestorePoint(const char *rpName) RecPtr = XLogInsert(RM_XLOG_ID, XLOG_RESTORE_POINT); ereport(LOG, - (errmsg("restore point \"%s\" created at %X/%X", - rpName, LSN_FORMAT_ARGS(RecPtr)))); + errmsg("restore point \"%s\" created at %X/%08X", + rpName, LSN_FORMAT_ARGS(RecPtr))); return RecPtr; } @@ -8530,11 +8530,6 @@ xlog_redo(XLogReaderState *record) ControlFile->checkPointCopy.nextXid = checkPoint.nextXid; LWLockRelease(ControlFileLock); - /* Update shared-memory copy of checkpoint XID/epoch */ - SpinLockAcquire(&XLogCtl->info_lck); - XLogCtl->ckptFullXid = checkPoint.nextXid; - SpinLockRelease(&XLogCtl->info_lck); - /* * We should've already switched to the new TLI before replaying this * record. @@ -8591,11 +8586,6 @@ xlog_redo(XLogReaderState *record) ControlFile->checkPointCopy.nextXid = checkPoint.nextXid; LWLockRelease(ControlFileLock); - /* Update shared-memory copy of checkpoint XID/epoch */ - SpinLockAcquire(&XLogCtl->info_lck); - XLogCtl->ckptFullXid = checkPoint.nextXid; - SpinLockRelease(&XLogCtl->info_lck); - /* TLI should not change in an on-line checkpoint */ (void) GetCurrentReplayRecPtr(&replayTLI); if (checkPoint.ThisTimeLineID != replayTLI) @@ -8943,9 +8933,8 @@ issue_xlog_fsync(int fd, XLogSegNo segno, TimeLineID tli) * backup state and tablespace map. * * Input parameters are "state" (the backup state), "fast" (if true, we do - * the checkpoint in immediate mode to make it faster), and "tablespaces" - * (if non-NULL, indicates a list of tablespaceinfo structs describing the - * cluster's tablespaces.). + * the checkpoint in fast mode), and "tablespaces" (if non-NULL, indicates a + * list of tablespaceinfo structs describing the cluster's tablespaces.). * * The tablespace map contents are appended to passed-in parameter * tablespace_map and the caller is responsible for including it in the backup @@ -9022,7 +9011,7 @@ do_pg_backup_start(const char *backupidstr, bool fast, List **tablespaces, * work correctly, it is critical that sessionBackupState is only updated * after this block is over. */ - PG_ENSURE_ERROR_CLEANUP(do_pg_abort_backup, DatumGetBool(true)); + PG_ENSURE_ERROR_CLEANUP(do_pg_abort_backup, BoolGetDatum(true)); { bool gotUniqueStartpoint = false; DIR *tblspcdir; @@ -9073,11 +9062,11 @@ do_pg_backup_start(const char *backupidstr, bool fast, List **tablespaces, * during recovery means that checkpointer is running, we can use * RequestCheckpoint() to establish a restartpoint. * - * We use CHECKPOINT_IMMEDIATE only if requested by user (via - * passing fast = true). Otherwise this can take awhile. + * We use CHECKPOINT_FAST only if requested by user (via passing + * fast = true). Otherwise this can take awhile. */ RequestCheckpoint(CHECKPOINT_FORCE | CHECKPOINT_WAIT | - (fast ? CHECKPOINT_IMMEDIATE : 0)); + (fast ? CHECKPOINT_FAST : 0)); /* * Now we need to fetch the checkpoint record location, and also @@ -9261,7 +9250,7 @@ do_pg_backup_start(const char *backupidstr, bool fast, List **tablespaces, state->starttime = (pg_time_t) time(NULL); } - PG_END_ENSURE_ERROR_CLEANUP(do_pg_abort_backup, DatumGetBool(true)); + PG_END_ENSURE_ERROR_CLEANUP(do_pg_abort_backup, BoolGetDatum(true)); state->started_in_recovery = backup_started_in_recovery; @@ -9601,7 +9590,7 @@ register_persistent_abort_backup_handler(void) if (already_done) return; - before_shmem_exit(do_pg_abort_backup, DatumGetBool(false)); + before_shmem_exit(do_pg_abort_backup, BoolGetDatum(false)); already_done = true; } diff --git a/src/backend/access/transam/xlogbackup.c b/src/backend/access/transam/xlogbackup.c index 342590e0a46..cda4b38b7d6 100644 --- a/src/backend/access/transam/xlogbackup.c +++ b/src/backend/access/transam/xlogbackup.c @@ -42,7 +42,7 @@ build_backup_content(BackupState *state, bool ishistoryfile) XLByteToSeg(state->startpoint, startsegno, wal_segment_size); XLogFileName(startxlogfile, state->starttli, startsegno, wal_segment_size); - appendStringInfo(result, "START WAL LOCATION: %X/%X (file %s)\n", + appendStringInfo(result, "START WAL LOCATION: %X/%08X (file %s)\n", LSN_FORMAT_ARGS(state->startpoint), startxlogfile); if (ishistoryfile) @@ -52,11 +52,11 @@ build_backup_content(BackupState *state, bool ishistoryfile) XLByteToSeg(state->stoppoint, stopsegno, wal_segment_size); XLogFileName(stopxlogfile, state->stoptli, stopsegno, wal_segment_size); - appendStringInfo(result, "STOP WAL LOCATION: %X/%X (file %s)\n", + appendStringInfo(result, "STOP WAL LOCATION: %X/%08X (file %s)\n", LSN_FORMAT_ARGS(state->stoppoint), stopxlogfile); } - appendStringInfo(result, "CHECKPOINT LOCATION: %X/%X\n", + appendStringInfo(result, "CHECKPOINT LOCATION: %X/%08X\n", LSN_FORMAT_ARGS(state->checkpointloc)); appendStringInfoString(result, "BACKUP METHOD: streamed\n"); appendStringInfo(result, "BACKUP FROM: %s\n", @@ -81,7 +81,7 @@ build_backup_content(BackupState *state, bool ishistoryfile) Assert(XLogRecPtrIsInvalid(state->istartpoint) == (state->istarttli == 0)); if (!XLogRecPtrIsInvalid(state->istartpoint)) { - appendStringInfo(result, "INCREMENTAL FROM LSN: %X/%X\n", + appendStringInfo(result, "INCREMENTAL FROM LSN: %X/%08X\n", LSN_FORMAT_ARGS(state->istartpoint)); appendStringInfo(result, "INCREMENTAL FROM TLI: %u\n", state->istarttli); diff --git a/src/backend/access/transam/xloginsert.c b/src/backend/access/transam/xloginsert.c index 5ee9d0b028e..c7571429e8e 100644 --- a/src/backend/access/transam/xloginsert.c +++ b/src/backend/access/transam/xloginsert.c @@ -530,6 +530,18 @@ XLogInsert(RmgrId rmid, uint8 info) } /* + * Simple wrapper to XLogInsert to insert a WAL record with elementary + * contents (only an int64 is supported as value currently). + */ +XLogRecPtr +XLogSimpleInsertInt64(RmgrId rmid, uint8 info, int64 value) +{ + XLogBeginInsert(); + XLogRegisterData(&value, sizeof(value)); + return XLogInsert(rmid, info); +} + +/* * Assemble a WAL record from the registered data and buffers into an * XLogRecData chain, ready for insertion with XLogInsertRecord(). * diff --git a/src/backend/access/transam/xlogprefetcher.c b/src/backend/access/transam/xlogprefetcher.c index 7735562db01..ed3aacabc98 100644 --- a/src/backend/access/transam/xlogprefetcher.c +++ b/src/backend/access/transam/xlogprefetcher.c @@ -546,7 +546,7 @@ XLogPrefetcherNextBlock(uintptr_t pgsr_private, XLogRecPtr *lsn) #ifdef XLOGPREFETCHER_DEBUG_LEVEL elog(XLOGPREFETCHER_DEBUG_LEVEL, - "suppressing all readahead until %X/%X is replayed due to possible TLI change", + "suppressing all readahead until %X/%08X is replayed due to possible TLI change", LSN_FORMAT_ARGS(record->lsn)); #endif @@ -579,7 +579,7 @@ XLogPrefetcherNextBlock(uintptr_t pgsr_private, XLogRecPtr *lsn) #ifdef XLOGPREFETCHER_DEBUG_LEVEL elog(XLOGPREFETCHER_DEBUG_LEVEL, - "suppressing prefetch in database %u until %X/%X is replayed due to raw file copy", + "suppressing prefetch in database %u until %X/%08X is replayed due to raw file copy", rlocator.dbOid, LSN_FORMAT_ARGS(record->lsn)); #endif @@ -607,7 +607,7 @@ XLogPrefetcherNextBlock(uintptr_t pgsr_private, XLogRecPtr *lsn) #ifdef XLOGPREFETCHER_DEBUG_LEVEL elog(XLOGPREFETCHER_DEBUG_LEVEL, - "suppressing prefetch in relation %u/%u/%u until %X/%X is replayed, which creates the relation", + "suppressing prefetch in relation %u/%u/%u until %X/%08X is replayed, which creates the relation", xlrec->rlocator.spcOid, xlrec->rlocator.dbOid, xlrec->rlocator.relNumber, @@ -630,7 +630,7 @@ XLogPrefetcherNextBlock(uintptr_t pgsr_private, XLogRecPtr *lsn) #ifdef XLOGPREFETCHER_DEBUG_LEVEL elog(XLOGPREFETCHER_DEBUG_LEVEL, - "suppressing prefetch in relation %u/%u/%u from block %u until %X/%X is replayed, which truncates the relation", + "suppressing prefetch in relation %u/%u/%u from block %u until %X/%08X is replayed, which truncates the relation", xlrec->rlocator.spcOid, xlrec->rlocator.dbOid, xlrec->rlocator.relNumber, @@ -729,7 +729,7 @@ XLogPrefetcherNextBlock(uintptr_t pgsr_private, XLogRecPtr *lsn) { #ifdef XLOGPREFETCHER_DEBUG_LEVEL elog(XLOGPREFETCHER_DEBUG_LEVEL, - "suppressing all prefetch in relation %u/%u/%u until %X/%X is replayed, because the relation does not exist on disk", + "suppressing all prefetch in relation %u/%u/%u until %X/%08X is replayed, because the relation does not exist on disk", reln->smgr_rlocator.locator.spcOid, reln->smgr_rlocator.locator.dbOid, reln->smgr_rlocator.locator.relNumber, @@ -750,7 +750,7 @@ XLogPrefetcherNextBlock(uintptr_t pgsr_private, XLogRecPtr *lsn) { #ifdef XLOGPREFETCHER_DEBUG_LEVEL elog(XLOGPREFETCHER_DEBUG_LEVEL, - "suppressing prefetch in relation %u/%u/%u from block %u until %X/%X is replayed, because the relation is too small", + "suppressing prefetch in relation %u/%u/%u from block %u until %X/%08X is replayed, because the relation is too small", reln->smgr_rlocator.locator.spcOid, reln->smgr_rlocator.locator.dbOid, reln->smgr_rlocator.locator.relNumber, @@ -928,7 +928,7 @@ XLogPrefetcherIsFiltered(XLogPrefetcher *prefetcher, RelFileLocator rlocator, { #ifdef XLOGPREFETCHER_DEBUG_LEVEL elog(XLOGPREFETCHER_DEBUG_LEVEL, - "prefetch of %u/%u/%u block %u suppressed; filtering until LSN %X/%X is replayed (blocks >= %u filtered)", + "prefetch of %u/%u/%u block %u suppressed; filtering until LSN %X/%08X is replayed (blocks >= %u filtered)", rlocator.spcOid, rlocator.dbOid, rlocator.relNumber, blockno, LSN_FORMAT_ARGS(filter->filter_until_replayed), filter->filter_from_block); @@ -944,7 +944,7 @@ XLogPrefetcherIsFiltered(XLogPrefetcher *prefetcher, RelFileLocator rlocator, { #ifdef XLOGPREFETCHER_DEBUG_LEVEL elog(XLOGPREFETCHER_DEBUG_LEVEL, - "prefetch of %u/%u/%u block %u suppressed; filtering until LSN %X/%X is replayed (whole database)", + "prefetch of %u/%u/%u block %u suppressed; filtering until LSN %X/%08X is replayed (whole database)", rlocator.spcOid, rlocator.dbOid, rlocator.relNumber, blockno, LSN_FORMAT_ARGS(filter->filter_until_replayed)); #endif diff --git a/src/backend/access/transam/xlogreader.c b/src/backend/access/transam/xlogreader.c index 2790ade1f91..dcc8d4f9c1b 100644 --- a/src/backend/access/transam/xlogreader.c +++ b/src/backend/access/transam/xlogreader.c @@ -617,7 +617,7 @@ restart: } else if (targetRecOff < pageHeaderSize) { - report_invalid_record(state, "invalid record offset at %X/%X: expected at least %u, got %u", + report_invalid_record(state, "invalid record offset at %X/%08X: expected at least %u, got %u", LSN_FORMAT_ARGS(RecPtr), pageHeaderSize, targetRecOff); goto err; @@ -626,7 +626,7 @@ restart: if ((((XLogPageHeader) state->readBuf)->xlp_info & XLP_FIRST_IS_CONTRECORD) && targetRecOff == pageHeaderSize) { - report_invalid_record(state, "contrecord is requested by %X/%X", + report_invalid_record(state, "contrecord is requested by %X/%08X", LSN_FORMAT_ARGS(RecPtr)); goto err; } @@ -667,7 +667,7 @@ restart: if (total_len < SizeOfXLogRecord) { report_invalid_record(state, - "invalid record length at %X/%X: expected at least %u, got %u", + "invalid record length at %X/%08X: expected at least %u, got %u", LSN_FORMAT_ARGS(RecPtr), (uint32) SizeOfXLogRecord, total_len); goto err; @@ -723,11 +723,12 @@ restart: /* Calculate pointer to beginning of next page */ targetPagePtr += XLOG_BLCKSZ; - /* Wait for the next page to become available */ - readOff = ReadPageInternal(state, targetPagePtr, - Min(total_len - gotlen + SizeOfXLogShortPHD, - XLOG_BLCKSZ)); - + /* + * Read the page header before processing the record data, so we + * can handle the case where the previous record ended as being a + * partial one. + */ + readOff = ReadPageInternal(state, targetPagePtr, SizeOfXLogShortPHD); if (readOff == XLREAD_WOULDBLOCK) return XLREAD_WOULDBLOCK; else if (readOff < 0) @@ -756,7 +757,7 @@ restart: if (!(pageHeader->xlp_info & XLP_FIRST_IS_CONTRECORD)) { report_invalid_record(state, - "there is no contrecord flag at %X/%X", + "there is no contrecord flag at %X/%08X", LSN_FORMAT_ARGS(RecPtr)); goto err; } @@ -769,13 +770,22 @@ restart: total_len != (pageHeader->xlp_rem_len + gotlen)) { report_invalid_record(state, - "invalid contrecord length %u (expected %lld) at %X/%X", + "invalid contrecord length %u (expected %lld) at %X/%08X", pageHeader->xlp_rem_len, ((long long) total_len) - gotlen, LSN_FORMAT_ARGS(RecPtr)); goto err; } + /* Wait for the next page to become available */ + readOff = ReadPageInternal(state, targetPagePtr, + Min(total_len - gotlen + SizeOfXLogShortPHD, + XLOG_BLCKSZ)); + if (readOff == XLREAD_WOULDBLOCK) + return XLREAD_WOULDBLOCK; + else if (readOff < 0) + goto err; + /* Append the continuation from this page to the buffer */ pageHeaderSize = XLogPageHeaderSize(pageHeader); @@ -1132,7 +1142,7 @@ ValidXLogRecordHeader(XLogReaderState *state, XLogRecPtr RecPtr, if (record->xl_tot_len < SizeOfXLogRecord) { report_invalid_record(state, - "invalid record length at %X/%X: expected at least %u, got %u", + "invalid record length at %X/%08X: expected at least %u, got %u", LSN_FORMAT_ARGS(RecPtr), (uint32) SizeOfXLogRecord, record->xl_tot_len); return false; @@ -1140,7 +1150,7 @@ ValidXLogRecordHeader(XLogReaderState *state, XLogRecPtr RecPtr, if (!RmgrIdIsValid(record->xl_rmid)) { report_invalid_record(state, - "invalid resource manager ID %u at %X/%X", + "invalid resource manager ID %u at %X/%08X", record->xl_rmid, LSN_FORMAT_ARGS(RecPtr)); return false; } @@ -1153,7 +1163,7 @@ ValidXLogRecordHeader(XLogReaderState *state, XLogRecPtr RecPtr, if (!(record->xl_prev < RecPtr)) { report_invalid_record(state, - "record with incorrect prev-link %X/%X at %X/%X", + "record with incorrect prev-link %X/%08X at %X/%08X", LSN_FORMAT_ARGS(record->xl_prev), LSN_FORMAT_ARGS(RecPtr)); return false; @@ -1169,7 +1179,7 @@ ValidXLogRecordHeader(XLogReaderState *state, XLogRecPtr RecPtr, if (record->xl_prev != PrevRecPtr) { report_invalid_record(state, - "record with incorrect prev-link %X/%X at %X/%X", + "record with incorrect prev-link %X/%08X at %X/%08X", LSN_FORMAT_ARGS(record->xl_prev), LSN_FORMAT_ARGS(RecPtr)); return false; @@ -1207,7 +1217,7 @@ ValidXLogRecord(XLogReaderState *state, XLogRecord *record, XLogRecPtr recptr) if (!EQ_CRC32C(record->xl_crc, crc)) { report_invalid_record(state, - "incorrect resource manager data checksum in record at %X/%X", + "incorrect resource manager data checksum in record at %X/%08X", LSN_FORMAT_ARGS(recptr)); return false; } @@ -1241,7 +1251,7 @@ XLogReaderValidatePageHeader(XLogReaderState *state, XLogRecPtr recptr, XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); report_invalid_record(state, - "invalid magic number %04X in WAL segment %s, LSN %X/%X, offset %u", + "invalid magic number %04X in WAL segment %s, LSN %X/%08X, offset %u", hdr->xlp_magic, fname, LSN_FORMAT_ARGS(recptr), @@ -1256,7 +1266,7 @@ XLogReaderValidatePageHeader(XLogReaderState *state, XLogRecPtr recptr, XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); report_invalid_record(state, - "invalid info bits %04X in WAL segment %s, LSN %X/%X, offset %u", + "invalid info bits %04X in WAL segment %s, LSN %X/%08X, offset %u", hdr->xlp_info, fname, LSN_FORMAT_ARGS(recptr), @@ -1298,7 +1308,7 @@ XLogReaderValidatePageHeader(XLogReaderState *state, XLogRecPtr recptr, /* hmm, first page of file doesn't have a long header? */ report_invalid_record(state, - "invalid info bits %04X in WAL segment %s, LSN %X/%X, offset %u", + "invalid info bits %04X in WAL segment %s, LSN %X/%08X, offset %u", hdr->xlp_info, fname, LSN_FORMAT_ARGS(recptr), @@ -1318,7 +1328,7 @@ XLogReaderValidatePageHeader(XLogReaderState *state, XLogRecPtr recptr, XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); report_invalid_record(state, - "unexpected pageaddr %X/%X in WAL segment %s, LSN %X/%X, offset %u", + "unexpected pageaddr %X/%08X in WAL segment %s, LSN %X/%08X, offset %u", LSN_FORMAT_ARGS(hdr->xlp_pageaddr), fname, LSN_FORMAT_ARGS(recptr), @@ -1344,7 +1354,7 @@ XLogReaderValidatePageHeader(XLogReaderState *state, XLogRecPtr recptr, XLogFileName(fname, state->seg.ws_tli, segno, state->segcxt.ws_segsize); report_invalid_record(state, - "out-of-sequence timeline ID %u (after %u) in WAL segment %s, LSN %X/%X, offset %u", + "out-of-sequence timeline ID %u (after %u) in WAL segment %s, LSN %X/%08X, offset %u", hdr->xlp_tli, state->latestPageTLI, fname, @@ -1756,7 +1766,7 @@ DecodeXLogRecord(XLogReaderState *state, if (block_id <= decoded->max_block_id) { report_invalid_record(state, - "out-of-order block_id %u at %X/%X", + "out-of-order block_id %u at %X/%08X", block_id, LSN_FORMAT_ARGS(state->ReadRecPtr)); goto err; @@ -1780,14 +1790,14 @@ DecodeXLogRecord(XLogReaderState *state, if (blk->has_data && blk->data_len == 0) { report_invalid_record(state, - "BKPBLOCK_HAS_DATA set, but no data included at %X/%X", + "BKPBLOCK_HAS_DATA set, but no data included at %X/%08X", LSN_FORMAT_ARGS(state->ReadRecPtr)); goto err; } if (!blk->has_data && blk->data_len != 0) { report_invalid_record(state, - "BKPBLOCK_HAS_DATA not set, but data length is %u at %X/%X", + "BKPBLOCK_HAS_DATA not set, but data length is %u at %X/%08X", (unsigned int) blk->data_len, LSN_FORMAT_ARGS(state->ReadRecPtr)); goto err; @@ -1823,7 +1833,7 @@ DecodeXLogRecord(XLogReaderState *state, blk->bimg_len == BLCKSZ)) { report_invalid_record(state, - "BKPIMAGE_HAS_HOLE set, but hole offset %u length %u block image length %u at %X/%X", + "BKPIMAGE_HAS_HOLE set, but hole offset %u length %u block image length %u at %X/%08X", (unsigned int) blk->hole_offset, (unsigned int) blk->hole_length, (unsigned int) blk->bimg_len, @@ -1839,7 +1849,7 @@ DecodeXLogRecord(XLogReaderState *state, (blk->hole_offset != 0 || blk->hole_length != 0)) { report_invalid_record(state, - "BKPIMAGE_HAS_HOLE not set, but hole offset %u length %u at %X/%X", + "BKPIMAGE_HAS_HOLE not set, but hole offset %u length %u at %X/%08X", (unsigned int) blk->hole_offset, (unsigned int) blk->hole_length, LSN_FORMAT_ARGS(state->ReadRecPtr)); @@ -1853,7 +1863,7 @@ DecodeXLogRecord(XLogReaderState *state, blk->bimg_len == BLCKSZ) { report_invalid_record(state, - "BKPIMAGE_COMPRESSED set, but block image length %u at %X/%X", + "BKPIMAGE_COMPRESSED set, but block image length %u at %X/%08X", (unsigned int) blk->bimg_len, LSN_FORMAT_ARGS(state->ReadRecPtr)); goto err; @@ -1868,7 +1878,7 @@ DecodeXLogRecord(XLogReaderState *state, blk->bimg_len != BLCKSZ) { report_invalid_record(state, - "neither BKPIMAGE_HAS_HOLE nor BKPIMAGE_COMPRESSED set, but block image length is %u at %X/%X", + "neither BKPIMAGE_HAS_HOLE nor BKPIMAGE_COMPRESSED set, but block image length is %u at %X/%08X", (unsigned int) blk->data_len, LSN_FORMAT_ARGS(state->ReadRecPtr)); goto err; @@ -1884,7 +1894,7 @@ DecodeXLogRecord(XLogReaderState *state, if (rlocator == NULL) { report_invalid_record(state, - "BKPBLOCK_SAME_REL set but no previous rel at %X/%X", + "BKPBLOCK_SAME_REL set but no previous rel at %X/%08X", LSN_FORMAT_ARGS(state->ReadRecPtr)); goto err; } @@ -1896,7 +1906,7 @@ DecodeXLogRecord(XLogReaderState *state, else { report_invalid_record(state, - "invalid block_id %u at %X/%X", + "invalid block_id %u at %X/%08X", block_id, LSN_FORMAT_ARGS(state->ReadRecPtr)); goto err; } @@ -1963,7 +1973,7 @@ DecodeXLogRecord(XLogReaderState *state, shortdata_err: report_invalid_record(state, - "record with invalid length at %X/%X", + "record with invalid length at %X/%08X", LSN_FORMAT_ARGS(state->ReadRecPtr)); err: *errormsg = state->errormsg_buf; @@ -2073,14 +2083,14 @@ RestoreBlockImage(XLogReaderState *record, uint8 block_id, char *page) !record->record->blocks[block_id].in_use) { report_invalid_record(record, - "could not restore image at %X/%X with invalid block %d specified", + "could not restore image at %X/%08X with invalid block %d specified", LSN_FORMAT_ARGS(record->ReadRecPtr), block_id); return false; } if (!record->record->blocks[block_id].has_image) { - report_invalid_record(record, "could not restore image at %X/%X with invalid state, block %d", + report_invalid_record(record, "could not restore image at %X/%08X with invalid state, block %d", LSN_FORMAT_ARGS(record->ReadRecPtr), block_id); return false; @@ -2107,7 +2117,7 @@ RestoreBlockImage(XLogReaderState *record, uint8 block_id, char *page) bkpb->bimg_len, BLCKSZ - bkpb->hole_length) <= 0) decomp_success = false; #else - report_invalid_record(record, "could not restore image at %X/%X compressed with %s not supported by build, block %d", + report_invalid_record(record, "could not restore image at %X/%08X compressed with %s not supported by build, block %d", LSN_FORMAT_ARGS(record->ReadRecPtr), "LZ4", block_id); @@ -2124,7 +2134,7 @@ RestoreBlockImage(XLogReaderState *record, uint8 block_id, char *page) if (ZSTD_isError(decomp_result)) decomp_success = false; #else - report_invalid_record(record, "could not restore image at %X/%X compressed with %s not supported by build, block %d", + report_invalid_record(record, "could not restore image at %X/%08X compressed with %s not supported by build, block %d", LSN_FORMAT_ARGS(record->ReadRecPtr), "zstd", block_id); @@ -2133,7 +2143,7 @@ RestoreBlockImage(XLogReaderState *record, uint8 block_id, char *page) } else { - report_invalid_record(record, "could not restore image at %X/%X compressed with unknown method, block %d", + report_invalid_record(record, "could not restore image at %X/%08X compressed with unknown method, block %d", LSN_FORMAT_ARGS(record->ReadRecPtr), block_id); return false; @@ -2141,7 +2151,7 @@ RestoreBlockImage(XLogReaderState *record, uint8 block_id, char *page) if (!decomp_success) { - report_invalid_record(record, "could not decompress image at %X/%X, block %d", + report_invalid_record(record, "could not decompress image at %X/%08X, block %d", LSN_FORMAT_ARGS(record->ReadRecPtr), block_id); return false; diff --git a/src/backend/access/transam/xlogrecovery.c b/src/backend/access/transam/xlogrecovery.c index 6ce979f2d8b..f23ec8969c2 100644 --- a/src/backend/access/transam/xlogrecovery.c +++ b/src/backend/access/transam/xlogrecovery.c @@ -620,10 +620,10 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, * than ControlFile->checkPoint is used. */ ereport(LOG, - (errmsg("starting backup recovery with redo LSN %X/%X, checkpoint LSN %X/%X, on timeline ID %u", - LSN_FORMAT_ARGS(RedoStartLSN), - LSN_FORMAT_ARGS(CheckPointLoc), - CheckPointTLI))); + errmsg("starting backup recovery with redo LSN %X/%08X, checkpoint LSN %X/%08X, on timeline ID %u", + LSN_FORMAT_ARGS(RedoStartLSN), + LSN_FORMAT_ARGS(CheckPointLoc), + CheckPointTLI)); /* * When a backup_label file is present, we want to roll forward from @@ -636,8 +636,8 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, memcpy(&checkPoint, XLogRecGetData(xlogreader), sizeof(CheckPoint)); wasShutdown = ((record->xl_info & ~XLR_INFO_MASK) == XLOG_CHECKPOINT_SHUTDOWN); ereport(DEBUG1, - (errmsg_internal("checkpoint record is at %X/%X", - LSN_FORMAT_ARGS(CheckPointLoc)))); + errmsg_internal("checkpoint record is at %X/%08X", + LSN_FORMAT_ARGS(CheckPointLoc))); InRecovery = true; /* force recovery even if SHUTDOWNED */ /* @@ -652,23 +652,23 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, if (!ReadRecord(xlogprefetcher, LOG, false, checkPoint.ThisTimeLineID)) ereport(FATAL, - (errmsg("could not find redo location %X/%X referenced by checkpoint record at %X/%X", - LSN_FORMAT_ARGS(checkPoint.redo), LSN_FORMAT_ARGS(CheckPointLoc)), - errhint("If you are restoring from a backup, touch \"%s/recovery.signal\" or \"%s/standby.signal\" and add required recovery options.\n" - "If you are not restoring from a backup, try removing the file \"%s/backup_label\".\n" - "Be careful: removing \"%s/backup_label\" will result in a corrupt cluster if restoring from a backup.", - DataDir, DataDir, DataDir, DataDir))); + errmsg("could not find redo location %X/%08X referenced by checkpoint record at %X/%08X", + LSN_FORMAT_ARGS(checkPoint.redo), LSN_FORMAT_ARGS(CheckPointLoc)), + errhint("If you are restoring from a backup, touch \"%s/recovery.signal\" or \"%s/standby.signal\" and add required recovery options.\n" + "If you are not restoring from a backup, try removing the file \"%s/backup_label\".\n" + "Be careful: removing \"%s/backup_label\" will result in a corrupt cluster if restoring from a backup.", + DataDir, DataDir, DataDir, DataDir)); } } else { ereport(FATAL, - (errmsg("could not locate required checkpoint record at %X/%X", - LSN_FORMAT_ARGS(CheckPointLoc)), - errhint("If you are restoring from a backup, touch \"%s/recovery.signal\" or \"%s/standby.signal\" and add required recovery options.\n" - "If you are not restoring from a backup, try removing the file \"%s/backup_label\".\n" - "Be careful: removing \"%s/backup_label\" will result in a corrupt cluster if restoring from a backup.", - DataDir, DataDir, DataDir, DataDir))); + errmsg("could not locate required checkpoint record at %X/%08X", + LSN_FORMAT_ARGS(CheckPointLoc)), + errhint("If you are restoring from a backup, touch \"%s/recovery.signal\" or \"%s/standby.signal\" and add required recovery options.\n" + "If you are not restoring from a backup, try removing the file \"%s/backup_label\".\n" + "Be careful: removing \"%s/backup_label\" will result in a corrupt cluster if restoring from a backup.", + DataDir, DataDir, DataDir, DataDir)); wasShutdown = false; /* keep compiler quiet */ } @@ -773,8 +773,8 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, */ if (!XLogRecPtrIsInvalid(ControlFile->backupStartPoint)) ereport(LOG, - (errmsg("restarting backup recovery with redo LSN %X/%X", - LSN_FORMAT_ARGS(ControlFile->backupStartPoint)))); + errmsg("restarting backup recovery with redo LSN %X/%08X", + LSN_FORMAT_ARGS(ControlFile->backupStartPoint))); /* Get the last valid checkpoint record. */ CheckPointLoc = ControlFile->checkPoint; @@ -786,8 +786,8 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, if (record != NULL) { ereport(DEBUG1, - (errmsg_internal("checkpoint record is at %X/%X", - LSN_FORMAT_ARGS(CheckPointLoc)))); + errmsg_internal("checkpoint record is at %X/%08X", + LSN_FORMAT_ARGS(CheckPointLoc))); } else { @@ -798,8 +798,8 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, * simplify processing around checkpoints. */ ereport(PANIC, - (errmsg("could not locate a valid checkpoint record at %X/%X", - LSN_FORMAT_ARGS(CheckPointLoc)))); + errmsg("could not locate a valid checkpoint record at %X/%08X", + LSN_FORMAT_ARGS(CheckPointLoc))); } memcpy(&checkPoint, XLogRecGetData(xlogreader), sizeof(CheckPoint)); wasShutdown = ((record->xl_info & ~XLR_INFO_MASK) == XLOG_CHECKPOINT_SHUTDOWN); @@ -824,8 +824,8 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, recoveryTargetName))); else if (recoveryTarget == RECOVERY_TARGET_LSN) ereport(LOG, - (errmsg("starting point-in-time recovery to WAL location (LSN) \"%X/%X\"", - LSN_FORMAT_ARGS(recoveryTargetLSN)))); + errmsg("starting point-in-time recovery to WAL location (LSN) \"%X/%08X\"", + LSN_FORMAT_ARGS(recoveryTargetLSN))); else if (recoveryTarget == RECOVERY_TARGET_IMMEDIATE) ereport(LOG, (errmsg("starting point-in-time recovery to earliest consistent point"))); @@ -855,7 +855,7 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, (errmsg("requested timeline %u is not a child of this server's history", recoveryTargetTLI), /* translator: %s is a backup_label file or a pg_control file */ - errdetail("Latest checkpoint in file \"%s\" is at %X/%X on timeline %u, but in the history of the requested timeline, the server forked off from that timeline at %X/%X.", + errdetail("Latest checkpoint in file \"%s\" is at %X/%08X on timeline %u, but in the history of the requested timeline, the server forked off from that timeline at %X/%08X.", haveBackupLabel ? "backup_label" : "pg_control", LSN_FORMAT_ARGS(CheckPointLoc), CheckPointTLI, @@ -870,15 +870,15 @@ InitWalRecovery(ControlFileData *ControlFile, bool *wasShutdown_ptr, tliOfPointInHistory(ControlFile->minRecoveryPoint - 1, expectedTLEs) != ControlFile->minRecoveryPointTLI) ereport(FATAL, - (errmsg("requested timeline %u does not contain minimum recovery point %X/%X on timeline %u", - recoveryTargetTLI, - LSN_FORMAT_ARGS(ControlFile->minRecoveryPoint), - ControlFile->minRecoveryPointTLI))); + errmsg("requested timeline %u does not contain minimum recovery point %X/%08X on timeline %u", + recoveryTargetTLI, + LSN_FORMAT_ARGS(ControlFile->minRecoveryPoint), + ControlFile->minRecoveryPointTLI)); ereport(DEBUG1, - (errmsg_internal("redo record is at %X/%X; shutdown %s", - LSN_FORMAT_ARGS(checkPoint.redo), - wasShutdown ? "true" : "false"))); + errmsg_internal("redo record is at %X/%08X; shutdown %s", + LSN_FORMAT_ARGS(checkPoint.redo), + wasShutdown ? "true" : "false")); ereport(DEBUG1, (errmsg_internal("next transaction ID: " UINT64_FORMAT "; next OID: %u", U64FromFullTransactionId(checkPoint.nextXid), @@ -1253,14 +1253,14 @@ read_backup_label(XLogRecPtr *checkPointLoc, TimeLineID *backupLabelTLI, * is pretty crude, but we are not expecting any variability in the file * format). */ - if (fscanf(lfp, "START WAL LOCATION: %X/%X (file %08X%16s)%c", + if (fscanf(lfp, "START WAL LOCATION: %X/%08X (file %08X%16s)%c", &hi, &lo, &tli_from_walseg, startxlogfilename, &ch) != 5 || ch != '\n') ereport(FATAL, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), errmsg("invalid data in file \"%s\"", BACKUP_LABEL_FILE))); RedoStartLSN = ((uint64) hi) << 32 | lo; RedoStartTLI = tli_from_walseg; - if (fscanf(lfp, "CHECKPOINT LOCATION: %X/%X%c", + if (fscanf(lfp, "CHECKPOINT LOCATION: %X/%08X%c", &hi, &lo, &ch) != 3 || ch != '\n') ereport(FATAL, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), @@ -1332,7 +1332,7 @@ read_backup_label(XLogRecPtr *checkPointLoc, TimeLineID *backupLabelTLI, tli_from_file, BACKUP_LABEL_FILE))); } - if (fscanf(lfp, "INCREMENTAL FROM LSN: %X/%X\n", &hi, &lo) > 0) + if (fscanf(lfp, "INCREMENTAL FROM LSN: %X/%08X\n", &hi, &lo) > 0) ereport(FATAL, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), errmsg("this is an incremental backup, not a data directory"), @@ -1626,6 +1626,7 @@ ShutdownWalRecovery(void) close(readFile); readFile = -1; } + pfree(xlogreader->private_data); XLogReaderFree(xlogreader); XLogPrefetcherFree(xlogprefetcher); @@ -1722,8 +1723,8 @@ PerformWalRecovery(void) if (record->xl_rmid != RM_XLOG_ID || (record->xl_info & ~XLR_INFO_MASK) != XLOG_CHECKPOINT_REDO) ereport(FATAL, - (errmsg("unexpected record type found at redo point %X/%X", - LSN_FORMAT_ARGS(xlogreader->ReadRecPtr)))); + errmsg("unexpected record type found at redo point %X/%08X", + LSN_FORMAT_ARGS(xlogreader->ReadRecPtr))); } else { @@ -1745,8 +1746,8 @@ PerformWalRecovery(void) RmgrStartup(); ereport(LOG, - (errmsg("redo starts at %X/%X", - LSN_FORMAT_ARGS(xlogreader->ReadRecPtr)))); + errmsg("redo starts at %X/%08X", + LSN_FORMAT_ARGS(xlogreader->ReadRecPtr))); /* Prepare to report progress of the redo phase. */ if (!StandbyMode) @@ -1758,7 +1759,7 @@ PerformWalRecovery(void) do { if (!StandbyMode) - ereport_startup_progress("redo in progress, elapsed time: %ld.%02d s, current LSN: %X/%X", + ereport_startup_progress("redo in progress, elapsed time: %ld.%02d s, current LSN: %X/%08X", LSN_FORMAT_ARGS(xlogreader->ReadRecPtr)); #ifdef WAL_DEBUG @@ -1767,7 +1768,7 @@ PerformWalRecovery(void) StringInfoData buf; initStringInfo(&buf); - appendStringInfo(&buf, "REDO @ %X/%X; LSN %X/%X: ", + appendStringInfo(&buf, "REDO @ %X/%08X; LSN %X/%08X: ", LSN_FORMAT_ARGS(xlogreader->ReadRecPtr), LSN_FORMAT_ARGS(xlogreader->EndRecPtr)); xlog_outrec(&buf, xlogreader); @@ -1880,9 +1881,9 @@ PerformWalRecovery(void) RmgrCleanup(); ereport(LOG, - (errmsg("redo done at %X/%X system usage: %s", - LSN_FORMAT_ARGS(xlogreader->ReadRecPtr), - pg_rusage_show(&ru0)))); + errmsg("redo done at %X/%08X system usage: %s", + LSN_FORMAT_ARGS(xlogreader->ReadRecPtr), + pg_rusage_show(&ru0))); xtime = GetLatestXTime(); if (xtime) ereport(LOG, @@ -2092,7 +2093,7 @@ xlogrecovery_redo(XLogReaderState *record, TimeLineID replayTLI) memcpy(&xlrec, XLogRecGetData(record), sizeof(xl_overwrite_contrecord)); if (xlrec.overwritten_lsn != record->overwrittenRecPtr) - elog(FATAL, "mismatching overwritten LSN %X/%X -> %X/%X", + elog(FATAL, "mismatching overwritten LSN %X/%08X -> %X/%08X", LSN_FORMAT_ARGS(xlrec.overwritten_lsn), LSN_FORMAT_ARGS(record->overwrittenRecPtr)); @@ -2101,9 +2102,9 @@ xlogrecovery_redo(XLogReaderState *record, TimeLineID replayTLI) missingContrecPtr = InvalidXLogRecPtr; ereport(LOG, - (errmsg("successfully skipped missing contrecord at %X/%X, overwritten at %s", - LSN_FORMAT_ARGS(xlrec.overwritten_lsn), - timestamptz_to_str(xlrec.overwrite_time)))); + errmsg("successfully skipped missing contrecord at %X/%08X, overwritten at %s", + LSN_FORMAT_ARGS(xlrec.overwritten_lsn), + timestamptz_to_str(xlrec.overwrite_time))); /* Verifying the record should only happen once */ record->overwrittenRecPtr = InvalidXLogRecPtr; @@ -2129,7 +2130,7 @@ xlogrecovery_redo(XLogReaderState *record, TimeLineID replayTLI) backupEndPoint = lsn; } else - elog(DEBUG1, "saw end-of-backup record for backup starting at %X/%X, waiting for %X/%X", + elog(DEBUG1, "saw end-of-backup record for backup starting at %X/%08X, waiting for %X/%08X", LSN_FORMAT_ARGS(startpoint), LSN_FORMAT_ARGS(backupStartPoint)); } } @@ -2224,9 +2225,9 @@ CheckRecoveryConsistency(void) backupEndRequired = false; ereport(LOG, - (errmsg("completed backup recovery with redo LSN %X/%X and end LSN %X/%X", - LSN_FORMAT_ARGS(saveBackupStartPoint), - LSN_FORMAT_ARGS(saveBackupEndPoint)))); + errmsg("completed backup recovery with redo LSN %X/%08X and end LSN %X/%08X", + LSN_FORMAT_ARGS(saveBackupStartPoint), + LSN_FORMAT_ARGS(saveBackupEndPoint))); } /* @@ -2255,8 +2256,8 @@ CheckRecoveryConsistency(void) reachedConsistency = true; SendPostmasterSignal(PMSIGNAL_RECOVERY_CONSISTENT); ereport(LOG, - (errmsg("consistent recovery state reached at %X/%X", - LSN_FORMAT_ARGS(lastReplayedEndRecPtr)))); + errmsg("consistent recovery state reached at %X/%08X", + LSN_FORMAT_ARGS(lastReplayedEndRecPtr))); } /* @@ -2293,7 +2294,7 @@ rm_redo_error_callback(void *arg) xlog_block_info(&buf, record); /* translator: %s is a WAL record description */ - errcontext("WAL redo at %X/%X for %s", + errcontext("WAL redo at %X/%08X for %s", LSN_FORMAT_ARGS(record->ReadRecPtr), buf.data); @@ -2328,7 +2329,7 @@ xlog_outdesc(StringInfo buf, XLogReaderState *record) static void xlog_outrec(StringInfo buf, XLogReaderState *record) { - appendStringInfo(buf, "prev %X/%X; xid %u", + appendStringInfo(buf, "prev %X/%08X; xid %u", LSN_FORMAT_ARGS(XLogRecGetPrev(record)), XLogRecGetXid(record)); @@ -2416,10 +2417,10 @@ checkTimeLineSwitch(XLogRecPtr lsn, TimeLineID newTLI, TimeLineID prevTLI, lsn < minRecoveryPoint && newTLI > minRecoveryPointTLI) ereport(PANIC, - (errmsg("unexpected timeline ID %u in checkpoint record, before reaching minimum recovery point %X/%X on timeline %u", - newTLI, - LSN_FORMAT_ARGS(minRecoveryPoint), - minRecoveryPointTLI))); + errmsg("unexpected timeline ID %u in checkpoint record, before reaching minimum recovery point %X/%08X on timeline %u", + newTLI, + LSN_FORMAT_ARGS(minRecoveryPoint), + minRecoveryPointTLI)); /* Looks good */ } @@ -2621,8 +2622,8 @@ recoveryStopsBefore(XLogReaderState *record) recoveryStopTime = 0; recoveryStopName[0] = '\0'; ereport(LOG, - (errmsg("recovery stopping before WAL location (LSN) \"%X/%X\"", - LSN_FORMAT_ARGS(recoveryStopLSN)))); + errmsg("recovery stopping before WAL location (LSN) \"%X/%08X\"", + LSN_FORMAT_ARGS(recoveryStopLSN))); return true; } @@ -2789,8 +2790,8 @@ recoveryStopsAfter(XLogReaderState *record) recoveryStopTime = 0; recoveryStopName[0] = '\0'; ereport(LOG, - (errmsg("recovery stopping after WAL location (LSN) \"%X/%X\"", - LSN_FORMAT_ARGS(recoveryStopLSN)))); + errmsg("recovery stopping after WAL location (LSN) \"%X/%08X\"", + LSN_FORMAT_ARGS(recoveryStopLSN))); return true; } @@ -2910,7 +2911,7 @@ getRecoveryStopReason(void) timestamptz_to_str(recoveryStopTime)); else if (recoveryTarget == RECOVERY_TARGET_LSN) snprintf(reason, sizeof(reason), - "%s LSN %X/%X\n", + "%s LSN %X/%08X\n", recoveryStopAfter ? "after" : "before", LSN_FORMAT_ARGS(recoveryStopLSN)); else if (recoveryTarget == RECOVERY_TARGET_NAME) @@ -3213,11 +3214,11 @@ ReadRecord(XLogPrefetcher *xlogprefetcher, int emode, XLogFileName(fname, xlogreader->seg.ws_tli, segno, wal_segment_size); ereport(emode_for_corrupt_record(emode, xlogreader->EndRecPtr), - (errmsg("unexpected timeline ID %u in WAL segment %s, LSN %X/%X, offset %u", - xlogreader->latestPageTLI, - fname, - LSN_FORMAT_ARGS(xlogreader->latestPagePtr), - offset))); + errmsg("unexpected timeline ID %u in WAL segment %s, LSN %X/%08X, offset %u", + xlogreader->latestPageTLI, + fname, + LSN_FORMAT_ARGS(xlogreader->latestPagePtr), + offset)); record = NULL; } @@ -3429,14 +3430,14 @@ retry: errno = save_errno; ereport(emode_for_corrupt_record(emode, targetPagePtr + reqLen), (errcode_for_file_access(), - errmsg("could not read from WAL segment %s, LSN %X/%X, offset %u: %m", + errmsg("could not read from WAL segment %s, LSN %X/%08X, offset %u: %m", fname, LSN_FORMAT_ARGS(targetPagePtr), readOff))); } else ereport(emode_for_corrupt_record(emode, targetPagePtr + reqLen), (errcode(ERRCODE_DATA_CORRUPTED), - errmsg("could not read from WAL segment %s, LSN %X/%X, offset %u: read %d of %zu", + errmsg("could not read from WAL segment %s, LSN %X/%08X, offset %u: read %d of %zu", fname, LSN_FORMAT_ARGS(targetPagePtr), readOff, r, (Size) XLOG_BLCKSZ))); goto next_record_is_invalid; @@ -3718,7 +3719,7 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess, wait_time = wal_retrieve_retry_interval - TimestampDifferenceMilliseconds(last_fail_time, now); - elog(LOG, "waiting for WAL to become available at %X/%X", + elog(LOG, "waiting for WAL to become available at %X/%08X", LSN_FORMAT_ARGS(RecPtr)); /* Do background tasks that might benefit us later. */ @@ -3864,7 +3865,7 @@ WaitForWALToBecomeAvailable(XLogRecPtr RecPtr, bool randAccess, tli = tliOfPointInHistory(tliRecPtr, expectedTLEs); if (curFileTLI > 0 && tli < curFileTLI) - elog(ERROR, "according to history file, WAL location %X/%X belongs to timeline %u, but previous recovered WAL file came from timeline %u", + elog(ERROR, "according to history file, WAL location %X/%08X belongs to timeline %u, but previous recovered WAL file came from timeline %u", LSN_FORMAT_ARGS(tliRecPtr), tli, curFileTLI); } @@ -4177,10 +4178,10 @@ rescanLatestTimeLine(TimeLineID replayTLI, XLogRecPtr replayLSN) if (currentTle->end < replayLSN) { ereport(LOG, - (errmsg("new timeline %u forked off current database system timeline %u before current recovery point %X/%X", - newtarget, - replayTLI, - LSN_FORMAT_ARGS(replayLSN)))); + errmsg("new timeline %u forked off current database system timeline %u before current recovery point %X/%08X", + newtarget, + replayTLI, + LSN_FORMAT_ARGS(replayLSN))); return false; } @@ -4760,7 +4761,7 @@ bool check_primary_slot_name(char **newval, void **extra, GucSource source) { if (*newval && strcmp(*newval, "") != 0 && - !ReplicationSlotValidateName(*newval, WARNING)) + !ReplicationSlotValidateName(*newval, false, WARNING)) return false; return true; @@ -4994,13 +4995,25 @@ check_recovery_target_timeline(char **newval, void **extra, GucSource source) rttg = RECOVERY_TARGET_TIMELINE_LATEST; else { + char *endp; + uint64 timeline; + rttg = RECOVERY_TARGET_TIMELINE_NUMERIC; errno = 0; - strtoul(*newval, NULL, 0); - if (errno == EINVAL || errno == ERANGE) + timeline = strtou64(*newval, &endp, 0); + + if (*endp != '\0' || errno == EINVAL || errno == ERANGE) + { + GUC_check_errdetail("\"%s\" is not a valid number.", + "recovery_target_timeline"); + return false; + } + + if (timeline < 1 || timeline > PG_UINT32_MAX) { - GUC_check_errdetail("\"recovery_target_timeline\" is not a valid number."); + GUC_check_errdetail("\"%s\" must be between %u and %u.", + "recovery_target_timeline", 1, UINT_MAX); return false; } } diff --git a/src/backend/access/transam/xlogutils.c b/src/backend/access/transam/xlogutils.c index c389b27f77d..27ea52fdfee 100644 --- a/src/backend/access/transam/xlogutils.c +++ b/src/backend/access/transam/xlogutils.c @@ -795,7 +795,7 @@ XLogReadDetermineTimeline(XLogReaderState *state, XLogRecPtr wantPage, list_free_deep(timelineHistory); - elog(DEBUG3, "switched to timeline %u valid until %X/%X", + elog(DEBUG3, "switched to timeline %u valid until %X/%08X", state->currTLI, LSN_FORMAT_ARGS(state->currTLIValidUntil)); } diff --git a/src/backend/backup/backup_manifest.c b/src/backend/backup/backup_manifest.c index 22e2be37c95..d05252f383c 100644 --- a/src/backend/backup/backup_manifest.c +++ b/src/backend/backup/backup_manifest.c @@ -281,7 +281,7 @@ AddWALInfoToBackupManifest(backup_manifest_info *manifest, XLogRecPtr startptr, } AppendToManifest(manifest, - "%s{ \"Timeline\": %u, \"Start-LSN\": \"%X/%X\", \"End-LSN\": \"%X/%X\" }", + "%s{ \"Timeline\": %u, \"Start-LSN\": \"%X/%08X\", \"End-LSN\": \"%X/%08X\" }", first_wal_range ? "" : ",\n", entry->tli, LSN_FORMAT_ARGS(tl_beginptr), diff --git a/src/backend/backup/basebackup_copy.c b/src/backend/backup/basebackup_copy.c index a284ce318ff..18b0b5a52d3 100644 --- a/src/backend/backup/basebackup_copy.c +++ b/src/backend/backup/basebackup_copy.c @@ -361,7 +361,7 @@ SendXlogRecPtrResult(XLogRecPtr ptr, TimeLineID tli) tstate = begin_tup_output_tupdesc(dest, tupdesc, &TTSOpsVirtual); /* Data row */ - values[0] = CStringGetTextDatum(psprintf("%X/%X", LSN_FORMAT_ARGS(ptr))); + values[0] = CStringGetTextDatum(psprintf("%X/%08X", LSN_FORMAT_ARGS(ptr))); values[1] = Int64GetDatum(tli); do_tup_output(tstate, values, nulls); diff --git a/src/backend/backup/basebackup_incremental.c b/src/backend/backup/basebackup_incremental.c index 28491b1e0ab..a0d48ff0fef 100644 --- a/src/backend/backup/basebackup_incremental.c +++ b/src/backend/backup/basebackup_incremental.c @@ -409,7 +409,7 @@ PrepareForIncrementalBackup(IncrementalBackupInfo *ib, if (range->start_lsn < tlep[i]->begin) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("manifest requires WAL from initial timeline %u starting at %X/%X, but that timeline begins at %X/%X", + errmsg("manifest requires WAL from initial timeline %u starting at %X/%08X, but that timeline begins at %X/%08X", range->tli, LSN_FORMAT_ARGS(range->start_lsn), LSN_FORMAT_ARGS(tlep[i]->begin)))); @@ -419,7 +419,7 @@ PrepareForIncrementalBackup(IncrementalBackupInfo *ib, if (range->start_lsn != tlep[i]->begin) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("manifest requires WAL from continuation timeline %u starting at %X/%X, but that timeline begins at %X/%X", + errmsg("manifest requires WAL from continuation timeline %u starting at %X/%08X, but that timeline begins at %X/%08X", range->tli, LSN_FORMAT_ARGS(range->start_lsn), LSN_FORMAT_ARGS(tlep[i]->begin)))); @@ -430,7 +430,7 @@ PrepareForIncrementalBackup(IncrementalBackupInfo *ib, if (range->end_lsn > backup_state->startpoint) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("manifest requires WAL from final timeline %u ending at %X/%X, but this backup starts at %X/%X", + errmsg("manifest requires WAL from final timeline %u ending at %X/%08X, but this backup starts at %X/%08X", range->tli, LSN_FORMAT_ARGS(range->end_lsn), LSN_FORMAT_ARGS(backup_state->startpoint)), @@ -441,7 +441,7 @@ PrepareForIncrementalBackup(IncrementalBackupInfo *ib, if (range->end_lsn != tlep[i]->end) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("manifest requires WAL from non-final timeline %u ending at %X/%X, but this server switched timelines at %X/%X", + errmsg("manifest requires WAL from non-final timeline %u ending at %X/%08X, but this server switched timelines at %X/%08X", range->tli, LSN_FORMAT_ARGS(range->end_lsn), LSN_FORMAT_ARGS(tlep[i]->end)))); @@ -522,18 +522,18 @@ PrepareForIncrementalBackup(IncrementalBackupInfo *ib, if (XLogRecPtrIsInvalid(tli_missing_lsn)) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("WAL summaries are required on timeline %u from %X/%X to %X/%X, but no summaries for that timeline and LSN range exist", + errmsg("WAL summaries are required on timeline %u from %X/%08X to %X/%08X, but no summaries for that timeline and LSN range exist", tle->tli, LSN_FORMAT_ARGS(tli_start_lsn), LSN_FORMAT_ARGS(tli_end_lsn)))); else ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("WAL summaries are required on timeline %u from %X/%X to %X/%X, but the summaries for that timeline and LSN range are incomplete", + errmsg("WAL summaries are required on timeline %u from %X/%08X to %X/%08X, but the summaries for that timeline and LSN range are incomplete", tle->tli, LSN_FORMAT_ARGS(tli_start_lsn), LSN_FORMAT_ARGS(tli_end_lsn)), - errdetail("The first unsummarized LSN in this range is %X/%X.", + errdetail("The first unsummarized LSN in this range is %X/%08X.", LSN_FORMAT_ARGS(tli_missing_lsn)))); } diff --git a/src/backend/bootstrap/bootstrap.c b/src/backend/bootstrap/bootstrap.c index 6db864892d0..fc8638c1b61 100644 --- a/src/backend/bootstrap/bootstrap.c +++ b/src/backend/bootstrap/bootstrap.c @@ -109,6 +109,8 @@ static const struct typinfo TypInfo[] = { F_REGROLEIN, F_REGROLEOUT}, {"regnamespace", REGNAMESPACEOID, 0, 4, true, TYPALIGN_INT, TYPSTORAGE_PLAIN, InvalidOid, F_REGNAMESPACEIN, F_REGNAMESPACEOUT}, + {"regdatabase", REGDATABASEOID, 0, 4, true, TYPALIGN_INT, TYPSTORAGE_PLAIN, InvalidOid, + F_REGDATABASEIN, F_REGDATABASEOUT}, {"text", TEXTOID, 0, -1, false, TYPALIGN_INT, TYPSTORAGE_EXTENDED, DEFAULT_COLLATION_OID, F_TEXTIN, F_TEXTOUT}, {"oid", OIDOID, 0, 4, true, TYPALIGN_INT, TYPSTORAGE_PLAIN, InvalidOid, diff --git a/src/backend/catalog/dependency.c b/src/backend/catalog/dependency.c index 18316a3968b..7dded634eb8 100644 --- a/src/backend/catalog/dependency.c +++ b/src/backend/catalog/dependency.c @@ -1850,6 +1850,17 @@ find_expr_references_walker(Node *node, errmsg("constant of the type %s cannot be used here", "regrole"))); break; + + /* + * Dependencies for regdatabase should be shared among all + * databases, so explicitly inhibit to have dependencies. + */ + case REGDATABASEOID: + ereport(ERROR, + (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("constant of the type %s cannot be used here", + "regdatabase"))); + break; } } return false; diff --git a/src/backend/catalog/heap.c b/src/backend/catalog/heap.c index fbaed5359ad..fd6537567ea 100644 --- a/src/backend/catalog/heap.c +++ b/src/backend/catalog/heap.c @@ -665,6 +665,15 @@ CheckAttributeType(const char *attname, } /* + * For consistency with check_virtual_generated_security(). + */ + if ((flags & CHKATYPE_IS_VIRTUAL) && atttypid >= FirstUnpinnedObjectId) + ereport(ERROR, + errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("virtual generated column \"%s\" cannot have a user-defined type", attname), + errdetail("Virtual generated columns that make use of user-defined types are not yet supported.")); + + /* * This might not be strictly invalid per SQL standard, but it is pretty * useless, and it cannot be dumped, so we must disallow it. */ @@ -1100,6 +1109,7 @@ AddNewRelationType(const char *typeName, * if false, relacl is always set NULL * allow_system_table_mods: true to allow creation in system namespaces * is_internal: is this a system-generated catalog? + * relrewrite: link to original relation during a table rewrite * * Output parameters: * typaddress: if not null, gets the object address of the new pg_type entry @@ -2996,7 +3006,7 @@ AddRelationNotNullConstraints(Relation rel, List *constraints, if (constr->is_no_inherit) ereport(ERROR, (errcode(ERRCODE_DATATYPE_MISMATCH), - errmsg("cannot define not-null constraint on column \"%s\" with NO INHERIT", + errmsg("cannot define not-null constraint with NO INHERIT on column \"%s\"", strVal(linitial(constr->keys))), errdetail("The column has an inherited not-null constraint."))); @@ -3215,6 +3225,86 @@ check_nested_generated(ParseState *pstate, Node *node) } /* + * Check security of virtual generated column expression. + * + * Just like selecting from a view is exploitable (CVE-2024-7348), selecting + * from a table with virtual generated columns is exploitable. Users who are + * concerned about this can avoid selecting from views, but telling them to + * avoid selecting from tables is less practical. + * + * To address this, this restricts generation expressions for virtual + * generated columns are restricted to using built-in functions and types. We + * assume that built-in functions and types cannot be exploited for this + * purpose. Note the overall security also requires that all functions in use + * a immutable. (For example, there are some built-in non-immutable functions + * that can run arbitrary SQL.) The immutability is checked elsewhere, since + * that is a property that needs to hold independent of security + * considerations. + * + * In the future, this could be expanded by some new mechanism to declare + * other functions and types as safe or trusted for this purpose, but that is + * to be designed. + */ + +/* + * Callback for check_functions_in_node() that determines whether a function + * is user-defined. + */ +static bool +contains_user_functions_checker(Oid func_id, void *context) +{ + return (func_id >= FirstUnpinnedObjectId); +} + +/* + * Checks for all the things we don't want in the generation expressions of + * virtual generated columns for security reasons. Errors out if it finds + * one. + */ +static bool +check_virtual_generated_security_walker(Node *node, void *context) +{ + ParseState *pstate = context; + + if (node == NULL) + return false; + + if (!IsA(node, List)) + { + if (check_functions_in_node(node, contains_user_functions_checker, NULL)) + ereport(ERROR, + errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("generation expression uses user-defined function"), + errdetail("Virtual generated columns that make use of user-defined functions are not yet supported."), + parser_errposition(pstate, exprLocation(node))); + + /* + * check_functions_in_node() doesn't check some node types (see + * comment there). We handle CoerceToDomain and MinMaxExpr by + * checking for built-in types. The other listed node types cannot + * call user-definable SQL-visible functions. + * + * We furthermore need this type check to handle built-in, immutable + * polymorphic functions such as array_eq(). + */ + if (exprType(node) >= FirstUnpinnedObjectId) + ereport(ERROR, + errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("generation expression uses user-defined type"), + errdetail("Virtual generated columns that make use of user-defined types are not yet supported."), + parser_errposition(pstate, exprLocation(node))); + } + + return expression_tree_walker(node, check_virtual_generated_security_walker, context); +} + +static void +check_virtual_generated_security(ParseState *pstate, Node *node) +{ + check_virtual_generated_security_walker(node, pstate); +} + +/* * Take a raw default and convert it to a cooked format ready for * storage. * @@ -3253,6 +3343,10 @@ cookDefault(ParseState *pstate, ereport(ERROR, (errcode(ERRCODE_INVALID_OBJECT_DEFINITION), errmsg("generation expression is not immutable"))); + + /* Check security of expressions for virtual generated column */ + if (attgenerated == ATTRIBUTE_GENERATED_VIRTUAL) + check_virtual_generated_security(pstate, expr); } else { diff --git a/src/backend/catalog/index.c b/src/backend/catalog/index.c index 739a92bdcc1..c4029a4f3d3 100644 --- a/src/backend/catalog/index.c +++ b/src/backend/catalog/index.c @@ -800,11 +800,11 @@ index_create(Relation heapRelation, errmsg("user-defined indexes on system catalog tables are not supported"))); /* - * Btree text_pattern_ops uses text_eq as the equality operator, which is - * fine as long as the collation is deterministic; text_eq then reduces to + * Btree text_pattern_ops uses texteq as the equality operator, which is + * fine as long as the collation is deterministic; texteq then reduces to * bitwise equality and so it is semantically compatible with the other * operators and functions in that opclass. But with a nondeterministic - * collation, text_eq could yield results that are incompatible with the + * collation, texteq could yield results that are incompatible with the * actual behavior of the index (which is determined by the opclass's * comparison function). We prevent such problems by refusing creation of * an index with that opclass and a nondeterministic collation. @@ -814,7 +814,7 @@ index_create(Relation heapRelation, * opclasses as incompatible with nondeterminism; but for now, this small * hack suffices. * - * Another solution is to use a special operator, not text_eq, as the + * Another solution is to use a special operator, not texteq, as the * equality opclass member; but that is undesirable because it would * prevent index usage in many queries that work fine today. */ @@ -3020,7 +3020,7 @@ index_build(Relation heapRelation, /* * Determine worker process details for parallel CREATE INDEX. Currently, - * only btree and BRIN have support for parallel builds. + * only btree, GIN, and BRIN have support for parallel builds. * * Note that planner considers parallel safety for us. */ diff --git a/src/backend/catalog/pg_subscription.c b/src/backend/catalog/pg_subscription.c index 1395032413e..244acf52f36 100644 --- a/src/backend/catalog/pg_subscription.c +++ b/src/backend/catalog/pg_subscription.c @@ -103,6 +103,7 @@ GetSubscription(Oid subid, bool missing_ok) sub->passwordrequired = subform->subpasswordrequired; sub->runasowner = subform->subrunasowner; sub->failover = subform->subfailover; + sub->retaindeadtuples = subform->subretaindeadtuples; /* Get conninfo */ datum = SysCacheGetAttrNotNull(SUBSCRIPTIONOID, @@ -319,7 +320,7 @@ AddSubscriptionRelState(Oid subid, Oid relid, char state, */ void UpdateSubscriptionRelState(Oid subid, Oid relid, char state, - XLogRecPtr sublsn) + XLogRecPtr sublsn, bool already_locked) { Relation rel; HeapTuple tup; @@ -327,9 +328,24 @@ UpdateSubscriptionRelState(Oid subid, Oid relid, char state, Datum values[Natts_pg_subscription_rel]; bool replaces[Natts_pg_subscription_rel]; - LockSharedObject(SubscriptionRelationId, subid, 0, AccessShareLock); + if (already_locked) + { +#ifdef USE_ASSERT_CHECKING + LOCKTAG tag; - rel = table_open(SubscriptionRelRelationId, RowExclusiveLock); + Assert(CheckRelationOidLockedByMe(SubscriptionRelRelationId, + RowExclusiveLock, true)); + SET_LOCKTAG_OBJECT(tag, InvalidOid, SubscriptionRelationId, subid, 0); + Assert(LockHeldByMe(&tag, AccessShareLock, true)); +#endif + + rel = table_open(SubscriptionRelRelationId, NoLock); + } + else + { + LockSharedObject(SubscriptionRelationId, subid, 0, AccessShareLock); + rel = table_open(SubscriptionRelRelationId, RowExclusiveLock); + } /* Try finding existing mapping. */ tup = SearchSysCacheCopy2(SUBSCRIPTIONRELMAP, diff --git a/src/backend/catalog/system_views.sql b/src/backend/catalog/system_views.sql index 08f780a2e63..77c693f630e 100644 --- a/src/backend/catalog/system_views.sql +++ b/src/backend/catalog/system_views.sql @@ -666,6 +666,14 @@ GRANT SELECT ON pg_shmem_allocations_numa TO pg_read_all_stats; REVOKE EXECUTE ON FUNCTION pg_get_shmem_allocations_numa() FROM PUBLIC; GRANT EXECUTE ON FUNCTION pg_get_shmem_allocations_numa() TO pg_read_all_stats; +CREATE VIEW pg_dsm_registry_allocations AS + SELECT * FROM pg_get_dsm_registry_allocations(); + +REVOKE ALL ON pg_dsm_registry_allocations FROM PUBLIC; +GRANT SELECT ON pg_dsm_registry_allocations TO pg_read_all_stats; +REVOKE EXECUTE ON FUNCTION pg_get_dsm_registry_allocations() FROM PUBLIC; +GRANT EXECUTE ON FUNCTION pg_get_dsm_registry_allocations() TO pg_read_all_stats; + CREATE VIEW pg_backend_memory_contexts AS SELECT * FROM pg_get_backend_memory_contexts(); @@ -895,7 +903,7 @@ CREATE VIEW pg_stat_activity AS S.wait_event, S.state, S.backend_xid, - s.backend_xmin, + S.backend_xmin, S.query_id, S.query, S.backend_type @@ -1378,7 +1386,8 @@ REVOKE ALL ON pg_subscription FROM public; GRANT SELECT (oid, subdbid, subskiplsn, subname, subowner, subenabled, subbinary, substream, subtwophasestate, subdisableonerr, subpasswordrequired, subrunasowner, subfailover, - subslotname, subsynccommit, subpublications, suborigin) + subretaindeadtuples, subslotname, subsynccommit, + subpublications, suborigin) ON pg_subscription TO public; CREATE VIEW pg_stat_subscription_stats AS @@ -1390,6 +1399,7 @@ CREATE VIEW pg_stat_subscription_stats AS ss.confl_insert_exists, ss.confl_update_origin_differs, ss.confl_update_exists, + ss.confl_update_deleted, ss.confl_update_missing, ss.confl_delete_origin_differs, ss.confl_delete_missing, diff --git a/src/backend/commands/analyze.c b/src/backend/commands/analyze.c index 4fffb76e557..40d66537ad7 100644 --- a/src/backend/commands/analyze.c +++ b/src/backend/commands/analyze.c @@ -76,7 +76,7 @@ static BufferAccessStrategy vac_strategy; static void do_analyze_rel(Relation onerel, - VacuumParams *params, List *va_cols, + const VacuumParams params, List *va_cols, AcquireSampleRowsFunc acquirefunc, BlockNumber relpages, bool inh, bool in_outer_xact, int elevel); static void compute_index_stats(Relation onerel, double totalrows, @@ -107,7 +107,7 @@ static Datum ind_fetch_func(VacAttrStatsP stats, int rownum, bool *isNull); */ void analyze_rel(Oid relid, RangeVar *relation, - VacuumParams *params, List *va_cols, bool in_outer_xact, + const VacuumParams params, List *va_cols, bool in_outer_xact, BufferAccessStrategy bstrategy) { Relation onerel; @@ -116,7 +116,7 @@ analyze_rel(Oid relid, RangeVar *relation, BlockNumber relpages = 0; /* Select logging level */ - if (params->options & VACOPT_VERBOSE) + if (params.options & VACOPT_VERBOSE) elevel = INFO; else elevel = DEBUG2; @@ -138,8 +138,8 @@ analyze_rel(Oid relid, RangeVar *relation, * * Make sure to generate only logs for ANALYZE in this case. */ - onerel = vacuum_open_relation(relid, relation, params->options & ~(VACOPT_VACUUM), - params->log_min_duration >= 0, + onerel = vacuum_open_relation(relid, relation, params.options & ~(VACOPT_VACUUM), + params.log_min_duration >= 0, ShareUpdateExclusiveLock); /* leave if relation could not be opened or locked */ @@ -155,7 +155,7 @@ analyze_rel(Oid relid, RangeVar *relation, */ if (!vacuum_is_permitted_for_relation(RelationGetRelid(onerel), onerel->rd_rel, - params->options & ~VACOPT_VACUUM)) + params.options & ~VACOPT_VACUUM)) { relation_close(onerel, ShareUpdateExclusiveLock); return; @@ -227,7 +227,7 @@ analyze_rel(Oid relid, RangeVar *relation, else { /* No need for a WARNING if we already complained during VACUUM */ - if (!(params->options & VACOPT_VACUUM)) + if (!(params.options & VACOPT_VACUUM)) ereport(WARNING, (errmsg("skipping \"%s\" --- cannot analyze non-tables or special system tables", RelationGetRelationName(onerel)))); @@ -275,7 +275,7 @@ analyze_rel(Oid relid, RangeVar *relation, * appropriate acquirefunc for each child table. */ static void -do_analyze_rel(Relation onerel, VacuumParams *params, +do_analyze_rel(Relation onerel, const VacuumParams params, List *va_cols, AcquireSampleRowsFunc acquirefunc, BlockNumber relpages, bool inh, bool in_outer_xact, int elevel) @@ -309,9 +309,9 @@ do_analyze_rel(Relation onerel, VacuumParams *params, PgStat_Counter startreadtime = 0; PgStat_Counter startwritetime = 0; - verbose = (params->options & VACOPT_VERBOSE) != 0; + verbose = (params.options & VACOPT_VERBOSE) != 0; instrument = (verbose || (AmAutoVacuumWorkerProcess() && - params->log_min_duration >= 0)); + params.log_min_duration >= 0)); if (inh) ereport(elevel, (errmsg("analyzing \"%s.%s\" inheritance tree", @@ -690,8 +690,8 @@ do_analyze_rel(Relation onerel, VacuumParams *params, * only do it for inherited stats. (We're never called for not-inherited * stats on partitioned tables anyway.) * - * Reset the changes_since_analyze counter only if we analyzed all - * columns; otherwise, there is still work for auto-analyze to do. + * Reset the mod_since_analyze counter only if we analyzed all columns; + * otherwise, there is still work for auto-analyze to do. */ if (!inh) pgstat_report_analyze(onerel, totalrows, totaldeadrows, @@ -706,7 +706,7 @@ do_analyze_rel(Relation onerel, VacuumParams *params, * amvacuumcleanup() when called in ANALYZE-only mode. The only exception * among core index AMs is GIN/ginvacuumcleanup(). */ - if (!(params->options & VACOPT_VACUUM)) + if (!(params.options & VACOPT_VACUUM)) { for (ind = 0; ind < nindexes; ind++) { @@ -736,9 +736,9 @@ do_analyze_rel(Relation onerel, VacuumParams *params, { TimestampTz endtime = GetCurrentTimestamp(); - if (verbose || params->log_min_duration == 0 || + if (verbose || params.log_min_duration == 0 || TimestampDifferenceExceeds(starttime, endtime, - params->log_min_duration)) + params.log_min_duration)) { long delay_in_ms; WalUsage walusage; diff --git a/src/backend/commands/cluster.c b/src/backend/commands/cluster.c index 54a08e4102e..b55221d44cd 100644 --- a/src/backend/commands/cluster.c +++ b/src/backend/commands/cluster.c @@ -917,7 +917,7 @@ copy_table_data(Relation NewHeap, Relation OldHeap, Relation OldIndex, bool verb * not to be aggressive about this. */ memset(¶ms, 0, sizeof(VacuumParams)); - vacuum_get_cutoffs(OldHeap, ¶ms, &cutoffs); + vacuum_get_cutoffs(OldHeap, params, &cutoffs); /* * FreezeXid will become the table's new relfrozenxid, and that mustn't go diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c index 74ae42b19a7..fae9c41db65 100644 --- a/src/backend/commands/copy.c +++ b/src/backend/commands/copy.c @@ -322,11 +322,13 @@ DoCopy(ParseState *pstate, const CopyStmt *stmt, } /* - * Extract a CopyHeaderChoice value from a DefElem. This is like - * defGetBoolean() but also accepts the special value "match". + * Extract the CopyFormatOptions.header_line value from a DefElem. + * + * Parses the HEADER option for COPY, which can be a boolean, a non-negative + * integer (number of lines to skip), or the special value "match". */ -static CopyHeaderChoice -defGetCopyHeaderChoice(DefElem *def, bool is_from) +static int +defGetCopyHeaderOption(DefElem *def, bool is_from) { /* * If no parameter value given, assume "true" is meant. @@ -335,20 +337,27 @@ defGetCopyHeaderChoice(DefElem *def, bool is_from) return COPY_HEADER_TRUE; /* - * Allow 0, 1, "true", "false", "on", "off", or "match". + * Allow 0, 1, "true", "false", "on", "off", a non-negative integer, or + * "match". */ switch (nodeTag(def->arg)) { case T_Integer: - switch (intVal(def->arg)) { - case 0: - return COPY_HEADER_FALSE; - case 1: - return COPY_HEADER_TRUE; - default: - /* otherwise, error out below */ - break; + int ival = intVal(def->arg); + + if (ival < 0) + ereport(ERROR, + (errcode(ERRCODE_INVALID_PARAMETER_VALUE), + errmsg("a negative integer value cannot be " + "specified for %s", def->defname))); + + if (!is_from && ival > 1) + ereport(ERROR, + (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("cannot use multi-line header in COPY TO"))); + + return ival; } break; default: @@ -381,7 +390,8 @@ defGetCopyHeaderChoice(DefElem *def, bool is_from) } ereport(ERROR, (errcode(ERRCODE_SYNTAX_ERROR), - errmsg("%s requires a Boolean value or \"match\"", + errmsg("%s requires a Boolean value, a non-negative integer, " + "or the string \"match\"", def->defname))); return COPY_HEADER_FALSE; /* keep compiler quiet */ } @@ -566,7 +576,7 @@ ProcessCopyOptions(ParseState *pstate, if (header_specified) errorConflictingDefElem(defel, pstate); header_specified = true; - opts_out->header_line = defGetCopyHeaderChoice(defel, is_from); + opts_out->header_line = defGetCopyHeaderOption(defel, is_from); } else if (strcmp(defel->defname, "quote") == 0) { @@ -769,7 +779,7 @@ ProcessCopyOptions(ParseState *pstate, errmsg("COPY delimiter cannot be \"%s\"", opts_out->delim))); /* Check header */ - if (opts_out->binary && opts_out->header_line) + if (opts_out->binary && opts_out->header_line != COPY_HEADER_FALSE) ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), /*- translator: %s is the name of a COPY option, e.g. ON_ERROR */ diff --git a/src/backend/commands/copyfromparse.c b/src/backend/commands/copyfromparse.c index f5fc346e201..b1ae97b833d 100644 --- a/src/backend/commands/copyfromparse.c +++ b/src/backend/commands/copyfromparse.c @@ -771,21 +771,30 @@ static pg_attribute_always_inline bool NextCopyFromRawFieldsInternal(CopyFromState cstate, char ***fields, int *nfields, bool is_csv) { int fldct; - bool done; + bool done = false; /* only available for text or csv input */ Assert(!cstate->opts.binary); /* on input check that the header line is correct if needed */ - if (cstate->cur_lineno == 0 && cstate->opts.header_line) + if (cstate->cur_lineno == 0 && cstate->opts.header_line != COPY_HEADER_FALSE) { ListCell *cur; TupleDesc tupDesc; + int lines_to_skip = cstate->opts.header_line; + + /* If set to "match", one header line is skipped */ + if (cstate->opts.header_line == COPY_HEADER_MATCH) + lines_to_skip = 1; tupDesc = RelationGetDescr(cstate->rel); - cstate->cur_lineno++; - done = CopyReadLine(cstate, is_csv); + for (int i = 0; i < lines_to_skip; i++) + { + cstate->cur_lineno++; + if ((done = CopyReadLine(cstate, is_csv))) + break; + } if (cstate->opts.header_line == COPY_HEADER_MATCH) { @@ -1538,7 +1547,7 @@ GetDecimalFromHex(char hex) if (isdigit((unsigned char) hex)) return hex - '0'; else - return tolower((unsigned char) hex) - 'a' + 10; + return pg_ascii_tolower((unsigned char) hex) - 'a' + 10; } /* diff --git a/src/backend/commands/copyto.c b/src/backend/commands/copyto.c index ea6f18f2c80..67b94b91cae 100644 --- a/src/backend/commands/copyto.c +++ b/src/backend/commands/copyto.c @@ -199,7 +199,7 @@ CopyToTextLikeStart(CopyToState cstate, TupleDesc tupDesc) cstate->file_encoding); /* if a header has been requested send the line */ - if (cstate->opts.header_line) + if (cstate->opts.header_line == COPY_HEADER_TRUE) { ListCell *cur; bool hdr_delim = false; diff --git a/src/backend/commands/dbcommands.c b/src/backend/commands/dbcommands.c index 5fbbcdaabb1..502a45163c8 100644 --- a/src/backend/commands/dbcommands.c +++ b/src/backend/commands/dbcommands.c @@ -570,8 +570,8 @@ CreateDatabaseUsingFileCopy(Oid src_dboid, Oid dst_dboid, Oid src_tsid, * any CREATE DATABASE commands. */ if (!IsBinaryUpgrade) - RequestCheckpoint(CHECKPOINT_IMMEDIATE | CHECKPOINT_FORCE | - CHECKPOINT_WAIT | CHECKPOINT_FLUSH_ALL); + RequestCheckpoint(CHECKPOINT_FAST | CHECKPOINT_FORCE | + CHECKPOINT_WAIT | CHECKPOINT_FLUSH_UNLOGGED); /* * Iterate through all tablespaces of the template database, and copy each @@ -673,7 +673,7 @@ CreateDatabaseUsingFileCopy(Oid src_dboid, Oid dst_dboid, Oid src_tsid, * strategy that avoids these problems. */ if (!IsBinaryUpgrade) - RequestCheckpoint(CHECKPOINT_IMMEDIATE | CHECKPOINT_FORCE | + RequestCheckpoint(CHECKPOINT_FAST | CHECKPOINT_FORCE | CHECKPOINT_WAIT); } @@ -1065,16 +1065,41 @@ createdb(ParseState *pstate, const CreatedbStmt *stmt) /* Check that the chosen locales are valid, and get canonical spellings */ if (!check_locale(LC_COLLATE, dbcollate, &canonname)) - ereport(ERROR, - (errcode(ERRCODE_WRONG_OBJECT_TYPE), - errmsg("invalid LC_COLLATE locale name: \"%s\"", dbcollate), - errhint("If the locale name is specific to ICU, use ICU_LOCALE."))); + { + if (dblocprovider == COLLPROVIDER_BUILTIN) + ereport(ERROR, + (errcode(ERRCODE_WRONG_OBJECT_TYPE), + errmsg("invalid LC_COLLATE locale name: \"%s\"", dbcollate), + errhint("If the locale name is specific to the builtin provider, use BUILTIN_LOCALE."))); + else if (dblocprovider == COLLPROVIDER_ICU) + ereport(ERROR, + (errcode(ERRCODE_WRONG_OBJECT_TYPE), + errmsg("invalid LC_COLLATE locale name: \"%s\"", dbcollate), + errhint("If the locale name is specific to the ICU provider, use ICU_LOCALE."))); + else + ereport(ERROR, + (errcode(ERRCODE_WRONG_OBJECT_TYPE), + errmsg("invalid LC_COLLATE locale name: \"%s\"", dbcollate))); + } dbcollate = canonname; if (!check_locale(LC_CTYPE, dbctype, &canonname)) - ereport(ERROR, - (errcode(ERRCODE_WRONG_OBJECT_TYPE), - errmsg("invalid LC_CTYPE locale name: \"%s\"", dbctype), - errhint("If the locale name is specific to ICU, use ICU_LOCALE."))); + { + if (dblocprovider == COLLPROVIDER_BUILTIN) + ereport(ERROR, + (errcode(ERRCODE_WRONG_OBJECT_TYPE), + errmsg("invalid LC_CTYPE locale name: \"%s\"", dbctype), + errhint("If the locale name is specific to the builtin provider, use BUILTIN_LOCALE."))); + else if (dblocprovider == COLLPROVIDER_ICU) + ereport(ERROR, + (errcode(ERRCODE_WRONG_OBJECT_TYPE), + errmsg("invalid LC_CTYPE locale name: \"%s\"", dbctype), + errhint("If the locale name is specific to the ICU provider, use ICU_LOCALE."))); + else + ereport(ERROR, + (errcode(ERRCODE_WRONG_OBJECT_TYPE), + errmsg("invalid LC_CTYPE locale name: \"%s\"", dbctype))); + } + dbctype = canonname; check_encoding_locale_matches(encoding, dbcollate, dbctype); @@ -1845,7 +1870,7 @@ dropdb(const char *dbname, bool missing_ok, bool force) * Force a checkpoint to make sure the checkpointer has received the * message sent by ForgetDatabaseSyncRequests. */ - RequestCheckpoint(CHECKPOINT_IMMEDIATE | CHECKPOINT_FORCE | CHECKPOINT_WAIT); + RequestCheckpoint(CHECKPOINT_FAST | CHECKPOINT_FORCE | CHECKPOINT_WAIT); /* Close all smgr fds in all backends. */ WaitForProcSignalBarrier(EmitProcSignalBarrier(PROCSIGNAL_BARRIER_SMGRRELEASE)); @@ -2095,8 +2120,8 @@ movedb(const char *dbname, const char *tblspcname) * On Windows, this also ensures that background procs don't hold any open * files, which would cause rmdir() to fail. */ - RequestCheckpoint(CHECKPOINT_IMMEDIATE | CHECKPOINT_FORCE | CHECKPOINT_WAIT - | CHECKPOINT_FLUSH_ALL); + RequestCheckpoint(CHECKPOINT_FAST | CHECKPOINT_FORCE | CHECKPOINT_WAIT + | CHECKPOINT_FLUSH_UNLOGGED); /* Close all smgr fds in all backends. */ WaitForProcSignalBarrier(EmitProcSignalBarrier(PROCSIGNAL_BARRIER_SMGRRELEASE)); @@ -2227,7 +2252,7 @@ movedb(const char *dbname, const char *tblspcname) * any unlogged operations done in the new DB tablespace before the * next checkpoint. */ - RequestCheckpoint(CHECKPOINT_IMMEDIATE | CHECKPOINT_FORCE | CHECKPOINT_WAIT); + RequestCheckpoint(CHECKPOINT_FAST | CHECKPOINT_FORCE | CHECKPOINT_WAIT); /* * Force synchronous commit, thus minimizing the window between diff --git a/src/backend/commands/explain.c b/src/backend/commands/explain.c index 7e2792ead71..8345bc0264b 100644 --- a/src/backend/commands/explain.c +++ b/src/backend/commands/explain.c @@ -3582,6 +3582,7 @@ static void show_memoize_info(MemoizeState *mstate, List *ancestors, ExplainState *es) { Plan *plan = ((PlanState *) mstate)->plan; + Memoize *mplan = (Memoize *) plan; ListCell *lc; List *context; StringInfoData keystr; @@ -3602,7 +3603,7 @@ show_memoize_info(MemoizeState *mstate, List *ancestors, ExplainState *es) plan, ancestors); - foreach(lc, ((Memoize *) plan)->param_exprs) + foreach(lc, mplan->param_exprs) { Node *expr = (Node *) lfirst(lc); @@ -3618,6 +3619,24 @@ show_memoize_info(MemoizeState *mstate, List *ancestors, ExplainState *es) pfree(keystr.data); + if (es->costs) + { + if (es->format == EXPLAIN_FORMAT_TEXT) + { + ExplainIndentText(es); + appendStringInfo(es->str, "Estimates: capacity=%u distinct keys=%.0f lookups=%.0f hit percent=%.2f%%\n", + mplan->est_entries, mplan->est_unique_keys, + mplan->est_calls, mplan->est_hit_ratio * 100.0); + } + else + { + ExplainPropertyUInteger("Estimated Capacity", NULL, mplan->est_entries, es); + ExplainPropertyFloat("Estimated Distinct Lookup Keys", NULL, mplan->est_unique_keys, 0, es); + ExplainPropertyFloat("Estimated Lookups", NULL, mplan->est_calls, 0, es); + ExplainPropertyFloat("Estimated Hit Percent", NULL, mplan->est_hit_ratio * 100.0, 2, es); + } + } + if (!es->analyze) return; diff --git a/src/backend/commands/foreigncmds.c b/src/backend/commands/foreigncmds.c index c14e038d54f..77f8461f42e 100644 --- a/src/backend/commands/foreigncmds.c +++ b/src/backend/commands/foreigncmds.c @@ -71,15 +71,26 @@ optionListToArray(List *options) foreach(cell, options) { DefElem *def = lfirst(cell); + const char *name; const char *value; Size len; text *t; + name = def->defname; value = defGetString(def); - len = VARHDRSZ + strlen(def->defname) + 1 + strlen(value); + + /* Insist that name not contain "=", else "a=b=c" is ambiguous */ + if (strchr(name, '=') != NULL) + ereport(ERROR, + (errcode(ERRCODE_INVALID_PARAMETER_VALUE), + errmsg("invalid option name \"%s\": must not contain \"=\"", + name))); + + len = VARHDRSZ + strlen(name) + 1 + strlen(value); + /* +1 leaves room for sprintf's trailing null */ t = palloc(len + 1); SET_VARSIZE(t, len); - sprintf(VARDATA(t), "%s=%s", def->defname, value); + sprintf(VARDATA(t), "%s=%s", name, value); astate = accumArrayResult(astate, PointerGetDatum(t), false, TEXTOID, @@ -1577,6 +1588,7 @@ ImportForeignSchema(ImportForeignSchemaStmt *stmt) pstmt->utilityStmt = (Node *) cstmt; pstmt->stmt_location = rs->stmt_location; pstmt->stmt_len = rs->stmt_len; + pstmt->planOrigin = PLAN_STMT_INTERNAL; /* Execute statement */ ProcessUtility(pstmt, cmd, false, diff --git a/src/backend/commands/indexcmds.c b/src/backend/commands/indexcmds.c index c3ec2076a52..6f753ab6d7a 100644 --- a/src/backend/commands/indexcmds.c +++ b/src/backend/commands/indexcmds.c @@ -2469,8 +2469,8 @@ GetOperatorFromCompareType(Oid opclass, Oid rhstype, CompareType cmptype, cmptype == COMPARE_EQ ? errmsg("could not identify an equality operator for type %s", format_type_be(opcintype)) : cmptype == COMPARE_OVERLAP ? errmsg("could not identify an overlaps operator for type %s", format_type_be(opcintype)) : cmptype == COMPARE_CONTAINED_BY ? errmsg("could not identify a contained-by operator for type %s", format_type_be(opcintype)) : 0, - errdetail("Could not translate compare type %d for operator family \"%s\", input type %s, access method \"%s\".", - cmptype, get_opfamily_name(opfamily, false), format_type_be(opcintype), get_am_name(amid))); + errdetail("Could not translate compare type %d for operator family \"%s\" of access method \"%s\".", + cmptype, get_opfamily_name(opfamily, false), get_am_name(amid))); /* * We parameterize rhstype so foreign keys can ask for a <@ operator @@ -2592,7 +2592,9 @@ makeObjectName(const char *name1, const char *name2, const char *label) * constraint names.) * * Note: it is theoretically possible to get a collision anyway, if someone - * else chooses the same name concurrently. This is fairly unlikely to be + * else chooses the same name concurrently. We shorten the race condition + * window by checking for conflicting relations using SnapshotDirty, but + * that doesn't close the window entirely. This is fairly unlikely to be * a problem in practice, especially if one is holding an exclusive lock on * the relation identified by name1. However, if choosing multiple names * within a single command, you'd better create the new object and do @@ -2608,15 +2610,45 @@ ChooseRelationName(const char *name1, const char *name2, int pass = 0; char *relname = NULL; char modlabel[NAMEDATALEN]; + SnapshotData SnapshotDirty; + Relation pgclassrel; + + /* prepare to search pg_class with a dirty snapshot */ + InitDirtySnapshot(SnapshotDirty); + pgclassrel = table_open(RelationRelationId, AccessShareLock); /* try the unmodified label first */ strlcpy(modlabel, label, sizeof(modlabel)); for (;;) { + ScanKeyData key[2]; + SysScanDesc scan; + bool collides; + relname = makeObjectName(name1, name2, modlabel); - if (!OidIsValid(get_relname_relid(relname, namespaceid))) + /* is there any conflicting relation name? */ + ScanKeyInit(&key[0], + Anum_pg_class_relname, + BTEqualStrategyNumber, F_NAMEEQ, + CStringGetDatum(relname)); + ScanKeyInit(&key[1], + Anum_pg_class_relnamespace, + BTEqualStrategyNumber, F_OIDEQ, + ObjectIdGetDatum(namespaceid)); + + scan = systable_beginscan(pgclassrel, ClassNameNspIndexId, + true /* indexOK */ , + &SnapshotDirty, + 2, key); + + collides = HeapTupleIsValid(systable_getnext(scan)); + + systable_endscan(scan); + + /* break out of loop if no conflict */ + if (!collides) { if (!isconstraint || !ConstraintNameExists(relname, namespaceid)) @@ -2628,6 +2660,8 @@ ChooseRelationName(const char *name1, const char *name2, snprintf(modlabel, sizeof(modlabel), "%s%d", label, ++pass); } + table_close(pgclassrel, AccessShareLock); + return relname; } diff --git a/src/backend/commands/matview.c b/src/backend/commands/matview.c index 27c2cb26ef5..188e26f0e6e 100644 --- a/src/backend/commands/matview.c +++ b/src/backend/commands/matview.c @@ -835,7 +835,8 @@ refresh_by_match_merge(Oid matviewOid, Oid tempOid, Oid relowner, if (!foundUniqueIndex) ereport(ERROR, errcode(ERRCODE_FEATURE_NOT_SUPPORTED), - errmsg("could not find suitable unique index on materialized view")); + errmsg("could not find suitable unique index on materialized view \"%s\"", + RelationGetRelationName(matviewRel))); appendStringInfoString(&querybuf, " AND newdata.* OPERATOR(pg_catalog.*=) mv.*) " diff --git a/src/backend/commands/publicationcmds.c b/src/backend/commands/publicationcmds.c index 0b23d94c38e..803c26ab216 100644 --- a/src/backend/commands/publicationcmds.c +++ b/src/backend/commands/publicationcmds.c @@ -2113,25 +2113,25 @@ AlterPublicationOwner_oid(Oid pubid, Oid newOwnerId) static char defGetGeneratedColsOption(DefElem *def) { - char *sval; + char *sval = ""; /* - * If no parameter value given, assume "stored" is meant. + * A parameter value is required. */ - if (!def->arg) - return PUBLISH_GENCOLS_STORED; - - sval = defGetString(def); + if (def->arg) + { + sval = defGetString(def); - if (pg_strcasecmp(sval, "none") == 0) - return PUBLISH_GENCOLS_NONE; - if (pg_strcasecmp(sval, "stored") == 0) - return PUBLISH_GENCOLS_STORED; + if (pg_strcasecmp(sval, "none") == 0) + return PUBLISH_GENCOLS_NONE; + if (pg_strcasecmp(sval, "stored") == 0) + return PUBLISH_GENCOLS_STORED; + } ereport(ERROR, errcode(ERRCODE_SYNTAX_ERROR), - errmsg("%s requires a \"none\" or \"stored\" value", - def->defname)); + errmsg("invalid value for publication parameter \"%s\": \"%s\"", def->defname, sval), + errdetail("Valid values are \"%s\" and \"%s\".", "none", "stored")); return PUBLISH_GENCOLS_NONE; /* keep compiler quiet */ } diff --git a/src/backend/commands/schemacmds.c b/src/backend/commands/schemacmds.c index 546160f0941..0f03d9743d2 100644 --- a/src/backend/commands/schemacmds.c +++ b/src/backend/commands/schemacmds.c @@ -215,6 +215,7 @@ CreateSchemaCommand(CreateSchemaStmt *stmt, const char *queryString, wrapper->utilityStmt = stmt; wrapper->stmt_location = stmt_location; wrapper->stmt_len = stmt_len; + wrapper->planOrigin = PLAN_STMT_INTERNAL; /* do this step */ ProcessUtility(wrapper, diff --git a/src/backend/commands/subscriptioncmds.c b/src/backend/commands/subscriptioncmds.c index 4aec73bcc6b..cd6c3684482 100644 --- a/src/backend/commands/subscriptioncmds.c +++ b/src/backend/commands/subscriptioncmds.c @@ -14,6 +14,7 @@ #include "postgres.h" +#include "access/commit_ts.h" #include "access/htup_details.h" #include "access/table.h" #include "access/twophase.h" @@ -71,8 +72,9 @@ #define SUBOPT_PASSWORD_REQUIRED 0x00000800 #define SUBOPT_RUN_AS_OWNER 0x00001000 #define SUBOPT_FAILOVER 0x00002000 -#define SUBOPT_LSN 0x00004000 -#define SUBOPT_ORIGIN 0x00008000 +#define SUBOPT_RETAIN_DEAD_TUPLES 0x00004000 +#define SUBOPT_LSN 0x00008000 +#define SUBOPT_ORIGIN 0x00010000 /* check if the 'val' has 'bits' set */ #define IsSet(val, bits) (((val) & (bits)) == (bits)) @@ -98,6 +100,7 @@ typedef struct SubOpts bool passwordrequired; bool runasowner; bool failover; + bool retaindeadtuples; char *origin; XLogRecPtr lsn; } SubOpts; @@ -105,8 +108,10 @@ typedef struct SubOpts static List *fetch_table_list(WalReceiverConn *wrconn, List *publications); static void check_publications_origin(WalReceiverConn *wrconn, List *publications, bool copydata, - char *origin, Oid *subrel_local_oids, - int subrel_count, char *subname); + bool retain_dead_tuples, char *origin, + Oid *subrel_local_oids, int subrel_count, + char *subname); +static void check_pub_dead_tuple_retention(WalReceiverConn *wrconn); static void check_duplicates_in_publist(List *publist, Datum *datums); static List *merge_publications(List *oldpublist, List *newpublist, bool addpub, const char *subname); static void ReportSlotConnectionError(List *rstates, Oid subid, char *slotname, char *err); @@ -162,6 +167,8 @@ parse_subscription_options(ParseState *pstate, List *stmt_options, opts->runasowner = false; if (IsSet(supported_opts, SUBOPT_FAILOVER)) opts->failover = false; + if (IsSet(supported_opts, SUBOPT_RETAIN_DEAD_TUPLES)) + opts->retaindeadtuples = false; if (IsSet(supported_opts, SUBOPT_ORIGIN)) opts->origin = pstrdup(LOGICALREP_ORIGIN_ANY); @@ -210,7 +217,7 @@ parse_subscription_options(ParseState *pstate, List *stmt_options, if (strcmp(opts->slot_name, "none") == 0) opts->slot_name = NULL; else - ReplicationSlotValidateName(opts->slot_name, ERROR); + ReplicationSlotValidateName(opts->slot_name, false, ERROR); } else if (IsSet(supported_opts, SUBOPT_COPY_DATA) && strcmp(defel->defname, "copy_data") == 0) @@ -307,6 +314,15 @@ parse_subscription_options(ParseState *pstate, List *stmt_options, opts->specified_opts |= SUBOPT_FAILOVER; opts->failover = defGetBoolean(defel); } + else if (IsSet(supported_opts, SUBOPT_RETAIN_DEAD_TUPLES) && + strcmp(defel->defname, "retain_dead_tuples") == 0) + { + if (IsSet(opts->specified_opts, SUBOPT_RETAIN_DEAD_TUPLES)) + errorConflictingDefElem(defel, pstate); + + opts->specified_opts |= SUBOPT_RETAIN_DEAD_TUPLES; + opts->retaindeadtuples = defGetBoolean(defel); + } else if (IsSet(supported_opts, SUBOPT_ORIGIN) && strcmp(defel->defname, "origin") == 0) { @@ -563,7 +579,8 @@ CreateSubscription(ParseState *pstate, CreateSubscriptionStmt *stmt, SUBOPT_SYNCHRONOUS_COMMIT | SUBOPT_BINARY | SUBOPT_STREAMING | SUBOPT_TWOPHASE_COMMIT | SUBOPT_DISABLE_ON_ERR | SUBOPT_PASSWORD_REQUIRED | - SUBOPT_RUN_AS_OWNER | SUBOPT_FAILOVER | SUBOPT_ORIGIN); + SUBOPT_RUN_AS_OWNER | SUBOPT_FAILOVER | + SUBOPT_RETAIN_DEAD_TUPLES | SUBOPT_ORIGIN); parse_subscription_options(pstate, stmt->options, supported_opts, &opts); /* @@ -630,6 +647,10 @@ CreateSubscription(ParseState *pstate, CreateSubscriptionStmt *stmt, stmt->subname))); } + /* Ensure that we can enable retain_dead_tuples */ + if (opts.retaindeadtuples) + CheckSubDeadTupleRetention(true, !opts.enabled, WARNING); + if (!IsSet(opts.specified_opts, SUBOPT_SLOT_NAME) && opts.slot_name == NULL) opts.slot_name = stmt->subname; @@ -670,6 +691,8 @@ CreateSubscription(ParseState *pstate, CreateSubscriptionStmt *stmt, values[Anum_pg_subscription_subpasswordrequired - 1] = BoolGetDatum(opts.passwordrequired); values[Anum_pg_subscription_subrunasowner - 1] = BoolGetDatum(opts.runasowner); values[Anum_pg_subscription_subfailover - 1] = BoolGetDatum(opts.failover); + values[Anum_pg_subscription_subretaindeadtuples - 1] = + BoolGetDatum(opts.retaindeadtuples); values[Anum_pg_subscription_subconninfo - 1] = CStringGetTextDatum(conninfo); if (opts.slot_name) @@ -722,7 +745,11 @@ CreateSubscription(ParseState *pstate, CreateSubscriptionStmt *stmt, { check_publications(wrconn, publications); check_publications_origin(wrconn, publications, opts.copy_data, - opts.origin, NULL, 0, stmt->subname); + opts.retaindeadtuples, opts.origin, + NULL, 0, stmt->subname); + + if (opts.retaindeadtuples) + check_pub_dead_tuple_retention(wrconn); /* * Set sync state based on if we were asked to do data copy or @@ -881,8 +908,8 @@ AlterSubscription_refresh(Subscription *sub, bool copy_data, sizeof(Oid), oid_cmp); check_publications_origin(wrconn, sub->publications, copy_data, - sub->origin, subrel_local_oids, - subrel_count, sub->name); + sub->retaindeadtuples, sub->origin, + subrel_local_oids, subrel_count, sub->name); /* * Rels that we want to remove from subscription and drop any slots @@ -1040,18 +1067,22 @@ AlterSubscription_refresh(Subscription *sub, bool copy_data, } /* - * Common checks for altering failover and two_phase options. + * Common checks for altering failover, two_phase, and retain_dead_tuples + * options. */ static void CheckAlterSubOption(Subscription *sub, const char *option, bool slot_needs_update, bool isTopLevel) { + Assert(strcmp(option, "failover") == 0 || + strcmp(option, "two_phase") == 0 || + strcmp(option, "retain_dead_tuples") == 0); + /* - * The checks in this function are required only for failover and - * two_phase options. + * Altering the retain_dead_tuples option does not update the slot on the + * publisher. */ - Assert(strcmp(option, "failover") == 0 || - strcmp(option, "two_phase") == 0); + Assert(!slot_needs_update || strcmp(option, "retain_dead_tuples") != 0); /* * Do not allow changing the option if the subscription is enabled. This @@ -1063,6 +1094,39 @@ CheckAlterSubOption(Subscription *sub, const char *option, * the publisher by the existing walsender, so we could have allowed that * even when the subscription is enabled. But we kept this restriction for * the sake of consistency and simplicity. + * + * Additionally, do not allow changing the retain_dead_tuples option when + * the subscription is enabled to prevent race conditions arising from the + * new option value being acknowledged asynchronously by the launcher and + * apply workers. + * + * Without the restriction, a race condition may arise when a user + * disables and immediately re-enables the retain_dead_tuples option. In + * this case, the launcher might drop the slot upon noticing the disabled + * action, while the apply worker may keep maintaining + * oldest_nonremovable_xid without noticing the option change. During this + * period, a transaction ID wraparound could falsely make this ID appear + * as if it originates from the future w.r.t the transaction ID stored in + * the slot maintained by launcher. + * + * Similarly, if the user enables retain_dead_tuples concurrently with the + * launcher starting the worker, the apply worker may start calculating + * oldest_nonremovable_xid before the launcher notices the enable action. + * Consequently, the launcher may update slot.xmin to a newer value than + * that maintained by the worker. In subsequent cycles, upon integrating + * the worker's oldest_nonremovable_xid, the launcher might detect a + * retreat in the calculated xmin, necessitating additional handling. + * + * XXX To address the above race conditions, we can define + * oldest_nonremovable_xid as FullTransactionID and adds the check to + * disallow retreating the conflict slot's xmin. For now, we kept the + * implementation simple by disallowing change to the retain_dead_tuples, + * but in the future we can change this after some more analysis. + * + * Note that we could restrict only the enabling of retain_dead_tuples to + * avoid the race conditions described above, but we maintain the + * restriction for both enable and disable operations for the sake of + * consistency. */ if (sub->enabled) ereport(ERROR, @@ -1110,6 +1174,9 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, bool update_tuple = false; bool update_failover = false; bool update_two_phase = false; + bool check_pub_rdt = false; + bool retain_dead_tuples; + char *origin; Subscription *sub; Form_pg_subscription form; bits32 supported_opts; @@ -1137,6 +1204,9 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, sub = GetSubscription(subid, false); + retain_dead_tuples = sub->retaindeadtuples; + origin = sub->origin; + /* * Don't allow non-superuser modification of a subscription with * password_required=false. @@ -1165,7 +1235,7 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, SUBOPT_DISABLE_ON_ERR | SUBOPT_PASSWORD_REQUIRED | SUBOPT_RUN_AS_OWNER | SUBOPT_FAILOVER | - SUBOPT_ORIGIN); + SUBOPT_RETAIN_DEAD_TUPLES | SUBOPT_ORIGIN); parse_subscription_options(pstate, stmt->options, supported_opts, &opts); @@ -1267,7 +1337,7 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, IsSet(opts.specified_opts, SUBOPT_SLOT_NAME)) ereport(ERROR, (errcode(ERRCODE_SYNTAX_ERROR), - errmsg("slot_name and two_phase cannot be altered at the same time"))); + errmsg("\"slot_name\" and \"two_phase\" cannot be altered at the same time"))); /* * Note that workers may still survive even if the @@ -1283,7 +1353,7 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, if (logicalrep_workers_find(subid, true, true)) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("cannot alter two_phase when logical replication worker is still running"), + errmsg("cannot alter \"two_phase\" when logical replication worker is still running"), errhint("Try again after some time."))); /* @@ -1297,7 +1367,7 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, LookupGXactBySubid(subid)) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("cannot disable two_phase when prepared transactions are present"), + errmsg("cannot disable \"two_phase\" when prepared transactions exist"), errhint("Resolve these transactions and try again."))); /* Change system catalog accordingly */ @@ -1325,11 +1395,62 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, replaces[Anum_pg_subscription_subfailover - 1] = true; } + if (IsSet(opts.specified_opts, SUBOPT_RETAIN_DEAD_TUPLES)) + { + values[Anum_pg_subscription_subretaindeadtuples - 1] = + BoolGetDatum(opts.retaindeadtuples); + replaces[Anum_pg_subscription_subretaindeadtuples - 1] = true; + + CheckAlterSubOption(sub, "retain_dead_tuples", false, isTopLevel); + + /* + * Workers may continue running even after the + * subscription has been disabled. + * + * To prevent race conditions (as described in + * CheckAlterSubOption()), ensure that all worker + * processes have already exited before proceeding. + */ + if (logicalrep_workers_find(subid, true, true)) + ereport(ERROR, + (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), + errmsg("cannot alter retain_dead_tuples when logical replication worker is still running"), + errhint("Try again after some time."))); + + /* + * Remind the user that enabling subscription will prevent + * the accumulation of dead tuples. + */ + if (opts.retaindeadtuples) + CheckSubDeadTupleRetention(true, !sub->enabled, NOTICE); + + /* + * Notify the launcher to manage the replication slot for + * conflict detection. This ensures that replication slot + * is efficiently handled (created, updated, or dropped) + * in response to any configuration changes. + */ + ApplyLauncherWakeupAtCommit(); + + check_pub_rdt = opts.retaindeadtuples; + retain_dead_tuples = opts.retaindeadtuples; + } + if (IsSet(opts.specified_opts, SUBOPT_ORIGIN)) { values[Anum_pg_subscription_suborigin - 1] = CStringGetTextDatum(opts.origin); replaces[Anum_pg_subscription_suborigin - 1] = true; + + /* + * Check if changes from different origins may be received + * from the publisher when the origin is changed to ANY + * and retain_dead_tuples is enabled. + */ + check_pub_rdt = retain_dead_tuples && + pg_strcasecmp(opts.origin, LOGICALREP_ORIGIN_ANY) == 0; + + origin = opts.origin; } update_tuple = true; @@ -1347,6 +1468,15 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), errmsg("cannot enable subscription that does not have a slot name"))); + /* + * Check track_commit_timestamp only when enabling the + * subscription in case it was disabled after creation. See + * comments atop CheckSubDeadTupleRetention() for details. + */ + if (sub->retaindeadtuples) + CheckSubDeadTupleRetention(opts.enabled, !opts.enabled, + WARNING); + values[Anum_pg_subscription_subenabled - 1] = BoolGetDatum(opts.enabled); replaces[Anum_pg_subscription_subenabled - 1] = true; @@ -1355,6 +1485,14 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, ApplyLauncherWakeupAtCommit(); update_tuple = true; + + /* + * The subscription might be initially created with + * connect=false and retain_dead_tuples=true, meaning the + * remote server's status may not be checked. Ensure this + * check is conducted now. + */ + check_pub_rdt = sub->retaindeadtuples && opts.enabled; break; } @@ -1369,6 +1507,13 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, CStringGetTextDatum(stmt->conninfo); replaces[Anum_pg_subscription_subconninfo - 1] = true; update_tuple = true; + + /* + * Since the remote server configuration might have changed, + * perform a check to ensure it permits enabling + * retain_dead_tuples. + */ + check_pub_rdt = sub->retaindeadtuples; break; case ALTER_SUBSCRIPTION_SET_PUBLICATION: @@ -1539,7 +1684,7 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, if (!XLogRecPtrIsInvalid(remote_lsn) && opts.lsn < remote_lsn) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), - errmsg("skip WAL location (LSN %X/%X) must be greater than origin LSN %X/%X", + errmsg("skip WAL location (LSN %X/%08X) must be greater than origin LSN %X/%08X", LSN_FORMAT_ARGS(opts.lsn), LSN_FORMAT_ARGS(remote_lsn)))); } @@ -1568,14 +1713,15 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, } /* - * Try to acquire the connection necessary for altering the slot, if - * needed. + * Try to acquire the connection necessary either for modifying the slot + * or for checking if the remote server permits enabling + * retain_dead_tuples. * * This has to be at the end because otherwise if there is an error while * doing the database operations we won't be able to rollback altered * slot. */ - if (update_failover || update_two_phase) + if (update_failover || update_two_phase || check_pub_rdt) { bool must_use_password; char *err; @@ -1584,10 +1730,14 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, /* Load the library providing us libpq calls. */ load_file("libpqwalreceiver", false); - /* Try to connect to the publisher. */ + /* + * Try to connect to the publisher, using the new connection string if + * available. + */ must_use_password = sub->passwordrequired && !sub->ownersuperuser; - wrconn = walrcv_connect(sub->conninfo, true, true, must_use_password, - sub->name, &err); + wrconn = walrcv_connect(stmt->conninfo ? stmt->conninfo : sub->conninfo, + true, true, must_use_password, sub->name, + &err); if (!wrconn) ereport(ERROR, (errcode(ERRCODE_CONNECTION_FAILURE), @@ -1596,9 +1746,17 @@ AlterSubscription(ParseState *pstate, AlterSubscriptionStmt *stmt, PG_TRY(); { - walrcv_alter_slot(wrconn, sub->slotname, - update_failover ? &opts.failover : NULL, - update_two_phase ? &opts.twophase : NULL); + if (retain_dead_tuples) + check_pub_dead_tuple_retention(wrconn); + + check_publications_origin(wrconn, sub->publications, false, + retain_dead_tuples, origin, NULL, 0, + sub->name); + + if (update_failover || update_two_phase) + walrcv_alter_slot(wrconn, sub->slotname, + update_failover ? &opts.failover : NULL, + update_two_phase ? &opts.twophase : NULL); } PG_FINALLY(); { @@ -2086,20 +2244,29 @@ AlterSubscriptionOwner_oid(Oid subid, Oid newOwnerId) * Check and log a warning if the publisher has subscribed to the same table, * its partition ancestors (if it's a partition), or its partition children (if * it's a partitioned table), from some other publishers. This check is - * required only if "copy_data = true" and "origin = none" for CREATE - * SUBSCRIPTION and ALTER SUBSCRIPTION ... REFRESH statements to notify the - * user that data having origin might have been copied. + * required in the following scenarios: * - * This check need not be performed on the tables that are already added - * because incremental sync for those tables will happen through WAL and the - * origin of the data can be identified from the WAL records. + * 1) For CREATE SUBSCRIPTION and ALTER SUBSCRIPTION ... REFRESH statements + * with "copy_data = true" and "origin = none": + * - Warn the user that data with an origin might have been copied. + * - This check is skipped for tables already added, as incremental sync via + * WAL allows origin tracking. The list of such tables is in + * subrel_local_oids. * - * subrel_local_oids contains the list of relation oids that are already - * present on the subscriber. + * 2) For CREATE SUBSCRIPTION and ALTER SUBSCRIPTION ... REFRESH statements + * with "retain_dead_tuples = true" and "origin = any", and for ALTER + * SUBSCRIPTION statements that modify retain_dead_tuples or origin, or + * when the publisher's status changes (e.g., due to a connection string + * update): + * - Warn the user that only conflict detection info for local changes on + * the publisher is retained. Data from other origins may lack sufficient + * details for reliable conflict detection. + * - See comments atop worker.c for more details. */ static void check_publications_origin(WalReceiverConn *wrconn, List *publications, - bool copydata, char *origin, Oid *subrel_local_oids, + bool copydata, bool retain_dead_tuples, + char *origin, Oid *subrel_local_oids, int subrel_count, char *subname) { WalRcvExecResult *res; @@ -2108,9 +2275,29 @@ check_publications_origin(WalReceiverConn *wrconn, List *publications, Oid tableRow[1] = {TEXTOID}; List *publist = NIL; int i; + bool check_rdt; + bool check_table_sync; + bool origin_none = origin && + pg_strcasecmp(origin, LOGICALREP_ORIGIN_NONE) == 0; + + /* + * Enable retain_dead_tuples checks only when origin is set to 'any', + * since with origin='none' only local changes are replicated to the + * subscriber. + */ + check_rdt = retain_dead_tuples && !origin_none; + + /* + * Enable table synchronization checks only when origin is 'none', to + * ensure that data from other origins is not inadvertently copied. + */ + check_table_sync = copydata && origin_none; - if (!copydata || !origin || - (pg_strcasecmp(origin, LOGICALREP_ORIGIN_NONE) != 0)) + /* retain_dead_tuples and table sync checks occur separately */ + Assert(!(check_rdt && check_table_sync)); + + /* Return if no checks are required */ + if (!check_rdt && !check_table_sync) return; initStringInfo(&cmd); @@ -2129,16 +2316,23 @@ check_publications_origin(WalReceiverConn *wrconn, List *publications, /* * In case of ALTER SUBSCRIPTION ... REFRESH, subrel_local_oids contains * the list of relation oids that are already present on the subscriber. - * This check should be skipped for these tables. + * This check should be skipped for these tables if checking for table + * sync scenario. However, when handling the retain_dead_tuples scenario, + * ensure all tables are checked, as some existing tables may now include + * changes from other origins due to newly created subscriptions on the + * publisher. */ - for (i = 0; i < subrel_count; i++) + if (check_table_sync) { - Oid relid = subrel_local_oids[i]; - char *schemaname = get_namespace_name(get_rel_namespace(relid)); - char *tablename = get_rel_name(relid); + for (i = 0; i < subrel_count; i++) + { + Oid relid = subrel_local_oids[i]; + char *schemaname = get_namespace_name(get_rel_namespace(relid)); + char *tablename = get_rel_name(relid); - appendStringInfo(&cmd, "AND NOT (N.nspname = '%s' AND C.relname = '%s')\n", - schemaname, tablename); + appendStringInfo(&cmd, "AND NOT (N.nspname = '%s' AND C.relname = '%s')\n", + schemaname, tablename); + } } res = walrcv_exec(wrconn, cmd.data, 1, tableRow); @@ -2173,22 +2367,37 @@ check_publications_origin(WalReceiverConn *wrconn, List *publications, * XXX: For simplicity, we don't check whether the table has any data or * not. If the table doesn't have any data then we don't need to * distinguish between data having origin and data not having origin so we - * can avoid logging a warning in that case. + * can avoid logging a warning for table sync scenario. */ if (publist) { StringInfo pubnames = makeStringInfo(); + StringInfo err_msg = makeStringInfo(); + StringInfo err_hint = makeStringInfo(); /* Prepare the list of publication(s) for warning message. */ GetPublicationsStr(publist, pubnames, false); + + if (check_table_sync) + { + appendStringInfo(err_msg, _("subscription \"%s\" requested copy_data with origin = NONE but might copy data that had a different origin"), + subname); + appendStringInfoString(err_hint, _("Verify that initial data copied from the publisher tables did not come from other origins.")); + } + else + { + appendStringInfo(err_msg, _("subscription \"%s\" enabled retain_dead_tuples but might not reliably detect conflicts for changes from different origins"), + subname); + appendStringInfoString(err_hint, _("Consider using origin = NONE or disabling retain_dead_tuples.")); + } + ereport(WARNING, errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("subscription \"%s\" requested copy_data with origin = NONE but might copy data that had a different origin", - subname), - errdetail_plural("The subscription being created subscribes to a publication (%s) that contains tables that are written to by other subscriptions.", - "The subscription being created subscribes to publications (%s) that contain tables that are written to by other subscriptions.", + errmsg_internal("%s", err_msg->data), + errdetail_plural("The subscription subscribes to a publication (%s) that contains tables that are written to by other subscriptions.", + "The subscription subscribes to publications (%s) that contain tables that are written to by other subscriptions.", list_length(publist), pubnames->data), - errhint("Verify that initial data copied from the publisher tables did not come from other origins.")); + errhint_internal("%s", err_hint->data)); } ExecDropSingleTupleTableSlot(slot); @@ -2197,6 +2406,101 @@ check_publications_origin(WalReceiverConn *wrconn, List *publications, } /* + * Determine whether the retain_dead_tuples can be enabled based on the + * publisher's status. + * + * This option is disallowed if the publisher is running a version earlier + * than the PG19, or if the publisher is in recovery (i.e., it is a standby + * server). + * + * See comments atop worker.c for a detailed explanation. + */ +static void +check_pub_dead_tuple_retention(WalReceiverConn *wrconn) +{ + WalRcvExecResult *res; + Oid RecoveryRow[1] = {BOOLOID}; + TupleTableSlot *slot; + bool isnull; + bool remote_in_recovery; + + if (walrcv_server_version(wrconn) < 19000) + ereport(ERROR, + errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), + errmsg("cannot enable retain_dead_tuples if the publisher is running a version earlier than PostgreSQL 19")); + + res = walrcv_exec(wrconn, "SELECT pg_is_in_recovery()", 1, RecoveryRow); + + if (res->status != WALRCV_OK_TUPLES) + ereport(ERROR, + (errcode(ERRCODE_CONNECTION_FAILURE), + errmsg("could not obtain recovery progress from the publisher: %s", + res->err))); + + slot = MakeSingleTupleTableSlot(res->tupledesc, &TTSOpsMinimalTuple); + if (!tuplestore_gettupleslot(res->tuplestore, true, false, slot)) + elog(ERROR, "failed to fetch tuple for the recovery progress"); + + remote_in_recovery = DatumGetBool(slot_getattr(slot, 1, &isnull)); + + if (remote_in_recovery) + ereport(ERROR, + errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("cannot enable retain_dead_tuples if the publisher is in recovery.")); + + ExecDropSingleTupleTableSlot(slot); + + walrcv_clear_result(res); +} + +/* + * Check if the subscriber's configuration is adequate to enable the + * retain_dead_tuples option. + * + * Issue an ERROR if the wal_level does not support the use of replication + * slots when check_guc is set to true. + * + * Issue a WARNING if track_commit_timestamp is not enabled when check_guc is + * set to true. This is only to highlight the importance of enabling + * track_commit_timestamp instead of catching all the misconfigurations, as + * this setting can be adjusted after subscription creation. Without it, the + * apply worker will simply skip conflict detection. + * + * Issue a WARNING or NOTICE if the subscription is disabled. Do not raise an + * ERROR since users can only modify retain_dead_tuples for disabled + * subscriptions. And as long as the subscription is enabled promptly, it will + * not pose issues. + */ +void +CheckSubDeadTupleRetention(bool check_guc, bool sub_disabled, + int elevel_for_sub_disabled) +{ + Assert(elevel_for_sub_disabled == NOTICE || + elevel_for_sub_disabled == WARNING); + + if (check_guc && wal_level < WAL_LEVEL_REPLICA) + ereport(ERROR, + errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), + errmsg("\"wal_level\" is insufficient to create the replication slot required by retain_dead_tuples"), + errhint("\"wal_level\" must be set to \"replica\" or \"logical\" at server start.")); + + if (check_guc && !track_commit_timestamp) + ereport(WARNING, + errcode(ERRCODE_INVALID_PARAMETER_VALUE), + errmsg("commit timestamp and origin data required for detecting conflicts won't be retained"), + errhint("Consider setting \"%s\" to true.", + "track_commit_timestamp")); + + if (sub_disabled) + ereport(elevel_for_sub_disabled, + errcode(ERRCODE_INVALID_PARAMETER_VALUE), + errmsg("deleted rows to detect conflicts would not be removed until the subscription is enabled"), + (elevel_for_sub_disabled > NOTICE) + ? errhint("Consider setting %s to false.", + "retain_dead_tuples") : 0); +} + +/* * Get the list of tables which belong to specified publications on the * publisher connection. * diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c index acf11e83c04..cb811520c29 100644 --- a/src/backend/commands/tablecmds.c +++ b/src/backend/commands/tablecmds.c @@ -430,8 +430,8 @@ static void AlterConstrUpdateConstraintEntry(ATAlterConstraint *cmdcon, Relation static ObjectAddress ATExecValidateConstraint(List **wqueue, Relation rel, char *constrName, bool recurse, bool recursing, LOCKMODE lockmode); -static void QueueFKConstraintValidation(List **wqueue, Relation conrel, Relation rel, - HeapTuple contuple, LOCKMODE lockmode); +static void QueueFKConstraintValidation(List **wqueue, Relation conrel, Relation fkrel, + Oid pkrelid, HeapTuple contuple, LOCKMODE lockmode); static void QueueCheckConstraintValidation(List **wqueue, Relation conrel, Relation rel, char *constrName, HeapTuple contuple, bool recurse, bool recursing, LOCKMODE lockmode); @@ -2711,8 +2711,7 @@ MergeAttributes(List *columns, const List *supers, char relpersistence, RelationGetRelationName(relation)))); /* If existing rel is temp, it must belong to this session */ - if (relation->rd_rel->relpersistence == RELPERSISTENCE_TEMP && - !relation->rd_islocaltemp) + if (RELATION_IS_OTHER_TEMP(relation)) ereport(ERROR, (errcode(ERRCODE_WRONG_OBJECT_TYPE), errmsg(!is_partition @@ -7374,7 +7373,7 @@ ATExecAddColumn(List **wqueue, AlteredTableInfo *tab, Relation rel, /* make sure datatype is legal for a column */ CheckAttributeType(NameStr(attribute->attname), attribute->atttypid, attribute->attcollation, list_make1_oid(rel->rd_rel->reltype), - 0); + (attribute->attgenerated == ATTRIBUTE_GENERATED_VIRTUAL ? CHKATYPE_IS_VIRTUAL : 0)); InsertPgAttributeTuples(attrdesc, tupdesc, myrelid, NULL, NULL); @@ -8609,7 +8608,7 @@ ATExecSetExpression(AlteredTableInfo *tab, Relation rel, const char *colName, rel->rd_att->constr && rel->rd_att->constr->num_check > 0) ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), - errmsg("ALTER TABLE / SET EXPRESSION is not supported for virtual generated columns on tables with check constraints"), + errmsg("ALTER TABLE / SET EXPRESSION is not supported for virtual generated columns in tables with check constraints"), errdetail("Column \"%s\" of relation \"%s\" is a virtual generated column.", colName, RelationGetRelationName(rel)))); @@ -8627,7 +8626,7 @@ ATExecSetExpression(AlteredTableInfo *tab, Relation rel, const char *colName, GetRelationPublications(RelationGetRelid(rel)) != NIL) ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), - errmsg("ALTER TABLE / SET EXPRESSION is not supported for virtual generated columns on tables that are part of a publication"), + errmsg("ALTER TABLE / SET EXPRESSION is not supported for virtual generated columns in tables that are part of a publication"), errdetail("Column \"%s\" of relation \"%s\" is a virtual generated column.", colName, RelationGetRelationName(rel)))); @@ -10189,7 +10188,7 @@ ATAddForeignKeyConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel, if (pk_has_without_overlaps && !with_period) ereport(ERROR, errcode(ERRCODE_INVALID_FOREIGN_KEY), - errmsg("foreign key must use PERIOD when referencing a primary using WITHOUT OVERLAPS")); + errmsg("foreign key must use PERIOD when referencing a primary key using WITHOUT OVERLAPS")); /* * Now we can check permissions. @@ -10330,8 +10329,8 @@ ATAddForeignKeyConstraint(List **wqueue, AlteredTableInfo *tab, Relation rel, for_overlaps ? errmsg("could not identify an overlaps operator for foreign key") : errmsg("could not identify an equality operator for foreign key"), - errdetail("Could not translate compare type %d for operator family \"%s\", input type %s, access method \"%s\".", - cmptype, get_opfamily_name(opfamily, false), format_type_be(opcintype), get_am_name(amid))); + errdetail("Could not translate compare type %d for operator family \"%s\" of access method \"%s\".", + cmptype, get_opfamily_name(opfamily, false), get_am_name(amid))); /* * There had better be a primary equality operator for the index. @@ -11858,6 +11857,7 @@ AttachPartitionForeignKey(List **wqueue, if (queueValidation) { Relation conrel; + Oid confrelid; conrel = table_open(ConstraintRelationId, RowExclusiveLock); @@ -11865,9 +11865,11 @@ AttachPartitionForeignKey(List **wqueue, if (!HeapTupleIsValid(partcontup)) elog(ERROR, "cache lookup failed for constraint %u", partConstrOid); + confrelid = ((Form_pg_constraint) GETSTRUCT(partcontup))->confrelid; + /* Use the same lock as for AT_ValidateConstraint */ - QueueFKConstraintValidation(wqueue, conrel, partition, partcontup, - ShareUpdateExclusiveLock); + QueueFKConstraintValidation(wqueue, conrel, partition, confrelid, + partcontup, ShareUpdateExclusiveLock); ReleaseSysCache(partcontup); table_close(conrel, RowExclusiveLock); } @@ -12463,9 +12465,12 @@ ATExecAlterConstrEnforceability(List **wqueue, ATAlterConstraint *cmdcon, /* * Tell Phase 3 to check that the constraint is satisfied by existing - * rows. + * rows. Only applies to leaf partitions, and (for constraints that + * reference a partitioned table) only if this is not one of the + * pg_constraint rows that exist solely to support action triggers. */ - if (rel->rd_rel->relkind == RELKIND_RELATION) + if (rel->rd_rel->relkind == RELKIND_RELATION && + currcon->confrelid == pkrelid) { AlteredTableInfo *tab; NewConstraint *newcon; @@ -12907,8 +12912,9 @@ ATExecValidateConstraint(List **wqueue, Relation rel, char *constrName, con->contype != CONSTRAINT_NOTNULL) ereport(ERROR, errcode(ERRCODE_WRONG_OBJECT_TYPE), - errmsg("constraint \"%s\" of relation \"%s\" is not a foreign key, check, or not-null constraint", - constrName, RelationGetRelationName(rel))); + errmsg("cannot validate constraint \"%s\" of relation \"%s\"", + constrName, RelationGetRelationName(rel)), + errdetail("This operation is not supported for this type of constraint.")); if (!con->conenforced) ereport(ERROR, @@ -12919,7 +12925,8 @@ ATExecValidateConstraint(List **wqueue, Relation rel, char *constrName, { if (con->contype == CONSTRAINT_FOREIGN) { - QueueFKConstraintValidation(wqueue, conrel, rel, tuple, lockmode); + QueueFKConstraintValidation(wqueue, conrel, rel, con->confrelid, + tuple, lockmode); } else if (con->contype == CONSTRAINT_CHECK) { @@ -12952,8 +12959,8 @@ ATExecValidateConstraint(List **wqueue, Relation rel, char *constrName, * for the specified relation and all its children. */ static void -QueueFKConstraintValidation(List **wqueue, Relation conrel, Relation rel, - HeapTuple contuple, LOCKMODE lockmode) +QueueFKConstraintValidation(List **wqueue, Relation conrel, Relation fkrel, + Oid pkrelid, HeapTuple contuple, LOCKMODE lockmode) { Form_pg_constraint con; AlteredTableInfo *tab; @@ -12964,7 +12971,17 @@ QueueFKConstraintValidation(List **wqueue, Relation conrel, Relation rel, Assert(con->contype == CONSTRAINT_FOREIGN); Assert(!con->convalidated); - if (rel->rd_rel->relkind == RELKIND_RELATION) + /* + * Add the validation to phase 3's queue; not needed for partitioned + * tables themselves, only for their partitions. + * + * When the referenced table (pkrelid) is partitioned, the referencing + * table (fkrel) has one pg_constraint row pointing to each partition + * thereof. These rows are there only to support action triggers and no + * table scan is needed, therefore skip this for them as well. + */ + if (fkrel->rd_rel->relkind == RELKIND_RELATION && + con->confrelid == pkrelid) { NewConstraint *newcon; Constraint *fkconstraint; @@ -12983,15 +13000,16 @@ QueueFKConstraintValidation(List **wqueue, Relation conrel, Relation rel, newcon->qual = (Node *) fkconstraint; /* Find or create work queue entry for this table */ - tab = ATGetQueueEntry(wqueue, rel); + tab = ATGetQueueEntry(wqueue, fkrel); tab->constraints = lappend(tab->constraints, newcon); } /* * If the table at either end of the constraint is partitioned, we need to - * recurse and handle every constraint that is a child of this constraint. + * recurse and handle every unvalidate constraint that is a child of this + * constraint. */ - if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE || + if (fkrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE || get_rel_relkind(con->confrelid) == RELKIND_PARTITIONED_TABLE) { ScanKeyData pkey; @@ -13023,8 +13041,12 @@ QueueFKConstraintValidation(List **wqueue, Relation conrel, Relation rel, childrel = table_open(childcon->conrelid, lockmode); - QueueFKConstraintValidation(wqueue, conrel, childrel, childtup, - lockmode); + /* + * NB: Note that pkrelid should be passed as-is during recursion, + * as it is required to identify the root referenced table. + */ + QueueFKConstraintValidation(wqueue, conrel, childrel, pkrelid, + childtup, lockmode); table_close(childrel, NoLock); } @@ -13032,7 +13054,11 @@ QueueFKConstraintValidation(List **wqueue, Relation conrel, Relation rel, } /* - * Now update the catalog, while we have the door open. + * Now mark the pg_constraint row as validated (even if we didn't check, + * notably the ones for partitions on the referenced side). + * + * We rely on transaction abort to roll back this change if phase 3 + * ultimately finds violating rows. This is a bit ugly. */ copyTuple = heap_copytuple(contuple); copy_con = (Form_pg_constraint) GETSTRUCT(copyTuple); @@ -14400,7 +14426,7 @@ ATPrepAlterColumnType(List **wqueue, /* make sure datatype is legal for a column */ CheckAttributeType(colName, targettype, targetcollid, list_make1_oid(rel->rd_rel->reltype), - 0); + (attTup->attgenerated == ATTRIBUTE_GENERATED_VIRTUAL ? CHKATYPE_IS_VIRTUAL : 0)); if (attTup->attgenerated == ATTRIBUTE_GENERATED_VIRTUAL) { @@ -14458,6 +14484,9 @@ ATPrepAlterColumnType(List **wqueue, /* Fix collations after all else */ assign_expr_collations(pstate, transform); + /* Expand virtual generated columns in the expr. */ + transform = expand_generated_columns_in_expr(transform, rel, 1); + /* Plan the expr now so we can accurately assess the need to rewrite. */ transform = (Node *) expression_planner((Expr *) transform); @@ -15385,9 +15414,12 @@ ATPostAlterTypeCleanup(List **wqueue, AlteredTableInfo *tab, LOCKMODE lockmode) /* * Re-parse the index and constraint definitions, and attach them to the * appropriate work queue entries. We do this before dropping because in - * the case of a FOREIGN KEY constraint, we might not yet have exclusive - * lock on the table the constraint is attached to, and we need to get - * that before reparsing/dropping. + * the case of a constraint on another table, we might not yet have + * exclusive lock on the table the constraint is attached to, and we need + * to get that before reparsing/dropping. (That's possible at least for + * FOREIGN KEY, CHECK, and EXCLUSION constraints; in non-FK cases it + * requires a dependency on the target table's composite type in the other + * table's constraint expressions.) * * We can't rely on the output of deparsing to tell us which relation to * operate on, because concurrent activity might have made the name @@ -15403,7 +15435,6 @@ ATPostAlterTypeCleanup(List **wqueue, AlteredTableInfo *tab, LOCKMODE lockmode) Form_pg_constraint con; Oid relid; Oid confrelid; - char contype; bool conislocal; tup = SearchSysCache1(CONSTROID, ObjectIdGetDatum(oldId)); @@ -15420,7 +15451,6 @@ ATPostAlterTypeCleanup(List **wqueue, AlteredTableInfo *tab, LOCKMODE lockmode) elog(ERROR, "could not identify relation associated with constraint %u", oldId); } confrelid = con->confrelid; - contype = con->contype; conislocal = con->conislocal; ReleaseSysCache(tup); @@ -15438,12 +15468,12 @@ ATPostAlterTypeCleanup(List **wqueue, AlteredTableInfo *tab, LOCKMODE lockmode) continue; /* - * When rebuilding an FK constraint that references the table we're - * modifying, we might not yet have any lock on the FK's table, so get - * one now. We'll need AccessExclusiveLock for the DROP CONSTRAINT - * step, so there's no value in asking for anything weaker. + * When rebuilding another table's constraint that references the + * table we're modifying, we might not yet have any lock on the other + * table, so get one now. We'll need AccessExclusiveLock for the DROP + * CONSTRAINT step, so there's no value in asking for anything weaker. */ - if (relid != tab->relid && contype == CONSTRAINT_FOREIGN) + if (relid != tab->relid) LockRelationOid(relid, AccessExclusiveLock); ATPostAlterTypeParse(oldId, relid, confrelid, @@ -15457,6 +15487,14 @@ ATPostAlterTypeCleanup(List **wqueue, AlteredTableInfo *tab, LOCKMODE lockmode) Oid relid; relid = IndexGetRelation(oldId, false); + + /* + * As above, make sure we have lock on the index's table if it's not + * the same table. + */ + if (relid != tab->relid) + LockRelationOid(relid, AccessExclusiveLock); + ATPostAlterTypeParse(oldId, relid, InvalidOid, (char *) lfirst(def_item), wqueue, lockmode, tab->rewrite); @@ -15473,6 +15511,20 @@ ATPostAlterTypeCleanup(List **wqueue, AlteredTableInfo *tab, LOCKMODE lockmode) Oid relid; relid = StatisticsGetRelation(oldId, false); + + /* + * As above, make sure we have lock on the statistics object's table + * if it's not the same table. However, we take + * ShareUpdateExclusiveLock here, aligning with the lock level used in + * CreateStatistics and RemoveStatisticsById. + * + * CAUTION: this should be done after all cases that grab + * AccessExclusiveLock, else we risk causing deadlock due to needing + * to promote our table lock. + */ + if (relid != tab->relid) + LockRelationOid(relid, ShareUpdateExclusiveLock); + ATPostAlterTypeParse(oldId, relid, InvalidOid, (char *) lfirst(def_item), wqueue, lockmode, tab->rewrite); @@ -15696,7 +15748,7 @@ ATPostAlterTypeParse(Oid oldId, Oid oldRelId, Oid refRelId, char *cmd, { AlterDomainStmt *stmt = (AlterDomainStmt *) stm; - if (stmt->subtype == 'C') /* ADD CONSTRAINT */ + if (stmt->subtype == AD_AddConstraint) { Constraint *con = castNode(Constraint, stmt->def); AlterTableCmd *cmd = makeNode(AlterTableCmd); @@ -17199,15 +17251,13 @@ ATExecAddInherit(Relation child_rel, RangeVar *parent, LOCKMODE lockmode) RelationGetRelationName(parent_rel)))); /* If parent rel is temp, it must belong to this session */ - if (parent_rel->rd_rel->relpersistence == RELPERSISTENCE_TEMP && - !parent_rel->rd_islocaltemp) + if (RELATION_IS_OTHER_TEMP(parent_rel)) ereport(ERROR, (errcode(ERRCODE_WRONG_OBJECT_TYPE), errmsg("cannot inherit from temporary relation of another session"))); /* Ditto for the child */ - if (child_rel->rd_rel->relpersistence == RELPERSISTENCE_TEMP && - !child_rel->rd_islocaltemp) + if (RELATION_IS_OTHER_TEMP(child_rel)) ereport(ERROR, (errcode(ERRCODE_WRONG_OBJECT_TYPE), errmsg("cannot inherit to temporary relation of another session"))); @@ -20278,15 +20328,13 @@ ATExecAttachPartition(List **wqueue, Relation rel, PartitionCmd *cmd, RelationGetRelationName(rel)))); /* If the parent is temp, it must belong to this session */ - if (rel->rd_rel->relpersistence == RELPERSISTENCE_TEMP && - !rel->rd_islocaltemp) + if (RELATION_IS_OTHER_TEMP(rel)) ereport(ERROR, (errcode(ERRCODE_WRONG_OBJECT_TYPE), errmsg("cannot attach as partition of temporary relation of another session"))); /* Ditto for the partition */ - if (attachrel->rd_rel->relpersistence == RELPERSISTENCE_TEMP && - !attachrel->rd_islocaltemp) + if (RELATION_IS_OTHER_TEMP(attachrel)) ereport(ERROR, (errcode(ERRCODE_WRONG_OBJECT_TYPE), errmsg("cannot attach temporary relation of another session as partition"))); diff --git a/src/backend/commands/tablespace.c b/src/backend/commands/tablespace.c index a9005cc7212..df31eace47a 100644 --- a/src/backend/commands/tablespace.c +++ b/src/backend/commands/tablespace.c @@ -500,7 +500,7 @@ DropTableSpace(DropTableSpaceStmt *stmt) * mustn't delete. So instead, we force a checkpoint which will clean * out any lingering files, and try again. */ - RequestCheckpoint(CHECKPOINT_IMMEDIATE | CHECKPOINT_FORCE | CHECKPOINT_WAIT); + RequestCheckpoint(CHECKPOINT_FAST | CHECKPOINT_FORCE | CHECKPOINT_WAIT); /* * On Windows, an unlinked file persists in the directory listing diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c index 67f8e70f9c1..7dc121f73f1 100644 --- a/src/backend/commands/trigger.c +++ b/src/backend/commands/trigger.c @@ -80,6 +80,7 @@ static bool GetTupleForTrigger(EState *estate, ItemPointer tid, LockTupleMode lockmode, TupleTableSlot *oldslot, + bool do_epq_recheck, TupleTableSlot **epqslot, TM_Result *tmresultp, TM_FailureData *tmfdp); @@ -2693,7 +2694,8 @@ ExecBRDeleteTriggers(EState *estate, EPQState *epqstate, HeapTuple fdw_trigtuple, TupleTableSlot **epqslot, TM_Result *tmresult, - TM_FailureData *tmfd) + TM_FailureData *tmfd, + bool is_merge_delete) { TupleTableSlot *slot = ExecGetTriggerOldSlot(estate, relinfo); TriggerDesc *trigdesc = relinfo->ri_TrigDesc; @@ -2708,9 +2710,17 @@ ExecBRDeleteTriggers(EState *estate, EPQState *epqstate, { TupleTableSlot *epqslot_candidate = NULL; + /* + * Get a copy of the on-disk tuple we are planning to delete. In + * general, if the tuple has been concurrently updated, we should + * recheck it using EPQ. However, if this is a MERGE DELETE action, + * we skip this EPQ recheck and leave it to the caller (it must do + * additional rechecking, and might end up executing a different + * action entirely). + */ if (!GetTupleForTrigger(estate, epqstate, relinfo, tupleid, - LockTupleExclusive, slot, &epqslot_candidate, - tmresult, tmfd)) + LockTupleExclusive, slot, !is_merge_delete, + &epqslot_candidate, tmresult, tmfd)) return false; /* @@ -2800,6 +2810,7 @@ ExecARDeleteTriggers(EState *estate, tupleid, LockTupleExclusive, slot, + false, NULL, NULL, NULL); @@ -2944,7 +2955,8 @@ ExecBRUpdateTriggers(EState *estate, EPQState *epqstate, HeapTuple fdw_trigtuple, TupleTableSlot *newslot, TM_Result *tmresult, - TM_FailureData *tmfd) + TM_FailureData *tmfd, + bool is_merge_update) { TriggerDesc *trigdesc = relinfo->ri_TrigDesc; TupleTableSlot *oldslot = ExecGetTriggerOldSlot(estate, relinfo); @@ -2965,10 +2977,17 @@ ExecBRUpdateTriggers(EState *estate, EPQState *epqstate, { TupleTableSlot *epqslot_candidate = NULL; - /* get a copy of the on-disk tuple we are planning to update */ + /* + * Get a copy of the on-disk tuple we are planning to update. In + * general, if the tuple has been concurrently updated, we should + * recheck it using EPQ. However, if this is a MERGE UPDATE action, + * we skip this EPQ recheck and leave it to the caller (it must do + * additional rechecking, and might end up executing a different + * action entirely). + */ if (!GetTupleForTrigger(estate, epqstate, relinfo, tupleid, - lockmode, oldslot, &epqslot_candidate, - tmresult, tmfd)) + lockmode, oldslot, !is_merge_update, + &epqslot_candidate, tmresult, tmfd)) return false; /* cancel the update action */ /* @@ -3142,6 +3161,7 @@ ExecARUpdateTriggers(EState *estate, ResultRelInfo *relinfo, tupleid, LockTupleExclusive, oldslot, + false, NULL, NULL, NULL); @@ -3298,6 +3318,7 @@ GetTupleForTrigger(EState *estate, ItemPointer tid, LockTupleMode lockmode, TupleTableSlot *oldslot, + bool do_epq_recheck, TupleTableSlot **epqslot, TM_Result *tmresultp, TM_FailureData *tmfdp) @@ -3357,29 +3378,30 @@ GetTupleForTrigger(EState *estate, if (tmfd.traversed) { /* - * Recheck the tuple using EPQ. For MERGE, we leave this - * to the caller (it must do additional rechecking, and - * might end up executing a different action entirely). + * Recheck the tuple using EPQ, if requested. Otherwise, + * just return that it was concurrently updated. */ - if (estate->es_plannedstmt->commandType == CMD_MERGE) + if (do_epq_recheck) { - if (tmresultp) - *tmresultp = TM_Updated; - return false; + *epqslot = EvalPlanQual(epqstate, + relation, + relinfo->ri_RangeTableIndex, + oldslot); + + /* + * If PlanQual failed for updated tuple - we must not + * process this tuple! + */ + if (TupIsNull(*epqslot)) + { + *epqslot = NULL; + return false; + } } - - *epqslot = EvalPlanQual(epqstate, - relation, - relinfo->ri_RangeTableIndex, - oldslot); - - /* - * If PlanQual failed for updated tuple - we must not - * process this tuple! - */ - if (TupIsNull(*epqslot)) + else { - *epqslot = NULL; + if (tmresultp) + *tmresultp = TM_Updated; return false; } } diff --git a/src/backend/commands/typecmds.c b/src/backend/commands/typecmds.c index 45ae7472ab5..26d985193ae 100644 --- a/src/backend/commands/typecmds.c +++ b/src/backend/commands/typecmds.c @@ -939,11 +939,19 @@ DefineDomain(ParseState *pstate, CreateDomainStmt *stmt) break; case CONSTR_NOTNULL: - if (nullDefined && !typNotNull) + if (nullDefined) + { + if (!typNotNull) + ereport(ERROR, + errcode(ERRCODE_SYNTAX_ERROR), + errmsg("conflicting NULL/NOT NULL constraints"), + parser_errposition(pstate, constr->location)); + ereport(ERROR, - errcode(ERRCODE_SYNTAX_ERROR), - errmsg("conflicting NULL/NOT NULL constraints"), + errcode(ERRCODE_INVALID_OBJECT_DEFINITION), + errmsg("redundant NOT NULL constraint definition"), parser_errposition(pstate, constr->location)); + } if (constr->is_no_inherit) ereport(ERROR, errcode(ERRCODE_INVALID_OBJECT_DEFINITION), diff --git a/src/backend/commands/vacuum.c b/src/backend/commands/vacuum.c index 33a33bf6b1c..733ef40ae7c 100644 --- a/src/backend/commands/vacuum.c +++ b/src/backend/commands/vacuum.c @@ -56,6 +56,7 @@ #include "utils/fmgroids.h" #include "utils/guc.h" #include "utils/guc_hooks.h" +#include "utils/injection_point.h" #include "utils/memutils.h" #include "utils/snapmgr.h" #include "utils/syscache.h" @@ -123,7 +124,7 @@ static void vac_truncate_clog(TransactionId frozenXID, MultiXactId minMulti, TransactionId lastSaneFrozenXid, MultiXactId lastSaneMinMulti); -static bool vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, +static bool vacuum_rel(Oid relid, RangeVar *relation, VacuumParams params, BufferAccessStrategy bstrategy); static double compute_parallel_delay(void); static VacOptValue get_vacoptval_from_boolean(DefElem *def); @@ -464,7 +465,7 @@ ExecVacuum(ParseState *pstate, VacuumStmt *vacstmt, bool isTopLevel) } /* Now go through the common routine */ - vacuum(vacstmt->rels, ¶ms, bstrategy, vac_context, isTopLevel); + vacuum(vacstmt->rels, params, bstrategy, vac_context, isTopLevel); /* Finally, clean up the vacuum memory context */ MemoryContextDelete(vac_context); @@ -493,7 +494,7 @@ ExecVacuum(ParseState *pstate, VacuumStmt *vacstmt, bool isTopLevel) * memory context that will not disappear at transaction commit. */ void -vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, +vacuum(List *relations, const VacuumParams params, BufferAccessStrategy bstrategy, MemoryContext vac_context, bool isTopLevel) { static bool in_vacuum = false; @@ -502,9 +503,7 @@ vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, volatile bool in_outer_xact, use_own_xacts; - Assert(params != NULL); - - stmttype = (params->options & VACOPT_VACUUM) ? "VACUUM" : "ANALYZE"; + stmttype = (params.options & VACOPT_VACUUM) ? "VACUUM" : "ANALYZE"; /* * We cannot run VACUUM inside a user transaction block; if we were inside @@ -514,7 +513,7 @@ vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, * * ANALYZE (without VACUUM) can run either way. */ - if (params->options & VACOPT_VACUUM) + if (params.options & VACOPT_VACUUM) { PreventInTransactionBlock(isTopLevel, stmttype); in_outer_xact = false; @@ -537,7 +536,7 @@ vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, * Build list of relation(s) to process, putting any new data in * vac_context for safekeeping. */ - if (params->options & VACOPT_ONLY_DATABASE_STATS) + if (params.options & VACOPT_ONLY_DATABASE_STATS) { /* We don't process any tables in this case */ Assert(relations == NIL); @@ -553,7 +552,7 @@ vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, List *sublist; MemoryContext old_context; - sublist = expand_vacuum_rel(vrel, vac_context, params->options); + sublist = expand_vacuum_rel(vrel, vac_context, params.options); old_context = MemoryContextSwitchTo(vac_context); newrels = list_concat(newrels, sublist); MemoryContextSwitchTo(old_context); @@ -561,7 +560,7 @@ vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, relations = newrels; } else - relations = get_all_vacuum_rels(vac_context, params->options); + relations = get_all_vacuum_rels(vac_context, params.options); /* * Decide whether we need to start/commit our own transactions. @@ -577,11 +576,11 @@ vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, * transaction block, and also in an autovacuum worker, use own * transactions so we can release locks sooner. */ - if (params->options & VACOPT_VACUUM) + if (params.options & VACOPT_VACUUM) use_own_xacts = true; else { - Assert(params->options & VACOPT_ANALYZE); + Assert(params.options & VACOPT_ANALYZE); if (AmAutoVacuumWorkerProcess()) use_own_xacts = true; else if (in_outer_xact) @@ -632,13 +631,13 @@ vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, { VacuumRelation *vrel = lfirst_node(VacuumRelation, cur); - if (params->options & VACOPT_VACUUM) + if (params.options & VACOPT_VACUUM) { if (!vacuum_rel(vrel->oid, vrel->relation, params, bstrategy)) continue; } - if (params->options & VACOPT_ANALYZE) + if (params.options & VACOPT_ANALYZE) { /* * If using separate xacts, start one for analyze. Otherwise, @@ -702,8 +701,8 @@ vacuum(List *relations, VacuumParams *params, BufferAccessStrategy bstrategy, StartTransactionCommand(); } - if ((params->options & VACOPT_VACUUM) && - !(params->options & VACOPT_SKIP_DATABASE_STATS)) + if ((params.options & VACOPT_VACUUM) && + !(params.options & VACOPT_SKIP_DATABASE_STATS)) { /* * Update pg_database.datfrozenxid, and truncate pg_xact if possible. @@ -1101,7 +1100,7 @@ get_all_vacuum_rels(MemoryContext vac_context, int options) * minimum). */ bool -vacuum_get_cutoffs(Relation rel, const VacuumParams *params, +vacuum_get_cutoffs(Relation rel, const VacuumParams params, struct VacuumCutoffs *cutoffs) { int freeze_min_age, @@ -1117,10 +1116,10 @@ vacuum_get_cutoffs(Relation rel, const VacuumParams *params, aggressiveMXIDCutoff; /* Use mutable copies of freeze age parameters */ - freeze_min_age = params->freeze_min_age; - multixact_freeze_min_age = params->multixact_freeze_min_age; - freeze_table_age = params->freeze_table_age; - multixact_freeze_table_age = params->multixact_freeze_table_age; + freeze_min_age = params.freeze_min_age; + multixact_freeze_min_age = params.multixact_freeze_min_age; + freeze_table_age = params.freeze_table_age; + multixact_freeze_table_age = params.multixact_freeze_table_age; /* Set pg_class fields in cutoffs */ cutoffs->relfrozenxid = rel->rd_rel->relfrozenxid; @@ -1997,7 +1996,7 @@ vac_truncate_clog(TransactionId frozenXID, * At entry and exit, we are not inside a transaction. */ static bool -vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, +vacuum_rel(Oid relid, RangeVar *relation, VacuumParams params, BufferAccessStrategy bstrategy) { LOCKMODE lmode; @@ -2008,13 +2007,18 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, Oid save_userid; int save_sec_context; int save_nestlevel; + VacuumParams toast_vacuum_params; - Assert(params != NULL); + /* + * This function scribbles on the parameters, so make a copy early to + * avoid affecting the TOAST table (if we do end up recursing to it). + */ + memcpy(&toast_vacuum_params, ¶ms, sizeof(VacuumParams)); /* Begin a transaction for vacuuming this relation */ StartTransactionCommand(); - if (!(params->options & VACOPT_FULL)) + if (!(params.options & VACOPT_FULL)) { /* * In lazy vacuum, we can set the PROC_IN_VACUUM flag, which lets @@ -2040,7 +2044,7 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, */ LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); MyProc->statusFlags |= PROC_IN_VACUUM; - if (params->is_wraparound) + if (params.is_wraparound) MyProc->statusFlags |= PROC_VACUUM_FOR_WRAPAROUND; ProcGlobal->statusFlags[MyProc->pgxactoff] = MyProc->statusFlags; LWLockRelease(ProcArrayLock); @@ -2064,12 +2068,12 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, * vacuum, but just ShareUpdateExclusiveLock for concurrent vacuum. Either * way, we can be sure that no other backend is vacuuming the same table. */ - lmode = (params->options & VACOPT_FULL) ? + lmode = (params.options & VACOPT_FULL) ? AccessExclusiveLock : ShareUpdateExclusiveLock; /* open the relation and get the appropriate lock on it */ - rel = vacuum_open_relation(relid, relation, params->options, - params->log_min_duration >= 0, lmode); + rel = vacuum_open_relation(relid, relation, params.options, + params.log_min_duration >= 0, lmode); /* leave if relation could not be opened or locked */ if (!rel) @@ -2084,8 +2088,8 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, * This is only safe to do because we hold a session lock on the main * relation that prevents concurrent deletion. */ - if (OidIsValid(params->toast_parent)) - priv_relid = params->toast_parent; + if (OidIsValid(params.toast_parent)) + priv_relid = params.toast_parent; else priv_relid = RelationGetRelid(rel); @@ -2098,7 +2102,7 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, */ if (!vacuum_is_permitted_for_relation(priv_relid, rel->rd_rel, - params->options & ~VACOPT_ANALYZE)) + params.options & ~VACOPT_ANALYZE)) { relation_close(rel, lmode); PopActiveSnapshot(); @@ -2169,7 +2173,7 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, * Set index_cleanup option based on index_cleanup reloption if it wasn't * specified in VACUUM command, or when running in an autovacuum worker */ - if (params->index_cleanup == VACOPTVALUE_UNSPECIFIED) + if (params.index_cleanup == VACOPTVALUE_UNSPECIFIED) { StdRdOptIndexCleanup vacuum_index_cleanup; @@ -2180,56 +2184,74 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, ((StdRdOptions *) rel->rd_options)->vacuum_index_cleanup; if (vacuum_index_cleanup == STDRD_OPTION_VACUUM_INDEX_CLEANUP_AUTO) - params->index_cleanup = VACOPTVALUE_AUTO; + params.index_cleanup = VACOPTVALUE_AUTO; else if (vacuum_index_cleanup == STDRD_OPTION_VACUUM_INDEX_CLEANUP_ON) - params->index_cleanup = VACOPTVALUE_ENABLED; + params.index_cleanup = VACOPTVALUE_ENABLED; else { Assert(vacuum_index_cleanup == STDRD_OPTION_VACUUM_INDEX_CLEANUP_OFF); - params->index_cleanup = VACOPTVALUE_DISABLED; + params.index_cleanup = VACOPTVALUE_DISABLED; } } +#ifdef USE_INJECTION_POINTS + if (params.index_cleanup == VACOPTVALUE_AUTO) + INJECTION_POINT("vacuum-index-cleanup-auto", NULL); + else if (params.index_cleanup == VACOPTVALUE_DISABLED) + INJECTION_POINT("vacuum-index-cleanup-disabled", NULL); + else if (params.index_cleanup == VACOPTVALUE_ENABLED) + INJECTION_POINT("vacuum-index-cleanup-enabled", NULL); +#endif + /* * Check if the vacuum_max_eager_freeze_failure_rate table storage * parameter was specified. This overrides the GUC value. */ if (rel->rd_options != NULL && ((StdRdOptions *) rel->rd_options)->vacuum_max_eager_freeze_failure_rate >= 0) - params->max_eager_freeze_failure_rate = + params.max_eager_freeze_failure_rate = ((StdRdOptions *) rel->rd_options)->vacuum_max_eager_freeze_failure_rate; /* * Set truncate option based on truncate reloption or GUC if it wasn't * specified in VACUUM command, or when running in an autovacuum worker */ - if (params->truncate == VACOPTVALUE_UNSPECIFIED) + if (params.truncate == VACOPTVALUE_UNSPECIFIED) { StdRdOptions *opts = (StdRdOptions *) rel->rd_options; if (opts && opts->vacuum_truncate_set) { if (opts->vacuum_truncate) - params->truncate = VACOPTVALUE_ENABLED; + params.truncate = VACOPTVALUE_ENABLED; else - params->truncate = VACOPTVALUE_DISABLED; + params.truncate = VACOPTVALUE_DISABLED; } else if (vacuum_truncate) - params->truncate = VACOPTVALUE_ENABLED; + params.truncate = VACOPTVALUE_ENABLED; else - params->truncate = VACOPTVALUE_DISABLED; + params.truncate = VACOPTVALUE_DISABLED; } +#ifdef USE_INJECTION_POINTS + if (params.truncate == VACOPTVALUE_AUTO) + INJECTION_POINT("vacuum-truncate-auto", NULL); + else if (params.truncate == VACOPTVALUE_DISABLED) + INJECTION_POINT("vacuum-truncate-disabled", NULL); + else if (params.truncate == VACOPTVALUE_ENABLED) + INJECTION_POINT("vacuum-truncate-enabled", NULL); +#endif + /* * Remember the relation's TOAST relation for later, if the caller asked * us to process it. In VACUUM FULL, though, the toast table is * automatically rebuilt by cluster_rel so we shouldn't recurse to it, * unless PROCESS_MAIN is disabled. */ - if ((params->options & VACOPT_PROCESS_TOAST) != 0 && - ((params->options & VACOPT_FULL) == 0 || - (params->options & VACOPT_PROCESS_MAIN) == 0)) + if ((params.options & VACOPT_PROCESS_TOAST) != 0 && + ((params.options & VACOPT_FULL) == 0 || + (params.options & VACOPT_PROCESS_MAIN) == 0)) toast_relid = rel->rd_rel->reltoastrelid; else toast_relid = InvalidOid; @@ -2252,16 +2274,16 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, * table is required (e.g., PROCESS_TOAST is set), we force PROCESS_MAIN * to be set when we recurse to the TOAST table. */ - if (params->options & VACOPT_PROCESS_MAIN) + if (params.options & VACOPT_PROCESS_MAIN) { /* * Do the actual work --- either FULL or "lazy" vacuum */ - if (params->options & VACOPT_FULL) + if (params.options & VACOPT_FULL) { ClusterParams cluster_params = {0}; - if ((params->options & VACOPT_VERBOSE) != 0) + if ((params.options & VACOPT_VERBOSE) != 0) cluster_params.options |= CLUOPT_VERBOSE; /* VACUUM FULL is now a variant of CLUSTER; see cluster.c */ @@ -2299,19 +2321,16 @@ vacuum_rel(Oid relid, RangeVar *relation, VacuumParams *params, */ if (toast_relid != InvalidOid) { - VacuumParams toast_vacuum_params; - /* * Force VACOPT_PROCESS_MAIN so vacuum_rel() processes it. Likewise, * set toast_parent so that the privilege checks are done on the main * relation. NB: This is only safe to do because we hold a session * lock on the main relation that prevents concurrent deletion. */ - memcpy(&toast_vacuum_params, params, sizeof(VacuumParams)); toast_vacuum_params.options |= VACOPT_PROCESS_MAIN; toast_vacuum_params.toast_parent = relid; - vacuum_rel(toast_relid, NULL, &toast_vacuum_params, bstrategy); + vacuum_rel(toast_relid, NULL, toast_vacuum_params, bstrategy); } /* diff --git a/src/backend/executor/execExprInterp.c b/src/backend/executor/execExprInterp.c index 8a72b5e70a4..1a37737d4a2 100644 --- a/src/backend/executor/execExprInterp.c +++ b/src/backend/executor/execExprInterp.c @@ -5228,7 +5228,6 @@ ExecEvalJsonCoercionFinish(ExprState *state, ExprEvalStep *op) * JsonBehavior expression. */ jsestate->escontext.error_occurred = false; - jsestate->escontext.error_occurred = false; jsestate->escontext.details_wanted = true; } } diff --git a/src/backend/executor/execGrouping.c b/src/backend/executor/execGrouping.c index 255bd795361..b5400749353 100644 --- a/src/backend/executor/execGrouping.c +++ b/src/backend/executor/execGrouping.c @@ -144,7 +144,7 @@ execTuplesHashPrepare(int numCols, * hashfunctions: FmgrInfos of datatype-specific hashing functions to use * collations: collations to use in comparisons * nbuckets: initial estimate of hashtable size - * additionalsize: size of data stored in ->additional + * additionalsize: size of data that may be stored along with the hash entry * metacxt: memory context for long-lived allocation, but not per-entry data * tablecxt: memory context in which to store table entries * tempcxt: short-lived context for evaluation hash and comparison functions @@ -288,7 +288,7 @@ ResetTupleHashTable(TupleHashTable hashtable) * * If isnew isn't NULL, then a new entry is created if no existing entry * matches. On return, *isnew is true if the entry is newly created, - * false if it existed already. ->additional_data in the new entry has + * false if it existed already. The additional data in the new entry has * been zeroed. */ TupleHashEntry diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c index bdf862b2406..ca33a854278 100644 --- a/src/backend/executor/execIndexing.c +++ b/src/backend/executor/execIndexing.c @@ -279,7 +279,7 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo) * executor is performing an UPDATE that could not use an * optimization like heapam's HOT (in more general terms a * call to table_tuple_update() took place and set - * 'update_indexes' to TUUI_All). Receiving this hint makes + * 'update_indexes' to TU_All). Receiving this hint makes * us consider if we should pass down the 'indexUnchanged' * hint in turn. That's something that we figure out for * each index_insert() call iff 'update' is true. @@ -290,7 +290,7 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo) * HOT has been applied and any updated columns are indexed * only by summarizing indexes (or in more general terms a * call to table_tuple_update() took place and set - * 'update_indexes' to TUUI_Summarizing). We can (and must) + * 'update_indexes' to TU_Summarizing). We can (and must) * therefore only update the indexes that have * 'amsummarizing' = true. * diff --git a/src/backend/executor/execParallel.c b/src/backend/executor/execParallel.c index f3e77bda279..f098a5557cf 100644 --- a/src/backend/executor/execParallel.c +++ b/src/backend/executor/execParallel.c @@ -189,6 +189,7 @@ ExecSerializePlan(Plan *plan, EState *estate) pstmt->permInfos = estate->es_rteperminfos; pstmt->resultRelations = NIL; pstmt->appendRelations = NIL; + pstmt->planOrigin = PLAN_STMT_INTERNAL; /* * Transfer only parallel-safe subplans, leaving a NULL "hole" in the list diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c index 53ddd25c42d..68184f5d671 100644 --- a/src/backend/executor/execReplication.c +++ b/src/backend/executor/execReplication.c @@ -14,12 +14,14 @@ #include "postgres.h" +#include "access/commit_ts.h" #include "access/genam.h" #include "access/gist.h" #include "access/relscan.h" #include "access/tableam.h" #include "access/transam.h" #include "access/xact.h" +#include "access/heapam.h" #include "catalog/pg_am_d.h" #include "commands/trigger.h" #include "executor/executor.h" @@ -36,7 +38,7 @@ static bool tuples_equal(TupleTableSlot *slot1, TupleTableSlot *slot2, - TypeCacheEntry **eq); + TypeCacheEntry **eq, Bitmapset *columns); /* * Setup a ScanKey for a search in the relation 'rel' for a tuple 'key' that @@ -221,7 +223,7 @@ retry: if (eq == NULL) eq = palloc0(sizeof(*eq) * outslot->tts_tupleDescriptor->natts); - if (!tuples_equal(outslot, searchslot, eq)) + if (!tuples_equal(outslot, searchslot, eq, NULL)) continue; } @@ -277,10 +279,13 @@ retry: /* * Compare the tuples in the slots by checking if they have equal values. + * + * If 'columns' is not null, only the columns specified within it will be + * considered for the equality check, ignoring all other columns. */ static bool tuples_equal(TupleTableSlot *slot1, TupleTableSlot *slot2, - TypeCacheEntry **eq) + TypeCacheEntry **eq, Bitmapset *columns) { int attrnum; @@ -306,6 +311,14 @@ tuples_equal(TupleTableSlot *slot1, TupleTableSlot *slot2, continue; /* + * Ignore columns that are not listed for checking. + */ + if (columns && + !bms_is_member(att->attnum - FirstLowInvalidHeapAttributeNumber, + columns)) + continue; + + /* * If one value is NULL and other is not, then they are certainly not * equal */ @@ -380,7 +393,7 @@ retry: /* Try to find the tuple */ while (table_scan_getnextslot(scan, ForwardScanDirection, scanslot)) { - if (!tuples_equal(scanslot, searchslot, eq)) + if (!tuples_equal(scanslot, searchslot, eq, NULL)) continue; found = true; @@ -456,6 +469,236 @@ BuildConflictIndexInfo(ResultRelInfo *resultRelInfo, Oid conflictindex) } /* + * If the tuple is recently dead and was deleted by a transaction with a newer + * commit timestamp than previously recorded, update the associated transaction + * ID, commit time, and origin. This helps ensure that conflict detection uses + * the most recent and relevant deletion metadata. + */ +static void +update_most_recent_deletion_info(TupleTableSlot *scanslot, + TransactionId oldestxmin, + TransactionId *delete_xid, + TimestampTz *delete_time, + RepOriginId *delete_origin) +{ + BufferHeapTupleTableSlot *hslot; + HeapTuple tuple; + Buffer buf; + bool recently_dead = false; + TransactionId xmax; + TimestampTz localts; + RepOriginId localorigin; + + hslot = (BufferHeapTupleTableSlot *) scanslot; + + tuple = ExecFetchSlotHeapTuple(scanslot, false, NULL); + buf = hslot->buffer; + + LockBuffer(buf, BUFFER_LOCK_SHARE); + + /* + * We do not consider HEAPTUPLE_DEAD status because it indicates either + * tuples whose inserting transaction was aborted (meaning there is no + * commit timestamp or origin), or tuples deleted by a transaction older + * than oldestxmin, making it safe to ignore them during conflict + * detection (See comments atop worker.c for details). + */ + if (HeapTupleSatisfiesVacuum(tuple, oldestxmin, buf) == HEAPTUPLE_RECENTLY_DEAD) + recently_dead = true; + + LockBuffer(buf, BUFFER_LOCK_UNLOCK); + + if (!recently_dead) + return; + + xmax = HeapTupleHeaderGetUpdateXid(tuple->t_data); + if (!TransactionIdIsValid(xmax)) + return; + + /* Select the dead tuple with the most recent commit timestamp */ + if (TransactionIdGetCommitTsData(xmax, &localts, &localorigin) && + TimestampDifferenceExceeds(*delete_time, localts, 0)) + { + *delete_xid = xmax; + *delete_time = localts; + *delete_origin = localorigin; + } +} + +/* + * Searches the relation 'rel' for the most recently deleted tuple that matches + * the values in 'searchslot' and is not yet removable by VACUUM. The function + * returns the transaction ID, origin, and commit timestamp of the transaction + * that deleted this tuple. + * + * 'oldestxmin' acts as a cutoff transaction ID. Tuples deleted by transactions + * with IDs >= 'oldestxmin' are considered recently dead and are eligible for + * conflict detection. + * + * Instead of stopping at the first match, we scan all matching dead tuples to + * identify most recent deletion. This is crucial because only the latest + * deletion is relevant for resolving conflicts. + * + * For example, consider a scenario on the subscriber where a row is deleted, + * re-inserted, and then deleted again only on the subscriber: + * + * - (pk, 1) - deleted at 9:00, + * - (pk, 1) - deleted at 9:02, + * + * Now, a remote update arrives: (pk, 1) -> (pk, 2), timestamped at 9:01. + * + * If we mistakenly return the older deletion (9:00), the system may wrongly + * apply the remote update using a last-update-wins strategy. Instead, we must + * recognize the more recent deletion at 9:02 and skip the update. See + * comments atop worker.c for details. Note, as of now, conflict resolution + * is not implemented. Consequently, the system may incorrectly report the + * older tuple as the conflicted one, leading to misleading results. + * + * The commit timestamp of the deleting transaction is used to determine which + * tuple was deleted most recently. + */ +bool +RelationFindDeletedTupleInfoSeq(Relation rel, TupleTableSlot *searchslot, + TransactionId oldestxmin, + TransactionId *delete_xid, + RepOriginId *delete_origin, + TimestampTz *delete_time) +{ + TupleTableSlot *scanslot; + TableScanDesc scan; + TypeCacheEntry **eq; + Bitmapset *indexbitmap; + TupleDesc desc PG_USED_FOR_ASSERTS_ONLY = RelationGetDescr(rel); + + Assert(equalTupleDescs(desc, searchslot->tts_tupleDescriptor)); + + *delete_xid = InvalidTransactionId; + *delete_origin = InvalidRepOriginId; + *delete_time = 0; + + /* + * If the relation has a replica identity key or a primary key that is + * unusable for locating deleted tuples (see + * IsIndexUsableForFindingDeletedTuple), a full table scan becomes + * necessary. In such cases, comparing the entire tuple is not required, + * since the remote tuple might not include all column values. Instead, + * the indexed columns alone are suffcient to identify the target tuple + * (see logicalrep_rel_mark_updatable). + */ + indexbitmap = RelationGetIndexAttrBitmap(rel, + INDEX_ATTR_BITMAP_IDENTITY_KEY); + + /* fallback to PK if no replica identity */ + if (!indexbitmap) + indexbitmap = RelationGetIndexAttrBitmap(rel, + INDEX_ATTR_BITMAP_PRIMARY_KEY); + + eq = palloc0(sizeof(*eq) * searchslot->tts_tupleDescriptor->natts); + + /* + * Start a heap scan using SnapshotAny to identify dead tuples that are + * not visible under a standard MVCC snapshot. Tuples from transactions + * not yet committed or those just committed prior to the scan are + * excluded in update_most_recent_deletion_info(). + */ + scan = table_beginscan(rel, SnapshotAny, 0, NULL); + scanslot = table_slot_create(rel, NULL); + + table_rescan(scan, NULL); + + /* Try to find the tuple */ + while (table_scan_getnextslot(scan, ForwardScanDirection, scanslot)) + { + if (!tuples_equal(scanslot, searchslot, eq, indexbitmap)) + continue; + + update_most_recent_deletion_info(scanslot, oldestxmin, delete_xid, + delete_time, delete_origin); + } + + table_endscan(scan); + ExecDropSingleTupleTableSlot(scanslot); + + return *delete_time != 0; +} + +/* + * Similar to RelationFindDeletedTupleInfoSeq() but using index scan to locate + * the deleted tuple. + */ +bool +RelationFindDeletedTupleInfoByIndex(Relation rel, Oid idxoid, + TupleTableSlot *searchslot, + TransactionId oldestxmin, + TransactionId *delete_xid, + RepOriginId *delete_origin, + TimestampTz *delete_time) +{ + Relation idxrel; + ScanKeyData skey[INDEX_MAX_KEYS]; + int skey_attoff; + IndexScanDesc scan; + TupleTableSlot *scanslot; + TypeCacheEntry **eq = NULL; + bool isIdxSafeToSkipDuplicates; + TupleDesc desc PG_USED_FOR_ASSERTS_ONLY = RelationGetDescr(rel); + + Assert(equalTupleDescs(desc, searchslot->tts_tupleDescriptor)); + Assert(OidIsValid(idxoid)); + + *delete_xid = InvalidTransactionId; + *delete_time = 0; + *delete_origin = InvalidRepOriginId; + + isIdxSafeToSkipDuplicates = (GetRelationIdentityOrPK(rel) == idxoid); + + scanslot = table_slot_create(rel, NULL); + + idxrel = index_open(idxoid, RowExclusiveLock); + + /* Build scan key. */ + skey_attoff = build_replindex_scan_key(skey, rel, idxrel, searchslot); + + /* + * Start an index scan using SnapshotAny to identify dead tuples that are + * not visible under a standard MVCC snapshot. Tuples from transactions + * not yet committed or those just committed prior to the scan are + * excluded in update_most_recent_deletion_info(). + */ + scan = index_beginscan(rel, idxrel, SnapshotAny, NULL, skey_attoff, 0); + + index_rescan(scan, skey, skey_attoff, NULL, 0); + + /* Try to find the tuple */ + while (index_getnext_slot(scan, ForwardScanDirection, scanslot)) + { + /* + * Avoid expensive equality check if the index is primary key or + * replica identity index. + */ + if (!isIdxSafeToSkipDuplicates) + { + if (eq == NULL) + eq = palloc0(sizeof(*eq) * scanslot->tts_tupleDescriptor->natts); + + if (!tuples_equal(scanslot, searchslot, eq, NULL)) + continue; + } + + update_most_recent_deletion_info(scanslot, oldestxmin, delete_xid, + delete_time, delete_origin); + } + + index_endscan(scan); + + index_close(idxrel, NoLock); + + ExecDropSingleTupleTableSlot(scanslot); + + return *delete_time != 0; +} + +/* * Find the tuple that violates the passed unique index (conflictindex). * * If the conflicting tuple is found return true, otherwise false. @@ -670,7 +913,7 @@ ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo, resultRelInfo->ri_TrigDesc->trig_update_before_row) { if (!ExecBRUpdateTriggers(estate, epqstate, resultRelInfo, - tid, NULL, slot, NULL, NULL)) + tid, NULL, slot, NULL, NULL, false)) skip_tuple = true; /* "do nothing" */ } @@ -746,7 +989,7 @@ ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo, resultRelInfo->ri_TrigDesc->trig_delete_before_row) { skip_tuple = !ExecBRDeleteTriggers(estate, epqstate, resultRelInfo, - tid, NULL, NULL, NULL, NULL); + tid, NULL, NULL, NULL, NULL, false); } if (!skip_tuple) diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c index 2bc89bf84dc..7c6c2c1f6e4 100644 --- a/src/backend/executor/nodeModifyTable.c +++ b/src/backend/executor/nodeModifyTable.c @@ -64,6 +64,7 @@ #include "nodes/nodeFuncs.h" #include "optimizer/optimizer.h" #include "rewrite/rewriteHandler.h" +#include "rewrite/rewriteManip.h" #include "storage/lmgr.h" #include "utils/builtins.h" #include "utils/datum.h" @@ -1473,7 +1474,8 @@ ExecDeletePrologue(ModifyTableContext *context, ResultRelInfo *resultRelInfo, return ExecBRDeleteTriggers(context->estate, context->epqstate, resultRelInfo, tupleid, oldtuple, - epqreturnslot, result, &context->tmfd); + epqreturnslot, result, &context->tmfd, + context->mtstate->operation == CMD_MERGE); } return true; @@ -2116,7 +2118,8 @@ ExecUpdatePrologue(ModifyTableContext *context, ResultRelInfo *resultRelInfo, return ExecBRUpdateTriggers(context->estate, context->epqstate, resultRelInfo, tupleid, oldtuple, slot, - result, &context->tmfd); + result, &context->tmfd, + context->mtstate->operation == CMD_MERGE); } return true; @@ -3735,6 +3738,7 @@ ExecInitMerge(ModifyTableState *mtstate, EState *estate) switch (action->commandType) { case CMD_INSERT: + /* INSERT actions always use rootRelInfo */ ExecCheckPlanOutput(rootRelInfo->ri_RelationDesc, action->targetList); @@ -3774,9 +3778,23 @@ ExecInitMerge(ModifyTableState *mtstate, EState *estate) } else { - /* not partitioned? use the stock relation and slot */ - tgtslot = resultRelInfo->ri_newTupleSlot; - tgtdesc = RelationGetDescr(resultRelInfo->ri_RelationDesc); + /* + * If the MERGE targets an inherited table, we insert + * into the root table, so we must initialize its + * "new" tuple slot, if not already done, and use its + * relation descriptor for the projection. + * + * For non-inherited tables, rootRelInfo and + * resultRelInfo are the same, and the "new" tuple + * slot will already have been initialized. + */ + if (rootRelInfo->ri_newTupleSlot == NULL) + rootRelInfo->ri_newTupleSlot = + table_slot_create(rootRelInfo->ri_RelationDesc, + &estate->es_tupleTable); + + tgtslot = rootRelInfo->ri_newTupleSlot; + tgtdesc = RelationGetDescr(rootRelInfo->ri_RelationDesc); } action_state->mas_proj = @@ -3809,6 +3827,114 @@ ExecInitMerge(ModifyTableState *mtstate, EState *estate) } } } + + /* + * If the MERGE targets an inherited table, any INSERT actions will use + * rootRelInfo, and rootRelInfo will not be in the resultRelInfo array. + * Therefore we must initialize its WITH CHECK OPTION constraints and + * RETURNING projection, as ExecInitModifyTable did for the resultRelInfo + * entries. + * + * Note that the planner does not build a withCheckOptionList or + * returningList for the root relation, but as in ExecInitPartitionInfo, + * we can use the first resultRelInfo entry as a reference to calculate + * the attno's for the root table. + */ + if (rootRelInfo != mtstate->resultRelInfo && + rootRelInfo->ri_RelationDesc->rd_rel->relkind != RELKIND_PARTITIONED_TABLE && + (mtstate->mt_merge_subcommands & MERGE_INSERT) != 0) + { + ModifyTable *node = (ModifyTable *) mtstate->ps.plan; + Relation rootRelation = rootRelInfo->ri_RelationDesc; + Relation firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc; + int firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex; + AttrMap *part_attmap = NULL; + bool found_whole_row; + + if (node->withCheckOptionLists != NIL) + { + List *wcoList; + List *wcoExprs = NIL; + + /* There should be as many WCO lists as result rels */ + Assert(list_length(node->withCheckOptionLists) == + list_length(node->resultRelations)); + + /* + * Use the first WCO list as a reference. In the most common case, + * this will be for the same relation as rootRelInfo, and so there + * will be no need to adjust its attno's. + */ + wcoList = linitial(node->withCheckOptionLists); + if (rootRelation != firstResultRel) + { + /* Convert any Vars in it to contain the root's attno's */ + part_attmap = + build_attrmap_by_name(RelationGetDescr(rootRelation), + RelationGetDescr(firstResultRel), + false); + + wcoList = (List *) + map_variable_attnos((Node *) wcoList, + firstVarno, 0, + part_attmap, + RelationGetForm(rootRelation)->reltype, + &found_whole_row); + } + + foreach(lc, wcoList) + { + WithCheckOption *wco = lfirst_node(WithCheckOption, lc); + ExprState *wcoExpr = ExecInitQual(castNode(List, wco->qual), + &mtstate->ps); + + wcoExprs = lappend(wcoExprs, wcoExpr); + } + + rootRelInfo->ri_WithCheckOptions = wcoList; + rootRelInfo->ri_WithCheckOptionExprs = wcoExprs; + } + + if (node->returningLists != NIL) + { + List *returningList; + + /* There should be as many returning lists as result rels */ + Assert(list_length(node->returningLists) == + list_length(node->resultRelations)); + + /* + * Use the first returning list as a reference. In the most common + * case, this will be for the same relation as rootRelInfo, and so + * there will be no need to adjust its attno's. + */ + returningList = linitial(node->returningLists); + if (rootRelation != firstResultRel) + { + /* Convert any Vars in it to contain the root's attno's */ + if (part_attmap == NULL) + part_attmap = + build_attrmap_by_name(RelationGetDescr(rootRelation), + RelationGetDescr(firstResultRel), + false); + + returningList = (List *) + map_variable_attnos((Node *) returningList, + firstVarno, 0, + part_attmap, + RelationGetForm(rootRelation)->reltype, + &found_whole_row); + } + rootRelInfo->ri_returningList = returningList; + + /* Initialize the RETURNING projection */ + rootRelInfo->ri_projectReturning = + ExecBuildProjectionInfo(returningList, econtext, + mtstate->ps.ps_ResultTupleSlot, + &mtstate->ps, + RelationGetDescr(rootRelation)); + } + } } /* diff --git a/src/backend/executor/nodeTidrangescan.c b/src/backend/executor/nodeTidrangescan.c index ab2eab9596e..26f7420b64b 100644 --- a/src/backend/executor/nodeTidrangescan.c +++ b/src/backend/executor/nodeTidrangescan.c @@ -128,9 +128,11 @@ TidExprListCreate(TidRangeScanState *tidrangestate) * TidRangeEval * * Compute and set node's block and offset range to scan by evaluating - * the trss_tidexprs. Returns false if we detect the range cannot + * node->trss_tidexprs. Returns false if we detect the range cannot * contain any tuples. Returns true if it's possible for the range to - * contain tuples. + * contain tuples. We don't bother validating that trss_mintid is less + * than or equal to trss_maxtid, as the scan_set_tidrange() table AM + * function will handle that. * ---------------------------------------------------------------- */ static bool diff --git a/src/backend/jit/llvm/Makefile b/src/backend/jit/llvm/Makefile index e8c12060b93..68677ba42e1 100644 --- a/src/backend/jit/llvm/Makefile +++ b/src/backend/jit/llvm/Makefile @@ -31,7 +31,7 @@ endif # All files in this directory use LLVM. CFLAGS += $(LLVM_CFLAGS) CXXFLAGS += $(LLVM_CXXFLAGS) -override CPPFLAGS := $(LLVM_CPPFLAGS) $(CPPFLAGS) +override CPPFLAGS += $(LLVM_CPPFLAGS) SHLIB_LINK += $(LLVM_LIBS) # Because this module includes C++ files, we need to use a C++ diff --git a/src/backend/jit/llvm/meson.build b/src/backend/jit/llvm/meson.build index c8e06dfbe35..805fbd69006 100644 --- a/src/backend/jit/llvm/meson.build +++ b/src/backend/jit/llvm/meson.build @@ -53,7 +53,7 @@ llvm_irgen_args = [ if ccache.found() llvm_irgen_command = ccache - llvm_irgen_args = [clang.path()] + llvm_irgen_args + llvm_irgen_args = [clang.full_path()] + llvm_irgen_args else llvm_irgen_command = clang endif diff --git a/src/backend/lib/README b/src/backend/lib/README index f2fb591237d..c28cbe356f0 100644 --- a/src/backend/lib/README +++ b/src/backend/lib/README @@ -1,8 +1,6 @@ This directory contains a general purpose data structures, for use anywhere in the backend: -binaryheap.c - a binary heap - bipartite_match.c - Hopcroft-Karp maximum cardinality algorithm for bipartite graphs bloomfilter.c - probabilistic, space-efficient set membership testing @@ -21,8 +19,6 @@ pairingheap.c - a pairing heap rbtree.c - a red-black tree -stringinfo.c - an extensible string type - Aside from the inherent characteristics of the data structures, there are a few practical differences between the binary heap and the pairing heap. The diff --git a/src/backend/libpq/auth.c b/src/backend/libpq/auth.c index 9f4d05ffbd4..4da46666439 100644 --- a/src/backend/libpq/auth.c +++ b/src/backend/libpq/auth.c @@ -94,8 +94,16 @@ static int auth_peer(hbaPort *port); #define PGSQL_PAM_SERVICE "postgresql" /* Service name passed to PAM */ +/* Work around original Solaris' lack of "const" in the conv_proc signature */ +#ifdef _PAM_LEGACY_NONCONST +#define PG_PAM_CONST +#else +#define PG_PAM_CONST const +#endif + static int CheckPAMAuth(Port *port, const char *user, const char *password); -static int pam_passwd_conv_proc(int num_msg, const struct pam_message **msg, +static int pam_passwd_conv_proc(int num_msg, + PG_PAM_CONST struct pam_message **msg, struct pam_response **resp, void *appdata_ptr); static struct pam_conv pam_passw_conv = { @@ -1917,7 +1925,7 @@ auth_peer(hbaPort *port) */ static int -pam_passwd_conv_proc(int num_msg, const struct pam_message **msg, +pam_passwd_conv_proc(int num_msg, PG_PAM_CONST struct pam_message **msg, struct pam_response **resp, void *appdata_ptr) { const char *passwd; diff --git a/src/backend/libpq/be-secure-gssapi.c b/src/backend/libpq/be-secure-gssapi.c index 3534f0b8111..5d98c58ffa8 100644 --- a/src/backend/libpq/be-secure-gssapi.c +++ b/src/backend/libpq/be-secure-gssapi.c @@ -121,9 +121,9 @@ be_gssapi_write(Port *port, const void *ptr, size_t len) * again, so if it offers a len less than that, something is wrong. * * Note: it may seem attractive to report partial write completion once - * we've successfully sent any encrypted packets. However, that can cause - * problems for callers; notably, pqPutMsgEnd's heuristic to send only - * full 8K blocks interacts badly with such a hack. We won't save much, + * we've successfully sent any encrypted packets. However, doing that + * expands the state space of this processing and has been responsible for + * bugs in the past (cf. commit d053a879b). We won't save much, * typically, by letting callers discard data early, so don't risk it. */ if (len < PqGSSSendConsumed) diff --git a/src/backend/libpq/be-secure-openssl.c b/src/backend/libpq/be-secure-openssl.c index 64ff3ce3d6a..c8b63ef8249 100644 --- a/src/backend/libpq/be-secure-openssl.c +++ b/src/backend/libpq/be-secure-openssl.c @@ -1436,10 +1436,10 @@ initialize_ecdh(SSL_CTX *context, bool isServerStart) */ ereport(isServerStart ? FATAL : LOG, errcode(ERRCODE_CONFIG_FILE_ERROR), - errmsg("failed to set group names specified in ssl_groups: %s", + errmsg("could not set group names specified in ssl_groups: %s", SSLerrmessageExt(ERR_get_error(), _("No valid groups found"))), - errhint("Ensure that each group name is spelled correctly and supported by the installed version of OpenSSL")); + errhint("Ensure that each group name is spelled correctly and supported by the installed version of OpenSSL.")); return false; } #endif diff --git a/src/backend/libpq/hba.c b/src/backend/libpq/hba.c index 332fad27835..fecee8224d0 100644 --- a/src/backend/libpq/hba.c +++ b/src/backend/libpq/hba.c @@ -2873,8 +2873,11 @@ check_ident_usermap(IdentLine *identLine, const char *usermap_name, !token_has_regexp(identLine->pg_user) && (ofs = strstr(identLine->pg_user->string, "\\1")) != NULL) { + const char *repl_str; + size_t repl_len; + char *old_pg_user; char *expanded_pg_user; - int offset; + size_t offset; /* substitution of the first argument requested */ if (matches[1].rm_so < 0) @@ -2886,18 +2889,33 @@ check_ident_usermap(IdentLine *identLine, const char *usermap_name, *error_p = true; return; } + repl_str = system_user + matches[1].rm_so; + repl_len = matches[1].rm_eo - matches[1].rm_so; /* - * length: original length minus length of \1 plus length of match - * plus null terminator + * It's allowed to have more than one \1 in the string, and we'll + * replace them all. But that's pretty unusual so we optimize on + * the assumption of only one occurrence, which motivates doing + * repeated replacements instead of making two passes over the + * string to determine the final length right away. */ - expanded_pg_user = palloc0(strlen(identLine->pg_user->string) - 2 + (matches[1].rm_eo - matches[1].rm_so) + 1); - offset = ofs - identLine->pg_user->string; - memcpy(expanded_pg_user, identLine->pg_user->string, offset); - memcpy(expanded_pg_user + offset, - system_user + matches[1].rm_so, - matches[1].rm_eo - matches[1].rm_so); - strcat(expanded_pg_user, ofs + 2); + old_pg_user = identLine->pg_user->string; + do + { + /* + * length: current length minus length of \1 plus length of + * replacement plus null terminator + */ + expanded_pg_user = palloc(strlen(old_pg_user) - 2 + repl_len + 1); + /* ofs points into the old_pg_user string at this point */ + offset = ofs - old_pg_user; + memcpy(expanded_pg_user, old_pg_user, offset); + memcpy(expanded_pg_user + offset, repl_str, repl_len); + strcpy(expanded_pg_user + offset + repl_len, ofs + 2); + if (old_pg_user != identLine->pg_user->string) + pfree(old_pg_user); + old_pg_user = expanded_pg_user; + } while ((ofs = strstr(old_pg_user + offset + repl_len, "\\1")) != NULL); /* * Mark the token as quoted, so it will only be compared literally diff --git a/src/backend/libpq/pg_ident.conf.sample b/src/backend/libpq/pg_ident.conf.sample index f5225f26cdf..8ee6c0ba315 100644 --- a/src/backend/libpq/pg_ident.conf.sample +++ b/src/backend/libpq/pg_ident.conf.sample @@ -13,25 +13,25 @@ # user names to their corresponding PostgreSQL user names. Records # are of the form: # -# MAPNAME SYSTEM-USERNAME PG-USERNAME +# MAPNAME SYSTEM-USERNAME DATABASE-USERNAME # # (The uppercase quantities must be replaced by actual values.) # # MAPNAME is the (otherwise freely chosen) map name that was used in # pg_hba.conf. SYSTEM-USERNAME is the detected user name of the -# client. PG-USERNAME is the requested PostgreSQL user name. The -# existence of a record specifies that SYSTEM-USERNAME may connect as -# PG-USERNAME. +# client. DATABASE-USERNAME is the requested PostgreSQL user name. +# The existence of a record specifies that SYSTEM-USERNAME may connect +# as DATABASE-USERNAME. # -# If SYSTEM-USERNAME starts with a slash (/), it will be treated as a -# regular expression. Optionally this can contain a capture (a -# parenthesized subexpression). The substring matching the capture -# will be substituted for \1 (backslash-one) if present in -# PG-USERNAME. +# If SYSTEM-USERNAME starts with a slash (/), the rest of it will be +# treated as a regular expression. Optionally this can contain a capture +# (a parenthesized subexpression). The substring matching the capture +# will be substituted for \1 (backslash-one) if that appears in +# DATABASE-USERNAME. # -# PG-USERNAME can be "all", a user name, a group name prefixed with "+", or -# a regular expression (if it starts with a slash (/)). If it is a regular -# expression, the substring matching with \1 has no effect. +# DATABASE-USERNAME can be "all", a user name, a group name prefixed with "+", +# or a regular expression (if it starts with a slash (/)). If it is a regular +# expression, no substitution for \1 will occur. # # Multiple maps may be specified in this file and used by pg_hba.conf. # @@ -69,4 +69,4 @@ # Put your actual configuration here # ---------------------------------- -# MAPNAME SYSTEM-USERNAME PG-USERNAME +# MAPNAME SYSTEM-USERNAME DATABASE-USERNAME diff --git a/src/backend/libpq/pqcomm.c b/src/backend/libpq/pqcomm.c index e5171467de1..25f739a6a17 100644 --- a/src/backend/libpq/pqcomm.c +++ b/src/backend/libpq/pqcomm.c @@ -858,7 +858,6 @@ RemoveSocketFiles(void) (void) unlink(sock_path); } /* Since we're about to exit, no need to reclaim storage */ - sock_paths = NIL; } diff --git a/src/backend/libpq/pqmq.c b/src/backend/libpq/pqmq.c index f1a08bc32ca..5f39949a367 100644 --- a/src/backend/libpq/pqmq.c +++ b/src/backend/libpq/pqmq.c @@ -23,7 +23,7 @@ #include "tcop/tcopprot.h" #include "utils/builtins.h" -static shm_mq_handle *pq_mq_handle; +static shm_mq_handle *pq_mq_handle = NULL; static bool pq_mq_busy = false; static pid_t pq_mq_parallel_leader_pid = 0; static ProcNumber pq_mq_parallel_leader_proc_number = INVALID_PROC_NUMBER; @@ -66,7 +66,11 @@ pq_redirect_to_shm_mq(dsm_segment *seg, shm_mq_handle *mqh) static void pq_cleanup_redirect_to_shm_mq(dsm_segment *seg, Datum arg) { - pq_mq_handle = NULL; + if (pq_mq_handle != NULL) + { + pfree(pq_mq_handle); + pq_mq_handle = NULL; + } whereToSendOutput = DestNone; } @@ -131,8 +135,11 @@ mq_putmessage(char msgtype, const char *s, size_t len) if (pq_mq_busy) { if (pq_mq_handle != NULL) + { shm_mq_detach(pq_mq_handle); - pq_mq_handle = NULL; + pfree(pq_mq_handle); + pq_mq_handle = NULL; + } return EOF; } @@ -152,8 +159,6 @@ mq_putmessage(char msgtype, const char *s, size_t len) iov[1].data = s; iov[1].len = len; - Assert(pq_mq_handle != NULL); - for (;;) { /* @@ -161,6 +166,7 @@ mq_putmessage(char msgtype, const char *s, size_t len) * that the shared memory value is updated before we send the parallel * message signal right after this. */ + Assert(pq_mq_handle != NULL); result = shm_mq_sendv(pq_mq_handle, iov, 2, true, true); if (pq_mq_parallel_leader_pid != 0) diff --git a/src/backend/main/main.c b/src/backend/main/main.c index 7d63cf94a6b..bdcb5e4f261 100644 --- a/src/backend/main/main.c +++ b/src/backend/main/main.c @@ -125,13 +125,17 @@ main(int argc, char *argv[]) set_pglocale_pgservice(argv[0], PG_TEXTDOMAIN("postgres")); /* - * In the postmaster, absorb the environment values for LC_COLLATE and - * LC_CTYPE. Individual backends will change these later to settings - * taken from pg_database, but the postmaster cannot do that. If we leave - * these set to "C" then message localization might not work well in the - * postmaster. + * Collation is handled by pg_locale.c, and the behavior is dependent on + * the provider. strcoll(), etc., should not be called directly. + */ + init_locale("LC_COLLATE", LC_COLLATE, "C"); + + /* + * In the postmaster, absorb the environment value for LC_CTYPE. + * Individual backends will change it later to pg_database.datctype, but + * the postmaster cannot do that. If we leave it set to "C" then message + * localization might not work well in the postmaster. */ - init_locale("LC_COLLATE", LC_COLLATE, ""); init_locale("LC_CTYPE", LC_CTYPE, ""); /* diff --git a/src/backend/nodes/gen_node_support.pl b/src/backend/nodes/gen_node_support.pl index c8595109b0e..9ecddb14231 100644 --- a/src/backend/nodes/gen_node_support.pl +++ b/src/backend/nodes/gen_node_support.pl @@ -1329,7 +1329,7 @@ _jumble${n}(JumbleState *jstate, Node *node) # Node type. Squash constants if requested. if ($query_jumble_squash) { - print $jff "\tJUMBLE_ELEMENTS($f);\n" + print $jff "\tJUMBLE_ELEMENTS($f, node);\n" unless $query_jumble_ignore; } else diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c index 25e08ba3426..eaf391fc2ab 100644 --- a/src/backend/nodes/outfuncs.c +++ b/src/backend/nodes/outfuncs.c @@ -653,6 +653,8 @@ _outA_Expr(StringInfo str, const A_Expr *node) WRITE_NODE_FIELD(lexpr); WRITE_NODE_FIELD(rexpr); + WRITE_LOCATION_FIELD(rexpr_list_start); + WRITE_LOCATION_FIELD(rexpr_list_end); WRITE_LOCATION_FIELD(location); } diff --git a/src/backend/nodes/queryjumblefuncs.c b/src/backend/nodes/queryjumblefuncs.c index ac3cb3d9caf..31f97151977 100644 --- a/src/backend/nodes/queryjumblefuncs.c +++ b/src/backend/nodes/queryjumblefuncs.c @@ -21,6 +21,11 @@ * tree(s) generated from the query. The executor can then use this value * to blame query costs on the proper queryId. * + * Arrays of two or more constants and PARAM_EXTERN parameters are "squashed" + * and contribute only once to the jumble. This has the effect that queries + * that differ only on the length of such lists have the same queryId. + * + * * Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group * Portions Copyright (c) 1994, Regents of the University of California * @@ -61,11 +66,13 @@ static void AppendJumble(JumbleState *jstate, const unsigned char *value, Size size); static void FlushPendingNulls(JumbleState *jstate); static void RecordConstLocation(JumbleState *jstate, - int location, bool squashed); + bool extern_param, + int location, int len); static void _jumbleNode(JumbleState *jstate, Node *node); -static void _jumbleElements(JumbleState *jstate, List *elements); -static void _jumbleA_Const(JumbleState *jstate, Node *node); static void _jumbleList(JumbleState *jstate, Node *node); +static void _jumbleElements(JumbleState *jstate, List *elements, Node *node); +static void _jumbleParam(JumbleState *jstate, Node *node); +static void _jumbleA_Const(JumbleState *jstate, Node *node); static void _jumbleVariableSetStmt(JumbleState *jstate, Node *node); static void _jumbleRangeTblEntry_eref(JumbleState *jstate, RangeTblEntry *rte, @@ -185,6 +192,7 @@ InitJumble(void) jstate->clocations_count = 0; jstate->highest_extern_param_id = 0; jstate->pending_nulls = 0; + jstate->has_squashed_lists = false; #ifdef USE_ASSERT_CHECKING jstate->total_jumble_len = 0; #endif @@ -207,6 +215,10 @@ DoJumble(JumbleState *jstate, Node *node) if (jstate->pending_nulls > 0) FlushPendingNulls(jstate); + /* Squashed list found, reset highest_extern_param_id */ + if (jstate->has_squashed_lists) + jstate->highest_extern_param_id = 0; + /* Process the jumble buffer and produce the hash value */ return DatumGetInt64(hash_any_extended(jstate->jumble, jstate->jumble_len, @@ -373,15 +385,17 @@ FlushPendingNulls(JumbleState *jstate) /* - * Record location of constant within query string of query tree that is - * currently being walked. + * Record the location of some kind of constant within a query string. + * These are not only bare constants but also expressions that ultimately + * constitute a constant, such as those inside casts and simple function + * calls; if extern_param, then it corresponds to a PARAM_EXTERN Param. * - * 'squashed' signals that the constant represents the first or the last - * element in a series of merged constants, and everything but the first/last - * element contributes nothing to the jumble hash. + * If length is -1, it indicates a single such constant element. If + * it's a positive integer, it indicates the length of a squashable + * list of them. */ static void -RecordConstLocation(JumbleState *jstate, int location, bool squashed) +RecordConstLocation(JumbleState *jstate, bool extern_param, int location, int len) { /* -1 indicates unknown or undefined location */ if (location >= 0) @@ -396,9 +410,15 @@ RecordConstLocation(JumbleState *jstate, int location, bool squashed) sizeof(LocationLen)); } jstate->clocations[jstate->clocations_count].location = location; - /* initialize lengths to -1 to simplify third-party module usage */ - jstate->clocations[jstate->clocations_count].squashed = squashed; - jstate->clocations[jstate->clocations_count].length = -1; + + /* + * Lengths are either positive integers (indicating a squashable + * list), or -1. + */ + Assert(len > -1 || len == -1); + jstate->clocations[jstate->clocations_count].length = len; + jstate->clocations[jstate->clocations_count].squashed = (len > -1); + jstate->clocations[jstate->clocations_count].extern_param = extern_param; jstate->clocations_count++; } } @@ -407,47 +427,74 @@ RecordConstLocation(JumbleState *jstate, int location, bool squashed) * Subroutine for _jumbleElements: Verify a few simple cases where we can * deduce that the expression is a constant: * - * - Ignore a possible wrapping RelabelType and CoerceViaIO. - * - If it's a FuncExpr, check that the function is an implicit + * - See through any wrapping RelabelType and CoerceViaIO layers. + * - If it's a FuncExpr, check that the function is a builtin * cast and its arguments are Const. - * - Otherwise test if the expression is a simple Const. + * - Otherwise test if the expression is a simple Const or a + * PARAM_EXTERN param. */ static bool -IsSquashableConst(Node *element) +IsSquashableConstant(Node *element) { - if (IsA(element, RelabelType)) - element = (Node *) ((RelabelType *) element)->arg; - - if (IsA(element, CoerceViaIO)) - element = (Node *) ((CoerceViaIO *) element)->arg; - - if (IsA(element, FuncExpr)) +restart: + switch (nodeTag(element)) { - FuncExpr *func = (FuncExpr *) element; - ListCell *temp; + case T_RelabelType: + /* Unwrap RelabelType */ + element = (Node *) ((RelabelType *) element)->arg; + goto restart; - if (func->funcformat != COERCE_IMPLICIT_CAST && - func->funcformat != COERCE_EXPLICIT_CAST) - return false; + case T_CoerceViaIO: + /* Unwrap CoerceViaIO */ + element = (Node *) ((CoerceViaIO *) element)->arg; + goto restart; - if (func->funcid > FirstGenbkiObjectId) - return false; + case T_Const: + return true; - foreach(temp, func->args) - { - Node *arg = lfirst(temp); + case T_Param: + return castNode(Param, element)->paramkind == PARAM_EXTERN; - if (!IsA(arg, Const)) /* XXX we could recurse here instead */ - return false; - } + case T_FuncExpr: + { + FuncExpr *func = (FuncExpr *) element; + ListCell *temp; - return true; - } + if (func->funcformat != COERCE_IMPLICIT_CAST && + func->funcformat != COERCE_EXPLICIT_CAST) + return false; - if (!IsA(element, Const)) - return false; + if (func->funcid > FirstGenbkiObjectId) + return false; - return true; + /* + * We can check function arguments recursively, being careful + * about recursing too deep. At each recursion level it's + * enough to test the stack on the first element. (Note that + * I wasn't able to hit this without bloating the stack + * artificially in this function: the parser errors out before + * stack size becomes a problem here.) + */ + foreach(temp, func->args) + { + Node *arg = lfirst(temp); + + if (!IsA(arg, Const)) + { + if (foreach_current_index(temp) == 0 && + stack_is_too_deep()) + return false; + else if (!IsSquashableConstant(arg)) + return false; + } + } + + return true; + } + + default: + return false; + } } /* @@ -457,39 +504,33 @@ IsSquashableConst(Node *element) * Return value indicates if squashing is possible. * * Note that this function searches only for explicit Const nodes with - * possibly very simple decorations on top, and does not try to simplify - * expressions. + * possibly very simple decorations on top and PARAM_EXTERN parameters, + * and does not try to simplify expressions. */ static bool -IsSquashableConstList(List *elements, Node **firstExpr, Node **lastExpr) +IsSquashableConstantList(List *elements) { ListCell *temp; - /* - * If squashing is disabled, or the list is too short, we don't try to - * squash it. - */ + /* If the list is too short, we don't try to squash it. */ if (list_length(elements) < 2) return false; foreach(temp, elements) { - if (!IsSquashableConst(lfirst(temp))) + if (!IsSquashableConstant(lfirst(temp))) return false; } - *firstExpr = linitial(elements); - *lastExpr = llast(elements); - return true; } #define JUMBLE_NODE(item) \ _jumbleNode(jstate, (Node *) expr->item) -#define JUMBLE_ELEMENTS(list) \ - _jumbleElements(jstate, (List *) expr->list) +#define JUMBLE_ELEMENTS(list, node) \ + _jumbleElements(jstate, (List *) expr->list, node) #define JUMBLE_LOCATION(location) \ - RecordConstLocation(jstate, expr->location, false) + RecordConstLocation(jstate, false, expr->location, -1) #define JUMBLE_FIELD(item) \ do { \ if (sizeof(expr->item) == 8) \ @@ -516,42 +557,6 @@ do { \ #include "queryjumblefuncs.funcs.c" -/* - * We jumble lists of constant elements as one individual item regardless - * of how many elements are in the list. This means different queries - * jumble to the same query_id, if the only difference is the number of - * elements in the list. - */ -static void -_jumbleElements(JumbleState *jstate, List *elements) -{ - Node *first, - *last; - - if (IsSquashableConstList(elements, &first, &last)) - { - /* - * If this list of elements is squashable, keep track of the location - * of its first and last elements. When reading back the locations - * array, we'll see two consecutive locations with ->squashed set to - * true, indicating the location of initial and final elements of this - * list. - * - * For the limited set of cases we support now (implicit coerce via - * FuncExpr, Const) it's fine to use exprLocation of the 'last' - * expression, but if more complex composite expressions are to be - * supported (e.g., OpExpr or FuncExpr as an explicit call), more - * sophisticated tracking will be needed. - */ - RecordConstLocation(jstate, exprLocation(first), true); - RecordConstLocation(jstate, exprLocation(last), true); - } - else - { - _jumbleNode(jstate, (Node *) elements); - } -} - static void _jumbleNode(JumbleState *jstate, Node *node) { @@ -593,26 +598,6 @@ _jumbleNode(JumbleState *jstate, Node *node) break; } - /* Special cases to handle outside the automated code */ - switch (nodeTag(expr)) - { - case T_Param: - { - Param *p = (Param *) node; - - /* - * Update the highest Param id seen, in order to start - * normalization correctly. - */ - if (p->paramkind == PARAM_EXTERN && - p->paramid > jstate->highest_extern_param_id) - jstate->highest_extern_param_id = p->paramid; - } - break; - default: - break; - } - /* Ensure we added something to the jumble buffer */ Assert(jstate->total_jumble_len > prev_jumble_len); } @@ -648,6 +633,79 @@ _jumbleList(JumbleState *jstate, Node *node) } } +/* + * We try to jumble lists of expressions as one individual item regardless + * of how many elements are in the list. This is know as squashing, which + * results in different queries jumbling to the same query_id, if the only + * difference is the number of elements in the list. + * + * We allow constants and PARAM_EXTERN parameters to be squashed. To normalize + * such queries, we use the start and end locations of the list of elements in + * a list. + */ +static void +_jumbleElements(JumbleState *jstate, List *elements, Node *node) +{ + bool normalize_list = false; + + if (IsSquashableConstantList(elements)) + { + if (IsA(node, ArrayExpr)) + { + ArrayExpr *aexpr = (ArrayExpr *) node; + + if (aexpr->list_start > 0 && aexpr->list_end > 0) + { + RecordConstLocation(jstate, + false, + aexpr->list_start + 1, + (aexpr->list_end - aexpr->list_start) - 1); + normalize_list = true; + jstate->has_squashed_lists = true; + } + } + } + + if (!normalize_list) + { + _jumbleNode(jstate, (Node *) elements); + } +} + +/* + * We store the highest param ID of extern params. This can later be used + * to start the numbering of the placeholder for squashed lists. + */ +static void +_jumbleParam(JumbleState *jstate, Node *node) +{ + Param *expr = (Param *) node; + + JUMBLE_FIELD(paramkind); + JUMBLE_FIELD(paramid); + JUMBLE_FIELD(paramtype); + /* paramtypmode and paramcollid are ignored */ + + if (expr->paramkind == PARAM_EXTERN) + { + /* + * At this point, only external parameter locations outside of + * squashable lists will be recorded. + */ + RecordConstLocation(jstate, true, expr->location, -1); + + /* + * Update the highest Param id seen, in order to start normalization + * correctly. + * + * Note: This value is reset at the end of jumbling if there exists a + * squashable list. See the comment in the definition of JumbleState. + */ + if (expr->paramid > jstate->highest_extern_param_id) + jstate->highest_extern_param_id = expr->paramid; + } +} + static void _jumbleA_Const(JumbleState *jstate, Node *node) { diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c index 8c90ab54af8..48b5d13b9b6 100644 --- a/src/backend/nodes/readfuncs.c +++ b/src/backend/nodes/readfuncs.c @@ -526,6 +526,8 @@ _readA_Expr(void) READ_NODE_FIELD(lexpr); READ_NODE_FIELD(rexpr); + READ_LOCATION_FIELD(rexpr_list_start); + READ_LOCATION_FIELD(rexpr_list_end); READ_LOCATION_FIELD(location); READ_DONE(); diff --git a/src/backend/optimizer/path/costsize.c b/src/backend/optimizer/path/costsize.c index 3d44815ed5a..344a3188317 100644 --- a/src/backend/optimizer/path/costsize.c +++ b/src/backend/optimizer/path/costsize.c @@ -2247,7 +2247,7 @@ append_nonpartial_cost(List *subpaths, int numpaths, int parallel_workers) * Determines and returns the cost of an Append node. */ void -cost_append(AppendPath *apath) +cost_append(AppendPath *apath, PlannerInfo *root) { ListCell *l; @@ -2309,26 +2309,52 @@ cost_append(AppendPath *apath) foreach(l, apath->subpaths) { Path *subpath = (Path *) lfirst(l); - Path sort_path; /* dummy for result of cost_sort */ + int presorted_keys; + Path sort_path; /* dummy for result of + * cost_sort/cost_incremental_sort */ - if (!pathkeys_contained_in(pathkeys, subpath->pathkeys)) + if (!pathkeys_count_contained_in(pathkeys, subpath->pathkeys, + &presorted_keys)) { /* * We'll need to insert a Sort node, so include costs for - * that. We can use the parent's LIMIT if any, since we + * that. We choose to use incremental sort if it is + * enabled and there are presorted keys; otherwise we use + * full sort. + * + * We can use the parent's LIMIT if any, since we * certainly won't pull more than that many tuples from * any child. */ - cost_sort(&sort_path, - NULL, /* doesn't currently need root */ - pathkeys, - subpath->disabled_nodes, - subpath->total_cost, - subpath->rows, - subpath->pathtarget->width, - 0.0, - work_mem, - apath->limit_tuples); + if (enable_incremental_sort && presorted_keys > 0) + { + cost_incremental_sort(&sort_path, + root, + pathkeys, + presorted_keys, + subpath->disabled_nodes, + subpath->startup_cost, + subpath->total_cost, + subpath->rows, + subpath->pathtarget->width, + 0.0, + work_mem, + apath->limit_tuples); + } + else + { + cost_sort(&sort_path, + root, + pathkeys, + subpath->disabled_nodes, + subpath->total_cost, + subpath->rows, + subpath->pathtarget->width, + 0.0, + work_mem, + apath->limit_tuples); + } + subpath = &sort_path; } @@ -2546,13 +2572,13 @@ cost_memoize_rescan(PlannerInfo *root, MemoizePath *mpath, Cost input_startup_cost = mpath->subpath->startup_cost; Cost input_total_cost = mpath->subpath->total_cost; double tuples = mpath->subpath->rows; - double calls = mpath->calls; + Cardinality est_calls = mpath->est_calls; int width = mpath->subpath->pathtarget->width; double hash_mem_bytes; double est_entry_bytes; - double est_cache_entries; - double ndistinct; + Cardinality est_cache_entries; + Cardinality ndistinct; double evict_ratio; double hit_ratio; Cost startup_cost; @@ -2578,7 +2604,7 @@ cost_memoize_rescan(PlannerInfo *root, MemoizePath *mpath, est_cache_entries = floor(hash_mem_bytes / est_entry_bytes); /* estimate on the distinct number of parameter values */ - ndistinct = estimate_num_groups(root, mpath->param_exprs, calls, NULL, + ndistinct = estimate_num_groups(root, mpath->param_exprs, est_calls, NULL, &estinfo); /* @@ -2590,7 +2616,10 @@ cost_memoize_rescan(PlannerInfo *root, MemoizePath *mpath, * certainly mean a MemoizePath will never survive add_path(). */ if ((estinfo.flags & SELFLAG_USED_DEFAULT) != 0) - ndistinct = calls; + ndistinct = est_calls; + + /* Remember the ndistinct estimate for EXPLAIN */ + mpath->est_unique_keys = ndistinct; /* * Since we've already estimated the maximum number of entries we can @@ -2618,9 +2647,12 @@ cost_memoize_rescan(PlannerInfo *root, MemoizePath *mpath, * must look at how many scans are estimated in total for this node and * how many of those scans we expect to get a cache hit. */ - hit_ratio = ((calls - ndistinct) / calls) * + hit_ratio = ((est_calls - ndistinct) / est_calls) * (est_cache_entries / Max(ndistinct, est_cache_entries)); + /* Remember the hit ratio estimate for EXPLAIN */ + mpath->est_hit_ratio = hit_ratio; + Assert(hit_ratio >= 0 && hit_ratio <= 1.0); /* diff --git a/src/backend/optimizer/path/joinpath.c b/src/backend/optimizer/path/joinpath.c index 26f0336f1e4..ebedc5574ca 100644 --- a/src/backend/optimizer/path/joinpath.c +++ b/src/backend/optimizer/path/joinpath.c @@ -154,13 +154,17 @@ add_paths_to_joinrel(PlannerInfo *root, /* * See if the inner relation is provably unique for this outer rel. * - * We have some special cases: for JOIN_SEMI and JOIN_ANTI, it doesn't - * matter since the executor can make the equivalent optimization anyway; - * we need not expend planner cycles on proofs. For JOIN_UNIQUE_INNER, we - * must be considering a semijoin whose inner side is not provably unique - * (else reduce_unique_semijoins would've simplified it), so there's no - * point in calling innerrel_is_unique. However, if the LHS covers all of - * the semijoin's min_lefthand, then it's appropriate to set inner_unique + * We have some special cases: for JOIN_SEMI, it doesn't matter since the + * executor can make the equivalent optimization anyway. It also doesn't + * help enable use of Memoize, since a semijoin with a provably unique + * inner side should have been reduced to an inner join in that case. + * Therefore, we need not expend planner cycles on proofs. (For + * JOIN_ANTI, although it doesn't help the executor for the same reason, + * it can benefit Memoize paths.) For JOIN_UNIQUE_INNER, we must be + * considering a semijoin whose inner side is not provably unique (else + * reduce_unique_semijoins would've simplified it), so there's no point in + * calling innerrel_is_unique. However, if the LHS covers all of the + * semijoin's min_lefthand, then it's appropriate to set inner_unique * because the path produced by create_unique_path will be unique relative * to the LHS. (If we have an LHS that's only part of the min_lefthand, * that is *not* true.) For JOIN_UNIQUE_OUTER, pass JOIN_INNER to avoid @@ -169,12 +173,6 @@ add_paths_to_joinrel(PlannerInfo *root, switch (jointype) { case JOIN_SEMI: - case JOIN_ANTI: - - /* - * XXX it may be worth proving this to allow a Memoize to be - * considered for Nested Loop Semi/Anti Joins. - */ extra.inner_unique = false; /* well, unproven */ break; case JOIN_UNIQUE_INNER: @@ -715,16 +713,21 @@ get_memoize_path(PlannerInfo *root, RelOptInfo *innerrel, return NULL; /* - * Currently we don't do this for SEMI and ANTI joins unless they're - * marked as inner_unique. This is because nested loop SEMI/ANTI joins - * don't scan the inner node to completion, which will mean memoize cannot - * mark the cache entry as complete. - * - * XXX Currently we don't attempt to mark SEMI/ANTI joins as inner_unique - * = true. Should we? See add_paths_to_joinrel() + * Currently we don't do this for SEMI and ANTI joins, because nested loop + * SEMI/ANTI joins don't scan the inner node to completion, which means + * memoize cannot mark the cache entry as complete. Nor can we mark the + * cache entry as complete after fetching the first inner tuple, because + * if that tuple and the current outer tuple don't satisfy the join + * clauses, a second inner tuple that satisfies the parameters would find + * the cache entry already marked as complete. The only exception is when + * the inner relation is provably unique, as in that case, there won't be + * a second matching tuple and we can safely mark the cache entry as + * complete after fetching the first inner tuple. Note that in such + * cases, the SEMI join should have been reduced to an inner join by + * reduce_unique_semijoins. */ - if (!extra->inner_unique && (jointype == JOIN_SEMI || - jointype == JOIN_ANTI)) + if ((jointype == JOIN_SEMI || jointype == JOIN_ANTI) && + !extra->inner_unique) return NULL; /* @@ -876,16 +879,13 @@ try_nestloop_path(PlannerInfo *root, /* * Check to see if proposed path is still parameterized, and reject if the * parameterization wouldn't be sensible --- unless allow_star_schema_join - * says to allow it anyway. Also, we must reject if have_dangerous_phv - * doesn't like the look of it, which could only happen if the nestloop is - * still parameterized. + * says to allow it anyway. */ required_outer = calc_nestloop_required_outer(outerrelids, outer_paramrels, innerrelids, inner_paramrels); if (required_outer && - ((!bms_overlap(required_outer, extra->param_source_rels) && - !allow_star_schema_join(root, outerrelids, inner_paramrels)) || - have_dangerous_phv(root, outerrelids, inner_paramrels))) + !bms_overlap(required_outer, extra->param_source_rels) && + !allow_star_schema_join(root, outerrelids, inner_paramrels)) { /* Waste no memory when we reject a path here */ bms_free(required_outer); diff --git a/src/backend/optimizer/path/joinrels.c b/src/backend/optimizer/path/joinrels.c index 60d65762b5d..aad41b94009 100644 --- a/src/backend/optimizer/path/joinrels.c +++ b/src/backend/optimizer/path/joinrels.c @@ -565,9 +565,6 @@ join_is_legal(PlannerInfo *root, RelOptInfo *rel1, RelOptInfo *rel2, * Also, if the lateral reference is only indirect, we should reject * the join; whatever rel(s) the reference chain goes through must be * joined to first. - * - * Another case that might keep us from building a valid plan is the - * implementation restriction described by have_dangerous_phv(). */ lateral_fwd = bms_overlap(rel1->relids, rel2->lateral_relids); lateral_rev = bms_overlap(rel2->relids, rel1->lateral_relids); @@ -584,9 +581,6 @@ join_is_legal(PlannerInfo *root, RelOptInfo *rel1, RelOptInfo *rel2, /* check there is a direct reference from rel2 to rel1 */ if (!bms_overlap(rel1->relids, rel2->direct_lateral_relids)) return false; /* only indirect refs, so reject */ - /* check we won't have a dangerous PHV */ - if (have_dangerous_phv(root, rel1->relids, rel2->lateral_relids)) - return false; /* might be unable to handle required PHV */ } else if (lateral_rev) { @@ -599,9 +593,6 @@ join_is_legal(PlannerInfo *root, RelOptInfo *rel1, RelOptInfo *rel2, /* check there is a direct reference from rel1 to rel2 */ if (!bms_overlap(rel2->relids, rel1->direct_lateral_relids)) return false; /* only indirect refs, so reject */ - /* check we won't have a dangerous PHV */ - if (have_dangerous_phv(root, rel2->relids, rel1->lateral_relids)) - return false; /* might be unable to handle required PHV */ } /* @@ -1279,57 +1270,6 @@ has_legal_joinclause(PlannerInfo *root, RelOptInfo *rel) /* - * There's a pitfall for creating parameterized nestloops: suppose the inner - * rel (call it A) has a parameter that is a PlaceHolderVar, and that PHV's - * minimum eval_at set includes the outer rel (B) and some third rel (C). - * We might think we could create a B/A nestloop join that's parameterized by - * C. But we would end up with a plan in which the PHV's expression has to be - * evaluated as a nestloop parameter at the B/A join; and the executor is only - * set up to handle simple Vars as NestLoopParams. Rather than add complexity - * and overhead to the executor for such corner cases, it seems better to - * forbid the join. (Note that we can still make use of A's parameterized - * path with pre-joined B+C as the outer rel. have_join_order_restriction() - * ensures that we will consider making such a join even if there are not - * other reasons to do so.) - * - * So we check whether any PHVs used in the query could pose such a hazard. - * We don't have any simple way of checking whether a risky PHV would actually - * be used in the inner plan, and the case is so unusual that it doesn't seem - * worth working very hard on it. - * - * This needs to be checked in two places. If the inner rel's minimum - * parameterization would trigger the restriction, then join_is_legal() should - * reject the join altogether, because there will be no workable paths for it. - * But joinpath.c has to check again for every proposed nestloop path, because - * the inner path might have more than the minimum parameterization, causing - * some PHV to be dangerous for it that otherwise wouldn't be. - */ -bool -have_dangerous_phv(PlannerInfo *root, - Relids outer_relids, Relids inner_params) -{ - ListCell *lc; - - foreach(lc, root->placeholder_list) - { - PlaceHolderInfo *phinfo = (PlaceHolderInfo *) lfirst(lc); - - if (!bms_is_subset(phinfo->ph_eval_at, inner_params)) - continue; /* ignore, could not be a nestloop param */ - if (!bms_overlap(phinfo->ph_eval_at, outer_relids)) - continue; /* ignore, not relevant to this join */ - if (bms_is_subset(phinfo->ph_eval_at, outer_relids)) - continue; /* safe, it can be eval'd within outerrel */ - /* Otherwise, it's potentially unsafe, so reject the join */ - return true; - } - - /* OK to perform the join */ - return false; -} - - -/* * is_dummy_rel --- has relation been proven empty? */ bool diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c index 4ad30b7627e..bfefc7dbea1 100644 --- a/src/backend/optimizer/plan/createplan.c +++ b/src/backend/optimizer/plan/createplan.c @@ -284,7 +284,10 @@ static Material *make_material(Plan *lefttree); static Memoize *make_memoize(Plan *lefttree, Oid *hashoperators, Oid *collations, List *param_exprs, bool singlerow, bool binary_mode, - uint32 est_entries, Bitmapset *keyparamids); + uint32 est_entries, Bitmapset *keyparamids, + Cardinality est_calls, + Cardinality est_unique_keys, + double est_hit_ratio); static WindowAgg *make_windowagg(List *tlist, WindowClause *wc, int partNumCols, AttrNumber *partColIdx, Oid *partOperators, Oid *partCollations, int ordNumCols, AttrNumber *ordColIdx, Oid *ordOperators, Oid *ordCollations, @@ -1318,6 +1321,7 @@ create_append_plan(PlannerInfo *root, AppendPath *best_path, int flags) Oid *sortOperators; Oid *collations; bool *nullsFirst; + int presorted_keys; /* * Compute sort column info, and adjust subplan's tlist as needed. @@ -1353,14 +1357,38 @@ create_append_plan(PlannerInfo *root, AppendPath *best_path, int flags) numsortkeys * sizeof(bool)) == 0); /* Now, insert a Sort node if subplan isn't sufficiently ordered */ - if (!pathkeys_contained_in(pathkeys, subpath->pathkeys)) + if (!pathkeys_count_contained_in(pathkeys, subpath->pathkeys, + &presorted_keys)) { - Sort *sort = make_sort(subplan, numsortkeys, + Plan *sort_plan; + + /* + * We choose to use incremental sort if it is enabled and + * there are presorted keys; otherwise we use full sort. + */ + if (enable_incremental_sort && presorted_keys > 0) + { + sort_plan = (Plan *) + make_incrementalsort(subplan, numsortkeys, presorted_keys, sortColIdx, sortOperators, collations, nullsFirst); - label_sort_with_costsize(root, sort, best_path->limit_tuples); - subplan = (Plan *) sort; + label_incrementalsort_with_costsize(root, + (IncrementalSort *) sort_plan, + pathkeys, + best_path->limit_tuples); + } + else + { + sort_plan = (Plan *) make_sort(subplan, numsortkeys, + sortColIdx, sortOperators, + collations, nullsFirst); + + label_sort_with_costsize(root, (Sort *) sort_plan, + best_path->limit_tuples); + } + + subplan = sort_plan; } } @@ -1491,6 +1519,7 @@ create_merge_append_plan(PlannerInfo *root, MergeAppendPath *best_path, Oid *sortOperators; Oid *collations; bool *nullsFirst; + int presorted_keys; /* Build the child plan */ /* Must insist that all children return the same tlist */ @@ -1525,14 +1554,38 @@ create_merge_append_plan(PlannerInfo *root, MergeAppendPath *best_path, numsortkeys * sizeof(bool)) == 0); /* Now, insert a Sort node if subplan isn't sufficiently ordered */ - if (!pathkeys_contained_in(pathkeys, subpath->pathkeys)) + if (!pathkeys_count_contained_in(pathkeys, subpath->pathkeys, + &presorted_keys)) { - Sort *sort = make_sort(subplan, numsortkeys, + Plan *sort_plan; + + /* + * We choose to use incremental sort if it is enabled and there + * are presorted keys; otherwise we use full sort. + */ + if (enable_incremental_sort && presorted_keys > 0) + { + sort_plan = (Plan *) + make_incrementalsort(subplan, numsortkeys, presorted_keys, sortColIdx, sortOperators, collations, nullsFirst); - label_sort_with_costsize(root, sort, best_path->limit_tuples); - subplan = (Plan *) sort; + label_incrementalsort_with_costsize(root, + (IncrementalSort *) sort_plan, + pathkeys, + best_path->limit_tuples); + } + else + { + sort_plan = (Plan *) make_sort(subplan, numsortkeys, + sortColIdx, sortOperators, + collations, nullsFirst); + + label_sort_with_costsize(root, (Sort *) sort_plan, + best_path->limit_tuples); + } + + subplan = sort_plan; } subplans = lappend(subplans, subplan); @@ -1703,7 +1756,8 @@ create_memoize_plan(PlannerInfo *root, MemoizePath *best_path, int flags) plan = make_memoize(subplan, operators, collations, param_exprs, best_path->singlerow, best_path->binary_mode, - best_path->est_entries, keyparamids); + best_path->est_entries, keyparamids, best_path->est_calls, + best_path->est_unique_keys, best_path->est_hit_ratio); copy_generic_path_info(&plan->plan, (Path *) best_path); @@ -4344,13 +4398,16 @@ create_nestloop_plan(PlannerInfo *root, NestLoop *join_plan; Plan *outer_plan; Plan *inner_plan; + Relids outerrelids; List *tlist = build_path_tlist(root, &best_path->jpath.path); List *joinrestrictclauses = best_path->jpath.joinrestrictinfo; List *joinclauses; List *otherclauses; - Relids outerrelids; List *nestParams; + List *outer_tlist; + bool outer_parallel_safe; Relids saveOuterRels = root->curOuterRels; + ListCell *lc; /* * If the inner path is parameterized by the topmost parent of the outer @@ -4372,8 +4429,8 @@ create_nestloop_plan(PlannerInfo *root, outer_plan = create_plan_recurse(root, best_path->jpath.outerjoinpath, 0); /* For a nestloop, include outer relids in curOuterRels for inner side */ - root->curOuterRels = bms_union(root->curOuterRels, - best_path->jpath.outerjoinpath->parent->relids); + outerrelids = best_path->jpath.outerjoinpath->parent->relids; + root->curOuterRels = bms_union(root->curOuterRels, outerrelids); inner_plan = create_plan_recurse(root, best_path->jpath.innerjoinpath, 0); @@ -4412,9 +4469,66 @@ create_nestloop_plan(PlannerInfo *root, * Identify any nestloop parameters that should be supplied by this join * node, and remove them from root->curOuterParams. */ - outerrelids = best_path->jpath.outerjoinpath->parent->relids; - nestParams = identify_current_nestloop_params(root, outerrelids); + nestParams = identify_current_nestloop_params(root, + outerrelids, + PATH_REQ_OUTER((Path *) best_path)); + + /* + * While nestloop parameters that are Vars had better be available from + * the outer_plan already, there are edge cases where nestloop parameters + * that are PHVs won't be. In such cases we must add them to the + * outer_plan's tlist, since the executor's NestLoopParam machinery + * requires the params to be simple outer-Var references to that tlist. + * (This is cheating a little bit, because the outer path's required-outer + * relids might not be enough to allow evaluating such a PHV. But in + * practice, if we could have evaluated the PHV at the nestloop node, we + * can do so in the outer plan too.) + */ + outer_tlist = outer_plan->targetlist; + outer_parallel_safe = outer_plan->parallel_safe; + foreach(lc, nestParams) + { + NestLoopParam *nlp = (NestLoopParam *) lfirst(lc); + PlaceHolderVar *phv; + TargetEntry *tle; + + if (IsA(nlp->paramval, Var)) + continue; /* nothing to do for simple Vars */ + /* Otherwise it must be a PHV */ + phv = castNode(PlaceHolderVar, nlp->paramval); + + if (tlist_member((Expr *) phv, outer_tlist)) + continue; /* already available */ + + /* + * It's possible that nestloop parameter PHVs selected to evaluate + * here contain references to surviving root->curOuterParams items + * (that is, they reference values that will be supplied by some + * higher-level nestloop). Those need to be converted to Params now. + * Note: it's safe to do this after the tlist_member() check, because + * equal() won't pay attention to phv->phexpr. + */ + phv->phexpr = (Expr *) replace_nestloop_params(root, + (Node *) phv->phexpr); + + /* Make a shallow copy of outer_tlist, if we didn't already */ + if (outer_tlist == outer_plan->targetlist) + outer_tlist = list_copy(outer_tlist); + /* ... and add the needed expression */ + tle = makeTargetEntry((Expr *) copyObject(phv), + list_length(outer_tlist) + 1, + NULL, + true); + outer_tlist = lappend(outer_tlist, tle); + /* ... and track whether tlist is (still) parallel-safe */ + if (outer_parallel_safe) + outer_parallel_safe = is_parallel_safe(root, (Node *) phv); + } + if (outer_tlist != outer_plan->targetlist) + outer_plan = change_plan_targetlist(outer_plan, outer_tlist, + outer_parallel_safe); + /* And finally, we can build the join plan node */ join_plan = make_nestloop(tlist, joinclauses, otherclauses, @@ -6639,7 +6753,9 @@ materialize_finished_plan(Plan *subplan) static Memoize * make_memoize(Plan *lefttree, Oid *hashoperators, Oid *collations, List *param_exprs, bool singlerow, bool binary_mode, - uint32 est_entries, Bitmapset *keyparamids) + uint32 est_entries, Bitmapset *keyparamids, + Cardinality est_calls, Cardinality est_unique_keys, + double est_hit_ratio) { Memoize *node = makeNode(Memoize); Plan *plan = &node->plan; @@ -6657,6 +6773,9 @@ make_memoize(Plan *lefttree, Oid *hashoperators, Oid *collations, node->binary_mode = binary_mode; node->est_entries = est_entries; node->keyparamids = keyparamids; + node->est_calls = est_calls; + node->est_unique_keys = est_unique_keys; + node->est_hit_ratio = est_hit_ratio; return node; } diff --git a/src/backend/optimizer/plan/initsplan.c b/src/backend/optimizer/plan/initsplan.c index 01804b085b3..3e3fec89252 100644 --- a/src/backend/optimizer/plan/initsplan.c +++ b/src/backend/optimizer/plan/initsplan.c @@ -3048,36 +3048,16 @@ add_base_clause_to_rel(PlannerInfo *root, Index relid, * expr_is_nonnullable * Check to see if the Expr cannot be NULL * - * If the Expr is a simple Var that is defined NOT NULL and meanwhile is not - * nulled by any outer joins, then we can know that it cannot be NULL. + * Currently we only support simple Vars. */ static bool expr_is_nonnullable(PlannerInfo *root, Expr *expr) { - RelOptInfo *rel; - Var *var; - /* For now only check simple Vars */ if (!IsA(expr, Var)) return false; - var = (Var *) expr; - - /* could the Var be nulled by any outer joins? */ - if (!bms_is_empty(var->varnullingrels)) - return false; - - /* system columns cannot be NULL */ - if (var->varattno < 0) - return true; - - /* is the column defined NOT NULL? */ - rel = find_base_rel(root, var->varno); - if (var->varattno > 0 && - bms_is_member(var->varattno, rel->notnullattnums)) - return true; - - return false; + return var_is_nonnullable(root, (Var *) expr, true); } /* diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c index ff65867eebe..d59d6e4c6a0 100644 --- a/src/backend/optimizer/plan/planner.c +++ b/src/backend/optimizer/plan/planner.c @@ -342,6 +342,7 @@ standard_planner(Query *parse, const char *query_string, int cursorOptions, glob->transientPlan = false; glob->dependsOnRole = false; glob->partition_directory = NULL; + glob->rel_notnullatts_hash = NULL; /* * Assess whether it's feasible to use parallel mode for this query. We @@ -557,6 +558,7 @@ standard_planner(Query *parse, const char *query_string, int cursorOptions, result->commandType = parse->commandType; result->queryId = parse->queryId; + result->planOrigin = PLAN_STMT_STANDARD; result->hasReturning = (parse->returningList != NIL); result->hasModifyingCTE = parse->hasModifyingCTE; result->canSetTag = parse->canSetTag; @@ -721,6 +723,18 @@ subquery_planner(PlannerGlobal *glob, Query *parse, PlannerInfo *parent_root, transform_MERGE_to_join(parse); /* + * Scan the rangetable for relation RTEs and retrieve the necessary + * catalog information for each relation. Using this information, clear + * the inh flag for any relation that has no children, collect not-null + * attribute numbers for any relation that has column not-null + * constraints, and expand virtual generated columns for any relation that + * contains them. Note that this step does not descend into sublinks and + * subqueries; if we pull up any sublinks or subqueries below, their + * relation RTEs are processed just before pulling them up. + */ + parse = root->parse = preprocess_relation_rtes(root); + + /* * If the FROM clause is empty, replace it with a dummy RTE_RESULT RTE, so * that we don't need so many special cases to deal with that situation. */ @@ -744,14 +758,6 @@ subquery_planner(PlannerGlobal *glob, Query *parse, PlannerInfo *parent_root, preprocess_function_rtes(root); /* - * Scan the rangetable for relations with virtual generated columns, and - * replace all Var nodes in the query that reference these columns with - * the generation expressions. Recursion issues here are handled in the - * same way as for SubLinks. - */ - parse = root->parse = expand_virtual_generated_columns(root); - - /* * Check to see if any subqueries in the jointree can be merged into this * query. */ @@ -787,23 +793,6 @@ subquery_planner(PlannerGlobal *glob, Query *parse, PlannerInfo *parent_root, switch (rte->rtekind) { - case RTE_RELATION: - if (rte->inh) - { - /* - * Check to see if the relation actually has any children; - * if not, clear the inh flag so we can treat it as a - * plain base relation. - * - * Note: this could give a false-positive result, if the - * rel once had children but no longer does. We used to - * be able to clear rte->inh later on when we discovered - * that, but no more; we have to handle such cases as - * full-fledged inheritance. - */ - rte->inh = has_subclass(rte->relid); - } - break; case RTE_JOIN: root->hasJoinRTEs = true; if (IS_OUTER_JOIN(rte->jointype)) @@ -6879,7 +6868,7 @@ plan_cluster_use_sort(Oid tableOid, Oid indexOid) * * tableOid is the table on which the index is to be built. indexOid is the * OID of an index to be created or reindexed (which must be an index with - * support for parallel builds - currently btree or BRIN). + * support for parallel builds - currently btree, GIN, or BRIN). * * Return value is the number of parallel worker processes to request. It * may be unsafe to proceed if this is 0. Note that this does not include the diff --git a/src/backend/optimizer/plan/subselect.c b/src/backend/optimizer/plan/subselect.c index e7cb3fede66..d71ed958e31 100644 --- a/src/backend/optimizer/plan/subselect.c +++ b/src/backend/optimizer/plan/subselect.c @@ -1454,6 +1454,7 @@ convert_EXISTS_sublink_to_join(PlannerInfo *root, SubLink *sublink, Query *parse = root->parse; Query *subselect = (Query *) sublink->subselect; Node *whereClause; + PlannerInfo subroot; int rtoffset; int varno; Relids clause_varnos; @@ -1516,6 +1517,35 @@ convert_EXISTS_sublink_to_join(PlannerInfo *root, SubLink *sublink, return NULL; /* + * Scan the rangetable for relation RTEs and retrieve the necessary + * catalog information for each relation. Using this information, clear + * the inh flag for any relation that has no children, collect not-null + * attribute numbers for any relation that has column not-null + * constraints, and expand virtual generated columns for any relation that + * contains them. + * + * Note: we construct up an entirely dummy PlannerInfo for use here. This + * is fine because only the "glob" and "parse" links will be used in this + * case. + * + * Note: we temporarily assign back the WHERE clause so that any virtual + * generated column references within it can be expanded. It should be + * separated out again afterward. + */ + MemSet(&subroot, 0, sizeof(subroot)); + subroot.type = T_PlannerInfo; + subroot.glob = root->glob; + subroot.parse = subselect; + subselect->jointree->quals = whereClause; + subselect = preprocess_relation_rtes(&subroot); + + /* + * Now separate out the WHERE clause again. + */ + whereClause = subselect->jointree->quals; + subselect->jointree->quals = NULL; + + /* * The subquery must have a nonempty jointree, but we can make it so. */ replace_empty_jointree(subselect); @@ -1732,6 +1762,7 @@ convert_EXISTS_to_ANY(PlannerInfo *root, Query *subselect, Node **testexpr, List **paramIds) { Node *whereClause; + PlannerInfo subroot; List *leftargs, *rightargs, *opids, @@ -1791,12 +1822,15 @@ convert_EXISTS_to_ANY(PlannerInfo *root, Query *subselect, * parent aliases were flattened already, and we're not going to pull any * child Vars (of any description) into the parent. * - * Note: passing the parent's root to eval_const_expressions is - * technically wrong, but we can get away with it since only the - * boundParams (if any) are used, and those would be the same in a - * subroot. - */ - whereClause = eval_const_expressions(root, whereClause); + * Note: we construct up an entirely dummy PlannerInfo to pass to + * eval_const_expressions. This is fine because only the "glob" and + * "parse" links are used by eval_const_expressions. + */ + MemSet(&subroot, 0, sizeof(subroot)); + subroot.type = T_PlannerInfo; + subroot.glob = root->glob; + subroot.parse = subselect; + whereClause = eval_const_expressions(&subroot, whereClause); whereClause = (Node *) canonicalize_qual((Expr *) whereClause, false); whereClause = (Node *) make_ands_implicit((Expr *) whereClause); diff --git a/src/backend/optimizer/prep/prepjointree.c b/src/backend/optimizer/prep/prepjointree.c index 87dc6f56b57..35e8d3c183b 100644 --- a/src/backend/optimizer/prep/prepjointree.c +++ b/src/backend/optimizer/prep/prepjointree.c @@ -4,10 +4,10 @@ * Planner preprocessing for subqueries and join tree manipulation. * * NOTE: the intended sequence for invoking these operations is + * preprocess_relation_rtes * replace_empty_jointree * pull_up_sublinks * preprocess_function_rtes - * expand_virtual_generated_columns * pull_up_subqueries * flatten_simple_union_all * do expression preprocessing (including flattening JOIN alias vars) @@ -36,6 +36,7 @@ #include "optimizer/clauses.h" #include "optimizer/optimizer.h" #include "optimizer/placeholder.h" +#include "optimizer/plancat.h" #include "optimizer/prep.h" #include "optimizer/subselect.h" #include "optimizer/tlist.h" @@ -102,6 +103,9 @@ typedef struct reduce_outer_joins_partial_state Relids unreduced_side; /* relids in its still-nullable side */ } reduce_outer_joins_partial_state; +static Query *expand_virtual_generated_columns(PlannerInfo *root, Query *parse, + RangeTblEntry *rte, int rt_index, + Relation relation); static Node *pull_up_sublinks_jointree_recurse(PlannerInfo *root, Node *jtnode, Relids *relids); static Node *pull_up_sublinks_qual_recurse(PlannerInfo *root, Node *node, @@ -393,6 +397,181 @@ transform_MERGE_to_join(Query *parse) } /* + * preprocess_relation_rtes + * Do the preprocessing work for any relation RTEs in the FROM clause. + * + * This scans the rangetable for relation RTEs and retrieves the necessary + * catalog information for each relation. Using this information, it clears + * the inh flag for any relation that has no children, collects not-null + * attribute numbers for any relation that has column not-null constraints, and + * expands virtual generated columns for any relation that contains them. + * + * Note that expanding virtual generated columns may cause the query tree to + * have new copies of rangetable entries. Therefore, we have to use list_nth + * instead of foreach when iterating over the query's rangetable. + * + * Returns a modified copy of the query tree, if any relations with virtual + * generated columns are present. + */ +Query * +preprocess_relation_rtes(PlannerInfo *root) +{ + Query *parse = root->parse; + int rtable_size; + int rt_index; + + rtable_size = list_length(parse->rtable); + + for (rt_index = 0; rt_index < rtable_size; rt_index++) + { + RangeTblEntry *rte = rt_fetch(rt_index + 1, parse->rtable); + Relation relation; + + /* We only care about relation RTEs. */ + if (rte->rtekind != RTE_RELATION) + continue; + + /* + * We need not lock the relation since it was already locked by the + * rewriter. + */ + relation = table_open(rte->relid, NoLock); + + /* + * Check to see if the relation actually has any children; if not, + * clear the inh flag so we can treat it as a plain base relation. + * + * Note: this could give a false-positive result, if the rel once had + * children but no longer does. We used to be able to clear rte->inh + * later on when we discovered that, but no more; we have to handle + * such cases as full-fledged inheritance. + */ + if (rte->inh) + rte->inh = relation->rd_rel->relhassubclass; + + /* + * Check to see if the relation has any column not-null constraints; + * if so, retrieve the constraint information and store it in a + * relation OID based hash table. + */ + get_relation_notnullatts(root, relation); + + /* + * Check to see if the relation has any virtual generated columns; if + * so, replace all Var nodes in the query that reference these columns + * with the generation expressions. + */ + parse = expand_virtual_generated_columns(root, parse, + rte, rt_index + 1, + relation); + + table_close(relation, NoLock); + } + + return parse; +} + +/* + * expand_virtual_generated_columns + * Expand virtual generated columns for the given relation. + * + * This checks whether the given relation has any virtual generated columns, + * and if so, replaces all Var nodes in the query that reference those columns + * with their generation expressions. + * + * Returns a modified copy of the query tree if the relation contains virtual + * generated columns. + */ +static Query * +expand_virtual_generated_columns(PlannerInfo *root, Query *parse, + RangeTblEntry *rte, int rt_index, + Relation relation) +{ + TupleDesc tupdesc; + + /* Only normal relations can have virtual generated columns */ + Assert(rte->rtekind == RTE_RELATION); + + tupdesc = RelationGetDescr(relation); + if (tupdesc->constr && tupdesc->constr->has_generated_virtual) + { + List *tlist = NIL; + pullup_replace_vars_context rvcontext; + + for (int i = 0; i < tupdesc->natts; i++) + { + Form_pg_attribute attr = TupleDescAttr(tupdesc, i); + TargetEntry *tle; + + if (attr->attgenerated == ATTRIBUTE_GENERATED_VIRTUAL) + { + Node *defexpr; + + defexpr = build_generation_expression(relation, i + 1); + ChangeVarNodes(defexpr, 1, rt_index, 0); + + tle = makeTargetEntry((Expr *) defexpr, i + 1, 0, false); + tlist = lappend(tlist, tle); + } + else + { + Var *var; + + var = makeVar(rt_index, + i + 1, + attr->atttypid, + attr->atttypmod, + attr->attcollation, + 0); + + tle = makeTargetEntry((Expr *) var, i + 1, 0, false); + tlist = lappend(tlist, tle); + } + } + + Assert(list_length(tlist) > 0); + Assert(!rte->lateral); + + /* + * The relation's targetlist items are now in the appropriate form to + * insert into the query, except that we may need to wrap them in + * PlaceHolderVars. Set up required context data for + * pullup_replace_vars. + */ + rvcontext.root = root; + rvcontext.targetlist = tlist; + rvcontext.target_rte = rte; + rvcontext.result_relation = parse->resultRelation; + /* won't need these values */ + rvcontext.relids = NULL; + rvcontext.nullinfo = NULL; + /* pass NULL for outer_hasSubLinks */ + rvcontext.outer_hasSubLinks = NULL; + rvcontext.varno = rt_index; + /* this flag will be set below, if needed */ + rvcontext.wrap_option = REPLACE_WRAP_NONE; + /* initialize cache array with indexes 0 .. length(tlist) */ + rvcontext.rv_cache = palloc0((list_length(tlist) + 1) * + sizeof(Node *)); + + /* + * If the query uses grouping sets, we need a PlaceHolderVar for each + * expression of the relation's targetlist items. (See comments in + * pull_up_simple_subquery().) + */ + if (parse->groupingSets) + rvcontext.wrap_option = REPLACE_WRAP_ALL; + + /* + * Apply pullup variable replacement throughout the query tree. + */ + parse = (Query *) pullup_replace_vars((Node *) parse, &rvcontext); + } + + return parse; +} + +/* * replace_empty_jointree * If the Query's jointree is empty, replace it with a dummy RTE_RESULT * relation. @@ -950,128 +1129,6 @@ preprocess_function_rtes(PlannerInfo *root) } /* - * expand_virtual_generated_columns - * Expand all virtual generated column references in a query. - * - * This scans the rangetable for relations with virtual generated columns, and - * replaces all Var nodes in the query that reference these columns with the - * generation expressions. Note that we do not descend into subqueries; that - * is taken care of when the subqueries are planned. - * - * This has to be done after we have pulled up any SubLinks within the query's - * quals; otherwise any virtual generated column references within the SubLinks - * that should be transformed into joins wouldn't get expanded. - * - * Returns a modified copy of the query tree, if any relations with virtual - * generated columns are present. - */ -Query * -expand_virtual_generated_columns(PlannerInfo *root) -{ - Query *parse = root->parse; - int rt_index; - ListCell *lc; - - rt_index = 0; - foreach(lc, parse->rtable) - { - RangeTblEntry *rte = (RangeTblEntry *) lfirst(lc); - Relation rel; - TupleDesc tupdesc; - - ++rt_index; - - /* - * Only normal relations can have virtual generated columns. - */ - if (rte->rtekind != RTE_RELATION) - continue; - - rel = table_open(rte->relid, NoLock); - - tupdesc = RelationGetDescr(rel); - if (tupdesc->constr && tupdesc->constr->has_generated_virtual) - { - List *tlist = NIL; - pullup_replace_vars_context rvcontext; - - for (int i = 0; i < tupdesc->natts; i++) - { - Form_pg_attribute attr = TupleDescAttr(tupdesc, i); - TargetEntry *tle; - - if (attr->attgenerated == ATTRIBUTE_GENERATED_VIRTUAL) - { - Node *defexpr; - - defexpr = build_generation_expression(rel, i + 1); - ChangeVarNodes(defexpr, 1, rt_index, 0); - - tle = makeTargetEntry((Expr *) defexpr, i + 1, 0, false); - tlist = lappend(tlist, tle); - } - else - { - Var *var; - - var = makeVar(rt_index, - i + 1, - attr->atttypid, - attr->atttypmod, - attr->attcollation, - 0); - - tle = makeTargetEntry((Expr *) var, i + 1, 0, false); - tlist = lappend(tlist, tle); - } - } - - Assert(list_length(tlist) > 0); - Assert(!rte->lateral); - - /* - * The relation's targetlist items are now in the appropriate form - * to insert into the query, except that we may need to wrap them - * in PlaceHolderVars. Set up required context data for - * pullup_replace_vars. - */ - rvcontext.root = root; - rvcontext.targetlist = tlist; - rvcontext.target_rte = rte; - rvcontext.result_relation = parse->resultRelation; - /* won't need these values */ - rvcontext.relids = NULL; - rvcontext.nullinfo = NULL; - /* pass NULL for outer_hasSubLinks */ - rvcontext.outer_hasSubLinks = NULL; - rvcontext.varno = rt_index; - /* this flag will be set below, if needed */ - rvcontext.wrap_option = REPLACE_WRAP_NONE; - /* initialize cache array with indexes 0 .. length(tlist) */ - rvcontext.rv_cache = palloc0((list_length(tlist) + 1) * - sizeof(Node *)); - - /* - * If the query uses grouping sets, we need a PlaceHolderVar for - * each expression of the relation's targetlist items. (See - * comments in pull_up_simple_subquery().) - */ - if (parse->groupingSets) - rvcontext.wrap_option = REPLACE_WRAP_ALL; - - /* - * Apply pullup variable replacement throughout the query tree. - */ - parse = (Query *) pullup_replace_vars((Node *) parse, &rvcontext); - } - - table_close(rel, NoLock); - } - - return parse; -} - -/* * pull_up_subqueries * Look for subqueries in the rangetable that can be pulled up into * the parent query. If the subquery has no special features like @@ -1334,6 +1391,16 @@ pull_up_simple_subquery(PlannerInfo *root, Node *jtnode, RangeTblEntry *rte, Assert(subquery->cteList == NIL); /* + * Scan the rangetable for relation RTEs and retrieve the necessary + * catalog information for each relation. Using this information, clear + * the inh flag for any relation that has no children, collect not-null + * attribute numbers for any relation that has column not-null + * constraints, and expand virtual generated columns for any relation that + * contains them. + */ + subquery = subroot->parse = preprocess_relation_rtes(subroot); + + /* * If the FROM clause is empty, replace it with a dummy RTE_RESULT RTE, so * that we don't need so many special cases to deal with that situation. */ @@ -1353,13 +1420,6 @@ pull_up_simple_subquery(PlannerInfo *root, Node *jtnode, RangeTblEntry *rte, preprocess_function_rtes(subroot); /* - * Scan the rangetable for relations with virtual generated columns, and - * replace all Var nodes in the query that reference these columns with - * the generation expressions. - */ - subquery = subroot->parse = expand_virtual_generated_columns(subroot); - - /* * Recursively pull up the subquery's subqueries, so that * pull_up_subqueries' processing is complete for its jointree and * rangetable. diff --git a/src/backend/optimizer/util/clauses.c b/src/backend/optimizer/util/clauses.c index 26a3e050086..6f0b338d2cd 100644 --- a/src/backend/optimizer/util/clauses.c +++ b/src/backend/optimizer/util/clauses.c @@ -20,6 +20,7 @@ #include "postgres.h" #include "access/htup_details.h" +#include "catalog/pg_class.h" #include "catalog/pg_language.h" #include "catalog/pg_operator.h" #include "catalog/pg_proc.h" @@ -36,6 +37,7 @@ #include "optimizer/clauses.h" #include "optimizer/cost.h" #include "optimizer/optimizer.h" +#include "optimizer/pathnode.h" #include "optimizer/plancat.h" #include "optimizer/planmain.h" #include "parser/analyze.h" @@ -43,6 +45,7 @@ #include "parser/parse_collate.h" #include "parser/parse_func.h" #include "parser/parse_oper.h" +#include "parser/parsetree.h" #include "rewrite/rewriteHandler.h" #include "rewrite/rewriteManip.h" #include "tcop/tcopprot.h" @@ -2242,7 +2245,8 @@ rowtype_field_matches(Oid rowtypeid, int fieldnum, * only operators and functions that are reasonable to try to execute. * * NOTE: "root" can be passed as NULL if the caller never wants to do any - * Param substitutions nor receive info about inlined functions. + * Param substitutions nor receive info about inlined functions nor reduce + * NullTest for Vars to constant true or constant false. * * NOTE: the planner assumes that this will always flatten nested AND and * OR clauses into N-argument form. See comments in prepqual.c. @@ -3333,6 +3337,13 @@ eval_const_expressions_mutator(Node *node, -1, coalesceexpr->coalescecollid); + /* + * If there's exactly one surviving argument, we no longer + * need COALESCE at all: the result is that argument + */ + if (list_length(newargs) == 1) + return (Node *) linitial(newargs); + newcoalesce = makeNode(CoalesceExpr); newcoalesce->coalescetype = coalesceexpr->coalescetype; newcoalesce->coalescecollid = coalesceexpr->coalescecollid; @@ -3537,6 +3548,31 @@ eval_const_expressions_mutator(Node *node, return makeBoolConst(result, false); } + if (!ntest->argisrow && arg && IsA(arg, Var) && context->root) + { + Var *varg = (Var *) arg; + bool result; + + if (var_is_nonnullable(context->root, varg, false)) + { + switch (ntest->nulltesttype) + { + case IS_NULL: + result = false; + break; + case IS_NOT_NULL: + result = true; + break; + default: + elog(ERROR, "unrecognized nulltesttype: %d", + (int) ntest->nulltesttype); + result = false; /* keep compiler quiet */ + break; + } + + return makeBoolConst(result, false); + } + } newntest = makeNode(NullTest); newntest->arg = (Expr *) arg; @@ -4156,6 +4192,67 @@ simplify_function(Oid funcid, Oid result_type, int32 result_typmod, } /* + * var_is_nonnullable: check to see if the Var cannot be NULL + * + * If the Var is defined NOT NULL and meanwhile is not nulled by any outer + * joins or grouping sets, then we can know that it cannot be NULL. + * + * use_rel_info indicates whether the corresponding RelOptInfo is available for + * use. + */ +bool +var_is_nonnullable(PlannerInfo *root, Var *var, bool use_rel_info) +{ + Relids notnullattnums = NULL; + + Assert(IsA(var, Var)); + + /* skip upper-level Vars */ + if (var->varlevelsup != 0) + return false; + + /* could the Var be nulled by any outer joins or grouping sets? */ + if (!bms_is_empty(var->varnullingrels)) + return false; + + /* system columns cannot be NULL */ + if (var->varattno < 0) + return true; + + /* + * Check if the Var is defined as NOT NULL. We retrieve the column NOT + * NULL constraint information from the corresponding RelOptInfo if it is + * available; otherwise, we search the hash table for this information. + */ + if (use_rel_info) + { + RelOptInfo *rel = find_base_rel(root, var->varno); + + notnullattnums = rel->notnullattnums; + } + else + { + RangeTblEntry *rte = planner_rt_fetch(var->varno, root); + + /* + * We must skip inheritance parent tables, as some child tables may + * have a NOT NULL constraint for a column while others may not. This + * cannot happen with partitioned tables, though. + */ + if (rte->inh && rte->relkind != RELKIND_PARTITIONED_TABLE) + return false; + + notnullattnums = find_relation_notnullatts(root, rte->relid); + } + + if (var->varattno > 0 && + bms_is_member(var->varattno, notnullattnums)) + return true; + + return false; +} + +/* * expand_function_arguments: convert named-notation args to positional args * and/or insert default args, as needed * diff --git a/src/backend/optimizer/util/inherit.c b/src/backend/optimizer/util/inherit.c index 17e51cd75d7..30d158069e3 100644 --- a/src/backend/optimizer/util/inherit.c +++ b/src/backend/optimizer/util/inherit.c @@ -466,8 +466,7 @@ expand_single_inheritance_child(PlannerInfo *root, RangeTblEntry *parentrte, Index *childRTindex_p) { Query *parse = root->parse; - Oid parentOID PG_USED_FOR_ASSERTS_ONLY = - RelationGetRelid(parentrel); + Oid parentOID = RelationGetRelid(parentrel); Oid childOID = RelationGetRelid(childrel); RangeTblEntry *childrte; Index childRTindex; @@ -514,6 +513,13 @@ expand_single_inheritance_child(PlannerInfo *root, RangeTblEntry *parentrte, *childRTindex_p = childRTindex; /* + * Retrieve column not-null constraint information for the child relation + * if its relation OID is different from the parent's. + */ + if (childOID != parentOID) + get_relation_notnullatts(root, childrel); + + /* * Build an AppendRelInfo struct for each parent/child pair. */ appinfo = make_append_rel_info(parentrel, childrel, diff --git a/src/backend/optimizer/util/paramassign.c b/src/backend/optimizer/util/paramassign.c index 3bd3ce37c8f..4c13c5931b4 100644 --- a/src/backend/optimizer/util/paramassign.c +++ b/src/backend/optimizer/util/paramassign.c @@ -599,38 +599,46 @@ process_subquery_nestloop_params(PlannerInfo *root, List *subplan_params) } /* - * Identify any NestLoopParams that should be supplied by a NestLoop plan - * node with the specified lefthand rels. Remove them from the active - * root->curOuterParams list and return them as the result list. + * Identify any NestLoopParams that should be supplied by a NestLoop + * plan node with the specified lefthand rels and required-outer rels. + * Remove them from the active root->curOuterParams list and return + * them as the result list. * - * XXX Here we also hack up the returned Vars and PHVs so that they do not - * contain nullingrel sets exceeding what is available from the outer side. - * This is needed if we have applied outer join identity 3, - * (A leftjoin B on (Pab)) leftjoin C on (Pb*c) - * = A leftjoin (B leftjoin C on (Pbc)) on (Pab) - * and C contains lateral references to B. It's still safe to apply the - * identity, but the parser will have created those references in the form - * "b*" (i.e., with varnullingrels listing the A/B join), while what we will - * have available from the nestloop's outer side is just "b". We deal with - * that here by stripping the nullingrels down to what is available from the - * outer side according to leftrelids. - * - * That fixes matters for the case of forward application of identity 3. - * If the identity was applied in the reverse direction, we will have - * parameter Vars containing too few nullingrel bits rather than too many. - * Currently, that causes no problems because setrefs.c applies only a - * subset check to nullingrels in NestLoopParams, but we'd have to work - * harder if we ever want to tighten that check. This is all pretty annoying - * because it greatly weakens setrefs.c's cross-check, but the alternative + * Vars and PHVs appearing in the result list must have nullingrel sets + * that could validly appear in the lefthand rel's output. Ordinarily that + * would be true already, but if we have applied outer join identity 3, + * there could be more or fewer nullingrel bits in the nodes appearing in + * curOuterParams than are in the nominal leftrelids. We deal with that by + * forcing their nullingrel sets to include exactly the outer-join relids + * that appear in leftrelids and can null the respective Var or PHV. + * This fix is a bit ad-hoc and intellectually unsatisfactory, because it's + * essentially jumping to the conclusion that we've placed evaluation of + * the nestloop parameters correctly, and thus it defeats the intent of the + * subsequent nullingrel cross-checks in setrefs.c. But the alternative * seems to be to generate multiple versions of each laterally-parameterized * subquery, which'd be unduly expensive. */ List * -identify_current_nestloop_params(PlannerInfo *root, Relids leftrelids) +identify_current_nestloop_params(PlannerInfo *root, + Relids leftrelids, + Relids outerrelids) { List *result; + Relids allleftrelids; ListCell *cell; + /* + * We'll be able to evaluate a PHV in the lefthand path if it uses the + * lefthand rels plus any available required-outer rels. But don't do so + * if it uses *only* required-outer rels; in that case it should be + * evaluated higher in the tree. For Vars, no such hair-splitting is + * necessary since they depend on only one relid. + */ + if (outerrelids) + allleftrelids = bms_union(leftrelids, outerrelids); + else + allleftrelids = leftrelids; + result = NIL; foreach(cell, root->curOuterParams) { @@ -646,25 +654,60 @@ identify_current_nestloop_params(PlannerInfo *root, Relids leftrelids) bms_is_member(nlp->paramval->varno, leftrelids)) { Var *var = (Var *) nlp->paramval; + RelOptInfo *rel = root->simple_rel_array[var->varno]; root->curOuterParams = foreach_delete_current(root->curOuterParams, cell); - var->varnullingrels = bms_intersect(var->varnullingrels, + var->varnullingrels = bms_intersect(rel->nulling_relids, leftrelids); result = lappend(result, nlp); } - else if (IsA(nlp->paramval, PlaceHolderVar) && - bms_is_subset(find_placeholder_info(root, - (PlaceHolderVar *) nlp->paramval)->ph_eval_at, - leftrelids)) + else if (IsA(nlp->paramval, PlaceHolderVar)) { PlaceHolderVar *phv = (PlaceHolderVar *) nlp->paramval; + PlaceHolderInfo *phinfo = find_placeholder_info(root, phv); + Relids eval_at = phinfo->ph_eval_at; - root->curOuterParams = foreach_delete_current(root->curOuterParams, - cell); - phv->phnullingrels = bms_intersect(phv->phnullingrels, - leftrelids); - result = lappend(result, nlp); + if (bms_is_subset(eval_at, allleftrelids) && + bms_overlap(eval_at, leftrelids)) + { + root->curOuterParams = foreach_delete_current(root->curOuterParams, + cell); + + /* + * Deal with an edge case: if the PHV was pulled up out of a + * subquery and it contains a subquery that was originally + * pushed down from this query level, then that will still be + * represented as a SubLink, because SS_process_sublinks won't + * recurse into outer PHVs, so it didn't get transformed + * during expression preprocessing in the subquery. We need a + * version of the PHV that has a SubPlan, which we can get + * from the current query level's placeholder_list. This is + * quite grotty of course, but dealing with it earlier in the + * handling of subplan params would be just as grotty, and it + * might end up being a waste of cycles if we don't decide to + * treat the PHV as a NestLoopParam. (Perhaps that whole + * mechanism should be redesigned someday, but today is not + * that day.) + */ + if (root->parse->hasSubLinks) + { + phv = copyObject(phinfo->ph_var); + + /* + * The ph_var will have empty nullingrels, but that + * doesn't matter since we're about to overwrite + * phv->phnullingrels. Other fields should be OK already. + */ + nlp->paramval = (Var *) phv; + } + + phv->phnullingrels = + bms_intersect(get_placeholder_nulling_relids(root, phinfo), + leftrelids); + + result = lappend(result, nlp); + } } } return result; diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c index e0192d4a491..a4c5867cdcb 100644 --- a/src/backend/optimizer/util/pathnode.c +++ b/src/backend/optimizer/util/pathnode.c @@ -1404,12 +1404,12 @@ create_append_path(PlannerInfo *root, pathnode->path.total_cost = child->total_cost; } else - cost_append(pathnode); + cost_append(pathnode, root); /* Must do this last, else cost_append complains */ pathnode->path.pathkeys = child->pathkeys; } else - cost_append(pathnode); + cost_append(pathnode, root); /* If the caller provided a row estimate, override the computed value. */ if (rows >= 0) @@ -1515,6 +1515,9 @@ create_merge_append_path(PlannerInfo *root, foreach(l, subpaths) { Path *subpath = (Path *) lfirst(l); + int presorted_keys; + Path sort_path; /* dummy for result of + * cost_sort/cost_incremental_sort */ /* All child paths should be unparameterized */ Assert(bms_is_empty(PATH_REQ_OUTER(subpath))); @@ -1523,32 +1526,52 @@ create_merge_append_path(PlannerInfo *root, pathnode->path.parallel_safe = pathnode->path.parallel_safe && subpath->parallel_safe; - if (pathkeys_contained_in(pathkeys, subpath->pathkeys)) + if (!pathkeys_count_contained_in(pathkeys, subpath->pathkeys, + &presorted_keys)) { - /* Subpath is adequately ordered, we won't need to sort it */ - input_disabled_nodes += subpath->disabled_nodes; - input_startup_cost += subpath->startup_cost; - input_total_cost += subpath->total_cost; - } - else - { - /* We'll need to insert a Sort node, so include cost for that */ - Path sort_path; /* dummy for result of cost_sort */ + /* + * We'll need to insert a Sort node, so include costs for that. We + * choose to use incremental sort if it is enabled and there are + * presorted keys; otherwise we use full sort. + * + * We can use the parent's LIMIT if any, since we certainly won't + * pull more than that many tuples from any child. + */ + if (enable_incremental_sort && presorted_keys > 0) + { + cost_incremental_sort(&sort_path, + root, + pathkeys, + presorted_keys, + subpath->disabled_nodes, + subpath->startup_cost, + subpath->total_cost, + subpath->rows, + subpath->pathtarget->width, + 0.0, + work_mem, + pathnode->limit_tuples); + } + else + { + cost_sort(&sort_path, + root, + pathkeys, + subpath->disabled_nodes, + subpath->total_cost, + subpath->rows, + subpath->pathtarget->width, + 0.0, + work_mem, + pathnode->limit_tuples); + } - cost_sort(&sort_path, - root, - pathkeys, - subpath->disabled_nodes, - subpath->total_cost, - subpath->rows, - subpath->pathtarget->width, - 0.0, - work_mem, - pathnode->limit_tuples); - input_disabled_nodes += sort_path.disabled_nodes; - input_startup_cost += sort_path.startup_cost; - input_total_cost += sort_path.total_cost; + subpath = &sort_path; } + + input_disabled_nodes += subpath->disabled_nodes; + input_startup_cost += subpath->startup_cost; + input_total_cost += subpath->total_cost; } /* @@ -1666,7 +1689,7 @@ create_material_path(RelOptInfo *rel, Path *subpath) MemoizePath * create_memoize_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath, List *param_exprs, List *hash_operators, - bool singlerow, bool binary_mode, double calls) + bool singlerow, bool binary_mode, Cardinality est_calls) { MemoizePath *pathnode = makeNode(MemoizePath); @@ -1687,7 +1710,6 @@ create_memoize_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath, pathnode->param_exprs = param_exprs; pathnode->singlerow = singlerow; pathnode->binary_mode = binary_mode; - pathnode->calls = clamp_row_est(calls); /* * For now we set est_entries to 0. cost_memoize_rescan() does all the @@ -1697,6 +1719,12 @@ create_memoize_path(PlannerInfo *root, RelOptInfo *rel, Path *subpath, */ pathnode->est_entries = 0; + pathnode->est_calls = clamp_row_est(est_calls); + + /* These will also be set later in cost_memoize_rescan() */ + pathnode->est_unique_keys = 0.0; + pathnode->est_hit_ratio = 0.0; + /* we should not generate this path type when enable_memoize=false */ Assert(enable_memoize); pathnode->path.disabled_nodes = subpath->disabled_nodes; @@ -4236,7 +4264,7 @@ reparameterize_path(PlannerInfo *root, Path *path, mpath->hash_operators, mpath->singlerow, mpath->binary_mode, - mpath->calls); + mpath->est_calls); } default: break; diff --git a/src/backend/optimizer/util/placeholder.c b/src/backend/optimizer/util/placeholder.c index 41a4c81e94a..e1cd00a72fb 100644 --- a/src/backend/optimizer/util/placeholder.c +++ b/src/backend/optimizer/util/placeholder.c @@ -545,3 +545,43 @@ contain_placeholder_references_walker(Node *node, return expression_tree_walker(node, contain_placeholder_references_walker, context); } + +/* + * Compute the set of outer-join relids that can null a placeholder. + * + * This is analogous to RelOptInfo.nulling_relids for Vars, but we compute it + * on-the-fly rather than saving it somewhere. Currently the value is needed + * at most once per query, so there's little value in doing otherwise. If it + * ever gains more widespread use, perhaps we should cache the result in + * PlaceHolderInfo. + */ +Relids +get_placeholder_nulling_relids(PlannerInfo *root, PlaceHolderInfo *phinfo) +{ + Relids result = NULL; + int relid = -1; + + /* + * Form the union of all potential nulling OJs for each baserel included + * in ph_eval_at. + */ + while ((relid = bms_next_member(phinfo->ph_eval_at, relid)) > 0) + { + RelOptInfo *rel = root->simple_rel_array[relid]; + + /* ignore the RTE_GROUP RTE */ + if (relid == root->group_rtindex) + continue; + + if (rel == NULL) /* must be an outer join */ + { + Assert(bms_is_member(relid, root->outer_join_rels)); + continue; + } + result = bms_add_members(result, rel->nulling_relids); + } + + /* Now remove any OJs already included in ph_eval_at, and we're done. */ + result = bms_del_members(result, phinfo->ph_eval_at); + return result; +} diff --git a/src/backend/optimizer/util/plancat.c b/src/backend/optimizer/util/plancat.c index 59233b64730..c6a58afc5e5 100644 --- a/src/backend/optimizer/util/plancat.c +++ b/src/backend/optimizer/util/plancat.c @@ -59,6 +59,12 @@ int constraint_exclusion = CONSTRAINT_EXCLUSION_PARTITION; /* Hook for plugins to get control in get_relation_info() */ get_relation_info_hook_type get_relation_info_hook = NULL; +typedef struct NotnullHashEntry +{ + Oid relid; /* OID of the relation */ + Relids notnullattnums; /* attnums of NOT NULL columns */ +} NotnullHashEntry; + static void get_relation_foreign_keys(PlannerInfo *root, RelOptInfo *rel, Relation relation, bool inhparent); @@ -172,27 +178,7 @@ get_relation_info(PlannerInfo *root, Oid relationObjectId, bool inhparent, * RangeTblEntry does get populated. */ if (!inhparent || relation->rd_rel->relkind == RELKIND_PARTITIONED_TABLE) - { - for (int i = 0; i < relation->rd_att->natts; i++) - { - CompactAttribute *attr = TupleDescCompactAttr(relation->rd_att, i); - - Assert(attr->attnullability != ATTNULLABLE_UNKNOWN); - - if (attr->attnullability == ATTNULLABLE_VALID) - { - rel->notnullattnums = bms_add_member(rel->notnullattnums, - i + 1); - - /* - * Per RemoveAttributeById(), dropped columns will have their - * attnotnull unset, so we needn't check for dropped columns - * in the above condition. - */ - Assert(!attr->attisdropped); - } - } - } + rel->notnullattnums = find_relation_notnullatts(root, relationObjectId); /* * Estimate relation size --- unless it's an inheritance parent, in which @@ -684,6 +670,105 @@ get_relation_foreign_keys(PlannerInfo *root, RelOptInfo *rel, } /* + * get_relation_notnullatts - + * Retrieves column not-null constraint information for a given relation. + * + * We do this while we have the relcache entry open, and store the column + * not-null constraint information in a hash table based on the relation OID. + */ +void +get_relation_notnullatts(PlannerInfo *root, Relation relation) +{ + Oid relid = RelationGetRelid(relation); + NotnullHashEntry *hentry; + bool found; + Relids notnullattnums = NULL; + + /* bail out if the relation has no not-null constraints */ + if (relation->rd_att->constr == NULL || + !relation->rd_att->constr->has_not_null) + return; + + /* create the hash table if it hasn't been created yet */ + if (root->glob->rel_notnullatts_hash == NULL) + { + HTAB *hashtab; + HASHCTL hash_ctl; + + hash_ctl.keysize = sizeof(Oid); + hash_ctl.entrysize = sizeof(NotnullHashEntry); + hash_ctl.hcxt = CurrentMemoryContext; + + hashtab = hash_create("Relation NOT NULL attnums", + 64L, /* arbitrary initial size */ + &hash_ctl, + HASH_ELEM | HASH_BLOBS | HASH_CONTEXT); + + root->glob->rel_notnullatts_hash = hashtab; + } + + /* + * Create a hash entry for this relation OID, if we don't have one + * already. + */ + hentry = (NotnullHashEntry *) hash_search(root->glob->rel_notnullatts_hash, + &relid, + HASH_ENTER, + &found); + + /* bail out if a hash entry already exists for this relation OID */ + if (found) + return; + + /* collect the column not-null constraint information for this relation */ + for (int i = 0; i < relation->rd_att->natts; i++) + { + CompactAttribute *attr = TupleDescCompactAttr(relation->rd_att, i); + + Assert(attr->attnullability != ATTNULLABLE_UNKNOWN); + + if (attr->attnullability == ATTNULLABLE_VALID) + { + notnullattnums = bms_add_member(notnullattnums, i + 1); + + /* + * Per RemoveAttributeById(), dropped columns will have their + * attnotnull unset, so we needn't check for dropped columns in + * the above condition. + */ + Assert(!attr->attisdropped); + } + } + + /* ... and initialize the new hash entry */ + hentry->notnullattnums = notnullattnums; +} + +/* + * find_relation_notnullatts - + * Searches the hash table and returns the column not-null constraint + * information for a given relation. + */ +Relids +find_relation_notnullatts(PlannerInfo *root, Oid relid) +{ + NotnullHashEntry *hentry; + bool found; + + if (root->glob->rel_notnullatts_hash == NULL) + return NULL; + + hentry = (NotnullHashEntry *) hash_search(root->glob->rel_notnullatts_hash, + &relid, + HASH_FIND, + &found); + if (!found) + return NULL; + + return hentry->notnullattnums; +} + +/* * infer_arbiter_indexes - * Determine the unique indexes used to arbitrate speculative insertion. * diff --git a/src/backend/parser/analyze.c b/src/backend/parser/analyze.c index a16fdd65601..34f7c17f576 100644 --- a/src/backend/parser/analyze.c +++ b/src/backend/parser/analyze.c @@ -239,102 +239,23 @@ parse_sub_analyze(Node *parseTree, ParseState *parentParseState, } /* - * setQueryLocationAndLength - * Set query's location and length from statement and ParseState - * - * Some statements, like PreparableStmt, can be located within parentheses. - * For example "(SELECT 1)" or "COPY (UPDATE ...) to x;". For those, we - * cannot use the whole string from the statement's location or the SQL - * string would yield incorrectly. The parser will set stmt_len, reflecting - * the size of the statement within the parentheses. Thus, when stmt_len is - * available, we need to use it for the Query's stmt_len. - * - * For other cases, the parser can't provide the length of individual - * statements. However, we have the statement's location plus the length - * (p_stmt_len) and location (p_stmt_location) of the top level RawStmt, - * stored in pstate. Thus, the statement's length is the RawStmt's length - * minus how much we've advanced in the RawStmt's string. If p_stmt_len - * is 0, the SQL string is used up to its end. - */ -static void -setQueryLocationAndLength(ParseState *pstate, Query *qry, Node *parseTree) -{ - ParseLoc stmt_len = 0; - - switch (nodeTag(parseTree)) - { - case T_InsertStmt: - qry->stmt_location = ((InsertStmt *) parseTree)->stmt_location; - stmt_len = ((InsertStmt *) parseTree)->stmt_len; - break; - - case T_DeleteStmt: - qry->stmt_location = ((DeleteStmt *) parseTree)->stmt_location; - stmt_len = ((DeleteStmt *) parseTree)->stmt_len; - break; - - case T_UpdateStmt: - qry->stmt_location = ((UpdateStmt *) parseTree)->stmt_location; - stmt_len = ((UpdateStmt *) parseTree)->stmt_len; - break; - - case T_MergeStmt: - qry->stmt_location = ((MergeStmt *) parseTree)->stmt_location; - stmt_len = ((MergeStmt *) parseTree)->stmt_len; - break; - - case T_SelectStmt: - qry->stmt_location = ((SelectStmt *) parseTree)->stmt_location; - stmt_len = ((SelectStmt *) parseTree)->stmt_len; - break; - - case T_PLAssignStmt: - qry->stmt_location = ((PLAssignStmt *) parseTree)->location; - break; - - default: - qry->stmt_location = pstate->p_stmt_location; - break; - } - - if (stmt_len > 0) - { - /* Statement's length is known, use it */ - qry->stmt_len = stmt_len; - } - else if (pstate->p_stmt_len > 0) - { - /* - * The top RawStmt's length is known, so calculate the statement's - * length from the statement's location and the RawStmt's length and - * location. - */ - qry->stmt_len = pstate->p_stmt_len - (qry->stmt_location - pstate->p_stmt_location); - } - - /* The calculated statement length should be calculated as positive. */ - Assert(qry->stmt_len >= 0); -} - -/* * transformTopLevelStmt - * transform a Parse tree into a Query tree. * - * This function is just responsible for storing location data - * from the RawStmt into the ParseState. + * This function is just responsible for transferring statement location data + * from the RawStmt into the finished Query. */ Query * transformTopLevelStmt(ParseState *pstate, RawStmt *parseTree) { Query *result; - /* Store RawStmt's length and location in pstate */ - pstate->p_stmt_len = parseTree->stmt_len; - pstate->p_stmt_location = parseTree->stmt_location; - /* We're at top level, so allow SELECT INTO */ result = transformOptionalSelectInto(pstate, parseTree->stmt); + result->stmt_location = parseTree->stmt_location; + result->stmt_len = parseTree->stmt_len; + return result; } @@ -503,7 +424,6 @@ transformStmt(ParseState *pstate, Node *parseTree) /* Mark as original query until we learn differently */ result->querySource = QSRC_ORIGINAL; result->canSetTag = true; - setQueryLocationAndLength(pstate, result, parseTree); return result; } diff --git a/src/backend/parser/gram.y b/src/backend/parser/gram.y index 0b5652071d1..db43034b9db 100644 --- a/src/backend/parser/gram.y +++ b/src/backend/parser/gram.y @@ -154,7 +154,6 @@ static void base_yyerror(YYLTYPE *yylloc, core_yyscan_t yyscanner, const char *msg); static RawStmt *makeRawStmt(Node *stmt, int stmt_location); static void updateRawStmtEnd(RawStmt *rs, int end_location); -static void updatePreparableStmtEnd(Node *n, int end_location); static Node *makeColumnRef(char *colname, List *indirection, int location, core_yyscan_t yyscanner); static Node *makeTypeCast(Node *arg, TypeName *typename, int location); @@ -178,13 +177,13 @@ static void insertSelectOptions(SelectStmt *stmt, SelectLimit *limitClause, WithClause *withClause, core_yyscan_t yyscanner); -static Node *makeSetOp(SetOperation op, bool all, Node *larg, Node *rarg, int location); +static Node *makeSetOp(SetOperation op, bool all, Node *larg, Node *rarg); static Node *doNegate(Node *n, int location); static void doNegateFloat(Float *v); static Node *makeAndExpr(Node *lexpr, Node *rexpr, int location); static Node *makeOrExpr(Node *lexpr, Node *rexpr, int location); static Node *makeNotExpr(Node *expr, int location); -static Node *makeAArrayExpr(List *elements, int location); +static Node *makeAArrayExpr(List *elements, int location, int end_location); static Node *makeSQLValueFunction(SQLValueFunctionOp op, int32 typmod, int location); static Node *makeXmlExpr(XmlExprOp op, char *name, List *named_args, @@ -319,6 +318,11 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query); %type <list> opt_qualified_name %type <boolean> opt_concurrently %type <dbehavior> opt_drop_behavior +%type <list> opt_utility_option_list +%type <list> utility_option_list +%type <defelt> utility_option_elem +%type <str> utility_option_name +%type <node> utility_option_arg %type <node> alter_column_default opclass_item opclass_drop alter_using %type <ival> add_drop opt_asc_desc opt_nulls_order @@ -339,10 +343,6 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query); create_extension_opt_item alter_extension_opt_item %type <ival> opt_lock lock_type cast_context -%type <str> utility_option_name -%type <defelt> utility_option_elem -%type <list> utility_option_list -%type <node> utility_option_arg %type <defelt> drop_option %type <boolean> opt_or_replace opt_no opt_grant_grant_option @@ -523,7 +523,7 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query); %type <defelt> def_elem reloption_elem old_aggr_elem operator_def_elem %type <node> def_arg columnElem where_clause where_or_current_clause a_expr b_expr c_expr AexprConst indirection_el opt_slice_bound - columnref in_expr having_clause func_table xmltable array_expr + columnref having_clause func_table xmltable array_expr OptWhereClause operator_def_arg %type <list> opt_column_and_period_list %type <list> rowsfrom_item rowsfrom_list opt_col_def_list @@ -557,7 +557,6 @@ static Node *makeRecursiveViewSelect(char *relname, List *aliases, Node *query); %type <list> generic_option_list alter_generic_option_list %type <ival> reindex_target_relation reindex_target_all -%type <list> opt_reindex_option_list %type <node> copy_generic_opt_arg copy_generic_opt_arg_list_item %type <defelt> copy_generic_opt_elem @@ -1142,6 +1141,41 @@ opt_drop_behavior: | /* EMPTY */ { $$ = DROP_RESTRICT; /* default */ } ; +opt_utility_option_list: + '(' utility_option_list ')' { $$ = $2; } + | /* EMPTY */ { $$ = NULL; } + ; + +utility_option_list: + utility_option_elem + { + $$ = list_make1($1); + } + | utility_option_list ',' utility_option_elem + { + $$ = lappend($1, $3); + } + ; + +utility_option_elem: + utility_option_name utility_option_arg + { + $$ = makeDefElem($1, $2, @1); + } + ; + +utility_option_name: + NonReservedWord { $$ = $1; } + | analyze_keyword { $$ = "analyze"; } + | FORMAT_LA { $$ = "format"; } + ; + +utility_option_arg: + opt_boolean_or_string { $$ = (Node *) makeString($1); } + | NumericOnly { $$ = (Node *) $1; } + | /* EMPTY */ { $$ = NULL; } + ; + /***************************************************************************** * * CALL statement @@ -2029,11 +2063,12 @@ constraints_set_mode: * Checkpoint statement */ CheckPointStmt: - CHECKPOINT + CHECKPOINT opt_utility_option_list { CheckPointStmt *n = makeNode(CheckPointStmt); $$ = (Node *) n; + n->options = $2; } ; @@ -2669,6 +2704,12 @@ alter_table_cmd: c->alterDeferrability = true; if ($4 & CAS_NO_INHERIT) c->alterInheritability = true; + /* handle unsupported case with specific error message */ + if ($4 & CAS_NOT_VALID) + ereport(ERROR, + errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("constraints cannot be altered to be NOT VALID"), + parser_errposition(@4)); processCASbits($4, @4, "FOREIGN KEY", &c->deferrable, &c->initdeferred, @@ -3417,7 +3458,6 @@ CopyStmt: COPY opt_binary qualified_name opt_column_list { CopyStmt *n = makeNode(CopyStmt); - updatePreparableStmtEnd($3, @4); n->relation = NULL; n->query = $3; n->attlist = NIL; @@ -6037,6 +6077,26 @@ CreateTrigStmt: EXECUTE FUNCTION_or_PROCEDURE func_name '(' TriggerFuncArgs ')' { CreateTrigStmt *n = makeNode(CreateTrigStmt); + bool dummy; + + if (($11 & CAS_NOT_VALID) != 0) + ereport(ERROR, + errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("constraint triggers cannot be marked %s", + "NOT VALID"), + parser_errposition(@11)); + if (($11 & CAS_NO_INHERIT) != 0) + ereport(ERROR, + errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("constraint triggers cannot be marked %s", + "NO INHERIT"), + parser_errposition(@11)); + if (($11 & CAS_NOT_ENFORCED) != 0) + ereport(ERROR, + errcode(ERRCODE_FEATURE_NOT_SUPPORTED), + errmsg("constraint triggers cannot be marked %s", + "NOT ENFORCED"), + parser_errposition(@11)); n->replace = $2; if (n->replace) /* not supported, see CreateTrigger */ @@ -6056,7 +6116,7 @@ CreateTrigStmt: n->whenClause = $15; n->transitionRels = NIL; processCASbits($11, @11, "TRIGGER", - &n->deferrable, &n->initdeferred, NULL, + &n->deferrable, &n->initdeferred, &dummy, NULL, NULL, yyscanner); n->constrrel = $10; $$ = (Node *) n; @@ -7479,6 +7539,8 @@ fetch_args: cursor_name n->portalname = $1; n->direction = FETCH_FORWARD; n->howMany = 1; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_NONE; $$ = (Node *) n; } | from_in cursor_name @@ -7488,6 +7550,19 @@ fetch_args: cursor_name n->portalname = $2; n->direction = FETCH_FORWARD; n->howMany = 1; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_NONE; + $$ = (Node *) n; + } + | SignedIconst opt_from_in cursor_name + { + FetchStmt *n = makeNode(FetchStmt); + + n->portalname = $3; + n->direction = FETCH_FORWARD; + n->howMany = $1; + n->location = @1; + n->direction_keyword = FETCH_KEYWORD_NONE; $$ = (Node *) n; } | NEXT opt_from_in cursor_name @@ -7497,6 +7572,8 @@ fetch_args: cursor_name n->portalname = $3; n->direction = FETCH_FORWARD; n->howMany = 1; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_NEXT; $$ = (Node *) n; } | PRIOR opt_from_in cursor_name @@ -7506,6 +7583,8 @@ fetch_args: cursor_name n->portalname = $3; n->direction = FETCH_BACKWARD; n->howMany = 1; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_PRIOR; $$ = (Node *) n; } | FIRST_P opt_from_in cursor_name @@ -7515,6 +7594,8 @@ fetch_args: cursor_name n->portalname = $3; n->direction = FETCH_ABSOLUTE; n->howMany = 1; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_FIRST; $$ = (Node *) n; } | LAST_P opt_from_in cursor_name @@ -7524,6 +7605,8 @@ fetch_args: cursor_name n->portalname = $3; n->direction = FETCH_ABSOLUTE; n->howMany = -1; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_LAST; $$ = (Node *) n; } | ABSOLUTE_P SignedIconst opt_from_in cursor_name @@ -7533,6 +7616,8 @@ fetch_args: cursor_name n->portalname = $4; n->direction = FETCH_ABSOLUTE; n->howMany = $2; + n->location = @2; + n->direction_keyword = FETCH_KEYWORD_ABSOLUTE; $$ = (Node *) n; } | RELATIVE_P SignedIconst opt_from_in cursor_name @@ -7542,15 +7627,8 @@ fetch_args: cursor_name n->portalname = $4; n->direction = FETCH_RELATIVE; n->howMany = $2; - $$ = (Node *) n; - } - | SignedIconst opt_from_in cursor_name - { - FetchStmt *n = makeNode(FetchStmt); - - n->portalname = $3; - n->direction = FETCH_FORWARD; - n->howMany = $1; + n->location = @2; + n->direction_keyword = FETCH_KEYWORD_RELATIVE; $$ = (Node *) n; } | ALL opt_from_in cursor_name @@ -7560,6 +7638,8 @@ fetch_args: cursor_name n->portalname = $3; n->direction = FETCH_FORWARD; n->howMany = FETCH_ALL; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_ALL; $$ = (Node *) n; } | FORWARD opt_from_in cursor_name @@ -7569,6 +7649,8 @@ fetch_args: cursor_name n->portalname = $3; n->direction = FETCH_FORWARD; n->howMany = 1; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_FORWARD; $$ = (Node *) n; } | FORWARD SignedIconst opt_from_in cursor_name @@ -7578,6 +7660,8 @@ fetch_args: cursor_name n->portalname = $4; n->direction = FETCH_FORWARD; n->howMany = $2; + n->location = @2; + n->direction_keyword = FETCH_KEYWORD_FORWARD; $$ = (Node *) n; } | FORWARD ALL opt_from_in cursor_name @@ -7587,6 +7671,8 @@ fetch_args: cursor_name n->portalname = $4; n->direction = FETCH_FORWARD; n->howMany = FETCH_ALL; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_FORWARD_ALL; $$ = (Node *) n; } | BACKWARD opt_from_in cursor_name @@ -7596,6 +7682,8 @@ fetch_args: cursor_name n->portalname = $3; n->direction = FETCH_BACKWARD; n->howMany = 1; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_BACKWARD; $$ = (Node *) n; } | BACKWARD SignedIconst opt_from_in cursor_name @@ -7605,6 +7693,8 @@ fetch_args: cursor_name n->portalname = $4; n->direction = FETCH_BACKWARD; n->howMany = $2; + n->location = @2; + n->direction_keyword = FETCH_KEYWORD_BACKWARD; $$ = (Node *) n; } | BACKWARD ALL opt_from_in cursor_name @@ -7614,6 +7704,8 @@ fetch_args: cursor_name n->portalname = $4; n->direction = FETCH_BACKWARD; n->howMany = FETCH_ALL; + n->location = -1; + n->direction_keyword = FETCH_KEYWORD_BACKWARD_ALL; $$ = (Node *) n; } ; @@ -9291,7 +9383,7 @@ DropTransformStmt: DROP TRANSFORM opt_if_exists FOR Typename LANGUAGE name opt_d *****************************************************************************/ ReindexStmt: - REINDEX opt_reindex_option_list reindex_target_relation opt_concurrently qualified_name + REINDEX opt_utility_option_list reindex_target_relation opt_concurrently qualified_name { ReindexStmt *n = makeNode(ReindexStmt); @@ -9304,7 +9396,7 @@ ReindexStmt: makeDefElem("concurrently", NULL, @4)); $$ = (Node *) n; } - | REINDEX opt_reindex_option_list SCHEMA opt_concurrently name + | REINDEX opt_utility_option_list SCHEMA opt_concurrently name { ReindexStmt *n = makeNode(ReindexStmt); @@ -9317,7 +9409,7 @@ ReindexStmt: makeDefElem("concurrently", NULL, @4)); $$ = (Node *) n; } - | REINDEX opt_reindex_option_list reindex_target_all opt_concurrently opt_single_name + | REINDEX opt_utility_option_list reindex_target_all opt_concurrently opt_single_name { ReindexStmt *n = makeNode(ReindexStmt); @@ -9339,10 +9431,6 @@ reindex_target_all: SYSTEM_P { $$ = REINDEX_OBJECT_SYSTEM; } | DATABASE { $$ = REINDEX_OBJECT_DATABASE; } ; -opt_reindex_option_list: - '(' utility_option_list ')' { $$ = $2; } - | /* EMPTY */ { $$ = NULL; } - ; /***************************************************************************** * @@ -11629,7 +11717,7 @@ AlterDomainStmt: { AlterDomainStmt *n = makeNode(AlterDomainStmt); - n->subtype = 'T'; + n->subtype = AD_AlterDefault; n->typeName = $3; n->def = $4; $$ = (Node *) n; @@ -11639,7 +11727,7 @@ AlterDomainStmt: { AlterDomainStmt *n = makeNode(AlterDomainStmt); - n->subtype = 'N'; + n->subtype = AD_DropNotNull; n->typeName = $3; $$ = (Node *) n; } @@ -11648,7 +11736,7 @@ AlterDomainStmt: { AlterDomainStmt *n = makeNode(AlterDomainStmt); - n->subtype = 'O'; + n->subtype = AD_SetNotNull; n->typeName = $3; $$ = (Node *) n; } @@ -11657,7 +11745,7 @@ AlterDomainStmt: { AlterDomainStmt *n = makeNode(AlterDomainStmt); - n->subtype = 'C'; + n->subtype = AD_AddConstraint; n->typeName = $3; n->def = $5; $$ = (Node *) n; @@ -11667,7 +11755,7 @@ AlterDomainStmt: { AlterDomainStmt *n = makeNode(AlterDomainStmt); - n->subtype = 'X'; + n->subtype = AD_DropConstraint; n->typeName = $3; n->name = $6; n->behavior = $7; @@ -11679,7 +11767,7 @@ AlterDomainStmt: { AlterDomainStmt *n = makeNode(AlterDomainStmt); - n->subtype = 'X'; + n->subtype = AD_DropConstraint; n->typeName = $3; n->name = $8; n->behavior = $9; @@ -11691,7 +11779,7 @@ AlterDomainStmt: { AlterDomainStmt *n = makeNode(AlterDomainStmt); - n->subtype = 'V'; + n->subtype = AD_ValidateConstraint; n->typeName = $3; n->name = $6; $$ = (Node *) n; @@ -11840,13 +11928,13 @@ ClusterStmt: n->params = $3; $$ = (Node *) n; } - | CLUSTER '(' utility_option_list ')' + | CLUSTER opt_utility_option_list { ClusterStmt *n = makeNode(ClusterStmt); n->relation = NULL; n->indexname = NULL; - n->params = $3; + n->params = $2; $$ = (Node *) n; } /* unparenthesized VERBOSE kept for pre-14 compatibility */ @@ -11856,21 +11944,18 @@ ClusterStmt: n->relation = $3; n->indexname = $4; - n->params = NIL; if ($2) - n->params = lappend(n->params, makeDefElem("verbose", NULL, @2)); + n->params = list_make1(makeDefElem("verbose", NULL, @2)); $$ = (Node *) n; } /* unparenthesized VERBOSE kept for pre-17 compatibility */ - | CLUSTER opt_verbose + | CLUSTER VERBOSE { ClusterStmt *n = makeNode(ClusterStmt); n->relation = NULL; n->indexname = NULL; - n->params = NIL; - if ($2) - n->params = lappend(n->params, makeDefElem("verbose", NULL, @2)); + n->params = list_make1(makeDefElem("verbose", NULL, @2)); $$ = (Node *) n; } /* kept for pre-8.3 compatibility */ @@ -11880,9 +11965,8 @@ ClusterStmt: n->relation = $5; n->indexname = $3; - n->params = NIL; if ($2) - n->params = lappend(n->params, makeDefElem("verbose", NULL, @2)); + n->params = list_make1(makeDefElem("verbose", NULL, @2)); $$ = (Node *) n; } ; @@ -11933,64 +12017,31 @@ VacuumStmt: VACUUM opt_full opt_freeze opt_verbose opt_analyze opt_vacuum_relati } ; -AnalyzeStmt: analyze_keyword opt_verbose opt_vacuum_relation_list +AnalyzeStmt: analyze_keyword opt_utility_option_list opt_vacuum_relation_list { VacuumStmt *n = makeNode(VacuumStmt); - n->options = NIL; - if ($2) - n->options = lappend(n->options, - makeDefElem("verbose", NULL, @2)); + n->options = $2; n->rels = $3; n->is_vacuumcmd = false; $$ = (Node *) n; } - | analyze_keyword '(' utility_option_list ')' opt_vacuum_relation_list + | analyze_keyword VERBOSE opt_vacuum_relation_list { VacuumStmt *n = makeNode(VacuumStmt); - n->options = $3; - n->rels = $5; + n->options = list_make1(makeDefElem("verbose", NULL, @2)); + n->rels = $3; n->is_vacuumcmd = false; $$ = (Node *) n; } ; -utility_option_list: - utility_option_elem - { - $$ = list_make1($1); - } - | utility_option_list ',' utility_option_elem - { - $$ = lappend($1, $3); - } - ; - analyze_keyword: ANALYZE | ANALYSE /* British */ ; -utility_option_elem: - utility_option_name utility_option_arg - { - $$ = makeDefElem($1, $2, @1); - } - ; - -utility_option_name: - NonReservedWord { $$ = $1; } - | analyze_keyword { $$ = "analyze"; } - | FORMAT_LA { $$ = "format"; } - ; - -utility_option_arg: - opt_boolean_or_string { $$ = (Node *) makeString($1); } - | NumericOnly { $$ = (Node *) $1; } - | /* EMPTY */ { $$ = NULL; } - ; - opt_analyze: analyze_keyword { $$ = true; } | /*EMPTY*/ { $$ = false; } @@ -12240,7 +12291,6 @@ InsertStmt: $5->onConflictClause = $6; $5->returningClause = $7; $5->withClause = $1; - $5->stmt_location = @$; $$ = (Node *) $5; } ; @@ -12431,7 +12481,6 @@ DeleteStmt: opt_with_clause DELETE_P FROM relation_expr_opt_alias n->whereClause = $6; n->returningClause = $7; n->withClause = $1; - n->stmt_location = @$; $$ = (Node *) n; } ; @@ -12506,7 +12555,6 @@ UpdateStmt: opt_with_clause UPDATE relation_expr_opt_alias n->whereClause = $7; n->returningClause = $8; n->withClause = $1; - n->stmt_location = @$; $$ = (Node *) n; } ; @@ -12584,7 +12632,6 @@ MergeStmt: m->joinCondition = $8; m->mergeWhenClauses = $9; m->returningClause = $10; - m->stmt_location = @$; $$ = (Node *) m; } @@ -12825,20 +12872,7 @@ SelectStmt: select_no_parens %prec UMINUS ; select_with_parens: - '(' select_no_parens ')' - { - SelectStmt *n = (SelectStmt *) $2; - - /* - * As SelectStmt's location starts at the SELECT keyword, - * we need to track the length of the SelectStmt within - * parentheses to be able to extract the relevant part - * of the query. Without this, the RawStmt's length would - * be used and would include the closing parenthesis. - */ - n->stmt_len = @3 - @2; - $$ = $2; - } + '(' select_no_parens ')' { $$ = $2; } | '(' select_with_parens ')' { $$ = $2; } ; @@ -12960,7 +12994,6 @@ simple_select: n->groupDistinct = ($7)->distinct; n->havingClause = $8; n->windowClause = $9; - n->stmt_location = @1; $$ = (Node *) n; } | SELECT distinct_clause target_list @@ -12978,7 +13011,6 @@ simple_select: n->groupDistinct = ($7)->distinct; n->havingClause = $8; n->windowClause = $9; - n->stmt_location = @1; $$ = (Node *) n; } | values_clause { $$ = $1; } @@ -12999,20 +13031,19 @@ simple_select: n->targetList = list_make1(rt); n->fromClause = list_make1($2); - n->stmt_location = @1; $$ = (Node *) n; } | select_clause UNION set_quantifier select_clause { - $$ = makeSetOp(SETOP_UNION, $3 == SET_QUANTIFIER_ALL, $1, $4, @1); + $$ = makeSetOp(SETOP_UNION, $3 == SET_QUANTIFIER_ALL, $1, $4); } | select_clause INTERSECT set_quantifier select_clause { - $$ = makeSetOp(SETOP_INTERSECT, $3 == SET_QUANTIFIER_ALL, $1, $4, @1); + $$ = makeSetOp(SETOP_INTERSECT, $3 == SET_QUANTIFIER_ALL, $1, $4); } | select_clause EXCEPT set_quantifier select_clause { - $$ = makeSetOp(SETOP_EXCEPT, $3 == SET_QUANTIFIER_ALL, $1, $4, @1); + $$ = makeSetOp(SETOP_EXCEPT, $3 == SET_QUANTIFIER_ALL, $1, $4); } ; @@ -13590,7 +13621,6 @@ values_clause: { SelectStmt *n = makeNode(SelectStmt); - n->stmt_location = @1; n->valuesLists = list_make1($3); $$ = (Node *) n; } @@ -15287,49 +15317,50 @@ a_expr: c_expr { $$ = $1; } (Node *) list_make2($5, $7), @2); } - | a_expr IN_P in_expr + | a_expr IN_P select_with_parens { - /* in_expr returns a SubLink or a list of a_exprs */ - if (IsA($3, SubLink)) - { - /* generate foo = ANY (subquery) */ - SubLink *n = (SubLink *) $3; + /* generate foo = ANY (subquery) */ + SubLink *n = makeNode(SubLink); - n->subLinkType = ANY_SUBLINK; - n->subLinkId = 0; - n->testexpr = $1; - n->operName = NIL; /* show it's IN not = ANY */ - n->location = @2; - $$ = (Node *) n; - } - else - { - /* generate scalar IN expression */ - $$ = (Node *) makeSimpleA_Expr(AEXPR_IN, "=", $1, $3, @2); - } + n->subselect = $3; + n->subLinkType = ANY_SUBLINK; + n->subLinkId = 0; + n->testexpr = $1; + n->operName = NIL; /* show it's IN not = ANY */ + n->location = @2; + $$ = (Node *) n; } - | a_expr NOT_LA IN_P in_expr %prec NOT_LA + | a_expr IN_P '(' expr_list ')' { - /* in_expr returns a SubLink or a list of a_exprs */ - if (IsA($4, SubLink)) - { - /* generate NOT (foo = ANY (subquery)) */ - /* Make an = ANY node */ - SubLink *n = (SubLink *) $4; - - n->subLinkType = ANY_SUBLINK; - n->subLinkId = 0; - n->testexpr = $1; - n->operName = NIL; /* show it's IN not = ANY */ - n->location = @2; - /* Stick a NOT on top; must have same parse location */ - $$ = makeNotExpr((Node *) n, @2); - } - else - { - /* generate scalar NOT IN expression */ - $$ = (Node *) makeSimpleA_Expr(AEXPR_IN, "<>", $1, $4, @2); - } + /* generate scalar IN expression */ + A_Expr *n = makeSimpleA_Expr(AEXPR_IN, "=", $1, (Node *) $4, @2); + + n->rexpr_list_start = @3; + n->rexpr_list_end = @5; + $$ = (Node *) n; + } + | a_expr NOT_LA IN_P select_with_parens %prec NOT_LA + { + /* generate NOT (foo = ANY (subquery)) */ + SubLink *n = makeNode(SubLink); + + n->subselect = $4; + n->subLinkType = ANY_SUBLINK; + n->subLinkId = 0; + n->testexpr = $1; + n->operName = NIL; /* show it's IN not = ANY */ + n->location = @2; + /* Stick a NOT on top; must have same parse location */ + $$ = makeNotExpr((Node *) n, @2); + } + | a_expr NOT_LA IN_P '(' expr_list ')' + { + /* generate scalar NOT IN expression */ + A_Expr *n = makeSimpleA_Expr(AEXPR_IN, "<>", $1, (Node *) $5, @2); + + n->rexpr_list_start = @4; + n->rexpr_list_end = @6; + $$ = (Node *) n; } | a_expr subquery_Op sub_type select_with_parens %prec Op { @@ -16764,15 +16795,15 @@ type_list: Typename { $$ = list_make1($1); } array_expr: '[' expr_list ']' { - $$ = makeAArrayExpr($2, @1); + $$ = makeAArrayExpr($2, @1, @3); } | '[' array_expr_list ']' { - $$ = makeAArrayExpr($2, @1); + $$ = makeAArrayExpr($2, @1, @3); } | '[' ']' { - $$ = makeAArrayExpr(NIL, @1); + $$ = makeAArrayExpr(NIL, @1, @2); } ; @@ -16894,17 +16925,6 @@ trim_list: a_expr FROM expr_list { $$ = lappend($3, $1); } | expr_list { $$ = $1; } ; -in_expr: select_with_parens - { - SubLink *n = makeNode(SubLink); - - n->subselect = $1; - /* other fields will be filled later */ - $$ = (Node *) n; - } - | '(' expr_list ')' { $$ = (Node *) $2; } - ; - /* * Define SQL-style CASE clause. * - Full specification @@ -18748,47 +18768,6 @@ updateRawStmtEnd(RawStmt *rs, int end_location) rs->stmt_len = end_location - rs->stmt_location; } -/* - * Adjust a PreparableStmt to reflect that it doesn't run to the end of the - * string. - */ -static void -updatePreparableStmtEnd(Node *n, int end_location) -{ - if (IsA(n, SelectStmt)) - { - SelectStmt *stmt = (SelectStmt *) n; - - stmt->stmt_len = end_location - stmt->stmt_location; - } - else if (IsA(n, InsertStmt)) - { - InsertStmt *stmt = (InsertStmt *) n; - - stmt->stmt_len = end_location - stmt->stmt_location; - } - else if (IsA(n, UpdateStmt)) - { - UpdateStmt *stmt = (UpdateStmt *) n; - - stmt->stmt_len = end_location - stmt->stmt_location; - } - else if (IsA(n, DeleteStmt)) - { - DeleteStmt *stmt = (DeleteStmt *) n; - - stmt->stmt_len = end_location - stmt->stmt_location; - } - else if (IsA(n, MergeStmt)) - { - MergeStmt *stmt = (MergeStmt *) n; - - stmt->stmt_len = end_location - stmt->stmt_location; - } - else - elog(ERROR, "unexpected node type %d", (int) n->type); -} - static Node * makeColumnRef(char *colname, List *indirection, int location, core_yyscan_t yyscanner) @@ -19167,14 +19146,11 @@ insertSelectOptions(SelectStmt *stmt, errmsg("multiple WITH clauses not allowed"), parser_errposition(exprLocation((Node *) withClause)))); stmt->withClause = withClause; - - /* Update SelectStmt's location to the start of the WITH clause */ - stmt->stmt_location = withClause->location; } } static Node * -makeSetOp(SetOperation op, bool all, Node *larg, Node *rarg, int location) +makeSetOp(SetOperation op, bool all, Node *larg, Node *rarg) { SelectStmt *n = makeNode(SelectStmt); @@ -19182,7 +19158,6 @@ makeSetOp(SetOperation op, bool all, Node *larg, Node *rarg, int location) n->all = all; n->larg = (SelectStmt *) larg; n->rarg = (SelectStmt *) rarg; - n->stmt_location = location; return (Node *) n; } @@ -19300,12 +19275,14 @@ makeNotExpr(Node *expr, int location) } static Node * -makeAArrayExpr(List *elements, int location) +makeAArrayExpr(List *elements, int location, int location_end) { A_ArrayExpr *n = makeNode(A_ArrayExpr); n->elements = elements; n->location = location; + n->list_start = location; + n->list_end = location_end; return (Node *) n; } diff --git a/src/backend/parser/parse_expr.c b/src/backend/parser/parse_expr.c index 1f8e2d54673..d66276801c6 100644 --- a/src/backend/parser/parse_expr.c +++ b/src/backend/parser/parse_expr.c @@ -1223,6 +1223,8 @@ transformAExprIn(ParseState *pstate, A_Expr *a) newa->element_typeid = scalar_type; newa->elements = aexprs; newa->multidims = false; + newa->list_start = a->rexpr_list_start; + newa->list_end = a->rexpr_list_end; newa->location = -1; result = (Node *) make_scalar_array_op(pstate, @@ -2165,6 +2167,8 @@ transformArrayExpr(ParseState *pstate, A_ArrayExpr *a, /* array_collid will be set by parse_collate.c */ newa->element_typeid = element_type; newa->elements = newcoercedelems; + newa->list_start = a->list_start; + newa->list_end = a->list_end; newa->location = a->location; return (Node *) newa; diff --git a/src/backend/parser/parse_utilcmd.c b/src/backend/parser/parse_utilcmd.c index 62015431fdf..afcf54169c3 100644 --- a/src/backend/parser/parse_utilcmd.c +++ b/src/backend/parser/parse_utilcmd.c @@ -1279,6 +1279,28 @@ transformTableLikeClause(CreateStmtContext *cxt, TableLikeClause *table_like_cla lst = RelationGetNotNullConstraints(RelationGetRelid(relation), false, true); cxt->nnconstraints = list_concat(cxt->nnconstraints, lst); + + /* Copy comments on not-null constraints */ + if (table_like_clause->options & CREATE_TABLE_LIKE_COMMENTS) + { + foreach_node(Constraint, nnconstr, lst) + { + if ((comment = GetComment(get_relation_constraint_oid(RelationGetRelid(relation), + nnconstr->conname, false), + ConstraintRelationId, + 0)) != NULL) + { + CommentStmt *stmt = makeNode(CommentStmt); + + stmt->objtype = OBJECT_TABCONSTRAINT; + stmt->object = (Node *) list_make3(makeString(cxt->relation->schemaname), + makeString(cxt->relation->relname), + makeString(nnconstr->conname)); + stmt->comment = comment; + cxt->alist = lappend(cxt->alist, stmt); + } + } + } } /* diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c index 4bdc2941efb..822cf4ec451 100644 --- a/src/backend/partitioning/partbounds.c +++ b/src/backend/partitioning/partbounds.c @@ -1007,9 +1007,6 @@ partition_bounds_copy(PartitionBoundInfo src, int ndatums; int nindexes; int partnatts; - bool hash_part; - int natts; - Datum *boundDatums; dest = (PartitionBoundInfo) palloc(sizeof(PartitionBoundInfoData)); @@ -1023,7 +1020,7 @@ partition_bounds_copy(PartitionBoundInfo src, dest->datums = (Datum **) palloc(sizeof(Datum *) * ndatums); - if (src->kind != NULL) + if (src->kind != NULL && ndatums > 0) { PartitionRangeDatumKind *boundKinds; @@ -1058,36 +1055,40 @@ partition_bounds_copy(PartitionBoundInfo src, * For hash partitioning, datums array will have two elements - modulus * and remainder. */ - hash_part = (key->strategy == PARTITION_STRATEGY_HASH); - natts = hash_part ? 2 : partnatts; - boundDatums = palloc(ndatums * natts * sizeof(Datum)); - - for (i = 0; i < ndatums; i++) + if (ndatums > 0) { - int j; - - dest->datums[i] = &boundDatums[i * natts]; + bool hash_part = (key->strategy == PARTITION_STRATEGY_HASH); + int natts = hash_part ? 2 : partnatts; + Datum *boundDatums = palloc(ndatums * natts * sizeof(Datum)); - for (j = 0; j < natts; j++) + for (i = 0; i < ndatums; i++) { - bool byval; - int typlen; + int j; - if (hash_part) - { - typlen = sizeof(int32); /* Always int4 */ - byval = true; /* int4 is pass-by-value */ - } - else + dest->datums[i] = &boundDatums[i * natts]; + + for (j = 0; j < natts; j++) { - byval = key->parttypbyval[j]; - typlen = key->parttyplen[j]; - } + if (dest->kind == NULL || + dest->kind[i][j] == PARTITION_RANGE_DATUM_VALUE) + { + bool byval; + int typlen; - if (dest->kind == NULL || - dest->kind[i][j] == PARTITION_RANGE_DATUM_VALUE) - dest->datums[i][j] = datumCopy(src->datums[i][j], - byval, typlen); + if (hash_part) + { + typlen = sizeof(int32); /* Always int4 */ + byval = true; /* int4 is pass-by-value */ + } + else + { + byval = key->parttypbyval[j]; + typlen = key->parttyplen[j]; + } + dest->datums[i][j] = datumCopy(src->datums[i][j], + byval, typlen); + } + } } } diff --git a/src/backend/postmaster/autovacuum.c b/src/backend/postmaster/autovacuum.c index 451fb90a610..ff96b36d710 100644 --- a/src/backend/postmaster/autovacuum.c +++ b/src/backend/postmaster/autovacuum.c @@ -310,6 +310,16 @@ static AutoVacuumShmemStruct *AutoVacuumShmem; static dlist_head DatabaseList = DLIST_STATIC_INIT(DatabaseList); static MemoryContext DatabaseListCxt = NULL; +/* + * Dummy pointer to persuade Valgrind that we've not leaked the array of + * avl_dbase structs. Make it global to ensure the compiler doesn't + * optimize it away. + */ +#ifdef USE_VALGRIND +extern avl_dbase *avl_dbase_array; +avl_dbase *avl_dbase_array; +#endif + /* Pointer to my own WorkerInfo, valid on each worker */ static WorkerInfo MyWorkerInfo = NULL; @@ -562,10 +572,10 @@ AutoVacLauncherMain(const void *startup_data, size_t startup_data_len) /* * Create the initial database list. The invariant we want this list to - * keep is that it's ordered by decreasing next_time. As soon as an entry - * is updated to a higher time, it will be moved to the front (which is - * correct because the only operation is to add autovacuum_naptime to the - * entry, and time always increases). + * keep is that it's ordered by decreasing next_worker. As soon as an + * entry is updated to a higher time, it will be moved to the front (which + * is correct because the only operation is to add autovacuum_naptime to + * the entry, and time always increases). */ rebuild_database_list(InvalidOid); @@ -1020,6 +1030,10 @@ rebuild_database_list(Oid newdb) /* put all the hash elements into an array */ dbary = palloc(nelems * sizeof(avl_dbase)); + /* keep Valgrind quiet */ +#ifdef USE_VALGRIND + avl_dbase_array = dbary; +#endif i = 0; hash_seq_init(&seq, dbhash); @@ -2565,8 +2579,18 @@ deleted: /* * We leak table_toast_map here (among other things), but since we're - * going away soon, it's not a problem. + * going away soon, it's not a problem normally. But when using Valgrind, + * release some stuff to reduce complaints about leaked storage. */ +#ifdef USE_VALGRIND + hash_destroy(table_toast_map); + FreeTupleDesc(pg_class_desc); + if (bstrategy) + pfree(bstrategy); +#endif + + /* Run the rest in xact context, mainly to avoid Valgrind leak warnings */ + MemoryContextSwitchTo(TopTransactionContext); /* * Update pg_database.datfrozenxid, and truncate pg_xact if possible. We @@ -3190,7 +3214,7 @@ autovacuum_do_vac_analyze(autovac_table *tab, BufferAccessStrategy bstrategy) rel_list = list_make1(rel); MemoryContextSwitchTo(old_context); - vacuum(rel_list, &tab->at_params, bstrategy, vac_context, true); + vacuum(rel_list, tab->at_params, bstrategy, vac_context, true); MemoryContextDelete(vac_context); } diff --git a/src/backend/postmaster/bgworker.c b/src/backend/postmaster/bgworker.c index 116ddf7b835..1ad65c237c3 100644 --- a/src/backend/postmaster/bgworker.c +++ b/src/backend/postmaster/bgworker.c @@ -613,6 +613,7 @@ ResetBackgroundWorkerCrashTimes(void) * resetting. */ rw->rw_crashed_at = 0; + rw->rw_pid = 0; /* * If there was anyone waiting for it, they're history. diff --git a/src/backend/postmaster/checkpointer.c b/src/backend/postmaster/checkpointer.c index fda91ffd1ce..8490148a47d 100644 --- a/src/backend/postmaster/checkpointer.c +++ b/src/backend/postmaster/checkpointer.c @@ -42,6 +42,8 @@ #include "access/xlog.h" #include "access/xlog_internal.h" #include "access/xlogrecovery.h" +#include "catalog/pg_authid.h" +#include "commands/defrem.h" #include "libpq/pqsignal.h" #include "miscadmin.h" #include "pgstat.h" @@ -61,6 +63,7 @@ #include "storage/shmem.h" #include "storage/smgr.h" #include "storage/spin.h" +#include "utils/acl.h" #include "utils/guc.h" #include "utils/memutils.h" #include "utils/resowner.h" @@ -127,6 +130,13 @@ typedef struct int num_requests; /* current # of requests */ int max_requests; /* allocated array size */ + + int head; /* Index of the first request in the ring + * buffer */ + int tail; /* Index of the last request in the ring + * buffer */ + + /* The ring buffer of pending checkpointer requests */ CheckpointerRequest requests[FLEXIBLE_ARRAY_MEMBER]; } CheckpointerShmemStruct; @@ -135,6 +145,12 @@ static CheckpointerShmemStruct *CheckpointerShmem; /* interval for calling AbsorbSyncRequests in CheckpointWriteDelay */ #define WRITES_PER_ABSORB 1000 +/* Maximum number of checkpointer requests to process in one batch */ +#define CKPT_REQ_BATCH_SIZE 10000 + +/* Max number of requests the checkpointer request queue can hold */ +#define MAX_CHECKPOINT_REQUESTS 10000000 + /* * GUC parameters */ @@ -161,7 +177,7 @@ static pg_time_t last_xlog_switch_time; static void ProcessCheckpointerInterrupts(void); static void CheckArchiveTimeout(void); static bool IsCheckpointOnSchedule(double progress); -static bool ImmediateCheckpointRequested(void); +static bool FastCheckpointRequested(void); static bool CompactCheckpointerRequestQueue(void); static void UpdateSharedMemoryConfig(void); @@ -734,12 +750,12 @@ CheckArchiveTimeout(void) } /* - * Returns true if an immediate checkpoint request is pending. (Note that - * this does not check the *current* checkpoint's IMMEDIATE flag, but whether - * there is one pending behind it.) + * Returns true if a fast checkpoint request is pending. (Note that this does + * not check the *current* checkpoint's FAST flag, but whether there is one + * pending behind it.) */ static bool -ImmediateCheckpointRequested(void) +FastCheckpointRequested(void) { volatile CheckpointerShmemStruct *cps = CheckpointerShmem; @@ -747,7 +763,7 @@ ImmediateCheckpointRequested(void) * We don't need to acquire the ckpt_lck in this case because we're only * looking at a single flag bit. */ - if (cps->ckpt_flags & CHECKPOINT_IMMEDIATE) + if (cps->ckpt_flags & CHECKPOINT_FAST) return true; return false; } @@ -760,7 +776,7 @@ ImmediateCheckpointRequested(void) * checkpoint_completion_target. * * The checkpoint request flags should be passed in; currently the only one - * examined is CHECKPOINT_IMMEDIATE, which disables delays between writes. + * examined is CHECKPOINT_FAST, which disables delays between writes. * * 'progress' is an estimate of how much of the work has been done, as a * fraction between 0.0 meaning none, and 1.0 meaning all done. @@ -778,10 +794,10 @@ CheckpointWriteDelay(int flags, double progress) * Perform the usual duties and take a nap, unless we're behind schedule, * in which case we just try to catch up as quickly as possible. */ - if (!(flags & CHECKPOINT_IMMEDIATE) && + if (!(flags & CHECKPOINT_FAST) && !ShutdownXLOGPending && !ShutdownRequestPending && - !ImmediateCheckpointRequested() && + !FastCheckpointRequested() && IsCheckpointOnSchedule(progress)) { if (ConfigReloadPending) @@ -970,24 +986,75 @@ CheckpointerShmemInit(void) */ MemSet(CheckpointerShmem, 0, size); SpinLockInit(&CheckpointerShmem->ckpt_lck); - CheckpointerShmem->max_requests = NBuffers; + CheckpointerShmem->max_requests = Min(NBuffers, MAX_CHECKPOINT_REQUESTS); + CheckpointerShmem->head = CheckpointerShmem->tail = 0; ConditionVariableInit(&CheckpointerShmem->start_cv); ConditionVariableInit(&CheckpointerShmem->done_cv); } } /* + * ExecCheckpoint + * Primary entry point for manual CHECKPOINT commands + * + * This is mainly a wrapper for RequestCheckpoint(). + */ +void +ExecCheckpoint(ParseState *pstate, CheckPointStmt *stmt) +{ + bool fast = true; + bool unlogged = false; + + foreach_ptr(DefElem, opt, stmt->options) + { + if (strcmp(opt->defname, "mode") == 0) + { + char *mode = defGetString(opt); + + if (strcmp(mode, "spread") == 0) + fast = false; + else if (strcmp(mode, "fast") != 0) + ereport(ERROR, + (errcode(ERRCODE_SYNTAX_ERROR), + errmsg("unrecognized MODE option \"%s\"", mode), + parser_errposition(pstate, opt->location))); + } + else if (strcmp(opt->defname, "flush_unlogged") == 0) + unlogged = defGetBoolean(opt); + else + ereport(ERROR, + (errcode(ERRCODE_SYNTAX_ERROR), + errmsg("unrecognized CHECKPOINT option \"%s\"", opt->defname), + parser_errposition(pstate, opt->location))); + } + + if (!has_privs_of_role(GetUserId(), ROLE_PG_CHECKPOINT)) + ereport(ERROR, + (errcode(ERRCODE_INSUFFICIENT_PRIVILEGE), + /* translator: %s is name of an SQL command (e.g., CHECKPOINT) */ + errmsg("permission denied to execute %s command", + "CHECKPOINT"), + errdetail("Only roles with privileges of the \"%s\" role may execute this command.", + "pg_checkpoint"))); + + RequestCheckpoint(CHECKPOINT_WAIT | + (fast ? CHECKPOINT_FAST : 0) | + (unlogged ? CHECKPOINT_FLUSH_UNLOGGED : 0) | + (RecoveryInProgress() ? 0 : CHECKPOINT_FORCE)); +} + +/* * RequestCheckpoint * Called in backend processes to request a checkpoint * * flags is a bitwise OR of the following: * CHECKPOINT_IS_SHUTDOWN: checkpoint is for database shutdown. * CHECKPOINT_END_OF_RECOVERY: checkpoint is for end of WAL recovery. - * CHECKPOINT_IMMEDIATE: finish the checkpoint ASAP, + * CHECKPOINT_FAST: finish the checkpoint ASAP, * ignoring checkpoint_completion_target parameter. * CHECKPOINT_FORCE: force a checkpoint even if no XLOG activity has occurred * since the last one (implied by CHECKPOINT_IS_SHUTDOWN or - * CHECKPOINT_END_OF_RECOVERY). + * CHECKPOINT_END_OF_RECOVERY, and the CHECKPOINT command). * CHECKPOINT_WAIT: wait for completion before returning (otherwise, * just signal checkpointer to do it, and return). * CHECKPOINT_CAUSE_XLOG: checkpoint is requested due to xlog filling. @@ -1009,7 +1076,7 @@ RequestCheckpoint(int flags) * There's no point in doing slow checkpoints in a standalone backend, * because there's no other backends the checkpoint could disrupt. */ - CreateCheckPoint(flags | CHECKPOINT_IMMEDIATE); + CreateCheckPoint(flags | CHECKPOINT_FAST); /* Free all smgr objects, as CheckpointerMain() normally would. */ smgrdestroyall(); @@ -1148,6 +1215,7 @@ ForwardSyncRequest(const FileTag *ftag, SyncRequestType type) { CheckpointerRequest *request; bool too_full; + int insert_pos; if (!IsUnderPostmaster) return false; /* probably shouldn't even get here */ @@ -1171,10 +1239,14 @@ ForwardSyncRequest(const FileTag *ftag, SyncRequestType type) } /* OK, insert request */ - request = &CheckpointerShmem->requests[CheckpointerShmem->num_requests++]; + insert_pos = CheckpointerShmem->tail; + request = &CheckpointerShmem->requests[insert_pos]; request->ftag = *ftag; request->type = type; + CheckpointerShmem->tail = (CheckpointerShmem->tail + 1) % CheckpointerShmem->max_requests; + CheckpointerShmem->num_requests++; + /* If queue is more than half full, nudge the checkpointer to empty it */ too_full = (CheckpointerShmem->num_requests >= CheckpointerShmem->max_requests / 2); @@ -1216,12 +1288,16 @@ CompactCheckpointerRequestQueue(void) struct CheckpointerSlotMapping { CheckpointerRequest request; - int slot; + int ring_idx; }; - int n, - preserve_count; + int n; int num_skipped = 0; + int head; + int max_requests; + int num_requests; + int read_idx, + write_idx; HASHCTL ctl; HTAB *htab; bool *skip_slot; @@ -1233,8 +1309,13 @@ CompactCheckpointerRequestQueue(void) if (CritSectionCount > 0) return false; + max_requests = CheckpointerShmem->max_requests; + num_requests = CheckpointerShmem->num_requests; + /* Initialize skip_slot array */ - skip_slot = palloc0(sizeof(bool) * CheckpointerShmem->num_requests); + skip_slot = palloc0(sizeof(bool) * max_requests); + + head = CheckpointerShmem->head; /* Initialize temporary hash table */ ctl.keysize = sizeof(CheckpointerRequest); @@ -1258,7 +1339,8 @@ CompactCheckpointerRequestQueue(void) * away preceding entries that would end up being canceled anyhow), but * it's not clear that the extra complexity would buy us anything. */ - for (n = 0; n < CheckpointerShmem->num_requests; n++) + read_idx = head; + for (n = 0; n < num_requests; n++) { CheckpointerRequest *request; struct CheckpointerSlotMapping *slotmap; @@ -1271,16 +1353,19 @@ CompactCheckpointerRequestQueue(void) * CheckpointerShmemInit. Note also that RelFileLocator had better * contain no pad bytes. */ - request = &CheckpointerShmem->requests[n]; + request = &CheckpointerShmem->requests[read_idx]; slotmap = hash_search(htab, request, HASH_ENTER, &found); if (found) { /* Duplicate, so mark the previous occurrence as skippable */ - skip_slot[slotmap->slot] = true; + skip_slot[slotmap->ring_idx] = true; num_skipped++; } /* Remember slot containing latest occurrence of this request value */ - slotmap->slot = n; + slotmap->ring_idx = read_idx; + + /* Move to the next request in the ring buffer */ + read_idx = (read_idx + 1) % max_requests; } /* Done with the hash table. */ @@ -1294,17 +1379,34 @@ CompactCheckpointerRequestQueue(void) } /* We found some duplicates; remove them. */ - preserve_count = 0; - for (n = 0; n < CheckpointerShmem->num_requests; n++) + read_idx = write_idx = head; + for (n = 0; n < num_requests; n++) { - if (skip_slot[n]) - continue; - CheckpointerShmem->requests[preserve_count++] = CheckpointerShmem->requests[n]; + /* If this slot is NOT skipped, keep it */ + if (!skip_slot[read_idx]) + { + /* If the read and write positions are different, copy the request */ + if (write_idx != read_idx) + CheckpointerShmem->requests[write_idx] = + CheckpointerShmem->requests[read_idx]; + + /* Advance the write position */ + write_idx = (write_idx + 1) % max_requests; + } + + read_idx = (read_idx + 1) % max_requests; } + + /* + * Update ring buffer state: head remains the same, tail moves, count + * decreases + */ + CheckpointerShmem->tail = write_idx; + CheckpointerShmem->num_requests -= num_skipped; + ereport(DEBUG1, (errmsg_internal("compacted fsync request queue from %d entries to %d entries", - CheckpointerShmem->num_requests, preserve_count))); - CheckpointerShmem->num_requests = preserve_count; + num_requests, CheckpointerShmem->num_requests))); /* Cleanup. */ pfree(skip_slot); @@ -1325,40 +1427,64 @@ AbsorbSyncRequests(void) { CheckpointerRequest *requests = NULL; CheckpointerRequest *request; - int n; + int n, + i; + bool loop; if (!AmCheckpointerProcess()) return; - LWLockAcquire(CheckpointerCommLock, LW_EXCLUSIVE); - - /* - * We try to avoid holding the lock for a long time by copying the request - * array, and processing the requests after releasing the lock. - * - * Once we have cleared the requests from shared memory, we have to PANIC - * if we then fail to absorb them (eg, because our hashtable runs out of - * memory). This is because the system cannot run safely if we are unable - * to fsync what we have been told to fsync. Fortunately, the hashtable - * is so small that the problem is quite unlikely to arise in practice. - */ - n = CheckpointerShmem->num_requests; - if (n > 0) + do { - requests = (CheckpointerRequest *) palloc(n * sizeof(CheckpointerRequest)); - memcpy(requests, CheckpointerShmem->requests, n * sizeof(CheckpointerRequest)); - } + LWLockAcquire(CheckpointerCommLock, LW_EXCLUSIVE); + + /*--- + * We try to avoid holding the lock for a long time by: + * 1. Copying the request array and processing the requests after + * releasing the lock; + * 2. Processing not the whole queue, but only batches of + * CKPT_REQ_BATCH_SIZE at once. + * + * Once we have cleared the requests from shared memory, we must + * PANIC if we then fail to absorb them (e.g., because our hashtable + * runs out of memory). This is because the system cannot run safely + * if we are unable to fsync what we have been told to fsync. + * Fortunately, the hashtable is so small that the problem is quite + * unlikely to arise in practice. + * + * Note: The maximum possible size of a ring buffer is + * MAX_CHECKPOINT_REQUESTS entries, which fit into a maximum palloc + * allocation size of 1Gb. Our maximum batch size, + * CKPT_REQ_BATCH_SIZE, is even smaller. + */ + n = Min(CheckpointerShmem->num_requests, CKPT_REQ_BATCH_SIZE); + if (n > 0) + { + if (!requests) + requests = (CheckpointerRequest *) palloc(n * sizeof(CheckpointerRequest)); - START_CRIT_SECTION(); + for (i = 0; i < n; i++) + { + requests[i] = CheckpointerShmem->requests[CheckpointerShmem->head]; + CheckpointerShmem->head = (CheckpointerShmem->head + 1) % CheckpointerShmem->max_requests; + } - CheckpointerShmem->num_requests = 0; + CheckpointerShmem->num_requests -= n; - LWLockRelease(CheckpointerCommLock); + } + + START_CRIT_SECTION(); + + /* Are there any requests in the queue? If so, keep going. */ + loop = CheckpointerShmem->num_requests != 0; + + LWLockRelease(CheckpointerCommLock); - for (request = requests; n > 0; request++, n--) - RememberSyncRequest(&request->ftag, request->type); + for (request = requests; n > 0; request++, n--) + RememberSyncRequest(&request->ftag, request->type); - END_CRIT_SECTION(); + END_CRIT_SECTION(); + } while (loop); if (requests) pfree(requests); diff --git a/src/backend/postmaster/pgarch.c b/src/backend/postmaster/pgarch.c index 7e622ae4bd2..78e39e5f866 100644 --- a/src/backend/postmaster/pgarch.c +++ b/src/backend/postmaster/pgarch.c @@ -718,15 +718,15 @@ pgarch_readyXlog(char *xlog) /* * Store the file in our max-heap if it has a high enough priority. */ - if (arch_files->arch_heap->bh_size < NUM_FILES_PER_DIRECTORY_SCAN) + if (binaryheap_size(arch_files->arch_heap) < NUM_FILES_PER_DIRECTORY_SCAN) { /* If the heap isn't full yet, quickly add it. */ - arch_file = arch_files->arch_filenames[arch_files->arch_heap->bh_size]; + arch_file = arch_files->arch_filenames[binaryheap_size(arch_files->arch_heap)]; strcpy(arch_file, basename); binaryheap_add_unordered(arch_files->arch_heap, CStringGetDatum(arch_file)); /* If we just filled the heap, make it a valid one. */ - if (arch_files->arch_heap->bh_size == NUM_FILES_PER_DIRECTORY_SCAN) + if (binaryheap_size(arch_files->arch_heap) == NUM_FILES_PER_DIRECTORY_SCAN) binaryheap_build(arch_files->arch_heap); } else if (ready_file_comparator(binaryheap_first(arch_files->arch_heap), @@ -744,21 +744,21 @@ pgarch_readyXlog(char *xlog) FreeDir(rldir); /* If no files were found, simply return. */ - if (arch_files->arch_heap->bh_size == 0) + if (binaryheap_empty(arch_files->arch_heap)) return false; /* * If we didn't fill the heap, we didn't make it a valid one. Do that * now. */ - if (arch_files->arch_heap->bh_size < NUM_FILES_PER_DIRECTORY_SCAN) + if (binaryheap_size(arch_files->arch_heap) < NUM_FILES_PER_DIRECTORY_SCAN) binaryheap_build(arch_files->arch_heap); /* * Fill arch_files array with the files to archive in ascending order of * priority. */ - arch_files->arch_files_size = arch_files->arch_heap->bh_size; + arch_files->arch_files_size = binaryheap_size(arch_files->arch_heap); for (int i = 0; i < arch_files->arch_files_size; i++) arch_files->arch_files[i] = DatumGetCString(binaryheap_remove_first(arch_files->arch_heap)); diff --git a/src/backend/postmaster/pmchild.c b/src/backend/postmaster/pmchild.c index cde1d23a4ca..584bb58c8ab 100644 --- a/src/backend/postmaster/pmchild.c +++ b/src/backend/postmaster/pmchild.c @@ -60,6 +60,17 @@ NON_EXEC_STATIC int num_pmchild_slots = 0; dlist_head ActiveChildList; /* + * Dummy pointer to persuade Valgrind that we've not leaked the array of + * PMChild structs. Make it global to ensure the compiler doesn't + * optimize it away. + */ +#ifdef USE_VALGRIND +extern PMChild *pmchild_array; +PMChild *pmchild_array; +#endif + + +/* * MaxLivePostmasterChildren * * This reports the number of postmaster child processes that can be active. @@ -125,8 +136,13 @@ InitPostmasterChildSlots(void) for (int i = 0; i < BACKEND_NUM_TYPES; i++) num_pmchild_slots += pmchild_pools[i].size; - /* Initialize them */ + /* Allocate enough slots, and make sure Valgrind doesn't complain */ slots = palloc(num_pmchild_slots * sizeof(PMChild)); +#ifdef USE_VALGRIND + pmchild_array = slots; +#endif + + /* Initialize them */ slotno = 0; for (int btype = 0; btype < BACKEND_NUM_TYPES; btype++) { diff --git a/src/backend/postmaster/postmaster.c b/src/backend/postmaster/postmaster.c index 490f7ce3664..e01d9f0cfe8 100644 --- a/src/backend/postmaster/postmaster.c +++ b/src/backend/postmaster/postmaster.c @@ -2630,6 +2630,13 @@ CleanupBackend(PMChild *bp, } bp = NULL; + /* + * In a crash case, exit immediately without resetting background worker + * state. However, if restart_after_crash is enabled, the background + * worker state (e.g., rw_pid) still needs be reset so the worker can + * restart after crash recovery. This reset is handled in + * ResetBackgroundWorkerCrashTimes(), not here. + */ if (crashed) { HandleChildCrash(bp_pid, exitstatus, procname); @@ -4337,15 +4344,15 @@ maybe_start_bgworkers(void) static bool maybe_reap_io_worker(int pid) { - for (int id = 0; id < MAX_IO_WORKERS; ++id) + for (int i = 0; i < MAX_IO_WORKERS; ++i) { - if (io_worker_children[id] && - io_worker_children[id]->pid == pid) + if (io_worker_children[i] && + io_worker_children[i]->pid == pid) { - ReleasePostmasterChildSlot(io_worker_children[id]); + ReleasePostmasterChildSlot(io_worker_children[i]); --io_worker_count; - io_worker_children[id] = NULL; + io_worker_children[i] = NULL; return true; } } @@ -4389,22 +4396,22 @@ maybe_adjust_io_workers(void) while (io_worker_count < io_workers) { PMChild *child; - int id; + int i; /* find unused entry in io_worker_children array */ - for (id = 0; id < MAX_IO_WORKERS; ++id) + for (i = 0; i < MAX_IO_WORKERS; ++i) { - if (io_worker_children[id] == NULL) + if (io_worker_children[i] == NULL) break; } - if (id == MAX_IO_WORKERS) - elog(ERROR, "could not find a free IO worker ID"); + if (i == MAX_IO_WORKERS) + elog(ERROR, "could not find a free IO worker slot"); /* Try to launch one. */ child = StartChildProcess(B_IO_WORKER); if (child != NULL) { - io_worker_children[id] = child; + io_worker_children[i] = child; ++io_worker_count; } else @@ -4415,11 +4422,11 @@ maybe_adjust_io_workers(void) if (io_worker_count > io_workers) { /* ask the IO worker in the highest slot to exit */ - for (int id = MAX_IO_WORKERS - 1; id >= 0; --id) + for (int i = MAX_IO_WORKERS - 1; i >= 0; --i) { - if (io_worker_children[id] != NULL) + if (io_worker_children[i] != NULL) { - kill(io_worker_children[id]->pid, SIGUSR2); + kill(io_worker_children[i]->pid, SIGUSR2); break; } } diff --git a/src/backend/postmaster/walsummarizer.c b/src/backend/postmaster/walsummarizer.c index 0fec4f1f871..777c9a8d555 100644 --- a/src/backend/postmaster/walsummarizer.c +++ b/src/backend/postmaster/walsummarizer.c @@ -385,7 +385,7 @@ WalSummarizerMain(const void *startup_data, size_t startup_data_len) switch_lsn = tliSwitchPoint(current_tli, tles, &switch_tli); ereport(DEBUG1, - errmsg_internal("switch point from TLI %u to TLI %u is at %X/%X", + errmsg_internal("switch point from TLI %u to TLI %u is at %X/%08X", current_tli, switch_tli, LSN_FORMAT_ARGS(switch_lsn))); } @@ -741,7 +741,7 @@ WaitForWalSummarization(XLogRecPtr lsn) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), errmsg("WAL summarization is not progressing"), - errdetail("Summarization is needed through %X/%X, but is stuck at %X/%X on disk and %X/%X in memory.", + errdetail("Summarization is needed through %X/%08X, but is stuck at %X/%08X on disk and %X/%08X in memory.", LSN_FORMAT_ARGS(lsn), LSN_FORMAT_ARGS(summarized_lsn), LSN_FORMAT_ARGS(pending_lsn)))); @@ -755,12 +755,12 @@ WaitForWalSummarization(XLogRecPtr lsn) current_time) / 1000; ereport(WARNING, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg_plural("still waiting for WAL summarization through %X/%X after %ld second", - "still waiting for WAL summarization through %X/%X after %ld seconds", + errmsg_plural("still waiting for WAL summarization through %X/%08X after %ld second", + "still waiting for WAL summarization through %X/%08X after %ld seconds", elapsed_seconds, LSN_FORMAT_ARGS(lsn), elapsed_seconds), - errdetail("Summarization has reached %X/%X on disk and %X/%X in memory.", + errdetail("Summarization has reached %X/%08X on disk and %X/%08X in memory.", LSN_FORMAT_ARGS(summarized_lsn), LSN_FORMAT_ARGS(pending_lsn)))); } @@ -981,7 +981,7 @@ SummarizeWAL(TimeLineID tli, XLogRecPtr start_lsn, bool exact, if (private_data->end_of_wal) { ereport(DEBUG1, - errmsg_internal("could not read WAL from timeline %u at %X/%X: end of WAL at %X/%X", + errmsg_internal("could not read WAL from timeline %u at %X/%08X: end of WAL at %X/%08X", tli, LSN_FORMAT_ARGS(start_lsn), LSN_FORMAT_ARGS(private_data->read_upto))); @@ -1000,8 +1000,8 @@ SummarizeWAL(TimeLineID tli, XLogRecPtr start_lsn, bool exact, } else ereport(ERROR, - (errmsg("could not find a valid record after %X/%X", - LSN_FORMAT_ARGS(start_lsn)))); + errmsg("could not find a valid record after %X/%08X", + LSN_FORMAT_ARGS(start_lsn))); } /* We shouldn't go backward. */ @@ -1034,7 +1034,7 @@ SummarizeWAL(TimeLineID tli, XLogRecPtr start_lsn, bool exact, * able to read a complete record. */ ereport(DEBUG1, - errmsg_internal("could not read WAL from timeline %u at %X/%X: end of WAL at %X/%X", + errmsg_internal("could not read WAL from timeline %u at %X/%08X: end of WAL at %X/%08X", tli, LSN_FORMAT_ARGS(xlogreader->EndRecPtr), LSN_FORMAT_ARGS(private_data->read_upto))); @@ -1045,13 +1045,13 @@ SummarizeWAL(TimeLineID tli, XLogRecPtr start_lsn, bool exact, if (errormsg) ereport(ERROR, (errcode_for_file_access(), - errmsg("could not read WAL from timeline %u at %X/%X: %s", + errmsg("could not read WAL from timeline %u at %X/%08X: %s", tli, LSN_FORMAT_ARGS(xlogreader->EndRecPtr), errormsg))); else ereport(ERROR, (errcode_for_file_access(), - errmsg("could not read WAL from timeline %u at %X/%X", + errmsg("could not read WAL from timeline %u at %X/%08X", tli, LSN_FORMAT_ARGS(xlogreader->EndRecPtr)))); } @@ -1222,7 +1222,7 @@ SummarizeWAL(TimeLineID tli, XLogRecPtr start_lsn, bool exact, /* Tell the user what we did. */ ereport(DEBUG1, - errmsg_internal("summarized WAL on TLI %u from %X/%X to %X/%X", + errmsg_internal("summarized WAL on TLI %u from %X/%08X to %X/%08X", tli, LSN_FORMAT_ARGS(summary_start_lsn), LSN_FORMAT_ARGS(summary_end_lsn))); @@ -1234,7 +1234,7 @@ SummarizeWAL(TimeLineID tli, XLogRecPtr start_lsn, bool exact, /* If we skipped a non-zero amount of WAL, log a debug message. */ if (summary_end_lsn > summary_start_lsn && fast_forward) ereport(DEBUG1, - errmsg_internal("skipped summarizing WAL on TLI %u from %X/%X to %X/%X", + errmsg_internal("skipped summarizing WAL on TLI %u from %X/%08X to %X/%08X", tli, LSN_FORMAT_ARGS(summary_start_lsn), LSN_FORMAT_ARGS(summary_end_lsn))); @@ -1580,7 +1580,7 @@ summarizer_read_local_xlog_page(XLogReaderState *state, /* Debugging output. */ ereport(DEBUG1, - errmsg_internal("timeline %u became historic, can read up to %X/%X", + errmsg_internal("timeline %u became historic, can read up to %X/%08X", private_data->tli, LSN_FORMAT_ARGS(private_data->read_upto))); } diff --git a/src/backend/regex/regc_pg_locale.c b/src/backend/regex/regc_pg_locale.c index 78193cfb964..d9eab5357bc 100644 --- a/src/backend/regex/regc_pg_locale.c +++ b/src/backend/regex/regc_pg_locale.c @@ -20,58 +20,13 @@ #include "common/unicode_category.h" #include "utils/pg_locale.h" -/* - * For the libc provider, to provide as much functionality as possible on a - * variety of platforms without going so far as to implement everything from - * scratch, we use several implementation strategies depending on the - * situation: - * - * 1. In C/POSIX collations, we use hard-wired code. We can't depend on - * the <ctype.h> functions since those will obey LC_CTYPE. Note that these - * collations don't give a fig about multibyte characters. - * - * 2. When working in UTF8 encoding, we use the <wctype.h> functions. - * This assumes that every platform uses Unicode codepoints directly - * as the wchar_t representation of Unicode. (XXX: ICU makes this assumption - * even for non-UTF8 encodings, which may be a problem.) On some platforms - * wchar_t is only 16 bits wide, so we have to punt for codepoints > 0xFFFF. - * - * 3. In all other encodings, we use the <ctype.h> functions for pg_wchar - * values up to 255, and punt for values above that. This is 100% correct - * only in single-byte encodings such as LATINn. However, non-Unicode - * multibyte encodings are mostly Far Eastern character sets for which the - * properties being tested here aren't very relevant for higher code values - * anyway. The difficulty with using the <wctype.h> functions with - * non-Unicode multibyte encodings is that we can have no certainty that - * the platform's wchar_t representation matches what we do in pg_wchar - * conversions. - * - * As a special case, in the "default" collation, (2) and (3) force ASCII - * letters to follow ASCII upcase/downcase rules, while in a non-default - * collation we just let the library functions do what they will. The case - * where this matters is treatment of I/i in Turkish, and the behavior is - * meant to match the upper()/lower() SQL functions. - * - * We store the active collation setting in static variables. In principle - * it could be passed down to here via the regex library's "struct vars" data - * structure; but that would require somewhat invasive changes in the regex - * library, and right now there's no real benefit to be gained from that. - * - * NB: the coding here assumes pg_wchar is an unsigned type. - */ - -typedef enum -{ - PG_REGEX_STRATEGY_C, /* C locale (encoding independent) */ - PG_REGEX_STRATEGY_BUILTIN, /* built-in Unicode semantics */ - PG_REGEX_STRATEGY_LIBC_WIDE, /* Use locale_t <wctype.h> functions */ - PG_REGEX_STRATEGY_LIBC_1BYTE, /* Use locale_t <ctype.h> functions */ - PG_REGEX_STRATEGY_ICU, /* Use ICU uchar.h functions */ -} PG_Locale_Strategy; - -static PG_Locale_Strategy pg_regex_strategy; static pg_locale_t pg_regex_locale; +static struct pg_locale_struct dummy_c_locale = { + .collate_is_c = true, + .ctype_is_c = true, +}; + /* * Hard-wired character properties for C locale */ @@ -228,7 +183,6 @@ void pg_set_regex_collation(Oid collation) { pg_locale_t locale = 0; - PG_Locale_Strategy strategy; if (!OidIsValid(collation)) { @@ -249,8 +203,7 @@ pg_set_regex_collation(Oid collation) * catalog access is available, so we can't call * pg_newlocale_from_collation(). */ - strategy = PG_REGEX_STRATEGY_C; - locale = 0; + locale = &dummy_c_locale; } else { @@ -267,113 +220,41 @@ pg_set_regex_collation(Oid collation) * C/POSIX collations use this path regardless of database * encoding */ - strategy = PG_REGEX_STRATEGY_C; - locale = 0; - } - else if (locale->provider == COLLPROVIDER_BUILTIN) - { - Assert(GetDatabaseEncoding() == PG_UTF8); - strategy = PG_REGEX_STRATEGY_BUILTIN; - } -#ifdef USE_ICU - else if (locale->provider == COLLPROVIDER_ICU) - { - strategy = PG_REGEX_STRATEGY_ICU; - } -#endif - else - { - Assert(locale->provider == COLLPROVIDER_LIBC); - if (GetDatabaseEncoding() == PG_UTF8) - strategy = PG_REGEX_STRATEGY_LIBC_WIDE; - else - strategy = PG_REGEX_STRATEGY_LIBC_1BYTE; + locale = &dummy_c_locale; } } - pg_regex_strategy = strategy; pg_regex_locale = locale; } static int pg_wc_isdigit(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISDIGIT)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_isdigit(c, !pg_regex_locale->info.builtin.casemap_full); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswdigit_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - isdigit_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_isdigit(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISDIGIT)); + else + return pg_regex_locale->ctype->wc_isdigit(c, pg_regex_locale); } static int pg_wc_isalpha(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISALPHA)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_isalpha(c); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswalpha_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - isalpha_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_isalpha(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISALPHA)); + else + return pg_regex_locale->ctype->wc_isalpha(c, pg_regex_locale); } static int pg_wc_isalnum(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISALNUM)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_isalnum(c, !pg_regex_locale->info.builtin.casemap_full); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswalnum_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - isalnum_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_isalnum(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISALNUM)); + else + return pg_regex_locale->ctype->wc_isalnum(c, pg_regex_locale); } static int @@ -388,231 +269,87 @@ pg_wc_isword(pg_wchar c) static int pg_wc_isupper(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISUPPER)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_isupper(c); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswupper_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - isupper_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_isupper(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISUPPER)); + else + return pg_regex_locale->ctype->wc_isupper(c, pg_regex_locale); } static int pg_wc_islower(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISLOWER)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_islower(c); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswlower_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - islower_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_islower(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISLOWER)); + else + return pg_regex_locale->ctype->wc_islower(c, pg_regex_locale); } static int pg_wc_isgraph(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISGRAPH)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_isgraph(c); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswgraph_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - isgraph_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_isgraph(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISGRAPH)); + else + return pg_regex_locale->ctype->wc_isgraph(c, pg_regex_locale); } static int pg_wc_isprint(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISPRINT)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_isprint(c); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswprint_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - isprint_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_isprint(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISPRINT)); + else + return pg_regex_locale->ctype->wc_isprint(c, pg_regex_locale); } static int pg_wc_ispunct(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISPUNCT)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_ispunct(c, !pg_regex_locale->info.builtin.casemap_full); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswpunct_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - ispunct_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_ispunct(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISPUNCT)); + else + return pg_regex_locale->ctype->wc_ispunct(c, pg_regex_locale); } static int pg_wc_isspace(pg_wchar c) { - switch (pg_regex_strategy) - { - case PG_REGEX_STRATEGY_C: - return (c <= (pg_wchar) 127 && - (pg_char_properties[c] & PG_ISSPACE)); - case PG_REGEX_STRATEGY_BUILTIN: - return pg_u_isspace(c); - case PG_REGEX_STRATEGY_LIBC_WIDE: - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return iswspace_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - return (c <= (pg_wchar) UCHAR_MAX && - isspace_l((unsigned char) c, pg_regex_locale->info.lt)); - break; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_isspace(c); -#endif - break; - } - return 0; /* can't get here, but keep compiler quiet */ + if (pg_regex_locale->ctype_is_c) + return (c <= (pg_wchar) 127 && + (pg_char_properties[c] & PG_ISSPACE)); + else + return pg_regex_locale->ctype->wc_isspace(c, pg_regex_locale); } static pg_wchar pg_wc_toupper(pg_wchar c) { - switch (pg_regex_strategy) + if (pg_regex_locale->ctype_is_c) { - case PG_REGEX_STRATEGY_C: - if (c <= (pg_wchar) 127) - return pg_ascii_toupper((unsigned char) c); - return c; - case PG_REGEX_STRATEGY_BUILTIN: - return unicode_uppercase_simple(c); - case PG_REGEX_STRATEGY_LIBC_WIDE: - /* force C behavior for ASCII characters, per comments above */ - if (pg_regex_locale->is_default && c <= (pg_wchar) 127) - return pg_ascii_toupper((unsigned char) c); - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return towupper_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - /* force C behavior for ASCII characters, per comments above */ - if (pg_regex_locale->is_default && c <= (pg_wchar) 127) - return pg_ascii_toupper((unsigned char) c); - if (c <= (pg_wchar) UCHAR_MAX) - return toupper_l((unsigned char) c, pg_regex_locale->info.lt); - return c; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_toupper(c); -#endif - break; + if (c <= (pg_wchar) 127) + return pg_ascii_toupper((unsigned char) c); + return c; } - return 0; /* can't get here, but keep compiler quiet */ + else + return pg_regex_locale->ctype->wc_toupper(c, pg_regex_locale); } static pg_wchar pg_wc_tolower(pg_wchar c) { - switch (pg_regex_strategy) + if (pg_regex_locale->ctype_is_c) { - case PG_REGEX_STRATEGY_C: - if (c <= (pg_wchar) 127) - return pg_ascii_tolower((unsigned char) c); - return c; - case PG_REGEX_STRATEGY_BUILTIN: - return unicode_lowercase_simple(c); - case PG_REGEX_STRATEGY_LIBC_WIDE: - /* force C behavior for ASCII characters, per comments above */ - if (pg_regex_locale->is_default && c <= (pg_wchar) 127) - return pg_ascii_tolower((unsigned char) c); - if (sizeof(wchar_t) >= 4 || c <= (pg_wchar) 0xFFFF) - return towlower_l((wint_t) c, pg_regex_locale->info.lt); - /* FALL THRU */ - case PG_REGEX_STRATEGY_LIBC_1BYTE: - /* force C behavior for ASCII characters, per comments above */ - if (pg_regex_locale->is_default && c <= (pg_wchar) 127) - return pg_ascii_tolower((unsigned char) c); - if (c <= (pg_wchar) UCHAR_MAX) - return tolower_l((unsigned char) c, pg_regex_locale->info.lt); - return c; - case PG_REGEX_STRATEGY_ICU: -#ifdef USE_ICU - return u_tolower(c); -#endif - break; + if (c <= (pg_wchar) 127) + return pg_ascii_tolower((unsigned char) c); + return c; } - return 0; /* can't get here, but keep compiler quiet */ + else + return pg_regex_locale->ctype->wc_tolower(c, pg_regex_locale); } @@ -738,37 +475,25 @@ pg_ctype_get_cache(pg_wc_probefunc probefunc, int cclasscode) * would always be true for production values of MAX_SIMPLE_CHR, but it's * useful to allow it to be small for testing purposes.) */ - switch (pg_regex_strategy) + if (pg_regex_locale->ctype_is_c) { - case PG_REGEX_STRATEGY_C: #if MAX_SIMPLE_CHR >= 127 - max_chr = (pg_wchar) 127; - pcc->cv.cclasscode = -1; + max_chr = (pg_wchar) 127; + pcc->cv.cclasscode = -1; #else - max_chr = (pg_wchar) MAX_SIMPLE_CHR; + max_chr = (pg_wchar) MAX_SIMPLE_CHR; #endif - break; - case PG_REGEX_STRATEGY_BUILTIN: - max_chr = (pg_wchar) MAX_SIMPLE_CHR; - break; - case PG_REGEX_STRATEGY_LIBC_WIDE: - max_chr = (pg_wchar) MAX_SIMPLE_CHR; - break; - case PG_REGEX_STRATEGY_LIBC_1BYTE: -#if MAX_SIMPLE_CHR >= UCHAR_MAX - max_chr = (pg_wchar) UCHAR_MAX; + } + else + { + if (pg_regex_locale->ctype->max_chr != 0 && + pg_regex_locale->ctype->max_chr <= MAX_SIMPLE_CHR) + { + max_chr = pg_regex_locale->ctype->max_chr; pcc->cv.cclasscode = -1; -#else - max_chr = (pg_wchar) MAX_SIMPLE_CHR; -#endif - break; - case PG_REGEX_STRATEGY_ICU: + } + else max_chr = (pg_wchar) MAX_SIMPLE_CHR; - break; - default: - Assert(false); - max_chr = 0; /* can't get here, but keep compiler quiet */ - break; } /* diff --git a/src/backend/replication/libpqwalreceiver/libpqwalreceiver.c b/src/backend/replication/libpqwalreceiver/libpqwalreceiver.c index 7b4ddf7a8f5..239641bfbb6 100644 --- a/src/backend/replication/libpqwalreceiver/libpqwalreceiver.c +++ b/src/backend/replication/libpqwalreceiver/libpqwalreceiver.c @@ -232,6 +232,9 @@ libpqrcv_connect(const char *conninfo, bool replication, bool logical, errhint("Target server's authentication method must be changed, or set password_required=false in the subscription parameters."))); } + PQsetNoticeReceiver(conn->streamConn, libpqsrv_notice_receiver, + "received message via replication"); + /* * Set always-secure search path for the cases where the connection is * used to run SQL queries, so malicious users can't get control. @@ -418,31 +421,22 @@ libpqrcv_identify_system(WalReceiverConn *conn, TimeLineID *primary_tli) "IDENTIFY_SYSTEM", WAIT_EVENT_LIBPQWALRECEIVER_RECEIVE); if (PQresultStatus(res) != PGRES_TUPLES_OK) - { - PQclear(res); ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("could not receive database system identifier and timeline ID from " "the primary server: %s", pchomp(PQerrorMessage(conn->streamConn))))); - } /* * IDENTIFY_SYSTEM returns 3 columns in 9.3 and earlier, and 4 columns in * 9.4 and onwards. */ if (PQnfields(res) < 3 || PQntuples(res) != 1) - { - int ntuples = PQntuples(res); - int nfields = PQnfields(res); - - PQclear(res); ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("invalid response from primary server"), errdetail("Could not identify system: got %d rows and %d fields, expected %d rows and %d or more fields.", - ntuples, nfields, 1, 3))); - } + PQntuples(res), PQnfields(res), 1, 3))); primary_sysid = pstrdup(PQgetvalue(res, 0, 0)); *primary_tli = pg_strtoint32(PQgetvalue(res, 0, 1)); PQclear(res); @@ -534,7 +528,7 @@ libpqrcv_startstreaming(WalReceiverConn *conn, if (options->logical) appendStringInfoString(&cmd, " LOGICAL"); - appendStringInfo(&cmd, " %X/%X", LSN_FORMAT_ARGS(options->startpoint)); + appendStringInfo(&cmd, " %X/%08X", LSN_FORMAT_ARGS(options->startpoint)); /* * Additional options are different depending on if we are doing logical @@ -604,13 +598,10 @@ libpqrcv_startstreaming(WalReceiverConn *conn, return false; } else if (PQresultStatus(res) != PGRES_COPY_BOTH) - { - PQclear(res); ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("could not start WAL streaming: %s", pchomp(PQerrorMessage(conn->streamConn))))); - } PQclear(res); return true; } @@ -718,26 +709,17 @@ libpqrcv_readtimelinehistoryfile(WalReceiverConn *conn, cmd, WAIT_EVENT_LIBPQWALRECEIVER_RECEIVE); if (PQresultStatus(res) != PGRES_TUPLES_OK) - { - PQclear(res); ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("could not receive timeline history file from " "the primary server: %s", pchomp(PQerrorMessage(conn->streamConn))))); - } if (PQnfields(res) != 2 || PQntuples(res) != 1) - { - int ntuples = PQntuples(res); - int nfields = PQnfields(res); - - PQclear(res); ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("invalid response from primary server"), errdetail("Expected 1 tuple with 2 fields, got %d tuples with %d fields.", - ntuples, nfields))); - } + PQntuples(res), PQnfields(res)))); *filename = pstrdup(PQgetvalue(res, 0, 0)); *len = PQgetlength(res, 0, 1); @@ -841,13 +823,10 @@ libpqrcv_receive(WalReceiverConn *conn, char **buffer, return -1; } else - { - PQclear(res); ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("could not receive data from WAL stream: %s", pchomp(PQerrorMessage(conn->streamConn))))); - } } if (rawlen < -1) ereport(ERROR, @@ -971,13 +950,10 @@ libpqrcv_create_slot(WalReceiverConn *conn, const char *slotname, pfree(cmd.data); if (PQresultStatus(res) != PGRES_TUPLES_OK) - { - PQclear(res); ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("could not create replication slot \"%s\": %s", slotname, pchomp(PQerrorMessage(conn->streamConn))))); - } if (lsn) *lsn = DatumGetLSN(DirectFunctionCall1Coll(pg_lsn_in, InvalidOid, diff --git a/src/backend/replication/logical/applyparallelworker.c b/src/backend/replication/logical/applyparallelworker.c index d25085d3515..1fa931a7422 100644 --- a/src/backend/replication/logical/applyparallelworker.c +++ b/src/backend/replication/logical/applyparallelworker.c @@ -441,7 +441,8 @@ pa_launch_parallel_worker(void) MySubscription->name, MyLogicalRepWorker->userid, InvalidOid, - dsm_segment_handle(winfo->dsm_seg)); + dsm_segment_handle(winfo->dsm_seg), + false); if (launched) { diff --git a/src/backend/replication/logical/conflict.c b/src/backend/replication/logical/conflict.c index 97c4e26b586..2fd3e8bbda5 100644 --- a/src/backend/replication/logical/conflict.c +++ b/src/backend/replication/logical/conflict.c @@ -29,6 +29,7 @@ static const char *const ConflictTypeNames[] = { [CT_UPDATE_EXISTS] = "update_exists", [CT_UPDATE_MISSING] = "update_missing", [CT_DELETE_ORIGIN_DIFFERS] = "delete_origin_differs", + [CT_UPDATE_DELETED] = "update_deleted", [CT_DELETE_MISSING] = "delete_missing", [CT_MULTIPLE_UNIQUE_CONFLICTS] = "multiple_unique_conflicts" }; @@ -176,6 +177,7 @@ errcode_apply_conflict(ConflictType type) case CT_UPDATE_ORIGIN_DIFFERS: case CT_UPDATE_MISSING: case CT_DELETE_ORIGIN_DIFFERS: + case CT_UPDATE_DELETED: case CT_DELETE_MISSING: return errcode(ERRCODE_T_R_SERIALIZATION_FAILURE); } @@ -261,6 +263,26 @@ errdetail_apply_conflict(EState *estate, ResultRelInfo *relinfo, break; + case CT_UPDATE_DELETED: + if (localts) + { + if (localorigin == InvalidRepOriginId) + appendStringInfo(&err_detail, _("The row to be updated was deleted locally in transaction %u at %s."), + localxmin, timestamptz_to_str(localts)); + else if (replorigin_by_oid(localorigin, true, &origin_name)) + appendStringInfo(&err_detail, _("The row to be updated was deleted by a different origin \"%s\" in transaction %u at %s."), + origin_name, localxmin, timestamptz_to_str(localts)); + + /* The origin that modified this row has been removed. */ + else + appendStringInfo(&err_detail, _("The row to be updated was deleted by a non-existent origin in transaction %u at %s."), + localxmin, timestamptz_to_str(localts)); + } + else + appendStringInfo(&err_detail, _("The row to be updated was deleted.")); + + break; + case CT_UPDATE_MISSING: appendStringInfoString(&err_detail, _("Could not find the row to be updated.")); break; diff --git a/src/backend/replication/logical/launcher.c b/src/backend/replication/logical/launcher.c index 10677da56b2..37377f7eb63 100644 --- a/src/backend/replication/logical/launcher.c +++ b/src/backend/replication/logical/launcher.c @@ -32,6 +32,7 @@ #include "postmaster/interrupt.h" #include "replication/logicallauncher.h" #include "replication/origin.h" +#include "replication/slot.h" #include "replication/walreceiver.h" #include "replication/worker_internal.h" #include "storage/ipc.h" @@ -91,7 +92,6 @@ static dshash_table *last_start_times = NULL; static bool on_commit_launcher_wakeup = false; -static void ApplyLauncherWakeup(void); static void logicalrep_launcher_onexit(int code, Datum arg); static void logicalrep_worker_onexit(int code, Datum arg); static void logicalrep_worker_detach(void); @@ -100,6 +100,9 @@ static int logicalrep_pa_worker_count(Oid subid); static void logicalrep_launcher_attach_dshmem(void); static void ApplyLauncherSetWorkerStartTime(Oid subid, TimestampTz start_time); static TimestampTz ApplyLauncherGetWorkerStartTime(Oid subid); +static void compute_min_nonremovable_xid(LogicalRepWorker *worker, TransactionId *xmin); +static bool acquire_conflict_slot_if_exists(void); +static void advance_conflict_slot_xmin(TransactionId new_xmin); /* @@ -148,6 +151,7 @@ get_subscription_list(void) sub->owner = subform->subowner; sub->enabled = subform->subenabled; sub->name = pstrdup(NameStr(subform->subname)); + sub->retaindeadtuples = subform->subretaindeadtuples; /* We don't fill fields we are not interested in. */ res = lappend(res, sub); @@ -175,12 +179,14 @@ WaitForReplicationWorkerAttach(LogicalRepWorker *worker, uint16 generation, BackgroundWorkerHandle *handle) { - BgwHandleStatus status; - int rc; + bool result = false; + bool dropped_latch = false; for (;;) { + BgwHandleStatus status; pid_t pid; + int rc; CHECK_FOR_INTERRUPTS(); @@ -189,8 +195,9 @@ WaitForReplicationWorkerAttach(LogicalRepWorker *worker, /* Worker either died or has started. Return false if died. */ if (!worker->in_use || worker->proc) { + result = worker->in_use; LWLockRelease(LogicalRepWorkerLock); - return worker->in_use; + break; } LWLockRelease(LogicalRepWorkerLock); @@ -205,7 +212,7 @@ WaitForReplicationWorkerAttach(LogicalRepWorker *worker, if (generation == worker->generation) logicalrep_worker_cleanup(worker); LWLockRelease(LogicalRepWorkerLock); - return false; + break; /* result is already false */ } /* @@ -220,8 +227,18 @@ WaitForReplicationWorkerAttach(LogicalRepWorker *worker, { ResetLatch(MyLatch); CHECK_FOR_INTERRUPTS(); + dropped_latch = true; } } + + /* + * If we had to clear a latch event in order to wait, be sure to restore + * it before exiting. Otherwise caller may miss events. + */ + if (dropped_latch) + SetLatch(MyLatch); + + return result; } /* @@ -296,7 +313,8 @@ logicalrep_workers_find(Oid subid, bool only_running, bool acquire_lock) bool logicalrep_worker_launch(LogicalRepWorkerType wtype, Oid dbid, Oid subid, const char *subname, Oid userid, - Oid relid, dsm_handle subworker_dsm) + Oid relid, dsm_handle subworker_dsm, + bool retain_dead_tuples) { BackgroundWorker bgw; BackgroundWorkerHandle *bgw_handle; @@ -315,10 +333,13 @@ logicalrep_worker_launch(LogicalRepWorkerType wtype, * - must be valid worker type * - tablesync workers are only ones to have relid * - parallel apply worker is the only kind of subworker + * - The replication slot used in conflict detection is created when + * retain_dead_tuples is enabled */ Assert(wtype != WORKERTYPE_UNKNOWN); Assert(is_tablesync_worker == OidIsValid(relid)); Assert(is_parallel_apply_worker == (subworker_dsm != DSM_HANDLE_INVALID)); + Assert(!retain_dead_tuples || MyReplicationSlot); ereport(DEBUG1, (errmsg_internal("starting logical replication worker for subscription \"%s\"", @@ -328,7 +349,7 @@ logicalrep_worker_launch(LogicalRepWorkerType wtype, if (max_active_replication_origins == 0) ereport(ERROR, (errcode(ERRCODE_CONFIGURATION_LIMIT_EXCEEDED), - errmsg("cannot start logical replication workers when \"max_active_replication_origins\"=0"))); + errmsg("cannot start logical replication workers when \"max_active_replication_origins\" is 0"))); /* * We need to do the modification of the shared memory under lock so that @@ -441,6 +462,9 @@ retry: worker->stream_fileset = NULL; worker->leader_pid = is_parallel_apply_worker ? MyProcPid : InvalidPid; worker->parallel_apply = is_parallel_apply_worker; + worker->oldest_nonremovable_xid = retain_dead_tuples + ? MyReplicationSlot->data.xmin + : InvalidTransactionId; worker->last_lsn = InvalidXLogRecPtr; TIMESTAMP_NOBEGIN(worker->last_send_time); TIMESTAMP_NOBEGIN(worker->last_recv_time); @@ -766,6 +790,8 @@ logicalrep_worker_detach(void) } LWLockRelease(LogicalRepWorkerLock); + + list_free(workers); } /* Block concurrent access. */ @@ -1016,7 +1042,7 @@ logicalrep_launcher_attach_dshmem(void) last_start_times_dsa = dsa_attach(LogicalRepCtx->last_start_dsa); dsa_pin_mapping(last_start_times_dsa); last_start_times = dshash_attach(last_start_times_dsa, &dsh_params, - LogicalRepCtx->last_start_dsh, 0); + LogicalRepCtx->last_start_dsh, NULL); } MemoryContextSwitchTo(oldcontext); @@ -1105,7 +1131,10 @@ ApplyLauncherWakeupAtCommit(void) on_commit_launcher_wakeup = true; } -static void +/* + * Wakeup the launcher immediately. + */ +void ApplyLauncherWakeup(void) { if (LogicalRepCtx->launcher_pid != 0) @@ -1137,6 +1166,12 @@ ApplyLauncherMain(Datum main_arg) */ BackgroundWorkerInitializeConnection(NULL, NULL, 0); + /* + * Acquire the conflict detection slot at startup to ensure it can be + * dropped if no longer needed after a restart. + */ + acquire_conflict_slot_if_exists(); + /* Enter main loop */ for (;;) { @@ -1146,6 +1181,9 @@ ApplyLauncherMain(Datum main_arg) MemoryContext subctx; MemoryContext oldctx; long wait_time = DEFAULT_NAPTIME_PER_CYCLE; + bool can_advance_xmin = true; + bool retain_dead_tuples = false; + TransactionId xmin = InvalidTransactionId; CHECK_FOR_INTERRUPTS(); @@ -1155,7 +1193,14 @@ ApplyLauncherMain(Datum main_arg) ALLOCSET_DEFAULT_SIZES); oldctx = MemoryContextSwitchTo(subctx); - /* Start any missing workers for enabled subscriptions. */ + /* + * Start any missing workers for enabled subscriptions. + * + * Also, during the iteration through all subscriptions, we compute + * the minimum XID required to protect deleted tuples for conflict + * detection if one of the subscription enables retain_dead_tuples + * option. + */ sublist = get_subscription_list(); foreach(lc, sublist) { @@ -1165,6 +1210,38 @@ ApplyLauncherMain(Datum main_arg) TimestampTz now; long elapsed; + if (sub->retaindeadtuples) + { + retain_dead_tuples = true; + + /* + * Can't advance xmin of the slot unless all the subscriptions + * with retain_dead_tuples are enabled. This is required to + * ensure that we don't advance the xmin of + * CONFLICT_DETECTION_SLOT if one of the subscriptions is not + * enabled. Otherwise, we won't be able to detect conflicts + * reliably for such a subscription even though it has set the + * retain_dead_tuples option. + */ + can_advance_xmin &= sub->enabled; + + /* + * Create a replication slot to retain information necessary + * for conflict detection such as dead tuples, commit + * timestamps, and origins. + * + * The slot is created before starting the apply worker to + * prevent it from unnecessarily maintaining its + * oldest_nonremovable_xid. + * + * The slot is created even for a disabled subscription to + * ensure that conflict-related information is available when + * applying remote changes that occurred before the + * subscription was enabled. + */ + CreateConflictDetectionSlot(); + } + if (!sub->enabled) continue; @@ -1173,7 +1250,27 @@ ApplyLauncherMain(Datum main_arg) LWLockRelease(LogicalRepWorkerLock); if (w != NULL) - continue; /* worker is running already */ + { + /* + * Compute the minimum xmin required to protect dead tuples + * required for conflict detection among all running apply + * workers that enables retain_dead_tuples. + */ + if (sub->retaindeadtuples && can_advance_xmin) + compute_min_nonremovable_xid(w, &xmin); + + /* worker is running already */ + continue; + } + + /* + * Can't advance xmin of the slot unless all the workers + * corresponding to subscriptions with retain_dead_tuples are + * running, disabling the further computation of the minimum + * nonremovable xid. + */ + if (sub->retaindeadtuples) + can_advance_xmin = false; /* * If the worker is eligible to start now, launch it. Otherwise, @@ -1194,10 +1291,22 @@ ApplyLauncherMain(Datum main_arg) (elapsed = TimestampDifferenceMilliseconds(last_start, now)) >= wal_retrieve_retry_interval) { ApplyLauncherSetWorkerStartTime(sub->oid, now); - logicalrep_worker_launch(WORKERTYPE_APPLY, - sub->dbid, sub->oid, sub->name, - sub->owner, InvalidOid, - DSM_HANDLE_INVALID); + if (!logicalrep_worker_launch(WORKERTYPE_APPLY, + sub->dbid, sub->oid, sub->name, + sub->owner, InvalidOid, + DSM_HANDLE_INVALID, + sub->retaindeadtuples)) + { + /* + * We get here either if we failed to launch a worker + * (perhaps for resource-exhaustion reasons) or if we + * launched one but it immediately quit. Either way, it + * seems appropriate to try again after + * wal_retrieve_retry_interval. + */ + wait_time = Min(wait_time, + wal_retrieve_retry_interval); + } } else { @@ -1206,6 +1315,20 @@ ApplyLauncherMain(Datum main_arg) } } + /* + * Drop the CONFLICT_DETECTION_SLOT slot if there is no subscription + * that requires us to retain dead tuples. Otherwise, if required, + * advance the slot's xmin to protect dead tuples required for the + * conflict detection. + */ + if (MyReplicationSlot) + { + if (!retain_dead_tuples) + ReplicationSlotDropAcquired(); + else if (can_advance_xmin) + advance_conflict_slot_xmin(xmin); + } + /* Switch back to original memory context. */ MemoryContextSwitchTo(oldctx); /* Clean the temporary memory. */ @@ -1234,6 +1357,125 @@ ApplyLauncherMain(Datum main_arg) } /* + * Determine the minimum non-removable transaction ID across all apply workers + * for subscriptions that have retain_dead_tuples enabled. Store the result + * in *xmin. + */ +static void +compute_min_nonremovable_xid(LogicalRepWorker *worker, TransactionId *xmin) +{ + TransactionId nonremovable_xid; + + Assert(worker != NULL); + + /* + * The replication slot for conflict detection must be created before the + * worker starts. + */ + Assert(MyReplicationSlot); + + SpinLockAcquire(&worker->relmutex); + nonremovable_xid = worker->oldest_nonremovable_xid; + SpinLockRelease(&worker->relmutex); + + Assert(TransactionIdIsValid(nonremovable_xid)); + + if (!TransactionIdIsValid(*xmin) || + TransactionIdPrecedes(nonremovable_xid, *xmin)) + *xmin = nonremovable_xid; +} + +/* + * Acquire the replication slot used to retain information for conflict + * detection, if it exists. + * + * Return true if successfully acquired, otherwise return false. + */ +static bool +acquire_conflict_slot_if_exists(void) +{ + if (!SearchNamedReplicationSlot(CONFLICT_DETECTION_SLOT, true)) + return false; + + ReplicationSlotAcquire(CONFLICT_DETECTION_SLOT, true, false); + return true; +} + +/* + * Advance the xmin the replication slot used to retain information required + * for conflict detection. + */ +static void +advance_conflict_slot_xmin(TransactionId new_xmin) +{ + Assert(MyReplicationSlot); + Assert(TransactionIdIsValid(new_xmin)); + Assert(TransactionIdPrecedesOrEquals(MyReplicationSlot->data.xmin, new_xmin)); + + /* Return if the xmin value of the slot cannot be advanced */ + if (TransactionIdEquals(MyReplicationSlot->data.xmin, new_xmin)) + return; + + SpinLockAcquire(&MyReplicationSlot->mutex); + MyReplicationSlot->effective_xmin = new_xmin; + MyReplicationSlot->data.xmin = new_xmin; + SpinLockRelease(&MyReplicationSlot->mutex); + + elog(DEBUG1, "updated xmin: %u", MyReplicationSlot->data.xmin); + + ReplicationSlotMarkDirty(); + ReplicationSlotsComputeRequiredXmin(false); + + /* + * Like PhysicalConfirmReceivedLocation(), do not save slot information + * each time. This is acceptable because all concurrent transactions on + * the publisher that require the data preceding the slot's xmin should + * have already been applied and flushed on the subscriber before the xmin + * is advanced. So, even if the slot's xmin regresses after a restart, it + * will be advanced again in the next cycle. Therefore, no data required + * for conflict detection will be prematurely removed. + */ + return; +} + +/* + * Create and acquire the replication slot used to retain information for + * conflict detection, if not yet. + */ +void +CreateConflictDetectionSlot(void) +{ + TransactionId xmin_horizon; + + /* Exit early, if the replication slot is already created and acquired */ + if (MyReplicationSlot) + return; + + ereport(LOG, + errmsg("creating replication conflict detection slot")); + + ReplicationSlotCreate(CONFLICT_DETECTION_SLOT, false, RS_PERSISTENT, false, + false, false); + + LWLockAcquire(ProcArrayLock, LW_EXCLUSIVE); + + xmin_horizon = GetOldestSafeDecodingTransactionId(false); + + SpinLockAcquire(&MyReplicationSlot->mutex); + MyReplicationSlot->effective_xmin = xmin_horizon; + MyReplicationSlot->data.xmin = xmin_horizon; + SpinLockRelease(&MyReplicationSlot->mutex); + + ReplicationSlotsComputeRequiredXmin(true); + + LWLockRelease(ProcArrayLock); + + /* Write this slot to disk */ + ReplicationSlotMarkDirty(); + ReplicationSlotSave(); +} + +/* * Is current process the logical replication launcher? */ bool diff --git a/src/backend/replication/logical/logical.c b/src/backend/replication/logical/logical.c index 1d56d0c4ef3..7e363a7c05b 100644 --- a/src/backend/replication/logical/logical.c +++ b/src/backend/replication/logical/logical.c @@ -29,6 +29,7 @@ #include "postgres.h" #include "access/xact.h" +#include "access/xlog_internal.h" #include "access/xlogutils.h" #include "fmgr.h" #include "miscadmin.h" @@ -41,6 +42,7 @@ #include "storage/proc.h" #include "storage/procarray.h" #include "utils/builtins.h" +#include "utils/injection_point.h" #include "utils/inval.h" #include "utils/memutils.h" @@ -565,7 +567,7 @@ CreateDecodingContext(XLogRecPtr start_lsn, * kinds of client errors; so the client may wish to check that * confirmed_flush_lsn matches its expectations. */ - elog(LOG, "%X/%X has been already streamed, forwarding to %X/%X", + elog(LOG, "%X/%08X has been already streamed, forwarding to %X/%08X", LSN_FORMAT_ARGS(start_lsn), LSN_FORMAT_ARGS(slot->data.confirmed_flush)); @@ -608,7 +610,7 @@ CreateDecodingContext(XLogRecPtr start_lsn, ereport(LOG, (errmsg("starting logical decoding for slot \"%s\"", NameStr(slot->data.name)), - errdetail("Streaming transactions committing after %X/%X, reading WAL from %X/%X.", + errdetail("Streaming transactions committing after %X/%08X, reading WAL from %X/%08X.", LSN_FORMAT_ARGS(slot->data.confirmed_flush), LSN_FORMAT_ARGS(slot->data.restart_lsn)))); @@ -635,7 +637,7 @@ DecodingContextFindStartpoint(LogicalDecodingContext *ctx) /* Initialize from where to start reading WAL. */ XLogBeginRead(ctx->reader, slot->data.restart_lsn); - elog(DEBUG1, "searching for logical decoding starting point, starting at %X/%X", + elog(DEBUG1, "searching for logical decoding starting point, starting at %X/%08X", LSN_FORMAT_ARGS(slot->data.restart_lsn)); /* Wait for a consistent starting point */ @@ -756,7 +758,7 @@ output_plugin_error_callback(void *arg) /* not all callbacks have an associated LSN */ if (state->report_location != InvalidXLogRecPtr) - errcontext("slot \"%s\", output plugin \"%s\", in the %s callback, associated LSN %X/%X", + errcontext("slot \"%s\", output plugin \"%s\", in the %s callback, associated LSN %X/%08X", NameStr(state->ctx->slot->data.name), NameStr(state->ctx->slot->data.plugin), state->callback_name, @@ -1723,7 +1725,7 @@ LogicalIncreaseXminForSlot(XLogRecPtr current_lsn, TransactionId xmin) SpinLockRelease(&slot->mutex); if (got_new_xmin) - elog(DEBUG1, "got new catalog xmin %u at %X/%X", xmin, + elog(DEBUG1, "got new catalog xmin %u at %X/%08X", xmin, LSN_FORMAT_ARGS(current_lsn)); /* candidate already valid with the current flush position, apply */ @@ -1783,7 +1785,7 @@ LogicalIncreaseRestartDecodingForSlot(XLogRecPtr current_lsn, XLogRecPtr restart slot->candidate_restart_lsn = restart_lsn; SpinLockRelease(&slot->mutex); - elog(DEBUG1, "got new restart lsn %X/%X at %X/%X", + elog(DEBUG1, "got new restart lsn %X/%08X at %X/%08X", LSN_FORMAT_ARGS(restart_lsn), LSN_FORMAT_ARGS(current_lsn)); } @@ -1798,7 +1800,7 @@ LogicalIncreaseRestartDecodingForSlot(XLogRecPtr current_lsn, XLogRecPtr restart confirmed_flush = slot->data.confirmed_flush; SpinLockRelease(&slot->mutex); - elog(DEBUG1, "failed to increase restart lsn: proposed %X/%X, after %X/%X, current candidate %X/%X, current after %X/%X, flushed up to %X/%X", + elog(DEBUG1, "failed to increase restart lsn: proposed %X/%08X, after %X/%08X, current candidate %X/%08X, current after %X/%08X, flushed up to %X/%08X", LSN_FORMAT_ARGS(restart_lsn), LSN_FORMAT_ARGS(current_lsn), LSN_FORMAT_ARGS(candidate_restart_lsn), @@ -1825,9 +1827,13 @@ LogicalConfirmReceivedLocation(XLogRecPtr lsn) { bool updated_xmin = false; bool updated_restart = false; + XLogRecPtr restart_lsn pg_attribute_unused(); SpinLockAcquire(&MyReplicationSlot->mutex); + /* remember the old restart lsn */ + restart_lsn = MyReplicationSlot->data.restart_lsn; + /* * Prevent moving the confirmed_flush backwards, as this could lead to * data duplication issues caused by replicating already replicated @@ -1881,6 +1887,18 @@ LogicalConfirmReceivedLocation(XLogRecPtr lsn) /* first write new xmin to disk, so we know what's up after a crash */ if (updated_xmin || updated_restart) { +#ifdef USE_INJECTION_POINTS + XLogSegNo seg1, + seg2; + + XLByteToSeg(restart_lsn, seg1, wal_segment_size); + XLByteToSeg(MyReplicationSlot->data.restart_lsn, seg2, wal_segment_size); + + /* trigger injection point, but only if segment changes */ + if (seg1 != seg2) + INJECTION_POINT("logical-replication-slot-advance-segment", NULL); +#endif + ReplicationSlotMarkDirty(); ReplicationSlotSave(); elog(DEBUG1, "updated xmin: %u restart: %u", updated_xmin, updated_restart); diff --git a/src/backend/replication/logical/origin.c b/src/backend/replication/logical/origin.c index a17bacf88e7..87f10e50dcc 100644 --- a/src/backend/replication/logical/origin.c +++ b/src/backend/replication/logical/origin.c @@ -826,9 +826,9 @@ StartupReplicationOrigin(void) last_state++; ereport(LOG, - (errmsg("recovered replication state of node %d to %X/%X", - disk_state.roident, - LSN_FORMAT_ARGS(disk_state.remote_lsn)))); + errmsg("recovered replication state of node %d to %X/%08X", + disk_state.roident, + LSN_FORMAT_ARGS(disk_state.remote_lsn))); } /* now check checksum */ diff --git a/src/backend/replication/logical/proto.c b/src/backend/replication/logical/proto.c index 1a352b542dc..1b3d9eb49dd 100644 --- a/src/backend/replication/logical/proto.c +++ b/src/backend/replication/logical/proto.c @@ -809,7 +809,7 @@ logicalrep_write_tuple(StringInfo out, Relation rel, TupleTableSlot *slot, continue; } - if (att->attlen == -1 && VARATT_IS_EXTERNAL_ONDISK(values[i])) + if (att->attlen == -1 && VARATT_IS_EXTERNAL_ONDISK(DatumGetPointer(values[i]))) { /* * Unchanged toasted datum. (Note that we don't promise to detect diff --git a/src/backend/replication/logical/reorderbuffer.c b/src/backend/replication/logical/reorderbuffer.c index 67655111875..34cf05668ae 100644 --- a/src/backend/replication/logical/reorderbuffer.c +++ b/src/backend/replication/logical/reorderbuffer.c @@ -109,10 +109,22 @@ #include "storage/procarray.h" #include "storage/sinval.h" #include "utils/builtins.h" +#include "utils/inval.h" #include "utils/memutils.h" #include "utils/rel.h" #include "utils/relfilenumbermap.h" +/* + * Each transaction has an 8MB limit for invalidation messages distributed from + * other transactions. This limit is set considering scenarios with many + * concurrent logical decoding operations. When the distributed invalidation + * messages reach this threshold, the transaction is marked as + * RBTXN_DISTR_INVAL_OVERFLOWED to invalidate the complete cache as we have lost + * some inval messages and hence don't know what needs to be invalidated. + */ +#define MAX_DISTR_INVAL_MSG_PER_TXN \ + ((8 * 1024 * 1024) / sizeof(SharedInvalidationMessage)) + /* entry for a hash table we use to map from xid to our transaction state */ typedef struct ReorderBufferTXNByIdEnt { @@ -472,6 +484,12 @@ ReorderBufferFreeTXN(ReorderBuffer *rb, ReorderBufferTXN *txn) txn->invalidations = NULL; } + if (txn->invalidations_distributed) + { + pfree(txn->invalidations_distributed); + txn->invalidations_distributed = NULL; + } + /* Reset the toast hash */ ReorderBufferToastReset(rb, txn); @@ -1397,7 +1415,7 @@ ReorderBufferIterTXNNext(ReorderBuffer *rb, ReorderBufferIterTXNState *state) int32 off; /* nothing there anymore */ - if (state->heap->bh_size == 0) + if (binaryheap_empty(state->heap)) return NULL; off = DatumGetInt32(binaryheap_first(state->heap)); @@ -2581,7 +2599,7 @@ ReorderBufferProcessTXN(ReorderBuffer *rb, ReorderBufferTXN *txn, if (++changes_count >= CHANGES_THRESHOLD) { - rb->update_progress_txn(rb, txn, change->lsn); + rb->update_progress_txn(rb, txn, prev_lsn); changes_count = 0; } } @@ -2661,7 +2679,17 @@ ReorderBufferProcessTXN(ReorderBuffer *rb, ReorderBufferTXN *txn, AbortCurrentTransaction(); /* make sure there's no cache pollution */ - ReorderBufferExecuteInvalidations(txn->ninvalidations, txn->invalidations); + if (rbtxn_distr_inval_overflowed(txn)) + { + Assert(txn->ninvalidations_distributed == 0); + InvalidateSystemCaches(); + } + else + { + ReorderBufferExecuteInvalidations(txn->ninvalidations, txn->invalidations); + ReorderBufferExecuteInvalidations(txn->ninvalidations_distributed, + txn->invalidations_distributed); + } if (using_subtxn) RollbackAndReleaseCurrentSubTransaction(); @@ -2710,8 +2738,17 @@ ReorderBufferProcessTXN(ReorderBuffer *rb, ReorderBufferTXN *txn, AbortCurrentTransaction(); /* make sure there's no cache pollution */ - ReorderBufferExecuteInvalidations(txn->ninvalidations, - txn->invalidations); + if (rbtxn_distr_inval_overflowed(txn)) + { + Assert(txn->ninvalidations_distributed == 0); + InvalidateSystemCaches(); + } + else + { + ReorderBufferExecuteInvalidations(txn->ninvalidations, txn->invalidations); + ReorderBufferExecuteInvalidations(txn->ninvalidations_distributed, + txn->invalidations_distributed); + } if (using_subtxn) RollbackAndReleaseCurrentSubTransaction(); @@ -3060,7 +3097,8 @@ ReorderBufferAbort(ReorderBuffer *rb, TransactionId xid, XLogRecPtr lsn, * We might have decoded changes for this transaction that could load * the cache as per the current transaction's view (consider DDL's * happened in this transaction). We don't want the decoding of future - * transactions to use those cache entries so execute invalidations. + * transactions to use those cache entries so execute only the inval + * messages in this transaction. */ if (txn->ninvalidations > 0) ReorderBufferImmediateInvalidation(rb, txn->ninvalidations, @@ -3147,9 +3185,10 @@ ReorderBufferForget(ReorderBuffer *rb, TransactionId xid, XLogRecPtr lsn) txn->final_lsn = lsn; /* - * Process cache invalidation messages if there are any. Even if we're not - * interested in the transaction's contents, it could have manipulated the - * catalog and we need to update the caches according to that. + * Process only cache invalidation messages in this transaction if there + * are any. Even if we're not interested in the transaction's contents, it + * could have manipulated the catalog and we need to update the caches + * according to that. */ if (txn->base_snapshot != NULL && txn->ninvalidations > 0) ReorderBufferImmediateInvalidation(rb, txn->ninvalidations, @@ -3422,6 +3461,57 @@ ReorderBufferAddNewTupleCids(ReorderBuffer *rb, TransactionId xid, } /* + * Add new invalidation messages to the reorder buffer queue. + */ +static void +ReorderBufferQueueInvalidations(ReorderBuffer *rb, TransactionId xid, + XLogRecPtr lsn, Size nmsgs, + SharedInvalidationMessage *msgs) +{ + ReorderBufferChange *change; + + change = ReorderBufferAllocChange(rb); + change->action = REORDER_BUFFER_CHANGE_INVALIDATION; + change->data.inval.ninvalidations = nmsgs; + change->data.inval.invalidations = (SharedInvalidationMessage *) + palloc(sizeof(SharedInvalidationMessage) * nmsgs); + memcpy(change->data.inval.invalidations, msgs, + sizeof(SharedInvalidationMessage) * nmsgs); + + ReorderBufferQueueChange(rb, xid, lsn, change, false); +} + +/* + * A helper function for ReorderBufferAddInvalidations() and + * ReorderBufferAddDistributedInvalidations() to accumulate the invalidation + * messages to the **invals_out. + */ +static void +ReorderBufferAccumulateInvalidations(SharedInvalidationMessage **invals_out, + uint32 *ninvals_out, + SharedInvalidationMessage *msgs_new, + Size nmsgs_new) +{ + if (*ninvals_out == 0) + { + *ninvals_out = nmsgs_new; + *invals_out = (SharedInvalidationMessage *) + palloc(sizeof(SharedInvalidationMessage) * nmsgs_new); + memcpy(*invals_out, msgs_new, sizeof(SharedInvalidationMessage) * nmsgs_new); + } + else + { + /* Enlarge the array of inval messages */ + *invals_out = (SharedInvalidationMessage *) + repalloc(*invals_out, sizeof(SharedInvalidationMessage) * + (*ninvals_out + nmsgs_new)); + memcpy(*invals_out + *ninvals_out, msgs_new, + nmsgs_new * sizeof(SharedInvalidationMessage)); + *ninvals_out += nmsgs_new; + } +} + +/* * Accumulate the invalidations for executing them later. * * This needs to be called for each XLOG_XACT_INVALIDATIONS message and @@ -3441,7 +3531,6 @@ ReorderBufferAddInvalidations(ReorderBuffer *rb, TransactionId xid, { ReorderBufferTXN *txn; MemoryContext oldcontext; - ReorderBufferChange *change; txn = ReorderBufferTXNByXid(rb, xid, true, NULL, lsn, true); @@ -3456,35 +3545,76 @@ ReorderBufferAddInvalidations(ReorderBuffer *rb, TransactionId xid, Assert(nmsgs > 0); - /* Accumulate invalidations. */ - if (txn->ninvalidations == 0) - { - txn->ninvalidations = nmsgs; - txn->invalidations = (SharedInvalidationMessage *) - palloc(sizeof(SharedInvalidationMessage) * nmsgs); - memcpy(txn->invalidations, msgs, - sizeof(SharedInvalidationMessage) * nmsgs); - } - else + ReorderBufferAccumulateInvalidations(&txn->invalidations, + &txn->ninvalidations, + msgs, nmsgs); + + ReorderBufferQueueInvalidations(rb, xid, lsn, nmsgs, msgs); + + MemoryContextSwitchTo(oldcontext); +} + +/* + * Accumulate the invalidations distributed by other committed transactions + * for executing them later. + * + * This function is similar to ReorderBufferAddInvalidations() but stores + * the given inval messages to the txn->invalidations_distributed with the + * overflow check. + * + * This needs to be called by committed transactions to distribute their + * inval messages to in-progress transactions. + */ +void +ReorderBufferAddDistributedInvalidations(ReorderBuffer *rb, TransactionId xid, + XLogRecPtr lsn, Size nmsgs, + SharedInvalidationMessage *msgs) +{ + ReorderBufferTXN *txn; + MemoryContext oldcontext; + + txn = ReorderBufferTXNByXid(rb, xid, true, NULL, lsn, true); + + oldcontext = MemoryContextSwitchTo(rb->context); + + /* + * Collect all the invalidations under the top transaction, if available, + * so that we can execute them all together. See comments + * ReorderBufferAddInvalidations. + */ + txn = rbtxn_get_toptxn(txn); + + Assert(nmsgs > 0); + + if (!rbtxn_distr_inval_overflowed(txn)) { - txn->invalidations = (SharedInvalidationMessage *) - repalloc(txn->invalidations, sizeof(SharedInvalidationMessage) * - (txn->ninvalidations + nmsgs)); + /* + * Check the transaction has enough space for storing distributed + * invalidation messages. + */ + if (txn->ninvalidations_distributed + nmsgs >= MAX_DISTR_INVAL_MSG_PER_TXN) + { + /* + * Mark the invalidation message as overflowed and free up the + * messages accumulated so far. + */ + txn->txn_flags |= RBTXN_DISTR_INVAL_OVERFLOWED; - memcpy(txn->invalidations + txn->ninvalidations, msgs, - nmsgs * sizeof(SharedInvalidationMessage)); - txn->ninvalidations += nmsgs; + if (txn->invalidations_distributed) + { + pfree(txn->invalidations_distributed); + txn->invalidations_distributed = NULL; + txn->ninvalidations_distributed = 0; + } + } + else + ReorderBufferAccumulateInvalidations(&txn->invalidations_distributed, + &txn->ninvalidations_distributed, + msgs, nmsgs); } - change = ReorderBufferAllocChange(rb); - change->action = REORDER_BUFFER_CHANGE_INVALIDATION; - change->data.inval.ninvalidations = nmsgs; - change->data.inval.invalidations = (SharedInvalidationMessage *) - palloc(sizeof(SharedInvalidationMessage) * nmsgs); - memcpy(change->data.inval.invalidations, msgs, - sizeof(SharedInvalidationMessage) * nmsgs); - - ReorderBufferQueueChange(rb, xid, lsn, change, false); + /* Queue the invalidation messages into the transaction */ + ReorderBufferQueueInvalidations(rb, xid, lsn, nmsgs, msgs); MemoryContextSwitchTo(oldcontext); } @@ -4787,7 +4917,7 @@ StartupReorderBuffer(void) continue; /* if it cannot be a slot, skip the directory */ - if (!ReplicationSlotValidateName(logical_de->d_name, DEBUG2)) + if (!ReplicationSlotValidateName(logical_de->d_name, true, DEBUG2)) continue; /* diff --git a/src/backend/replication/logical/slotsync.c b/src/backend/replication/logical/slotsync.c index 656e66e0ae0..37738440113 100644 --- a/src/backend/replication/logical/slotsync.c +++ b/src/backend/replication/logical/slotsync.c @@ -211,9 +211,9 @@ update_local_synced_slot(RemoteSlot *remote_slot, Oid remote_dbid, * impact the users, so we used DEBUG1 level to log the message. */ ereport(slot->data.persistency == RS_TEMPORARY ? LOG : DEBUG1, - errmsg("could not synchronize replication slot \"%s\" because remote slot precedes local slot", + errmsg("could not synchronize replication slot \"%s\"", remote_slot->name), - errdetail("The remote slot has LSN %X/%X and catalog xmin %u, but the local slot has LSN %X/%X and catalog xmin %u.", + errdetail("Synchronization could lead to data loss, because the remote slot needs WAL at LSN %X/%08X and catalog xmin %u, but the standby has LSN %X/%08X and catalog xmin %u.", LSN_FORMAT_ARGS(remote_slot->restart_lsn), remote_slot->catalog_xmin, LSN_FORMAT_ARGS(slot->data.restart_lsn), @@ -275,7 +275,7 @@ update_local_synced_slot(RemoteSlot *remote_slot, Oid remote_dbid, ereport(ERROR, errmsg_internal("synchronized confirmed_flush for slot \"%s\" differs from remote slot", remote_slot->name), - errdetail_internal("Remote slot has LSN %X/%X but local slot has LSN %X/%X.", + errdetail_internal("Remote slot has LSN %X/%08X but local slot has LSN %X/%08X.", LSN_FORMAT_ARGS(remote_slot->confirmed_lsn), LSN_FORMAT_ARGS(slot->data.confirmed_flush))); } @@ -593,7 +593,7 @@ update_and_persist_local_synced_slot(RemoteSlot *remote_slot, Oid remote_dbid) { ereport(LOG, errmsg("could not synchronize replication slot \"%s\"", remote_slot->name), - errdetail("Logical decoding could not find consistent point from local slot's LSN %X/%X.", + errdetail("Synchronization could lead to data loss, because the standby could not build a consistent snapshot to decode WALs at LSN %X/%08X.", LSN_FORMAT_ARGS(slot->data.restart_lsn))); return false; @@ -642,7 +642,7 @@ synchronize_one_slot(RemoteSlot *remote_slot, Oid remote_dbid) ereport(AmLogicalSlotSyncWorkerProcess() ? LOG : ERROR, errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), errmsg("skipping slot synchronization because the received slot sync" - " LSN %X/%X for slot \"%s\" is ahead of the standby position %X/%X", + " LSN %X/%08X for slot \"%s\" is ahead of the standby position %X/%08X", LSN_FORMAT_ARGS(remote_slot->confirmed_lsn), remote_slot->name, LSN_FORMAT_ARGS(latestFlushPtr))); @@ -733,7 +733,7 @@ synchronize_one_slot(RemoteSlot *remote_slot, Oid remote_dbid) ereport(ERROR, errmsg_internal("cannot synchronize local slot \"%s\"", remote_slot->name), - errdetail_internal("Local slot's start streaming location LSN(%X/%X) is ahead of remote slot's LSN(%X/%X).", + errdetail_internal("Local slot's start streaming location LSN(%X/%08X) is ahead of remote slot's LSN(%X/%08X).", LSN_FORMAT_ARGS(slot->data.confirmed_flush), LSN_FORMAT_ARGS(remote_slot->confirmed_lsn))); @@ -1059,14 +1059,14 @@ ValidateSlotSyncParams(int elevel) { /* * Logical slot sync/creation requires wal_level >= logical. - * - * Since altering the wal_level requires a server restart, so error out in - * this case regardless of elevel provided by caller. */ if (wal_level < WAL_LEVEL_LOGICAL) - ereport(ERROR, + { + ereport(elevel, errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("replication slot synchronization requires \"wal_level\" >= \"logical\"")); + return false; + } /* * A physical replication slot(primary_slot_name) is required on the diff --git a/src/backend/replication/logical/snapbuild.c b/src/backend/replication/logical/snapbuild.c index 0d7bddbe4ed..8532bfd27e5 100644 --- a/src/backend/replication/logical/snapbuild.c +++ b/src/backend/replication/logical/snapbuild.c @@ -774,7 +774,7 @@ SnapBuildDistributeSnapshotAndInval(SnapBuild *builder, XLogRecPtr lsn, Transact if (rbtxn_is_prepared(txn)) continue; - elog(DEBUG2, "adding a new snapshot and invalidations to %u at %X/%X", + elog(DEBUG2, "adding a new snapshot and invalidations to %u at %X/%08X", txn->xid, LSN_FORMAT_ARGS(lsn)); /* @@ -794,6 +794,13 @@ SnapBuildDistributeSnapshotAndInval(SnapBuild *builder, XLogRecPtr lsn, Transact * contents built by the current transaction even after its decoding, * which should have been invalidated due to concurrent catalog * changing transaction. + * + * Distribute only the invalidation messages generated by the current + * committed transaction. Invalidation messages received from other + * transactions would have already been propagated to the relevant + * in-progress transactions. This transaction would have processed + * those invalidations, ensuring that subsequent transactions observe + * a consistent cache state. */ if (txn->xid != xid) { @@ -807,8 +814,9 @@ SnapBuildDistributeSnapshotAndInval(SnapBuild *builder, XLogRecPtr lsn, Transact { Assert(msgs != NULL); - ReorderBufferAddInvalidations(builder->reorder, txn->xid, lsn, - ninvalidations, msgs); + ReorderBufferAddDistributedInvalidations(builder->reorder, + txn->xid, lsn, + ninvalidations, msgs); } } } @@ -1263,10 +1271,10 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn builder->initial_xmin_horizon)) { ereport(DEBUG1, - (errmsg_internal("skipping snapshot at %X/%X while building logical decoding snapshot, xmin horizon too low", - LSN_FORMAT_ARGS(lsn)), - errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", - builder->initial_xmin_horizon, running->oldestRunningXid))); + errmsg_internal("skipping snapshot at %X/%08X while building logical decoding snapshot, xmin horizon too low", + LSN_FORMAT_ARGS(lsn)), + errdetail_internal("initial xmin horizon of %u vs the snapshot's %u", + builder->initial_xmin_horizon, running->oldestRunningXid)); SnapBuildWaitSnapshot(running, builder->initial_xmin_horizon); @@ -1302,9 +1310,9 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn builder->next_phase_at = InvalidTransactionId; ereport(LOG, - (errmsg("logical decoding found consistent point at %X/%X", - LSN_FORMAT_ARGS(lsn)), - errdetail("There are no running transactions."))); + errmsg("logical decoding found consistent point at %X/%08X", + LSN_FORMAT_ARGS(lsn)), + errdetail("There are no running transactions.")); return false; } @@ -1351,10 +1359,10 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn Assert(TransactionIdIsNormal(builder->xmax)); ereport(LOG, - (errmsg("logical decoding found initial starting point at %X/%X", - LSN_FORMAT_ARGS(lsn)), - errdetail("Waiting for transactions (approximately %d) older than %u to end.", - running->xcnt, running->nextXid))); + errmsg("logical decoding found initial starting point at %X/%08X", + LSN_FORMAT_ARGS(lsn)), + errdetail("Waiting for transactions (approximately %d) older than %u to end.", + running->xcnt, running->nextXid)); SnapBuildWaitSnapshot(running, running->nextXid); } @@ -1375,10 +1383,10 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn builder->next_phase_at = running->nextXid; ereport(LOG, - (errmsg("logical decoding found initial consistent point at %X/%X", - LSN_FORMAT_ARGS(lsn)), - errdetail("Waiting for transactions (approximately %d) older than %u to end.", - running->xcnt, running->nextXid))); + errmsg("logical decoding found initial consistent point at %X/%08X", + LSN_FORMAT_ARGS(lsn)), + errdetail("Waiting for transactions (approximately %d) older than %u to end.", + running->xcnt, running->nextXid)); SnapBuildWaitSnapshot(running, running->nextXid); } @@ -1399,9 +1407,9 @@ SnapBuildFindSnapshot(SnapBuild *builder, XLogRecPtr lsn, xl_running_xacts *runn builder->next_phase_at = InvalidTransactionId; ereport(LOG, - (errmsg("logical decoding found consistent point at %X/%X", - LSN_FORMAT_ARGS(lsn)), - errdetail("There are no old transactions anymore."))); + errmsg("logical decoding found consistent point at %X/%08X", + LSN_FORMAT_ARGS(lsn)), + errdetail("There are no old transactions anymore.")); } /* @@ -1905,9 +1913,9 @@ SnapBuildRestore(SnapBuild *builder, XLogRecPtr lsn) Assert(builder->state == SNAPBUILD_CONSISTENT); ereport(LOG, - (errmsg("logical decoding found consistent point at %X/%X", - LSN_FORMAT_ARGS(lsn)), - errdetail("Logical decoding will begin using saved snapshot."))); + errmsg("logical decoding found consistent point at %X/%08X", + LSN_FORMAT_ARGS(lsn)), + errdetail("Logical decoding will begin using saved snapshot.")); return true; snapshot_not_interesting: @@ -2053,7 +2061,7 @@ SnapBuildSnapshotExists(XLogRecPtr lsn) int ret; struct stat stat_buf; - sprintf(path, "%s/%X-%X.snap", + sprintf(path, "%s/%08X-%08X.snap", PG_LOGICAL_SNAPSHOTS_DIR, LSN_FORMAT_ARGS(lsn)); diff --git a/src/backend/replication/logical/tablesync.c b/src/backend/replication/logical/tablesync.c index 8e1e8762f62..d3356bc84ee 100644 --- a/src/backend/replication/logical/tablesync.c +++ b/src/backend/replication/logical/tablesync.c @@ -316,7 +316,8 @@ process_syncing_tables_for_sync(XLogRecPtr current_lsn) UpdateSubscriptionRelState(MyLogicalRepWorker->subid, MyLogicalRepWorker->relid, MyLogicalRepWorker->relstate, - MyLogicalRepWorker->relstate_lsn); + MyLogicalRepWorker->relstate_lsn, + false); /* * End streaming so that LogRepWorkerWalRcvConn can be used to drop @@ -425,6 +426,7 @@ process_syncing_tables_for_apply(XLogRecPtr current_lsn) ListCell *lc; bool started_tx = false; bool should_exit = false; + Relation rel = NULL; Assert(!IsTransactionState()); @@ -492,7 +494,17 @@ process_syncing_tables_for_apply(XLogRecPtr current_lsn) * worker to remove the origin tracking as if there is any * error while dropping we won't restart it to drop the * origin. So passing missing_ok = true. + * + * Lock the subscription and origin in the same order as we + * are doing during DDL commands to avoid deadlocks. See + * AlterSubscription_refresh. */ + LockSharedObject(SubscriptionRelationId, MyLogicalRepWorker->subid, + 0, AccessShareLock); + + if (!rel) + rel = table_open(SubscriptionRelRelationId, RowExclusiveLock); + ReplicationOriginNameForLogicalRep(MyLogicalRepWorker->subid, rstate->relid, originname, @@ -504,7 +516,7 @@ process_syncing_tables_for_apply(XLogRecPtr current_lsn) */ UpdateSubscriptionRelState(MyLogicalRepWorker->subid, rstate->relid, rstate->state, - rstate->lsn); + rstate->lsn, true); } } else @@ -555,7 +567,14 @@ process_syncing_tables_for_apply(XLogRecPtr current_lsn) * This is required to avoid any undetected deadlocks * due to any existing lock as deadlock detector won't * be able to detect the waits on the latch. + * + * Also close any tables prior to the commit. */ + if (rel) + { + table_close(rel, NoLock); + rel = NULL; + } CommitTransactionCommand(); pgstat_report_stat(false); } @@ -603,20 +622,31 @@ process_syncing_tables_for_apply(XLogRecPtr current_lsn) TimestampDifferenceExceeds(hentry->last_start_time, now, wal_retrieve_retry_interval)) { - logicalrep_worker_launch(WORKERTYPE_TABLESYNC, - MyLogicalRepWorker->dbid, - MySubscription->oid, - MySubscription->name, - MyLogicalRepWorker->userid, - rstate->relid, - DSM_HANDLE_INVALID); + /* + * Set the last_start_time even if we fail to start + * the worker, so that we won't retry until + * wal_retrieve_retry_interval has elapsed. + */ hentry->last_start_time = now; + (void) logicalrep_worker_launch(WORKERTYPE_TABLESYNC, + MyLogicalRepWorker->dbid, + MySubscription->oid, + MySubscription->name, + MyLogicalRepWorker->userid, + rstate->relid, + DSM_HANDLE_INVALID, + false); } } } } } + /* Close table if opened */ + if (rel) + table_close(rel, NoLock); + + if (started_tx) { /* @@ -1408,7 +1438,8 @@ LogicalRepSyncTableStart(XLogRecPtr *origin_startpos) UpdateSubscriptionRelState(MyLogicalRepWorker->subid, MyLogicalRepWorker->relid, MyLogicalRepWorker->relstate, - MyLogicalRepWorker->relstate_lsn); + MyLogicalRepWorker->relstate_lsn, + false); CommitTransactionCommand(); pgstat_report_stat(true); @@ -1541,14 +1572,15 @@ LogicalRepSyncTableStart(XLogRecPtr *origin_startpos) UpdateSubscriptionRelState(MyLogicalRepWorker->subid, MyLogicalRepWorker->relid, SUBREL_STATE_FINISHEDCOPY, - MyLogicalRepWorker->relstate_lsn); + MyLogicalRepWorker->relstate_lsn, + false); CommitTransactionCommand(); copy_table_done: elog(DEBUG1, - "LogicalRepSyncTableStart: '%s' origin_startpos lsn %X/%X", + "LogicalRepSyncTableStart: '%s' origin_startpos lsn %X/%08X", originname, LSN_FORMAT_ARGS(*origin_startpos)); /* diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c index a23262957ac..89e241c8392 100644 --- a/src/backend/replication/logical/worker.c +++ b/src/backend/replication/logical/worker.c @@ -109,13 +109,6 @@ * If ever a user needs to be aware of the tri-state value, they can fetch it * from the pg_subscription catalog (see column subtwophasestate). * - * We don't allow to toggle two_phase option of a subscription because it can - * lead to an inconsistent replica. Consider, initially, it was on and we have - * received some prepare then we turn it off, now at commit time the server - * will send the entire transaction data along with the commit. With some more - * analysis, we can allow changing this option from off to on but not sure if - * that alone would be useful. - * * Finally, to avoid problems mentioned in previous paragraphs from any * subsequent (not READY) tablesyncs (need to toggle two_phase option from 'on' * to 'off' and then again back to 'on') there is a restriction for @@ -139,6 +132,96 @@ * failover = true when creating the subscription. Enabling failover allows us * to smoothly transition to the promoted standby, ensuring that we can * subscribe to the new primary without losing any data. + * + * RETAIN DEAD TUPLES + * ---------------------- + * Each apply worker that enabled retain_dead_tuples option maintains a + * non-removable transaction ID (oldest_nonremovable_xid) in shared memory to + * prevent dead rows from being removed prematurely when the apply worker still + * needs them to detect update_deleted conflicts. Additionally, this helps to + * retain the required commit_ts module information, which further helps to + * detect update_origin_differs and delete_origin_differs conflicts reliably, as + * otherwise, vacuum freeze could remove the required information. + * + * The logical replication launcher manages an internal replication slot named + * "pg_conflict_detection". It asynchronously aggregates the non-removable + * transaction ID from all apply workers to determine the appropriate xmin for + * the slot, thereby retaining necessary tuples. + * + * The non-removable transaction ID in the apply worker is advanced to the + * oldest running transaction ID once all concurrent transactions on the + * publisher have been applied and flushed locally. The process involves: + * + * - RDT_GET_CANDIDATE_XID: + * Call GetOldestActiveTransactionId() to take oldestRunningXid as the + * candidate xid. + * + * - RDT_REQUEST_PUBLISHER_STATUS: + * Send a message to the walsender requesting the publisher status, which + * includes the latest WAL write position and information about transactions + * that are in the commit phase. + * + * - RDT_WAIT_FOR_PUBLISHER_STATUS: + * Wait for the status from the walsender. After receiving the first status, + * do not proceed if there are concurrent remote transactions that are still + * in the commit phase. These transactions might have been assigned an + * earlier commit timestamp but have not yet written the commit WAL record. + * Continue to request the publisher status (RDT_REQUEST_PUBLISHER_STATUS) + * until all these transactions have completed. + * + * - RDT_WAIT_FOR_LOCAL_FLUSH: + * Advance the non-removable transaction ID if the current flush location has + * reached or surpassed the last received WAL position. + * + * The overall state progression is: GET_CANDIDATE_XID -> + * REQUEST_PUBLISHER_STATUS -> WAIT_FOR_PUBLISHER_STATUS -> (loop to + * REQUEST_PUBLISHER_STATUS till concurrent remote transactions end) -> + * WAIT_FOR_LOCAL_FLUSH -> loop back to GET_CANDIDATE_XID. + * + * Retaining the dead tuples for this period is sufficient for ensuring + * eventual consistency using last-update-wins strategy, as dead tuples are + * useful for detecting conflicts only during the application of concurrent + * transactions from remote nodes. After applying and flushing all remote + * transactions that occurred concurrently with the tuple DELETE, any + * subsequent UPDATE from a remote node should have a later timestamp. In such + * cases, it is acceptable to detect an update_missing scenario and convert the + * UPDATE to an INSERT when applying it. But, for concurrent remote + * transactions with earlier timestamps than the DELETE, detecting + * update_deleted is necessary, as the UPDATEs in remote transactions should be + * ignored if their timestamp is earlier than that of the dead tuples. + * + * Note that advancing the non-removable transaction ID is not supported if the + * publisher is also a physical standby. This is because the logical walsender + * on the standby can only get the WAL replay position but there may be more + * WALs that are being replicated from the primary and those WALs could have + * earlier commit timestamp. + * + * Similarly, when the publisher has subscribed to another publisher, + * information necessary for conflict detection cannot be retained for + * changes from origins other than the publisher. This is because publisher + * lacks the information on concurrent transactions of other publishers to + * which it subscribes. As the information on concurrent transactions is + * unavailable beyond subscriber's immediate publishers, the non-removable + * transaction ID might be advanced prematurely before changes from other + * origins have been fully applied. + * + * XXX Retaining information for changes from other origins might be possible + * by requesting the subscription on that origin to enable retain_dead_tuples + * and fetching the conflict detection slot.xmin along with the publisher's + * status. In the RDT_WAIT_FOR_PUBLISHER_STATUS phase, the apply worker could + * wait for the remote slot's xmin to reach the oldest active transaction ID, + * ensuring that all transactions from other origins have been applied on the + * publisher, thereby getting the latest WAL position that includes all + * concurrent changes. However, this approach may impact performance, so it + * might not worth the effort. + * + * XXX It seems feasible to get the latest commit's WAL location from the + * publisher and wait till that is applied. However, we can't do that + * because commit timestamps can regress as a commit with a later LSN is not + * guaranteed to have a later timestamp than those with earlier LSNs. Having + * said that, even if that is possible, it won't improve performance much as + * the apply always lag and moves slowly as compared with the transactions + * on the publisher. *------------------------------------------------------------------------- */ @@ -147,6 +230,7 @@ #include <sys/stat.h> #include <unistd.h> +#include "access/commit_ts.h" #include "access/table.h" #include "access/tableam.h" #include "access/twophase.h" @@ -155,6 +239,7 @@ #include "catalog/pg_inherits.h" #include "catalog/pg_subscription.h" #include "catalog/pg_subscription_rel.h" +#include "commands/subscriptioncmds.h" #include "commands/tablecmds.h" #include "commands/trigger.h" #include "executor/executor.h" @@ -173,12 +258,14 @@ #include "replication/logicalrelation.h" #include "replication/logicalworker.h" #include "replication/origin.h" +#include "replication/slot.h" #include "replication/walreceiver.h" #include "replication/worker_internal.h" #include "rewrite/rewriteHandler.h" #include "storage/buffile.h" #include "storage/ipc.h" #include "storage/lmgr.h" +#include "storage/procarray.h" #include "tcop/tcopprot.h" #include "utils/acl.h" #include "utils/dynahash.h" @@ -275,6 +362,78 @@ typedef enum TRANS_PARALLEL_APPLY, } TransApplyAction; +/* + * The phases involved in advancing the non-removable transaction ID. + * + * See comments atop worker.c for details of the transition between these + * phases. + */ +typedef enum +{ + RDT_GET_CANDIDATE_XID, + RDT_REQUEST_PUBLISHER_STATUS, + RDT_WAIT_FOR_PUBLISHER_STATUS, + RDT_WAIT_FOR_LOCAL_FLUSH +} RetainDeadTuplesPhase; + +/* + * Critical information for managing phase transitions within the + * RetainDeadTuplesPhase. + */ +typedef struct RetainDeadTuplesData +{ + RetainDeadTuplesPhase phase; /* current phase */ + XLogRecPtr remote_lsn; /* WAL write position on the publisher */ + + /* + * Oldest transaction ID that was in the commit phase on the publisher. + * Use FullTransactionId to prevent issues with transaction ID wraparound, + * where a new remote_oldestxid could falsely appear to originate from the + * past and block advancement. + */ + FullTransactionId remote_oldestxid; + + /* + * Next transaction ID to be assigned on the publisher. Use + * FullTransactionId for consistency and to allow straightforward + * comparisons with remote_oldestxid. + */ + FullTransactionId remote_nextxid; + + TimestampTz reply_time; /* when the publisher responds with status */ + + /* + * Publisher transaction ID that must be awaited to complete before + * entering the final phase (RDT_WAIT_FOR_LOCAL_FLUSH). Use + * FullTransactionId for the same reason as remote_nextxid. + */ + FullTransactionId remote_wait_for; + + TransactionId candidate_xid; /* candidate for the non-removable + * transaction ID */ + TimestampTz flushpos_update_time; /* when the remote flush position was + * updated in final phase + * (RDT_WAIT_FOR_LOCAL_FLUSH) */ + + /* + * The following fields are used to determine the timing for the next + * round of transaction ID advancement. + */ + TimestampTz last_recv_time; /* when the last message was received */ + TimestampTz candidate_xid_time; /* when the candidate_xid is decided */ + int xid_advance_interval; /* how much time (ms) to wait before + * attempting to advance the + * non-removable transaction ID */ +} RetainDeadTuplesData; + +/* + * The minimum (100ms) and maximum (3 minutes) intervals for advancing + * non-removable transaction IDs. The maximum interval is a bit arbitrary but + * is sufficient to not cause any undue network traffic. + */ +#define MIN_XID_ADVANCE_INTERVAL 100 +#define MAX_XID_ADVANCE_INTERVAL 180000 + /* errcontext tracker */ static ApplyErrorCallbackArg apply_error_callback_arg = { @@ -339,6 +498,13 @@ static XLogRecPtr skip_xact_finish_lsn = InvalidXLogRecPtr; /* BufFile handle of the current streaming file */ static BufFile *stream_fd = NULL; +/* + * The remote WAL position that has been applied and flushed locally. We record + * and use this information both while sending feedback to the server and + * advancing oldest_nonremovable_xid. + */ +static XLogRecPtr last_flushpos = InvalidXLogRecPtr; + typedef struct SubXactInfo { TransactionId xid; /* XID of the subxact */ @@ -379,6 +545,19 @@ static void stream_close_file(void); static void send_feedback(XLogRecPtr recvpos, bool force, bool requestReply); +static void maybe_advance_nonremovable_xid(RetainDeadTuplesData *rdt_data, + bool status_received); +static bool can_advance_nonremovable_xid(RetainDeadTuplesData *rdt_data); +static void process_rdt_phase_transition(RetainDeadTuplesData *rdt_data, + bool status_received); +static void get_candidate_xid(RetainDeadTuplesData *rdt_data); +static void request_publisher_status(RetainDeadTuplesData *rdt_data); +static void wait_for_publisher_status(RetainDeadTuplesData *rdt_data, + bool status_received); +static void wait_for_local_flush(RetainDeadTuplesData *rdt_data); +static void adjust_xid_advance_interval(RetainDeadTuplesData *rdt_data, + bool new_xid_found); + static void apply_handle_commit_internal(LogicalRepCommitData *commit_data); static void apply_handle_insert_internal(ApplyExecutionData *edata, ResultRelInfo *relinfo, @@ -397,6 +576,12 @@ static bool FindReplTupleInLocalRel(ApplyExecutionData *edata, Relation localrel Oid localidxoid, TupleTableSlot *remoteslot, TupleTableSlot **localslot); +static bool FindDeletedTupleInLocalRel(Relation localrel, + Oid localidxoid, + TupleTableSlot *remoteslot, + TransactionId *delete_xid, + RepOriginId *delete_origin, + TimestampTz *delete_time); static void apply_handle_tuple_routing(ApplyExecutionData *edata, TupleTableSlot *remoteslot, LogicalRepTupleData *newtup, @@ -1023,7 +1208,7 @@ apply_handle_commit(StringInfo s) if (commit_data.commit_lsn != remote_final_lsn) ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), - errmsg_internal("incorrect commit LSN %X/%X in commit message (expected %X/%X)", + errmsg_internal("incorrect commit LSN %X/%08X in commit message (expected %X/%08X)", LSN_FORMAT_ARGS(commit_data.commit_lsn), LSN_FORMAT_ARGS(remote_final_lsn)))); @@ -1115,7 +1300,7 @@ apply_handle_prepare(StringInfo s) if (prepare_data.prepare_lsn != remote_final_lsn) ereport(ERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), - errmsg_internal("incorrect prepare LSN %X/%X in prepare message (expected %X/%X)", + errmsg_internal("incorrect prepare LSN %X/%08X in prepare message (expected %X/%08X)", LSN_FORMAT_ARGS(prepare_data.prepare_lsn), LSN_FORMAT_ARGS(remote_final_lsn)))); @@ -2733,17 +2918,31 @@ apply_handle_update_internal(ApplyExecutionData *edata, } else { + ConflictType type; TupleTableSlot *newslot = localslot; + /* + * Detecting whether the tuple was recently deleted or never existed + * is crucial to avoid misleading the user during confict handling. + */ + if (FindDeletedTupleInLocalRel(localrel, localindexoid, remoteslot, + &conflicttuple.xmin, + &conflicttuple.origin, + &conflicttuple.ts) && + conflicttuple.origin != replorigin_session_origin) + type = CT_UPDATE_DELETED; + else + type = CT_UPDATE_MISSING; + /* Store the new tuple for conflict reporting */ slot_store_data(newslot, relmapentry, newtup); /* - * The tuple to be updated could not be found. Do nothing except for - * emitting a log message. + * The tuple to be updated could not be found or was deleted. Do + * nothing except for emitting a log message. */ - ReportApplyConflict(estate, relinfo, LOG, CT_UPDATE_MISSING, - remoteslot, newslot, list_make1(&conflicttuple)); + ReportApplyConflict(estate, relinfo, LOG, type, remoteslot, newslot, + list_make1(&conflicttuple)); } /* Cleanup. */ @@ -2964,6 +3163,112 @@ FindReplTupleInLocalRel(ApplyExecutionData *edata, Relation localrel, } /* + * Determine whether the index can reliably locate the deleted tuple in the + * local relation. + * + * An index may exclude deleted tuples if it was re-indexed or re-created during + * change application. Therefore, an index is considered usable only if the + * conflict detection slot.xmin (conflict_detection_xmin) is greater than the + * index tuple's xmin. This ensures that any tuples deleted prior to the index + * creation or re-indexing are not relevant for conflict detection in the + * current apply worker. + * + * Note that indexes may also be excluded if they were modified by other DDL + * operations, such as ALTER INDEX. However, this is acceptable, as the + * likelihood of such DDL changes coinciding with the need to scan dead + * tuples for the update_deleted is low. + */ +static bool +IsIndexUsableForFindingDeletedTuple(Oid localindexoid, + TransactionId conflict_detection_xmin) +{ + HeapTuple index_tuple; + TransactionId index_xmin; + + index_tuple = SearchSysCache1(INDEXRELID, ObjectIdGetDatum(localindexoid)); + + if (!HeapTupleIsValid(index_tuple)) /* should not happen */ + elog(ERROR, "cache lookup failed for index %u", localindexoid); + + /* + * No need to check for a frozen transaction ID, as + * TransactionIdPrecedes() manages it internally, treating it as falling + * behind the conflict_detection_xmin. + */ + index_xmin = HeapTupleHeaderGetXmin(index_tuple->t_data); + + ReleaseSysCache(index_tuple); + + return TransactionIdPrecedes(index_xmin, conflict_detection_xmin); +} + +/* + * Attempts to locate a deleted tuple in the local relation that matches the + * values of the tuple received from the publication side (in 'remoteslot'). + * The search is performed using either the replica identity index, primary + * key, other available index, or a sequential scan if necessary. + * + * Returns true if the deleted tuple is found. If found, the transaction ID, + * origin, and commit timestamp of the deletion are stored in '*delete_xid', + * '*delete_origin', and '*delete_time' respectively. + */ +static bool +FindDeletedTupleInLocalRel(Relation localrel, Oid localidxoid, + TupleTableSlot *remoteslot, + TransactionId *delete_xid, RepOriginId *delete_origin, + TimestampTz *delete_time) +{ + TransactionId oldestxmin; + ReplicationSlot *slot; + + /* + * Return false if either dead tuples are not retained or commit timestamp + * data is not available. + */ + if (!MySubscription->retaindeadtuples || !track_commit_timestamp) + return false; + + /* + * For conflict detection, we use the conflict slot's xmin value instead + * of invoking GetOldestNonRemovableTransactionId(). The slot.xmin acts as + * a threshold to identify tuples that were recently deleted. These tuples + * are not visible to concurrent transactions, but we log an + * update_deleted conflict if such a tuple matches the remote update being + * applied. + * + * Although GetOldestNonRemovableTransactionId() can return a value older + * than the slot's xmin, for our current purpose it is acceptable to treat + * tuples deleted by transactions prior to slot.xmin as update_missing + * conflicts. + * + * Ideally, we would use oldest_nonremovable_xid, which is directly + * maintained by the leader apply worker. However, this value is not + * available to table synchronization or parallel apply workers, making + * slot.xmin a practical alternative in those contexts. + */ + slot = SearchNamedReplicationSlot(CONFLICT_DETECTION_SLOT, true); + + Assert(slot); + + SpinLockAcquire(&slot->mutex); + oldestxmin = slot->data.xmin; + SpinLockRelease(&slot->mutex); + + Assert(TransactionIdIsValid(oldestxmin)); + + if (OidIsValid(localidxoid) && + IsIndexUsableForFindingDeletedTuple(localidxoid, oldestxmin)) + return RelationFindDeletedTupleInfoByIndex(localrel, localidxoid, + remoteslot, oldestxmin, + delete_xid, delete_origin, + delete_time); + else + return RelationFindDeletedTupleInfoSeq(localrel, remoteslot, + oldestxmin, delete_xid, + delete_origin, delete_time); +} + +/* * This handles insert, update, delete on a partitioned table. */ static void @@ -3081,18 +3386,35 @@ apply_handle_tuple_routing(ApplyExecutionData *edata, remoteslot_part, &localslot); if (!found) { + ConflictType type; TupleTableSlot *newslot = localslot; + /* + * Detecting whether the tuple was recently deleted or + * never existed is crucial to avoid misleading the user + * during confict handling. + */ + if (FindDeletedTupleInLocalRel(partrel, + part_entry->localindexoid, + remoteslot_part, + &conflicttuple.xmin, + &conflicttuple.origin, + &conflicttuple.ts) && + conflicttuple.origin != replorigin_session_origin) + type = CT_UPDATE_DELETED; + else + type = CT_UPDATE_MISSING; + /* Store the new tuple for conflict reporting */ slot_store_data(newslot, part_entry, newtup); /* - * The tuple to be updated could not be found. Do nothing - * except for emitting a log message. + * The tuple to be updated could not be found or was + * deleted. Do nothing except for emitting a log message. */ ReportApplyConflict(estate, partrelinfo, LOG, - CT_UPDATE_MISSING, remoteslot_part, - newslot, list_make1(&conflicttuple)); + type, remoteslot_part, newslot, + list_make1(&conflicttuple)); return; } @@ -3584,6 +3906,7 @@ LogicalRepApplyLoop(XLogRecPtr last_received) bool ping_sent = false; TimeLineID tli; ErrorContextCallback errcallback; + RetainDeadTuplesData rdt_data = {0}; /* * Init the ApplyMessageContext which we clean up after each replication @@ -3662,6 +3985,8 @@ LogicalRepApplyLoop(XLogRecPtr last_received) last_recv_timestamp = GetCurrentTimestamp(); ping_sent = false; + rdt_data.last_recv_time = last_recv_timestamp; + /* Ensure we are reading the data into our memory context. */ MemoryContextSwitchTo(ApplyMessageContext); @@ -3688,6 +4013,8 @@ LogicalRepApplyLoop(XLogRecPtr last_received) UpdateWorkerStats(last_received, send_time, false); apply_dispatch(&s); + + maybe_advance_nonremovable_xid(&rdt_data, false); } else if (c == 'k') { @@ -3703,8 +4030,31 @@ LogicalRepApplyLoop(XLogRecPtr last_received) last_received = end_lsn; send_feedback(last_received, reply_requested, false); + + maybe_advance_nonremovable_xid(&rdt_data, false); + UpdateWorkerStats(last_received, timestamp, true); } + else if (c == 's') /* Primary status update */ + { + rdt_data.remote_lsn = pq_getmsgint64(&s); + rdt_data.remote_oldestxid = FullTransactionIdFromU64((uint64) pq_getmsgint64(&s)); + rdt_data.remote_nextxid = FullTransactionIdFromU64((uint64) pq_getmsgint64(&s)); + rdt_data.reply_time = pq_getmsgint64(&s); + + /* + * This should never happen, see + * ProcessStandbyPSRequestMessage. But if it happens + * due to a bug, we don't want to proceed as it can + * incorrectly advance oldest_nonremovable_xid. + */ + if (XLogRecPtrIsInvalid(rdt_data.remote_lsn)) + elog(ERROR, "cannot get the latest WAL position from the publisher"); + + maybe_advance_nonremovable_xid(&rdt_data, true); + + UpdateWorkerStats(last_received, rdt_data.reply_time, false); + } /* other message types are purposefully ignored */ MemoryContextReset(ApplyMessageContext); @@ -3717,6 +4067,11 @@ LogicalRepApplyLoop(XLogRecPtr last_received) /* confirm all writes so far */ send_feedback(last_received, false, false); + /* Reset the timestamp if no message was received */ + rdt_data.last_recv_time = 0; + + maybe_advance_nonremovable_xid(&rdt_data, false); + if (!in_remote_transaction && !in_streamed_transaction) { /* @@ -3751,6 +4106,14 @@ LogicalRepApplyLoop(XLogRecPtr last_received) else wait_time = NAPTIME_PER_CYCLE; + /* + * Ensure to wake up when it's possible to advance the non-removable + * transaction ID. + */ + if (rdt_data.phase == RDT_GET_CANDIDATE_XID && + rdt_data.xid_advance_interval) + wait_time = Min(wait_time, rdt_data.xid_advance_interval); + rc = WaitLatchOrSocket(MyLatch, WL_SOCKET_READABLE | WL_LATCH_SET | WL_TIMEOUT | WL_EXIT_ON_PM_DEATH, @@ -3814,6 +4177,8 @@ LogicalRepApplyLoop(XLogRecPtr last_received) send_feedback(last_received, requestReply, requestReply); + maybe_advance_nonremovable_xid(&rdt_data, false); + /* * Force reporting to ensure long idle periods don't lead to * arbitrarily delayed stats. Stats can only be reported outside @@ -3849,7 +4214,6 @@ send_feedback(XLogRecPtr recvpos, bool force, bool requestReply) static XLogRecPtr last_recvpos = InvalidXLogRecPtr; static XLogRecPtr last_writepos = InvalidXLogRecPtr; - static XLogRecPtr last_flushpos = InvalidXLogRecPtr; XLogRecPtr writepos; XLogRecPtr flushpos; @@ -3910,7 +4274,7 @@ send_feedback(XLogRecPtr recvpos, bool force, bool requestReply) pq_sendint64(reply_message, now); /* sendTime */ pq_sendbyte(reply_message, requestReply); /* replyRequested */ - elog(DEBUG2, "sending feedback (force %d) to recv %X/%X, write %X/%X, flush %X/%X", + elog(DEBUG2, "sending feedback (force %d) to recv %X/%08X, write %X/%08X, flush %X/%08X", force, LSN_FORMAT_ARGS(recvpos), LSN_FORMAT_ARGS(writepos), @@ -3928,6 +4292,368 @@ send_feedback(XLogRecPtr recvpos, bool force, bool requestReply) } /* + * Attempt to advance the non-removable transaction ID. + * + * See comments atop worker.c for details. + */ +static void +maybe_advance_nonremovable_xid(RetainDeadTuplesData *rdt_data, + bool status_received) +{ + if (!can_advance_nonremovable_xid(rdt_data)) + return; + + process_rdt_phase_transition(rdt_data, status_received); +} + +/* + * Preliminary check to determine if advancing the non-removable transaction ID + * is allowed. + */ +static bool +can_advance_nonremovable_xid(RetainDeadTuplesData *rdt_data) +{ + /* + * It is sufficient to manage non-removable transaction ID for a + * subscription by the main apply worker to detect update_deleted reliably + * even for table sync or parallel apply workers. + */ + if (!am_leader_apply_worker()) + return false; + + /* No need to advance if retaining dead tuples is not required */ + if (!MySubscription->retaindeadtuples) + return false; + + return true; +} + +/* + * Process phase transitions during the non-removable transaction ID + * advancement. See comments atop worker.c for details of the transition. + */ +static void +process_rdt_phase_transition(RetainDeadTuplesData *rdt_data, + bool status_received) +{ + switch (rdt_data->phase) + { + case RDT_GET_CANDIDATE_XID: + get_candidate_xid(rdt_data); + break; + case RDT_REQUEST_PUBLISHER_STATUS: + request_publisher_status(rdt_data); + break; + case RDT_WAIT_FOR_PUBLISHER_STATUS: + wait_for_publisher_status(rdt_data, status_received); + break; + case RDT_WAIT_FOR_LOCAL_FLUSH: + wait_for_local_flush(rdt_data); + break; + } +} + +/* + * Workhorse for the RDT_GET_CANDIDATE_XID phase. + */ +static void +get_candidate_xid(RetainDeadTuplesData *rdt_data) +{ + TransactionId oldest_running_xid; + TimestampTz now; + + /* + * Use last_recv_time when applying changes in the loop to avoid + * unnecessary system time retrieval. If last_recv_time is not available, + * obtain the current timestamp. + */ + now = rdt_data->last_recv_time ? rdt_data->last_recv_time : GetCurrentTimestamp(); + + /* + * Compute the candidate_xid and request the publisher status at most once + * per xid_advance_interval. Refer to adjust_xid_advance_interval() for + * details on how this value is dynamically adjusted. This is to avoid + * using CPU and network resources without making much progress. + */ + if (!TimestampDifferenceExceeds(rdt_data->candidate_xid_time, now, + rdt_data->xid_advance_interval)) + return; + + /* + * Immediately update the timer, even if the function returns later + * without setting candidate_xid due to inactivity on the subscriber. This + * avoids frequent calls to GetOldestActiveTransactionId. + */ + rdt_data->candidate_xid_time = now; + + /* + * Consider transactions in the current database, as only dead tuples from + * this database are required for conflict detection. + */ + oldest_running_xid = GetOldestActiveTransactionId(false, false); + + /* + * Oldest active transaction ID (oldest_running_xid) can't be behind any + * of its previously computed value. + */ + Assert(TransactionIdPrecedesOrEquals(MyLogicalRepWorker->oldest_nonremovable_xid, + oldest_running_xid)); + + /* Return if the oldest_nonremovable_xid cannot be advanced */ + if (TransactionIdEquals(MyLogicalRepWorker->oldest_nonremovable_xid, + oldest_running_xid)) + { + adjust_xid_advance_interval(rdt_data, false); + return; + } + + adjust_xid_advance_interval(rdt_data, true); + + rdt_data->candidate_xid = oldest_running_xid; + rdt_data->phase = RDT_REQUEST_PUBLISHER_STATUS; + + /* process the next phase */ + process_rdt_phase_transition(rdt_data, false); +} + +/* + * Workhorse for the RDT_REQUEST_PUBLISHER_STATUS phase. + */ +static void +request_publisher_status(RetainDeadTuplesData *rdt_data) +{ + static StringInfo request_message = NULL; + + if (!request_message) + { + MemoryContext oldctx = MemoryContextSwitchTo(ApplyContext); + + request_message = makeStringInfo(); + MemoryContextSwitchTo(oldctx); + } + else + resetStringInfo(request_message); + + /* + * Send the current time to update the remote walsender's latest reply + * message received time. + */ + pq_sendbyte(request_message, 'p'); + pq_sendint64(request_message, GetCurrentTimestamp()); + + elog(DEBUG2, "sending publisher status request message"); + + /* Send a request for the publisher status */ + walrcv_send(LogRepWorkerWalRcvConn, + request_message->data, request_message->len); + + rdt_data->phase = RDT_WAIT_FOR_PUBLISHER_STATUS; + + /* + * Skip calling maybe_advance_nonremovable_xid() since further transition + * is possible only once we receive the publisher status message. + */ +} + +/* + * Workhorse for the RDT_WAIT_FOR_PUBLISHER_STATUS phase. + */ +static void +wait_for_publisher_status(RetainDeadTuplesData *rdt_data, + bool status_received) +{ + /* + * Return if we have requested but not yet received the publisher status. + */ + if (!status_received) + return; + + if (!FullTransactionIdIsValid(rdt_data->remote_wait_for)) + rdt_data->remote_wait_for = rdt_data->remote_nextxid; + + /* + * Check if all remote concurrent transactions that were active at the + * first status request have now completed. If completed, proceed to the + * next phase; otherwise, continue checking the publisher status until + * these transactions finish. + * + * It's possible that transactions in the commit phase during the last + * cycle have now finished committing, but remote_oldestxid remains older + * than remote_wait_for. This can happen if some old transaction came in + * the commit phase when we requested status in this cycle. We do not + * handle this case explicitly as it's rare and the benefit doesn't + * justify the required complexity. Tracking would require either caching + * all xids at the publisher or sending them to subscribers. The condition + * will resolve naturally once the remaining transactions are finished. + * + * Directly advancing the non-removable transaction ID is possible if + * there are no activities on the publisher since the last advancement + * cycle. However, it requires maintaining two fields, last_remote_nextxid + * and last_remote_lsn, within the structure for comparison with the + * current cycle's values. Considering the minimal cost of continuing in + * RDT_WAIT_FOR_LOCAL_FLUSH without awaiting changes, we opted not to + * advance the transaction ID here. + */ + if (FullTransactionIdPrecedesOrEquals(rdt_data->remote_wait_for, + rdt_data->remote_oldestxid)) + rdt_data->phase = RDT_WAIT_FOR_LOCAL_FLUSH; + else + rdt_data->phase = RDT_REQUEST_PUBLISHER_STATUS; + + /* process the next phase */ + process_rdt_phase_transition(rdt_data, false); +} + +/* + * Workhorse for the RDT_WAIT_FOR_LOCAL_FLUSH phase. + */ +static void +wait_for_local_flush(RetainDeadTuplesData *rdt_data) +{ + Assert(!XLogRecPtrIsInvalid(rdt_data->remote_lsn) && + TransactionIdIsValid(rdt_data->candidate_xid)); + + /* + * We expect the publisher and subscriber clocks to be in sync using time + * sync service like NTP. Otherwise, we will advance this worker's + * oldest_nonremovable_xid prematurely, leading to the removal of rows + * required to detect update_deleted reliably. This check primarily + * addresses scenarios where the publisher's clock falls behind; if the + * publisher's clock is ahead, subsequent transactions will naturally bear + * later commit timestamps, conforming to the design outlined atop + * worker.c. + * + * XXX Consider waiting for the publisher's clock to catch up with the + * subscriber's before proceeding to the next phase. + */ + if (TimestampDifferenceExceeds(rdt_data->reply_time, + rdt_data->candidate_xid_time, 0)) + ereport(ERROR, + errmsg_internal("oldest_nonremovable_xid transaction ID could be advanced prematurely"), + errdetail_internal("The clock on the publisher is behind that of the subscriber.")); + + /* + * Do not attempt to advance the non-removable transaction ID when table + * sync is in progress. During this time, changes from a single + * transaction may be applied by multiple table sync workers corresponding + * to the target tables. So, it's necessary for all table sync workers to + * apply and flush the corresponding changes before advancing the + * transaction ID, otherwise, dead tuples that are still needed for + * conflict detection in table sync workers could be removed prematurely. + * However, confirming the apply and flush progress across all table sync + * workers is complex and not worth the effort, so we simply return if not + * all tables are in the READY state. + * + * It is safe to add new tables with initial states to the subscription + * after this check because any changes applied to these tables should + * have a WAL position greater than the rdt_data->remote_lsn. + */ + if (!AllTablesyncsReady()) + return; + + /* + * Update and check the remote flush position if we are applying changes + * in a loop. This is done at most once per WalWriterDelay to avoid + * performing costly operations in get_flush_position() too frequently + * during change application. + */ + if (last_flushpos < rdt_data->remote_lsn && rdt_data->last_recv_time && + TimestampDifferenceExceeds(rdt_data->flushpos_update_time, + rdt_data->last_recv_time, WalWriterDelay)) + { + XLogRecPtr writepos; + XLogRecPtr flushpos; + bool have_pending_txes; + + /* Fetch the latest remote flush position */ + get_flush_position(&writepos, &flushpos, &have_pending_txes); + + if (flushpos > last_flushpos) + last_flushpos = flushpos; + + rdt_data->flushpos_update_time = rdt_data->last_recv_time; + } + + /* Return to wait for the changes to be applied */ + if (last_flushpos < rdt_data->remote_lsn) + return; + + /* + * Reaching here means the remote WAL position has been received, and all + * transactions up to that position on the publisher have been applied and + * flushed locally. So, we can advance the non-removable transaction ID. + */ + SpinLockAcquire(&MyLogicalRepWorker->relmutex); + MyLogicalRepWorker->oldest_nonremovable_xid = rdt_data->candidate_xid; + SpinLockRelease(&MyLogicalRepWorker->relmutex); + + elog(DEBUG2, "confirmed flush up to remote lsn %X/%X: new oldest_nonremovable_xid %u", + LSN_FORMAT_ARGS(rdt_data->remote_lsn), + rdt_data->candidate_xid); + + /* Notify launcher to update the xmin of the conflict slot */ + ApplyLauncherWakeup(); + + /* + * Reset all data fields except those used to determine the timing for the + * next round of transaction ID advancement. We can even use + * flushpos_update_time in the next round to decide whether to get the + * latest flush position. + */ + rdt_data->phase = RDT_GET_CANDIDATE_XID; + rdt_data->remote_lsn = InvalidXLogRecPtr; + rdt_data->remote_oldestxid = InvalidFullTransactionId; + rdt_data->remote_nextxid = InvalidFullTransactionId; + rdt_data->reply_time = 0; + rdt_data->remote_wait_for = InvalidFullTransactionId; + rdt_data->candidate_xid = InvalidTransactionId; + + /* process the next phase */ + process_rdt_phase_transition(rdt_data, false); +} + +/* + * Adjust the interval for advancing non-removable transaction IDs. + * + * We double the interval to try advancing the non-removable transaction IDs + * if there is no activity on the node. The maximum value of the interval is + * capped by wal_receiver_status_interval if it is not zero, otherwise to a + * 3 minutes which should be sufficient to avoid using CPU or network + * resources without much benefit. + * + * The interval is reset to a minimum value of 100ms once there is some + * activity on the node. + * + * XXX The use of wal_receiver_status_interval is a bit arbitrary so we can + * consider the other interval or a separate GUC if the need arises. + */ +static void +adjust_xid_advance_interval(RetainDeadTuplesData *rdt_data, bool new_xid_found) +{ + if (!new_xid_found && rdt_data->xid_advance_interval) + { + int max_interval = wal_receiver_status_interval + ? wal_receiver_status_interval * 1000 + : MAX_XID_ADVANCE_INTERVAL; + + /* + * No new transaction ID has been assigned since the last check, so + * double the interval, but not beyond the maximum allowable value. + */ + rdt_data->xid_advance_interval = Min(rdt_data->xid_advance_interval * 2, + max_interval); + } + else + { + /* + * A new transaction ID was found or the interval is not yet + * initialized, so set the interval to the minimum value. + */ + rdt_data->xid_advance_interval = MIN_XID_ADVANCE_INTERVAL; + } +} + +/* * Exit routine for apply workers due to subscription parameter changes. */ static void @@ -4715,6 +5441,30 @@ InitializeLogRepWorker(void) apply_worker_exit(); } + /* + * Restart the worker if retain_dead_tuples was enabled during startup. + * + * At this point, the replication slot used for conflict detection might + * not exist yet, or could be dropped soon if the launcher perceives + * retain_dead_tuples as disabled. To avoid unnecessary tracking of + * oldest_nonremovable_xid when the slot is absent or at risk of being + * dropped, a restart is initiated. + * + * The oldest_nonremovable_xid should be initialized only when the + * retain_dead_tuples is enabled before launching the worker. See + * logicalrep_worker_launch. + */ + if (am_leader_apply_worker() && + MySubscription->retaindeadtuples && + !TransactionIdIsValid(MyLogicalRepWorker->oldest_nonremovable_xid)) + { + ereport(LOG, + errmsg("logical replication worker for subscription \"%s\" will restart because the option %s was enabled during startup", + MySubscription->name, "retain_dead_tuples")); + + apply_worker_exit(); + } + /* Setup synchronous commit according to the user's wishes */ SetConfigOption("synchronous_commit", MySubscription->synccommit, PGC_BACKEND, PGC_S_OVERRIDE); @@ -4871,6 +5621,14 @@ DisableSubscriptionAndExit(void) errmsg("subscription \"%s\" has been disabled because of an error", MySubscription->name)); + /* + * Skip the track_commit_timestamp check when disabling the worker due to + * an error, as verifying commit timestamps is unnecessary in this + * context. + */ + if (MySubscription->retaindeadtuples) + CheckSubDeadTupleRetention(false, true, WARNING); + proc_exit(0); } @@ -4916,7 +5674,7 @@ maybe_start_skipping_changes(XLogRecPtr finish_lsn) skip_xact_finish_lsn = finish_lsn; ereport(LOG, - errmsg("logical replication starts skipping transaction at LSN %X/%X", + errmsg("logical replication starts skipping transaction at LSN %X/%08X", LSN_FORMAT_ARGS(skip_xact_finish_lsn))); } @@ -4930,8 +5688,8 @@ stop_skipping_changes(void) return; ereport(LOG, - (errmsg("logical replication completed skipping transaction at LSN %X/%X", - LSN_FORMAT_ARGS(skip_xact_finish_lsn)))); + errmsg("logical replication completed skipping transaction at LSN %X/%08X", + LSN_FORMAT_ARGS(skip_xact_finish_lsn))); /* Stop skipping changes */ skip_xact_finish_lsn = InvalidXLogRecPtr; @@ -5019,7 +5777,7 @@ clear_subscription_skip_lsn(XLogRecPtr finish_lsn) if (myskiplsn != finish_lsn) ereport(WARNING, errmsg("skip-LSN of subscription \"%s\" cleared", MySubscription->name), - errdetail("Remote transaction's finish WAL location (LSN) %X/%X did not match skip-LSN %X/%X.", + errdetail("Remote transaction's finish WAL location (LSN) %X/%08X did not match skip-LSN %X/%08X.", LSN_FORMAT_ARGS(finish_lsn), LSN_FORMAT_ARGS(myskiplsn))); } @@ -5056,7 +5814,7 @@ apply_error_callback(void *arg) logicalrep_message_type(errarg->command), errarg->remote_xid); else - errcontext("processing remote data for replication origin \"%s\" during message type \"%s\" in transaction %u, finished at %X/%X", + errcontext("processing remote data for replication origin \"%s\" during message type \"%s\" in transaction %u, finished at %X/%08X", errarg->origin_name, logicalrep_message_type(errarg->command), errarg->remote_xid, @@ -5074,7 +5832,7 @@ apply_error_callback(void *arg) errarg->rel->remoterel.relname, errarg->remote_xid); else - errcontext("processing remote data for replication origin \"%s\" during message type \"%s\" for replication target relation \"%s.%s\" in transaction %u, finished at %X/%X", + errcontext("processing remote data for replication origin \"%s\" during message type \"%s\" for replication target relation \"%s.%s\" in transaction %u, finished at %X/%08X", errarg->origin_name, logicalrep_message_type(errarg->command), errarg->rel->remoterel.nspname, @@ -5093,7 +5851,7 @@ apply_error_callback(void *arg) errarg->rel->remoterel.attnames[errarg->remote_attnum], errarg->remote_xid); else - errcontext("processing remote data for replication origin \"%s\" during message type \"%s\" for replication target relation \"%s.%s\" column \"%s\" in transaction %u, finished at %X/%X", + errcontext("processing remote data for replication origin \"%s\" during message type \"%s\" for replication target relation \"%s.%s\" column \"%s\" in transaction %u, finished at %X/%08X", errarg->origin_name, logicalrep_message_type(errarg->command), errarg->rel->remoterel.nspname, diff --git a/src/backend/replication/pgoutput/pgoutput.c b/src/backend/replication/pgoutput/pgoutput.c index 693a766e6d7..80540c017bd 100644 --- a/src/backend/replication/pgoutput/pgoutput.c +++ b/src/backend/replication/pgoutput/pgoutput.c @@ -297,10 +297,12 @@ parse_output_parameters(List *options, PGOutputData *data) bool two_phase_option_given = false; bool origin_option_given = false; + /* Initialize optional parameters to defaults */ data->binary = false; data->streaming = LOGICALREP_STREAM_OFF; data->messages = false; data->two_phase = false; + data->publish_no_origin = false; foreach(lc, options) { @@ -1372,8 +1374,8 @@ pgoutput_row_filter(Relation relation, TupleTableSlot *old_slot, * VARTAG_INDIRECT. See ReorderBufferToastReplace. */ if (att->attlen == -1 && - VARATT_IS_EXTERNAL_ONDISK(new_slot->tts_values[i]) && - !VARATT_IS_EXTERNAL_ONDISK(old_slot->tts_values[i])) + VARATT_IS_EXTERNAL_ONDISK(DatumGetPointer(new_slot->tts_values[i])) && + !VARATT_IS_EXTERNAL_ONDISK(DatumGetPointer(old_slot->tts_values[i]))) { if (!tmp_new_slot) { @@ -1789,7 +1791,7 @@ LoadPublications(List *pubnames) else ereport(WARNING, errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("skipped loading publication: %s", pubname), + errmsg("skipped loading publication \"%s\"", pubname), errdetail("The publication does not exist at this point in the WAL."), errhint("Create the publication if it does not exist.")); } diff --git a/src/backend/replication/repl_gram.y b/src/backend/replication/repl_gram.y index 7440aae5a1a..8a649199ec6 100644 --- a/src/backend/replication/repl_gram.y +++ b/src/backend/replication/repl_gram.y @@ -279,7 +279,7 @@ alter_replication_slot: ; /* - * START_REPLICATION [SLOT slot] [PHYSICAL] %X/%X [TIMELINE %u] + * START_REPLICATION [SLOT slot] [PHYSICAL] %X/%08X [TIMELINE %u] */ start_replication: K_START_REPLICATION opt_slot opt_physical RECPTR opt_timeline @@ -295,7 +295,7 @@ start_replication: } ; -/* START_REPLICATION SLOT slot LOGICAL %X/%X options */ +/* START_REPLICATION SLOT slot LOGICAL %X/%08X options */ start_logical_replication: K_START_REPLICATION K_SLOT IDENT K_LOGICAL RECPTR plugin_options { diff --git a/src/backend/replication/repl_scanner.l b/src/backend/replication/repl_scanner.l index 014ea8d25c6..b6930e28659 100644 --- a/src/backend/replication/repl_scanner.l +++ b/src/backend/replication/repl_scanner.l @@ -155,7 +155,7 @@ UPLOAD_MANIFEST { return K_UPLOAD_MANIFEST; } {hexdigit}+\/{hexdigit}+ { uint32 hi, lo; - if (sscanf(yytext, "%X/%X", &hi, &lo) != 2) + if (sscanf(yytext, "%X/%08X", &hi, &lo) != 2) replication_yyerror(NULL, yyscanner, "invalid streaming start location"); yylval->recptr = ((uint64) hi) << 32 | lo; return RECPTR; diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c index 600b87fa9cb..8605776ad86 100644 --- a/src/backend/replication/slot.c +++ b/src/backend/replication/slot.c @@ -47,6 +47,7 @@ #include "miscadmin.h" #include "pgstat.h" #include "postmaster/interrupt.h" +#include "replication/logicallauncher.h" #include "replication/slotsync.h" #include "replication/slot.h" #include "replication/walsender_private.h" @@ -154,7 +155,7 @@ int max_replication_slots = 10; /* the maximum number of replication * Invalidate replication slots that have remained idle longer than this * duration; '0' disables it. */ -int idle_replication_slot_timeout_mins = 0; +int idle_replication_slot_timeout_secs = 0; /* * This GUC lists streaming replication standby server slot names that @@ -172,6 +173,7 @@ static SyncStandbySlotsConfigData *synchronized_standby_slots_config; static XLogRecPtr ss_oldest_flush_lsn = InvalidXLogRecPtr; static void ReplicationSlotShmemExit(int code, Datum arg); +static bool IsSlotForConflictCheck(const char *name); static void ReplicationSlotDropPtr(ReplicationSlot *slot); /* internal persistency functions */ @@ -258,13 +260,17 @@ ReplicationSlotShmemExit(int code, Datum arg) /* * Check whether the passed slot name is valid and report errors at elevel. * + * An error will be reported for a reserved replication slot name if + * allow_reserved_name is set to false. + * * Slot names may consist out of [a-z0-9_]{1,NAMEDATALEN-1} which should allow * the name to be used as a directory name on every supported OS. * * Returns whether the directory name is valid or not if elevel < ERROR. */ bool -ReplicationSlotValidateName(const char *name, int elevel) +ReplicationSlotValidateName(const char *name, bool allow_reserved_name, + int elevel) { const char *cp; @@ -300,10 +306,32 @@ ReplicationSlotValidateName(const char *name, int elevel) return false; } } + + if (!allow_reserved_name && IsSlotForConflictCheck(name)) + { + ereport(elevel, + errcode(ERRCODE_RESERVED_NAME), + errmsg("replication slot name \"%s\" is reserved", + name), + errdetail("The name \"%s\" is reserved for the conflict detection slot.", + CONFLICT_DETECTION_SLOT)); + + return false; + } + return true; } /* + * Return true if the replication slot name is "pg_conflict_detection". + */ +static bool +IsSlotForConflictCheck(const char *name) +{ + return (strcmp(name, CONFLICT_DETECTION_SLOT) == 0); +} + +/* * Create a new replication slot and mark it as used by this backend. * * name: Name of the slot @@ -330,7 +358,12 @@ ReplicationSlotCreate(const char *name, bool db_specific, Assert(MyReplicationSlot == NULL); - ReplicationSlotValidateName(name, ERROR); + /* + * The logical launcher or pg_upgrade may create or migrate an internal + * slot, so using a reserved name is allowed in these cases. + */ + ReplicationSlotValidateName(name, IsBinaryUpgrade || IsLogicalLauncher(), + ERROR); if (failover) { @@ -424,6 +457,7 @@ ReplicationSlotCreate(const char *name, bool db_specific, slot->candidate_restart_valid = InvalidXLogRecPtr; slot->candidate_restart_lsn = InvalidXLogRecPtr; slot->last_saved_confirmed_flush = InvalidXLogRecPtr; + slot->last_saved_restart_lsn = InvalidXLogRecPtr; slot->inactive_since = 0; /* @@ -581,6 +615,17 @@ retry: } /* + * Do not allow users to acquire the reserved slot. This scenario may + * occur if the launcher that owns the slot has terminated unexpectedly + * due to an error, and a backend process attempts to reuse the slot. + */ + if (!IsLogicalLauncher() && IsSlotForConflictCheck(name)) + ereport(ERROR, + errcode(ERRCODE_UNDEFINED_OBJECT), + errmsg("cannot acquire replication slot \"%s\"", name), + errdetail("The slot is reserved for conflict detection and can only be acquired by logical replication launcher.")); + + /* * This is the slot we want; check if it's active under some other * process. In single user mode, we don't need this check. */ @@ -1165,20 +1210,41 @@ ReplicationSlotsComputeRequiredLSN(void) { ReplicationSlot *s = &ReplicationSlotCtl->replication_slots[i]; XLogRecPtr restart_lsn; + XLogRecPtr last_saved_restart_lsn; bool invalidated; + ReplicationSlotPersistency persistency; if (!s->in_use) continue; SpinLockAcquire(&s->mutex); + persistency = s->data.persistency; restart_lsn = s->data.restart_lsn; invalidated = s->data.invalidated != RS_INVAL_NONE; + last_saved_restart_lsn = s->last_saved_restart_lsn; SpinLockRelease(&s->mutex); /* invalidated slots need not apply */ if (invalidated) continue; + /* + * For persistent slot use last_saved_restart_lsn to compute the + * oldest LSN for removal of WAL segments. The segments between + * last_saved_restart_lsn and restart_lsn might be needed by a + * persistent slot in the case of database crash. Non-persistent + * slots can't survive the database crash, so we don't care about + * last_saved_restart_lsn for them. + */ + if (persistency == RS_PERSISTENT) + { + if (last_saved_restart_lsn != InvalidXLogRecPtr && + restart_lsn > last_saved_restart_lsn) + { + restart_lsn = last_saved_restart_lsn; + } + } + if (restart_lsn != InvalidXLogRecPtr && (min_required == InvalidXLogRecPtr || restart_lsn < min_required)) @@ -1216,7 +1282,9 @@ ReplicationSlotsComputeLogicalRestartLSN(void) { ReplicationSlot *s; XLogRecPtr restart_lsn; + XLogRecPtr last_saved_restart_lsn; bool invalidated; + ReplicationSlotPersistency persistency; s = &ReplicationSlotCtl->replication_slots[i]; @@ -1230,14 +1298,33 @@ ReplicationSlotsComputeLogicalRestartLSN(void) /* read once, it's ok if it increases while we're checking */ SpinLockAcquire(&s->mutex); + persistency = s->data.persistency; restart_lsn = s->data.restart_lsn; invalidated = s->data.invalidated != RS_INVAL_NONE; + last_saved_restart_lsn = s->last_saved_restart_lsn; SpinLockRelease(&s->mutex); /* invalidated slots need not apply */ if (invalidated) continue; + /* + * For persistent slot use last_saved_restart_lsn to compute the + * oldest LSN for removal of WAL segments. The segments between + * last_saved_restart_lsn and restart_lsn might be needed by a + * persistent slot in the case of database crash. Non-persistent + * slots can't survive the database crash, so we don't care about + * last_saved_restart_lsn for them. + */ + if (persistency == RS_PERSISTENT) + { + if (last_saved_restart_lsn != InvalidXLogRecPtr && + restart_lsn > last_saved_restart_lsn) + { + restart_lsn = last_saved_restart_lsn; + } + } + if (restart_lsn == InvalidXLogRecPtr) continue; @@ -1455,6 +1542,7 @@ ReplicationSlotReserveWal(void) Assert(slot != NULL); Assert(slot->data.restart_lsn == InvalidXLogRecPtr); + Assert(slot->last_saved_restart_lsn == InvalidXLogRecPtr); /* * The replication slot mechanism is used to prevent removal of required @@ -1547,8 +1635,8 @@ ReportSlotInvalidation(ReplicationSlotInvalidationCause cause, uint64 ex = oldestLSN - restart_lsn; appendStringInfo(&err_detail, - ngettext("The slot's restart_lsn %X/%X exceeds the limit by %" PRIu64 " byte.", - "The slot's restart_lsn %X/%X exceeds the limit by %" PRIu64 " bytes.", + ngettext("The slot's restart_lsn %X/%08X exceeds the limit by %" PRIu64 " byte.", + "The slot's restart_lsn %X/%08X exceeds the limit by %" PRIu64 " bytes.", ex), LSN_FORMAT_ARGS(restart_lsn), ex); @@ -1568,13 +1656,10 @@ ReportSlotInvalidation(ReplicationSlotInvalidationCause cause, case RS_INVAL_IDLE_TIMEOUT: { - int minutes = slot_idle_seconds / SECS_PER_MINUTE; - int secs = slot_idle_seconds % SECS_PER_MINUTE; - /* translator: %s is a GUC variable name */ - appendStringInfo(&err_detail, _("The slot's idle time of %dmin %02ds exceeds the configured \"%s\" duration of %dmin."), - minutes, secs, "idle_replication_slot_timeout", - idle_replication_slot_timeout_mins); + appendStringInfo(&err_detail, _("The slot's idle time of %lds exceeds the configured \"%s\" duration of %ds."), + slot_idle_seconds, "idle_replication_slot_timeout", + idle_replication_slot_timeout_secs); /* translator: %s is a GUC variable name */ appendStringInfo(&err_hint, _("You might need to increase \"%s\"."), "idle_replication_slot_timeout"); @@ -1612,7 +1697,7 @@ ReportSlotInvalidation(ReplicationSlotInvalidationCause cause, static inline bool CanInvalidateIdleSlot(ReplicationSlot *s) { - return (idle_replication_slot_timeout_mins != 0 && + return (idle_replication_slot_timeout_secs != 0 && !XLogRecPtrIsInvalid(s->data.restart_lsn) && s->inactive_since > 0 && !(RecoveryInProgress() && s->data.synced)); @@ -1673,9 +1758,9 @@ DetermineSlotInvalidationCause(uint32 possible_causes, ReplicationSlot *s, if (CanInvalidateIdleSlot(s)) { /* - * We simulate the invalidation due to idle_timeout as the minimum - * time idle time is one minute which makes tests take a long - * time. + * Simulate the invalidation due to idle_timeout to test the + * timeout behavior promptly, without waiting for it to trigger + * naturally. */ #ifdef USE_INJECTION_POINTS if (IS_INJECTION_POINT_ATTACHED("slot-timeout-inval")) @@ -1690,7 +1775,7 @@ DetermineSlotInvalidationCause(uint32 possible_causes, ReplicationSlot *s, * idle_replication_slot_timeout GUC. */ if (TimestampDifferenceExceedsSeconds(s->inactive_since, now, - idle_replication_slot_timeout_mins * SECS_PER_MINUTE)) + idle_replication_slot_timeout_secs)) { *inactive_since = s->inactive_since; return RS_INVAL_IDLE_TIMEOUT; @@ -1835,7 +1920,10 @@ InvalidatePossiblyObsoleteSlot(uint32 possible_causes, * just rely on .invalidated. */ if (invalidation_cause == RS_INVAL_WAL_REMOVED) + { s->data.restart_lsn = InvalidXLogRecPtr; + s->last_saved_restart_lsn = InvalidXLogRecPtr; + } /* Let caller know */ *invalidated = true; @@ -1844,15 +1932,6 @@ InvalidatePossiblyObsoleteSlot(uint32 possible_causes, SpinLockRelease(&s->mutex); /* - * The logical replication slots shouldn't be invalidated as GUC - * max_slot_wal_keep_size is set to -1 and - * idle_replication_slot_timeout is set to 0 during the binary - * upgrade. See check_old_cluster_for_valid_slots() where we ensure - * that no invalidated before the upgrade. - */ - Assert(!(*invalidated && SlotIsLogical(s) && IsBinaryUpgrade)); - - /* * Calculate the idle time duration of the slot if slot is marked * invalidated with RS_INVAL_IDLE_TIMEOUT. */ @@ -1998,6 +2077,10 @@ restart: if (!s->in_use) continue; + /* Prevent invalidation of logical slots during binary upgrade */ + if (SlotIsLogical(s) && IsBinaryUpgrade) + continue; + if (InvalidatePossiblyObsoleteSlot(possible_causes, s, oldestLSN, dboid, snapshotConflictHorizon, &invalidated)) @@ -2032,6 +2115,7 @@ void CheckPointReplicationSlots(bool is_shutdown) { int i; + bool last_saved_restart_lsn_updated = false; elog(DEBUG1, "performing replication slot checkpoint"); @@ -2076,9 +2160,23 @@ CheckPointReplicationSlots(bool is_shutdown) SpinLockRelease(&s->mutex); } + /* + * Track if we're going to update slot's last_saved_restart_lsn. We + * need this to know if we need to recompute the required LSN. + */ + if (s->last_saved_restart_lsn != s->data.restart_lsn) + last_saved_restart_lsn_updated = true; + SaveSlotToPath(s, path, LOG); } LWLockRelease(ReplicationSlotAllocationLock); + + /* + * Recompute the required LSN if SaveSlotToPath() updated + * last_saved_restart_lsn for any slot. + */ + if (last_saved_restart_lsn_updated) + ReplicationSlotsComputeRequiredLSN(); } /* @@ -2354,6 +2452,7 @@ SaveSlotToPath(ReplicationSlot *slot, const char *dir, int elevel) if (!slot->just_dirtied) slot->dirty = false; slot->last_saved_confirmed_flush = cp.slotdata.confirmed_flush; + slot->last_saved_restart_lsn = cp.slotdata.restart_lsn; SpinLockRelease(&slot->mutex); LWLockRelease(&slot->io_in_progress_lock); @@ -2569,6 +2668,7 @@ RestoreSlotFromDisk(const char *name) slot->effective_xmin = cp.slotdata.xmin; slot->effective_catalog_xmin = cp.slotdata.catalog_xmin; slot->last_saved_confirmed_flush = cp.slotdata.confirmed_flush; + slot->last_saved_restart_lsn = cp.slotdata.restart_lsn; slot->candidate_catalog_xmin = InvalidTransactionId; slot->candidate_xmin_lsn = InvalidXLogRecPtr; @@ -2993,22 +3093,3 @@ WaitForStandbyConfirmation(XLogRecPtr wait_for_lsn) ConditionVariableCancelSleep(); } - -/* - * GUC check_hook for idle_replication_slot_timeout - * - * The value of idle_replication_slot_timeout must be set to 0 during - * a binary upgrade. See start_postmaster() in pg_upgrade for more details. - */ -bool -check_idle_replication_slot_timeout(int *newval, void **extra, GucSource source) -{ - if (IsBinaryUpgrade && *newval != 0) - { - GUC_check_errdetail("\"%s\" must be set to 0 during binary upgrade mode.", - "idle_replication_slot_timeout"); - return false; - } - - return true; -} diff --git a/src/backend/replication/slotfuncs.c b/src/backend/replication/slotfuncs.c index 36cc2ed4e44..69f4c6157c5 100644 --- a/src/backend/replication/slotfuncs.c +++ b/src/backend/replication/slotfuncs.c @@ -566,7 +566,7 @@ pg_replication_slot_advance(PG_FUNCTION_ARGS) if (moveto < minlsn) ereport(ERROR, (errcode(ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE), - errmsg("cannot advance replication slot to %X/%X, minimum is %X/%X", + errmsg("cannot advance replication slot to %X/%08X, minimum is %X/%08X", LSN_FORMAT_ARGS(moveto), LSN_FORMAT_ARGS(minlsn)))); /* Do the actual slot update, depending on the slot type */ diff --git a/src/backend/replication/syncrep.c b/src/backend/replication/syncrep.c index cc35984ad00..32cf3a48b89 100644 --- a/src/backend/replication/syncrep.c +++ b/src/backend/replication/syncrep.c @@ -258,7 +258,7 @@ SyncRepWaitForLSN(XLogRecPtr lsn, bool commit) { char buffer[32]; - sprintf(buffer, "waiting for %X/%X", LSN_FORMAT_ARGS(lsn)); + sprintf(buffer, "waiting for %X/%08X", LSN_FORMAT_ARGS(lsn)); set_ps_display_suffix(buffer); } @@ -566,7 +566,7 @@ SyncRepReleaseWaiters(void) LWLockRelease(SyncRepLock); - elog(DEBUG3, "released %d procs up to write %X/%X, %d procs up to flush %X/%X, %d procs up to apply %X/%X", + elog(DEBUG3, "released %d procs up to write %X/%08X, %d procs up to flush %X/%08X, %d procs up to apply %X/%08X", numwrite, LSN_FORMAT_ARGS(writePtr), numflush, LSN_FORMAT_ARGS(flushPtr), numapply, LSN_FORMAT_ARGS(applyPtr)); diff --git a/src/backend/replication/syncrep_scanner.l b/src/backend/replication/syncrep_scanner.l index 7dec1f869c7..02004d621e7 100644 --- a/src/backend/replication/syncrep_scanner.l +++ b/src/backend/replication/syncrep_scanner.l @@ -157,17 +157,16 @@ syncrep_yyerror(SyncRepConfigData **syncrep_parse_result_p, char **syncrep_parse { struct yyguts_t *yyg = (struct yyguts_t *) yyscanner; /* needed for yytext * macro */ - char *syncrep_parse_error_msg = *syncrep_parse_error_msg_p; /* report only the first error in a parse operation */ - if (syncrep_parse_error_msg) + if (*syncrep_parse_error_msg_p) return; if (yytext[0]) - syncrep_parse_error_msg = psprintf("%s at or near \"%s\"", - message, yytext); + *syncrep_parse_error_msg_p = psprintf("%s at or near \"%s\"", + message, yytext); else - syncrep_parse_error_msg = psprintf("%s at end of input", - message); + *syncrep_parse_error_msg_p = psprintf("%s at end of input", + message); } void diff --git a/src/backend/replication/walreceiver.c b/src/backend/replication/walreceiver.c index 8c4d0fd9aed..b6281101711 100644 --- a/src/backend/replication/walreceiver.c +++ b/src/backend/replication/walreceiver.c @@ -386,12 +386,12 @@ WalReceiverMain(const void *startup_data, size_t startup_data_len) { if (first_stream) ereport(LOG, - (errmsg("started streaming WAL from primary at %X/%X on timeline %u", - LSN_FORMAT_ARGS(startpoint), startpointTLI))); + errmsg("started streaming WAL from primary at %X/%08X on timeline %u", + LSN_FORMAT_ARGS(startpoint), startpointTLI)); else ereport(LOG, - (errmsg("restarted WAL streaming at %X/%X on timeline %u", - LSN_FORMAT_ARGS(startpoint), startpointTLI))); + errmsg("restarted WAL streaming at %X/%08X on timeline %u", + LSN_FORMAT_ARGS(startpoint), startpointTLI)); first_stream = false; /* Initialize LogstreamResult and buffers for processing messages */ @@ -470,7 +470,7 @@ WalReceiverMain(const void *startup_data, size_t startup_data_len) { ereport(LOG, (errmsg("replication terminated by primary server"), - errdetail("End of WAL reached on timeline %u at %X/%X.", + errdetail("End of WAL reached on timeline %u at %X/%08X.", startpointTLI, LSN_FORMAT_ARGS(LogstreamResult.Write)))); endofwal = true; @@ -711,7 +711,7 @@ WalRcvWaitForStartPosition(XLogRecPtr *startpoint, TimeLineID *startpointTLI) { char activitymsg[50]; - snprintf(activitymsg, sizeof(activitymsg), "restarting at %X/%X", + snprintf(activitymsg, sizeof(activitymsg), "restarting at %X/%08X", LSN_FORMAT_ARGS(*startpoint)); set_ps_display(activitymsg); } @@ -1014,7 +1014,7 @@ XLogWalRcvFlush(bool dying, TimeLineID tli) { char activitymsg[50]; - snprintf(activitymsg, sizeof(activitymsg), "streaming %X/%X", + snprintf(activitymsg, sizeof(activitymsg), "streaming %X/%08X", LSN_FORMAT_ARGS(LogstreamResult.Write)); set_ps_display(activitymsg); } @@ -1138,7 +1138,7 @@ XLogWalRcvSendReply(bool force, bool requestReply) pq_sendbyte(&reply_message, requestReply ? 1 : 0); /* Send it */ - elog(DEBUG2, "sending write %X/%X flush %X/%X apply %X/%X%s", + elog(DEBUG2, "sending write %X/%08X flush %X/%08X apply %X/%08X%s", LSN_FORMAT_ARGS(writePtr), LSN_FORMAT_ARGS(flushPtr), LSN_FORMAT_ARGS(applyPtr), diff --git a/src/backend/replication/walsender.c b/src/backend/replication/walsender.c index 9fa8beb6103..ee911394a23 100644 --- a/src/backend/replication/walsender.c +++ b/src/backend/replication/walsender.c @@ -65,6 +65,7 @@ #include "funcapi.h" #include "libpq/libpq.h" #include "libpq/pqformat.h" +#include "libpq/protocol.h" #include "miscadmin.h" #include "nodes/replnodes.h" #include "pgstat.h" @@ -84,6 +85,7 @@ #include "storage/ipc.h" #include "storage/pmsignal.h" #include "storage/proc.h" +#include "storage/procarray.h" #include "tcop/dest.h" #include "tcop/tcopprot.h" #include "utils/acl.h" @@ -258,6 +260,7 @@ static void StartLogicalReplication(StartReplicationCmd *cmd); static void ProcessStandbyMessage(void); static void ProcessStandbyReplyMessage(void); static void ProcessStandbyHSFeedbackMessage(void); +static void ProcessStandbyPSRequestMessage(void); static void ProcessRepliesIfAny(void); static void ProcessPendingWrites(void); static void WalSndKeepalive(bool requestReply, XLogRecPtr writePtr); @@ -408,7 +411,7 @@ IdentifySystem(void) else logptr = GetFlushRecPtr(&currTLI); - snprintf(xloc, sizeof(xloc), "%X/%X", LSN_FORMAT_ARGS(logptr)); + snprintf(xloc, sizeof(xloc), "%X/%08X", LSN_FORMAT_ARGS(logptr)); if (MyDatabaseId != InvalidOid) { @@ -515,7 +518,7 @@ ReadReplicationSlot(ReadReplicationSlotCmd *cmd) { char xloc[64]; - snprintf(xloc, sizeof(xloc), "%X/%X", + snprintf(xloc, sizeof(xloc), "%X/%08X", LSN_FORMAT_ARGS(slot_contents.data.restart_lsn)); values[i] = CStringGetTextDatum(xloc); nulls[i] = false; @@ -733,13 +736,13 @@ HandleUploadManifestPacket(StringInfo buf, off_t *offset, switch (mtype) { - case 'd': /* CopyData */ + case PqMsg_CopyData: maxmsglen = PQ_LARGE_MESSAGE_LIMIT; break; - case 'c': /* CopyDone */ - case 'f': /* CopyFail */ - case 'H': /* Flush */ - case 'S': /* Sync */ + case PqMsg_CopyDone: + case PqMsg_CopyFail: + case PqMsg_Flush: + case PqMsg_Sync: maxmsglen = PQ_SMALL_MESSAGE_LIMIT; break; default: @@ -761,19 +764,19 @@ HandleUploadManifestPacket(StringInfo buf, off_t *offset, /* Process the message */ switch (mtype) { - case 'd': /* CopyData */ + case PqMsg_CopyData: AppendIncrementalManifestData(ib, buf->data, buf->len); return true; - case 'c': /* CopyDone */ + case PqMsg_CopyDone: return false; - case 'H': /* Sync */ - case 'S': /* Flush */ + case PqMsg_Sync: + case PqMsg_Flush: /* Ignore these while in CopyOut mode as we do elsewhere. */ return true; - case 'f': + case PqMsg_CopyFail: ereport(ERROR, (errcode(ERRCODE_QUERY_CANCELED), errmsg("COPY from stdin failed: %s", @@ -892,12 +895,12 @@ StartReplication(StartReplicationCmd *cmd) switchpoint < cmd->startpoint) { ereport(ERROR, - (errmsg("requested starting point %X/%X on timeline %u is not in this server's history", - LSN_FORMAT_ARGS(cmd->startpoint), - cmd->timeline), - errdetail("This server's history forked from timeline %u at %X/%X.", - cmd->timeline, - LSN_FORMAT_ARGS(switchpoint)))); + errmsg("requested starting point %X/%08X on timeline %u is not in this server's history", + LSN_FORMAT_ARGS(cmd->startpoint), + cmd->timeline), + errdetail("This server's history forked from timeline %u at %X/%08X.", + cmd->timeline, + LSN_FORMAT_ARGS(switchpoint))); } sendTimeLineValidUpto = switchpoint; } @@ -939,9 +942,9 @@ StartReplication(StartReplicationCmd *cmd) if (FlushPtr < cmd->startpoint) { ereport(ERROR, - (errmsg("requested starting point %X/%X is ahead of the WAL flush position of this server %X/%X", - LSN_FORMAT_ARGS(cmd->startpoint), - LSN_FORMAT_ARGS(FlushPtr)))); + errmsg("requested starting point %X/%08X is ahead of the WAL flush position of this server %X/%08X", + LSN_FORMAT_ARGS(cmd->startpoint), + LSN_FORMAT_ARGS(FlushPtr))); } /* Start streaming from the requested point */ @@ -983,7 +986,7 @@ StartReplication(StartReplicationCmd *cmd) Datum values[2]; bool nulls[2] = {0}; - snprintf(startpos_str, sizeof(startpos_str), "%X/%X", + snprintf(startpos_str, sizeof(startpos_str), "%X/%08X", LSN_FORMAT_ARGS(sendTimeLineValidUpto)); dest = CreateDestReceiver(DestRemoteSimple); @@ -1324,7 +1327,7 @@ CreateReplicationSlot(CreateReplicationSlotCmd *cmd) ReplicationSlotPersist(); } - snprintf(xloc, sizeof(xloc), "%X/%X", + snprintf(xloc, sizeof(xloc), "%X/%08X", LSN_FORMAT_ARGS(MyReplicationSlot->data.confirmed_flush)); dest = CreateDestReceiver(DestRemoteSimple); @@ -1567,7 +1570,7 @@ WalSndWriteData(LogicalDecodingContext *ctx, XLogRecPtr lsn, TransactionId xid, tmpbuf.data, sizeof(int64)); /* output previously gathered data in a CopyData packet */ - pq_putmessage_noblock('d', ctx->out->data, ctx->out->len); + pq_putmessage_noblock(PqMsg_CopyData, ctx->out->data, ctx->out->len); CHECK_FOR_INTERRUPTS(); @@ -2303,7 +2306,7 @@ ProcessRepliesIfAny(void) case PqMsg_CopyDone: if (!streamingDoneSending) { - pq_putmessage_noblock('c', NULL, 0); + pq_putmessage_noblock(PqMsg_CopyDone, NULL, 0); streamingDoneSending = true; } @@ -2355,6 +2358,10 @@ ProcessStandbyMessage(void) ProcessStandbyHSFeedbackMessage(); break; + case 'p': + ProcessStandbyPSRequestMessage(); + break; + default: ereport(COMMERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), @@ -2429,7 +2436,7 @@ ProcessStandbyReplyMessage(void) /* Copy because timestamptz_to_str returns a static buffer */ replyTimeStr = pstrdup(timestamptz_to_str(replyTime)); - elog(DEBUG2, "write %X/%X flush %X/%X apply %X/%X%s reply_time %s", + elog(DEBUG2, "write %X/%08X flush %X/%08X apply %X/%08X%s reply_time %s", LSN_FORMAT_ARGS(writePtr), LSN_FORMAT_ARGS(flushPtr), LSN_FORMAT_ARGS(applyPtr), @@ -2702,6 +2709,60 @@ ProcessStandbyHSFeedbackMessage(void) } /* + * Process the request for a primary status update message. + */ +static void +ProcessStandbyPSRequestMessage(void) +{ + XLogRecPtr lsn = InvalidXLogRecPtr; + TransactionId oldestXidInCommit; + FullTransactionId nextFullXid; + FullTransactionId fullOldestXidInCommit; + WalSnd *walsnd = MyWalSnd; + TimestampTz replyTime; + + /* + * This shouldn't happen because we don't support getting primary status + * message from standby. + */ + if (RecoveryInProgress()) + elog(ERROR, "the primary status is unavailable during recovery"); + + replyTime = pq_getmsgint64(&reply_message); + + /* + * Update shared state for this WalSender process based on reply data from + * standby. + */ + SpinLockAcquire(&walsnd->mutex); + walsnd->replyTime = replyTime; + SpinLockRelease(&walsnd->mutex); + + /* + * Consider transactions in the current database, as only these are the + * ones replicated. + */ + oldestXidInCommit = GetOldestActiveTransactionId(true, false); + nextFullXid = ReadNextFullTransactionId(); + fullOldestXidInCommit = FullTransactionIdFromAllowableAt(nextFullXid, + oldestXidInCommit); + lsn = GetXLogWriteRecPtr(); + + elog(DEBUG2, "sending primary status"); + + /* construct the message... */ + resetStringInfo(&output_message); + pq_sendbyte(&output_message, 's'); + pq_sendint64(&output_message, lsn); + pq_sendint64(&output_message, (int64) U64FromFullTransactionId(fullOldestXidInCommit)); + pq_sendint64(&output_message, (int64) U64FromFullTransactionId(nextFullXid)); + pq_sendint64(&output_message, GetCurrentTimestamp()); + + /* ... and send it wrapped in CopyData */ + pq_putmessage_noblock(PqMsg_CopyData, output_message.data, output_message.len); +} + +/* * Compute how long send/receive loops should sleep. * * If wal_sender_timeout is enabled we want to wake up in time to send @@ -3246,12 +3307,12 @@ XLogSendPhysical(void) wal_segment_close(xlogreader); /* Send CopyDone */ - pq_putmessage_noblock('c', NULL, 0); + pq_putmessage_noblock(PqMsg_CopyDone, NULL, 0); streamingDoneSending = true; WalSndCaughtUp = true; - elog(DEBUG1, "walsender reached end of timeline at %X/%X (sent up to %X/%X)", + elog(DEBUG1, "walsender reached end of timeline at %X/%08X (sent up to %X/%08X)", LSN_FORMAT_ARGS(sendTimeLineValidUpto), LSN_FORMAT_ARGS(sentPtr)); return; @@ -3374,7 +3435,7 @@ retry: memcpy(&output_message.data[1 + sizeof(int64) + sizeof(int64)], tmpbuf.data, sizeof(int64)); - pq_putmessage_noblock('d', output_message.data, output_message.len); + pq_putmessage_noblock(PqMsg_CopyData, output_message.data, output_message.len); sentPtr = endptr; @@ -3392,7 +3453,7 @@ retry: { char activitymsg[50]; - snprintf(activitymsg, sizeof(activitymsg), "streaming %X/%X", + snprintf(activitymsg, sizeof(activitymsg), "streaming %X/%08X", LSN_FORMAT_ARGS(sentPtr)); set_ps_display(activitymsg); } @@ -3449,8 +3510,16 @@ XLogSendLogical(void) if (flushPtr == InvalidXLogRecPtr || logical_decoding_ctx->reader->EndRecPtr >= flushPtr) { + /* + * For cascading logical WAL senders, we use the replay LSN instead of + * the flush LSN, since logical decoding on a standby only processes + * WAL that has been replayed. This distinction becomes particularly + * important during shutdown, as new WAL is no longer replayed and the + * last replayed LSN marks the furthest point up to which decoding can + * proceed. + */ if (am_cascading_walsender) - flushPtr = GetStandbyFlushRecPtr(NULL); + flushPtr = GetXLogReplayRecPtr(NULL); else flushPtr = GetFlushRecPtr(NULL); } @@ -4072,7 +4141,7 @@ WalSndKeepalive(bool requestReply, XLogRecPtr writePtr) pq_sendbyte(&output_message, requestReply ? 1 : 0); /* ... and send it wrapped in CopyData */ - pq_putmessage_noblock('d', output_message.data, output_message.len); + pq_putmessage_noblock(PqMsg_CopyData, output_message.data, output_message.len); /* Set local flag */ if (requestReply) diff --git a/src/backend/rewrite/rewriteHandler.c b/src/backend/rewrite/rewriteHandler.c index 2ef0e7fbf3a..adc9e7600e1 100644 --- a/src/backend/rewrite/rewriteHandler.c +++ b/src/backend/rewrite/rewriteHandler.c @@ -923,8 +923,9 @@ rewriteTargetListIU(List *targetList, apply_default = true; /* - * Can only insert DEFAULT into generated columns, regardless of - * any OVERRIDING clauses. + * Can only insert DEFAULT into generated columns. (The + * OVERRIDING clause does not apply to generated columns, so we + * don't consider it here.) */ if (att_tup->attgenerated && !apply_default) { diff --git a/src/backend/statistics/mcv.c b/src/backend/statistics/mcv.c index d98cda698d9..f59fb821543 100644 --- a/src/backend/statistics/mcv.c +++ b/src/backend/statistics/mcv.c @@ -767,7 +767,7 @@ statext_mcv_serialize(MCVList *mcvlist, VacAttrStats **stats) values[dim][i] = PointerGetDatum(PG_DETOAST_DATUM(values[dim][i])); /* serialized length (uint32 length + data) */ - len = VARSIZE_ANY_EXHDR(values[dim][i]); + len = VARSIZE_ANY_EXHDR(DatumGetPointer(values[dim][i])); info[dim].nbytes += sizeof(uint32); /* length */ info[dim].nbytes += len; /* value (no header) */ diff --git a/src/backend/storage/aio/README.md b/src/backend/storage/aio/README.md index f10b5c7e31e..72ae3b3737d 100644 --- a/src/backend/storage/aio/README.md +++ b/src/backend/storage/aio/README.md @@ -94,7 +94,7 @@ pgaio_io_register_callbacks(ioh, PGAIO_HCB_SHARED_BUFFER_READV, 0); * * In this example we're reading only a single buffer, hence the 1. */ -pgaio_io_set_handle_data_32(ioh, (uint32 *) buffer, 1); +pgaio_io_set_handle_data_32(ioh, (uint32 *) &buffer, 1); /* * Pass the AIO handle to lower-level function. When operating on the level of @@ -119,8 +119,9 @@ pgaio_io_set_handle_data_32(ioh, (uint32 *) buffer, 1); * e.g. due to reaching a limit on the number of unsubmitted IOs, and even * complete before smgrstartreadv() returns. */ +void *page = BufferGetBlock(buffer); smgrstartreadv(ioh, operation->smgr, forknum, blkno, - BufferGetBlock(buffer), 1); + &page, 1); /* * To benefit from AIO, it is beneficial to perform other work, including diff --git a/src/backend/storage/aio/aio.c b/src/backend/storage/aio/aio.c index c64d815ebd1..3643f27ad6e 100644 --- a/src/backend/storage/aio/aio.c +++ b/src/backend/storage/aio/aio.c @@ -556,6 +556,13 @@ bool pgaio_io_was_recycled(PgAioHandle *ioh, uint64 ref_generation, PgAioHandleState *state) { *state = ioh->state; + + /* + * Ensure that we don't see an earlier state of the handle than ioh->state + * due to compiler or CPU reordering. This protects both ->generation as + * directly used here, and other fields in the handle accessed in the + * caller if the handle was not reused. + */ pg_read_barrier(); return ioh->generation != ref_generation; @@ -752,7 +759,7 @@ pgaio_io_wait_for_free(void) { int reclaimed = 0; - pgaio_debug(DEBUG2, "waiting for free IO with %d pending, %d in-flight, %d idle IOs", + pgaio_debug(DEBUG2, "waiting for free IO with %d pending, %u in-flight, %u idle IOs", pgaio_my_backend->num_staged_ios, dclist_count(&pgaio_my_backend->in_flight_ios), dclist_count(&pgaio_my_backend->idle_ios)); @@ -773,7 +780,12 @@ pgaio_io_wait_for_free(void) * Note that no interrupts are processed between the state check * and the call to reclaim - that's important as otherwise an * interrupt could have already reclaimed the handle. + * + * Need to ensure that there's no reordering, in the more common + * paths, where we wait for IO, that's done by + * pgaio_io_was_recycled(). */ + pg_read_barrier(); pgaio_io_reclaim(ioh); reclaimed++; } @@ -797,7 +809,7 @@ pgaio_io_wait_for_free(void) if (dclist_count(&pgaio_my_backend->in_flight_ios) == 0) ereport(ERROR, errmsg_internal("no free IOs despite no in-flight IOs"), - errdetail_internal("%d pending, %d in-flight, %d idle IOs", + errdetail_internal("%d pending, %u in-flight, %u idle IOs", pgaio_my_backend->num_staged_ios, dclist_count(&pgaio_my_backend->in_flight_ios), dclist_count(&pgaio_my_backend->idle_ios))); @@ -828,7 +840,7 @@ pgaio_io_wait_for_free(void) case PGAIO_HS_COMPLETED_IO: case PGAIO_HS_SUBMITTED: pgaio_debug_io(DEBUG2, ioh, - "waiting for free io with %d in flight", + "waiting for free io with %u in flight", dclist_count(&pgaio_my_backend->in_flight_ios)); /* @@ -852,7 +864,12 @@ pgaio_io_wait_for_free(void) * check and the call to reclaim - that's important as * otherwise an interrupt could have already reclaimed the * handle. + * + * Need to ensure that there's no reordering, in the more + * common paths, where we wait for IO, that's done by + * pgaio_io_was_recycled(). */ + pg_read_barrier(); pgaio_io_reclaim(ioh); break; } @@ -1252,7 +1269,7 @@ pgaio_closing_fd(int fd) break; pgaio_debug_io(DEBUG2, ioh, - "waiting for IO before FD %d gets closed, %d in-flight IOs", + "waiting for IO before FD %d gets closed, %u in-flight IOs", fd, dclist_count(&pgaio_my_backend->in_flight_ios)); /* see comment in pgaio_io_wait_for_free() about raciness */ @@ -1288,7 +1305,7 @@ pgaio_shutdown(int code, Datum arg) uint64 generation = ioh->generation; pgaio_debug_io(DEBUG2, ioh, - "waiting for IO to complete during shutdown, %d in-flight IOs", + "waiting for IO to complete during shutdown, %u in-flight IOs", dclist_count(&pgaio_my_backend->in_flight_ios)); /* see comment in pgaio_io_wait_for_free() about raciness */ diff --git a/src/backend/storage/aio/aio_callback.c b/src/backend/storage/aio/aio_callback.c index 0ad9795bb7e..03c9bba0802 100644 --- a/src/backend/storage/aio/aio_callback.c +++ b/src/backend/storage/aio/aio_callback.c @@ -256,6 +256,9 @@ pgaio_io_call_complete_shared(PgAioHandle *ioh) pgaio_result_status_string(result.status), result.id, result.error_data, result.result); result = ce->cb->complete_shared(ioh, result, cb_data); + + /* the callback should never transition to unknown */ + Assert(result.status != PGAIO_RS_UNKNOWN); } ioh->distilled_result = result; @@ -290,6 +293,7 @@ pgaio_io_call_complete_local(PgAioHandle *ioh) /* start with distilled result from shared callback */ result = ioh->distilled_result; + Assert(result.status != PGAIO_RS_UNKNOWN); for (int i = ioh->num_callbacks; i > 0; i--) { @@ -306,6 +310,9 @@ pgaio_io_call_complete_local(PgAioHandle *ioh) pgaio_result_status_string(result.status), result.id, result.error_data, result.result); result = ce->cb->complete_local(ioh, result, cb_data); + + /* the callback should never transition to unknown */ + Assert(result.status != PGAIO_RS_UNKNOWN); } /* diff --git a/src/backend/storage/aio/method_io_uring.c b/src/backend/storage/aio/method_io_uring.c index c719ba2727a..0a8c054162f 100644 --- a/src/backend/storage/aio/method_io_uring.c +++ b/src/backend/storage/aio/method_io_uring.c @@ -29,6 +29,9 @@ #ifdef IOMETHOD_IO_URING_ENABLED +#include <sys/mman.h> +#include <unistd.h> + #include <liburing.h> #include "miscadmin.h" @@ -94,12 +97,32 @@ PgAioUringContext struct io_uring io_uring_ring; } PgAioUringContext; +/* + * Information about the capabilities that io_uring has. + * + * Depending on liburing and kernel version different features are + * supported. At least for the kernel a kernel version check does not suffice + * as various vendors do backport features to older kernels :(. + */ +typedef struct PgAioUringCaps +{ + bool checked; + /* -1 if io_uring_queue_init_mem() is unsupported */ + int mem_init_size; +} PgAioUringCaps; + + /* PgAioUringContexts for all backends */ static PgAioUringContext *pgaio_uring_contexts; /* the current backend's context */ static PgAioUringContext *pgaio_my_uring_context; +static PgAioUringCaps pgaio_uring_caps = +{ + .checked = false, + .mem_init_size = -1, +}; static uint32 pgaio_uring_procs(void) @@ -111,30 +134,184 @@ pgaio_uring_procs(void) return MaxBackends + NUM_AUXILIARY_PROCS - MAX_IO_WORKERS; } -static Size +/* + * Initializes pgaio_uring_caps, unless that's already done. + */ +static void +pgaio_uring_check_capabilities(void) +{ + if (pgaio_uring_caps.checked) + return; + + /* + * By default io_uring creates a shared memory mapping for each io_uring + * instance, leading to a large number of memory mappings. Unfortunately a + * large number of memory mappings slows things down, backend exit is + * particularly affected. To address that, newer kernels (6.5) support + * using user-provided memory for the memory, by putting the relevant + * memory into shared memory we don't need any additional mappings. + * + * To know whether this is supported, we unfortunately need to probe the + * kernel by trying to create a ring with userspace-provided memory. This + * also has a secondary benefit: We can determine precisely how much + * memory we need for each io_uring instance. + */ +#if defined(HAVE_LIBURING_QUEUE_INIT_MEM) && defined(IORING_SETUP_NO_MMAP) + { + struct io_uring test_ring; + size_t ring_size; + void *ring_ptr; + struct io_uring_params p = {0}; + int ret; + + /* + * Liburing does not yet provide an API to query how much memory a + * ring will need. So we over-estimate it here. As the memory is freed + * just below that's small temporary waste of memory. + * + * 1MB is more than enough for rings within io_max_concurrency's + * range. + */ + ring_size = 1024 * 1024; + + /* + * Hard to believe a system exists where 1MB would not be a multiple + * of the page size. But it's cheap to ensure... + */ + ring_size -= ring_size % sysconf(_SC_PAGESIZE); + + ring_ptr = mmap(NULL, ring_size, PROT_READ | PROT_WRITE, MAP_SHARED | MAP_ANONYMOUS, -1, 0); + if (ring_ptr == MAP_FAILED) + elog(ERROR, + "mmap(%zu) to determine io_uring_queue_init_mem() support failed: %m", + ring_size); + + ret = io_uring_queue_init_mem(io_max_concurrency, &test_ring, &p, ring_ptr, ring_size); + if (ret > 0) + { + pgaio_uring_caps.mem_init_size = ret; + + elog(DEBUG1, + "can use combined memory mapping for io_uring, each ring needs %d bytes", + ret); + + /* clean up the created ring, it was just for a test */ + io_uring_queue_exit(&test_ring); + } + else + { + /* + * There are different reasons for ring creation to fail, but it's + * ok to treat that just as io_uring_queue_init_mem() not being + * supported. We'll report a more detailed error in + * pgaio_uring_shmem_init(). + */ + errno = -ret; + elog(DEBUG1, + "cannot use combined memory mapping for io_uring, ring creation failed: %m"); + + } + + if (munmap(ring_ptr, ring_size) != 0) + elog(ERROR, "munmap() failed: %m"); + } +#else + { + elog(DEBUG1, + "can't use combined memory mapping for io_uring, kernel or liburing too old"); + } +#endif + + pgaio_uring_caps.checked = true; +} + +/* + * Memory for all PgAioUringContext instances + */ +static size_t pgaio_uring_context_shmem_size(void) { return mul_size(pgaio_uring_procs(), sizeof(PgAioUringContext)); } +/* + * Memory for the combined memory used by io_uring instances. Returns 0 if + * that is not supported by kernel/liburing. + */ +static size_t +pgaio_uring_ring_shmem_size(void) +{ + size_t sz = 0; + + if (pgaio_uring_caps.mem_init_size > 0) + { + /* + * Memory for rings needs to be allocated to the page boundary, + * reserve space. Luckily it does not need to be aligned to hugepage + * boundaries, even if huge pages are used. + */ + sz = add_size(sz, sysconf(_SC_PAGESIZE)); + sz = add_size(sz, mul_size(pgaio_uring_procs(), + pgaio_uring_caps.mem_init_size)); + } + + return sz; +} + static size_t pgaio_uring_shmem_size(void) { - return pgaio_uring_context_shmem_size(); + size_t sz; + + /* + * Kernel and liburing support for various features influences how much + * shmem we need, perform the necessary checks. + */ + pgaio_uring_check_capabilities(); + + sz = pgaio_uring_context_shmem_size(); + sz = add_size(sz, pgaio_uring_ring_shmem_size()); + + return sz; } static void pgaio_uring_shmem_init(bool first_time) { - int TotalProcs = MaxBackends + NUM_AUXILIARY_PROCS - MAX_IO_WORKERS; + int TotalProcs = pgaio_uring_procs(); bool found; + char *shmem; + size_t ring_mem_remain = 0; + char *ring_mem_next = 0; - pgaio_uring_contexts = (PgAioUringContext *) - ShmemInitStruct("AioUring", pgaio_uring_shmem_size(), &found); - + /* + * We allocate memory for all PgAioUringContext instances and, if + * supported, the memory required for each of the io_uring instances, in + * one ShmemInitStruct(). + */ + shmem = ShmemInitStruct("AioUringContext", pgaio_uring_shmem_size(), &found); if (found) return; + pgaio_uring_contexts = (PgAioUringContext *) shmem; + shmem += pgaio_uring_context_shmem_size(); + + /* if supported, handle memory alignment / sizing for io_uring memory */ + if (pgaio_uring_caps.mem_init_size > 0) + { + ring_mem_remain = pgaio_uring_ring_shmem_size(); + ring_mem_next = (char *) shmem; + + /* align to page boundary, see also pgaio_uring_ring_shmem_size() */ + ring_mem_next = (char *) TYPEALIGN(sysconf(_SC_PAGESIZE), ring_mem_next); + + /* account for alignment */ + ring_mem_remain -= ring_mem_next - shmem; + shmem += ring_mem_next - shmem; + + shmem += ring_mem_remain; + } + for (int contextno = 0; contextno < TotalProcs; contextno++) { PgAioUringContext *context = &pgaio_uring_contexts[contextno]; @@ -158,7 +335,28 @@ pgaio_uring_shmem_init(bool first_time) * be worth using that - also need to evaluate if that causes * noticeable additional contention? */ - ret = io_uring_queue_init(io_max_concurrency, &context->io_uring_ring, 0); + + /* + * If supported (c.f. pgaio_uring_check_capabilities()), create ring + * with its data in shared memory. Otherwise fall back io_uring + * creating a memory mapping for each ring. + */ +#if defined(HAVE_LIBURING_QUEUE_INIT_MEM) && defined(IORING_SETUP_NO_MMAP) + if (pgaio_uring_caps.mem_init_size > 0) + { + struct io_uring_params p = {0}; + + ret = io_uring_queue_init_mem(io_max_concurrency, &context->io_uring_ring, &p, ring_mem_next, ring_mem_remain); + + ring_mem_remain -= ret; + ring_mem_next += ret; + } + else +#endif + { + ret = io_uring_queue_init(io_max_concurrency, &context->io_uring_ring, 0); + } + if (ret < 0) { char *hint = NULL; @@ -400,9 +598,9 @@ pgaio_uring_wait_one(PgAioHandle *ioh, uint64 ref_generation) while (true) { pgaio_debug_io(DEBUG3, ioh, - "wait_one io_gen: %llu, ref_gen: %llu, cycle %d", - (long long unsigned) ioh->generation, - (long long unsigned) ref_generation, + "wait_one io_gen: %" PRIu64 ", ref_gen: %" PRIu64 ", cycle %d", + ioh->generation, + ref_generation, waited); if (pgaio_io_was_recycled(ioh, ref_generation, &state) || diff --git a/src/backend/storage/aio/method_worker.c b/src/backend/storage/aio/method_worker.c index 743cccc2acd..bf8f77e6ff6 100644 --- a/src/backend/storage/aio/method_worker.c +++ b/src/backend/storage/aio/method_worker.c @@ -52,26 +52,26 @@ #define IO_WORKER_WAKEUP_FANOUT 2 -typedef struct AioWorkerSubmissionQueue +typedef struct PgAioWorkerSubmissionQueue { uint32 size; uint32 mask; uint32 head; uint32 tail; - uint32 ios[FLEXIBLE_ARRAY_MEMBER]; -} AioWorkerSubmissionQueue; + uint32 sqes[FLEXIBLE_ARRAY_MEMBER]; +} PgAioWorkerSubmissionQueue; -typedef struct AioWorkerSlot +typedef struct PgAioWorkerSlot { Latch *latch; bool in_use; -} AioWorkerSlot; +} PgAioWorkerSlot; -typedef struct AioWorkerControl +typedef struct PgAioWorkerControl { uint64 idle_worker_mask; - AioWorkerSlot workers[FLEXIBLE_ARRAY_MEMBER]; -} AioWorkerControl; + PgAioWorkerSlot workers[FLEXIBLE_ARRAY_MEMBER]; +} PgAioWorkerControl; static size_t pgaio_worker_shmem_size(void); @@ -96,8 +96,8 @@ int io_workers = 3; static int io_worker_queue_size = 64; static int MyIoWorkerId; -static AioWorkerSubmissionQueue *io_worker_submission_queue; -static AioWorkerControl *io_worker_control; +static PgAioWorkerSubmissionQueue *io_worker_submission_queue; +static PgAioWorkerControl *io_worker_control; static size_t @@ -106,15 +106,15 @@ pgaio_worker_queue_shmem_size(int *queue_size) /* Round size up to next power of two so we can make a mask. */ *queue_size = pg_nextpower2_32(io_worker_queue_size); - return offsetof(AioWorkerSubmissionQueue, ios) + + return offsetof(PgAioWorkerSubmissionQueue, sqes) + sizeof(uint32) * *queue_size; } static size_t pgaio_worker_control_shmem_size(void) { - return offsetof(AioWorkerControl, workers) + - sizeof(AioWorkerSlot) * MAX_IO_WORKERS; + return offsetof(PgAioWorkerControl, workers) + + sizeof(PgAioWorkerSlot) * MAX_IO_WORKERS; } static size_t @@ -162,7 +162,7 @@ pgaio_worker_shmem_init(bool first_time) } static int -pgaio_choose_idle_worker(void) +pgaio_worker_choose_idle(void) { int worker; @@ -172,6 +172,7 @@ pgaio_choose_idle_worker(void) /* Find the lowest bit position, and clear it. */ worker = pg_rightmost_one_pos64(io_worker_control->idle_worker_mask); io_worker_control->idle_worker_mask &= ~(UINT64_C(1) << worker); + Assert(io_worker_control->workers[worker].in_use); return worker; } @@ -179,7 +180,7 @@ pgaio_choose_idle_worker(void) static bool pgaio_worker_submission_queue_insert(PgAioHandle *ioh) { - AioWorkerSubmissionQueue *queue; + PgAioWorkerSubmissionQueue *queue; uint32 new_head; queue = io_worker_submission_queue; @@ -191,7 +192,7 @@ pgaio_worker_submission_queue_insert(PgAioHandle *ioh) return false; /* full */ } - queue->ios[queue->head] = pgaio_io_get_id(ioh); + queue->sqes[queue->head] = pgaio_io_get_id(ioh); queue->head = new_head; return true; @@ -200,14 +201,14 @@ pgaio_worker_submission_queue_insert(PgAioHandle *ioh) static uint32 pgaio_worker_submission_queue_consume(void) { - AioWorkerSubmissionQueue *queue; + PgAioWorkerSubmissionQueue *queue; uint32 result; queue = io_worker_submission_queue; if (queue->tail == queue->head) return UINT32_MAX; /* empty */ - result = queue->ios[queue->tail]; + result = queue->sqes[queue->tail]; queue->tail = (queue->tail + 1) & (queue->size - 1); return result; @@ -240,37 +241,37 @@ pgaio_worker_needs_synchronous_execution(PgAioHandle *ioh) } static void -pgaio_worker_submit_internal(int nios, PgAioHandle *ios[]) +pgaio_worker_submit_internal(int num_staged_ios, PgAioHandle **staged_ios) { PgAioHandle *synchronous_ios[PGAIO_SUBMIT_BATCH_SIZE]; int nsync = 0; Latch *wakeup = NULL; int worker; - Assert(nios <= PGAIO_SUBMIT_BATCH_SIZE); + Assert(num_staged_ios <= PGAIO_SUBMIT_BATCH_SIZE); LWLockAcquire(AioWorkerSubmissionQueueLock, LW_EXCLUSIVE); - for (int i = 0; i < nios; ++i) + for (int i = 0; i < num_staged_ios; ++i) { - Assert(!pgaio_worker_needs_synchronous_execution(ios[i])); - if (!pgaio_worker_submission_queue_insert(ios[i])) + Assert(!pgaio_worker_needs_synchronous_execution(staged_ios[i])); + if (!pgaio_worker_submission_queue_insert(staged_ios[i])) { /* * We'll do it synchronously, but only after we've sent as many as * we can to workers, to maximize concurrency. */ - synchronous_ios[nsync++] = ios[i]; + synchronous_ios[nsync++] = staged_ios[i]; continue; } if (wakeup == NULL) { /* Choose an idle worker to wake up if we haven't already. */ - worker = pgaio_choose_idle_worker(); + worker = pgaio_worker_choose_idle(); if (worker >= 0) wakeup = io_worker_control->workers[worker].latch; - pgaio_debug_io(DEBUG4, ios[i], + pgaio_debug_io(DEBUG4, staged_ios[i], "choosing worker %d", worker); } @@ -316,6 +317,7 @@ pgaio_worker_die(int code, Datum arg) Assert(io_worker_control->workers[MyIoWorkerId].in_use); Assert(io_worker_control->workers[MyIoWorkerId].latch == MyLatch); + io_worker_control->idle_worker_mask &= ~(UINT64_C(1) << MyIoWorkerId); io_worker_control->workers[MyIoWorkerId].in_use = false; io_worker_control->workers[MyIoWorkerId].latch = NULL; LWLockRelease(AioWorkerSubmissionQueueLock); @@ -461,7 +463,12 @@ IoWorkerMain(const void *startup_data, size_t startup_data_len) int nwakeups = 0; int worker; - /* Try to get a job to do. */ + /* + * Try to get a job to do. + * + * The lwlock acquisition also provides the necessary memory barrier + * to ensure that we don't see an outdated data in the handle. + */ LWLockAcquire(AioWorkerSubmissionQueueLock, LW_EXCLUSIVE); if ((io_index = pgaio_worker_submission_queue_consume()) == UINT32_MAX) { @@ -483,7 +490,7 @@ IoWorkerMain(const void *startup_data, size_t startup_data_len) IO_WORKER_WAKEUP_FANOUT); for (int i = 0; i < nwakeups; ++i) { - if ((worker = pgaio_choose_idle_worker()) < 0) + if ((worker = pgaio_worker_choose_idle()) < 0) break; latches[nlatches++] = io_worker_control->workers[worker].latch; } @@ -568,6 +575,12 @@ IoWorkerMain(const void *startup_data, size_t startup_data_len) } CHECK_FOR_INTERRUPTS(); + + if (ConfigReloadPending) + { + ConfigReloadPending = false; + ProcessConfigFile(PGC_SIGHUP); + } } error_context_stack = errcallback.previous; diff --git a/src/backend/storage/buffer/bufmgr.c b/src/backend/storage/buffer/bufmgr.c index f93131a645e..67431208e7f 100644 --- a/src/backend/storage/buffer/bufmgr.c +++ b/src/backend/storage/buffer/bufmgr.c @@ -2743,12 +2743,10 @@ ExtendBufferedRelShared(BufferManagerRelation bmr, * because mdread doesn't complain about reads beyond EOF (when * zero_damaged_pages is ON) and so a previous attempt to read a block * beyond EOF could have left a "valid" zero-filled buffer. - * Unfortunately, we have also seen this case occurring because of - * buggy Linux kernels that sometimes return an lseek(SEEK_END) result - * that doesn't account for a recent write. In that situation, the - * pre-existing buffer would contain valid data that we don't want to - * overwrite. Since the legitimate cases should always have left a - * zero-filled buffer, complain if not PageIsNew. + * + * This has also been observed when relation was overwritten by + * external process. Since the legitimate cases should always have + * left a zero-filled buffer, complain if not PageIsNew. */ if (existing_id >= 0) { @@ -2778,8 +2776,7 @@ ExtendBufferedRelShared(BufferManagerRelation bmr, ereport(ERROR, (errmsg("unexpected data beyond EOF in block %u of relation %s", existing_hdr->tag.blockNum, - relpath(bmr.smgr->smgr_rlocator, fork).str), - errhint("This has been seen to occur with buggy kernels; consider updating your system."))); + relpath(bmr.smgr->smgr_rlocator, fork).str))); /* * We *must* do smgr[zero]extend before succeeding, else the page @@ -3339,10 +3336,10 @@ UnpinBufferNoOwner(BufferDesc *buf) * BufferSync -- Write out all dirty buffers in the pool. * * This is called at checkpoint time to write out all dirty shared buffers. - * The checkpoint request flags should be passed in. If CHECKPOINT_IMMEDIATE - * is set, we disable delays between writes; if CHECKPOINT_IS_SHUTDOWN, - * CHECKPOINT_END_OF_RECOVERY or CHECKPOINT_FLUSH_ALL is set, we write even - * unlogged buffers, which are otherwise skipped. The remaining flags + * The checkpoint request flags should be passed in. If CHECKPOINT_FAST is + * set, we disable delays between writes; if CHECKPOINT_IS_SHUTDOWN, + * CHECKPOINT_END_OF_RECOVERY or CHECKPOINT_FLUSH_UNLOGGED is set, we write + * even unlogged buffers, which are otherwise skipped. The remaining flags * currently have no effect here. */ static void @@ -3367,7 +3364,7 @@ BufferSync(int flags) * recovery, we write all dirty buffers. */ if (!((flags & (CHECKPOINT_IS_SHUTDOWN | CHECKPOINT_END_OF_RECOVERY | - CHECKPOINT_FLUSH_ALL)))) + CHECKPOINT_FLUSH_UNLOGGED)))) mask |= BM_PERMANENT; /* @@ -4550,11 +4547,9 @@ DropRelationBuffers(SMgrRelation smgr_reln, ForkNumber *forkNum, if (RelFileLocatorBackendIsTemp(rlocator)) { if (rlocator.backend == MyProcNumber) - { - for (j = 0; j < nforks; j++) - DropRelationLocalBuffers(rlocator.locator, forkNum[j], - firstDelBlock[j]); - } + DropRelationLocalBuffers(rlocator.locator, forkNum, nforks, + firstDelBlock); + return; } @@ -7320,7 +7315,7 @@ buffer_readv_report(PgAioResult result, const PgAioTargetData *td, affected_count > 1 ? errdetail("Block %u held first zeroed page.", first + first_off) : 0, - errhint("See server log for details about the other %u invalid block(s).", + errhint("See server log for details about the other %d invalid block(s).", affected_count + checkfail_count - 1)); return; } diff --git a/src/backend/storage/buffer/localbuf.c b/src/backend/storage/buffer/localbuf.c index 63101d56a07..3c0d20f4659 100644 --- a/src/backend/storage/buffer/localbuf.c +++ b/src/backend/storage/buffer/localbuf.c @@ -629,7 +629,7 @@ InvalidateLocalBuffer(BufferDesc *bufHdr, bool check_unreferenced) */ if (check_unreferenced && (LocalRefCount[bufid] != 0 || BUF_STATE_GET_REFCOUNT(buf_state) != 0)) - elog(ERROR, "block %u of %s is still referenced (local %u)", + elog(ERROR, "block %u of %s is still referenced (local %d)", bufHdr->tag.blockNum, relpathbackend(BufTagGetRelFileLocator(&bufHdr->tag), MyProcNumber, @@ -660,10 +660,11 @@ InvalidateLocalBuffer(BufferDesc *bufHdr, bool check_unreferenced) * See DropRelationBuffers in bufmgr.c for more notes. */ void -DropRelationLocalBuffers(RelFileLocator rlocator, ForkNumber forkNum, - BlockNumber firstDelBlock) +DropRelationLocalBuffers(RelFileLocator rlocator, ForkNumber *forkNum, + int nforks, BlockNumber *firstDelBlock) { int i; + int j; for (i = 0; i < NLocBuffer; i++) { @@ -672,12 +673,18 @@ DropRelationLocalBuffers(RelFileLocator rlocator, ForkNumber forkNum, buf_state = pg_atomic_read_u32(&bufHdr->state); - if ((buf_state & BM_TAG_VALID) && - BufTagMatchesRelFileLocator(&bufHdr->tag, &rlocator) && - BufTagGetForkNum(&bufHdr->tag) == forkNum && - bufHdr->tag.blockNum >= firstDelBlock) + if (!(buf_state & BM_TAG_VALID) || + !BufTagMatchesRelFileLocator(&bufHdr->tag, &rlocator)) + continue; + + for (j = 0; j < nforks; j++) { - InvalidateLocalBuffer(bufHdr, true); + if (BufTagGetForkNum(&bufHdr->tag) == forkNum[j] && + bufHdr->tag.blockNum >= firstDelBlock[j]) + { + InvalidateLocalBuffer(bufHdr, true); + break; + } } } } @@ -925,10 +932,11 @@ GetLocalBufferStorage(void) num_bufs = Min(num_bufs, MaxAllocSize / BLCKSZ); /* Buffers should be I/O aligned. */ - cur_block = (char *) - TYPEALIGN(PG_IO_ALIGN_SIZE, - MemoryContextAlloc(LocalBufferContext, - num_bufs * BLCKSZ + PG_IO_ALIGN_SIZE)); + cur_block = MemoryContextAllocAligned(LocalBufferContext, + num_bufs * BLCKSZ, + PG_IO_ALIGN_SIZE, + 0); + next_buf_in_block = 0; num_bufs_in_block = num_bufs; } diff --git a/src/backend/storage/file/fd.c b/src/backend/storage/file/fd.c index 0e8299dd556..a4ec7959f31 100644 --- a/src/backend/storage/file/fd.c +++ b/src/backend/storage/file/fd.c @@ -400,25 +400,22 @@ pg_fsync(int fd) * portable, even if it runs ok on the current system. * * We assert here that a descriptor for a file was opened with write - * permissions (either O_RDWR or O_WRONLY) and for a directory without - * write permissions (O_RDONLY). + * permissions (i.e., not O_RDONLY) and for a directory without write + * permissions (O_RDONLY). Notice that the assertion check is made even + * if fsync() is disabled. * - * Ignore any fstat errors and let the follow-up fsync() do its work. - * Doing this sanity check here counts for the case where fsync() is - * disabled. + * If fstat() fails, ignore it and let the follow-up fsync() complain. */ if (fstat(fd, &st) == 0) { int desc_flags = fcntl(fd, F_GETFL); - /* - * O_RDONLY is historically 0, so just make sure that for directories - * no write flags are used. - */ + desc_flags &= O_ACCMODE; + if (S_ISDIR(st.st_mode)) - Assert((desc_flags & (O_RDWR | O_WRONLY)) == 0); + Assert(desc_flags == O_RDONLY); else - Assert((desc_flags & (O_RDWR | O_WRONLY)) != 0); + Assert(desc_flags != O_RDONLY); } errno = 0; #endif diff --git a/src/backend/storage/file/fileset.c b/src/backend/storage/file/fileset.c index 64141c7cb91..4d5ee353fd7 100644 --- a/src/backend/storage/file/fileset.c +++ b/src/backend/storage/file/fileset.c @@ -185,7 +185,7 @@ FileSetPath(char *path, FileSet *fileset, Oid tablespace) static Oid ChooseTablespace(const FileSet *fileset, const char *name) { - uint32 hash = hash_any((const unsigned char *) name, strlen(name)); + uint32 hash = hash_bytes((const unsigned char *) name, strlen(name)); return fileset->tablespaces[hash % fileset->ntablespaces]; } diff --git a/src/backend/storage/ipc/dsm_registry.c b/src/backend/storage/ipc/dsm_registry.c index 1d4fd31ffed..1682cc6d34c 100644 --- a/src/backend/storage/ipc/dsm_registry.c +++ b/src/backend/storage/ipc/dsm_registry.c @@ -15,6 +15,20 @@ * current backend. This function guarantees that only one backend * initializes the segment and that all other backends just attach it. * + * A DSA can be created in or retrieved from the registry by calling + * GetNamedDSA(). As with GetNamedDSMSegment(), if a DSA with the provided + * name does not yet exist, it is created. Otherwise, GetNamedDSA() + * ensures the DSA is attached to the current backend. This function + * guarantees that only one backend initializes the DSA and that all other + * backends just attach it. + * + * A dshash table can be created in or retrieved from the registry by + * calling GetNamedDSHash(). As with GetNamedDSMSegment(), if a hash + * table with the provided name does not yet exist, it is created. + * Otherwise, GetNamedDSHash() ensures the hash table is attached to the + * current backend. This function guarantees that only one backend + * initializes the table and that all other backends just attach it. + * * Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group * Portions Copyright (c) 1994, Regents of the University of California * @@ -26,12 +40,20 @@ #include "postgres.h" +#include "funcapi.h" #include "lib/dshash.h" #include "storage/dsm_registry.h" #include "storage/lwlock.h" #include "storage/shmem.h" +#include "utils/builtins.h" #include "utils/memutils.h" +#define DSMR_NAME_LEN 128 + +#define DSMR_DSA_TRANCHE_SUFFIX " DSA" +#define DSMR_DSA_TRANCHE_SUFFIX_LEN (sizeof(DSMR_DSA_TRANCHE_SUFFIX) - 1) +#define DSMR_DSA_TRANCHE_NAME_LEN (DSMR_NAME_LEN + DSMR_DSA_TRANCHE_SUFFIX_LEN) + typedef struct DSMRegistryCtxStruct { dsa_handle dsah; @@ -40,15 +62,55 @@ typedef struct DSMRegistryCtxStruct static DSMRegistryCtxStruct *DSMRegistryCtx; -typedef struct DSMRegistryEntry +typedef struct NamedDSMState { - char name[64]; dsm_handle handle; size_t size; +} NamedDSMState; + +typedef struct NamedDSAState +{ + dsa_handle handle; + int tranche; + char tranche_name[DSMR_DSA_TRANCHE_NAME_LEN]; +} NamedDSAState; + +typedef struct NamedDSHState +{ + NamedDSAState dsa; + dshash_table_handle handle; + int tranche; + char tranche_name[DSMR_NAME_LEN]; +} NamedDSHState; + +typedef enum DSMREntryType +{ + DSMR_ENTRY_TYPE_DSM, + DSMR_ENTRY_TYPE_DSA, + DSMR_ENTRY_TYPE_DSH, +} DSMREntryType; + +static const char *const DSMREntryTypeNames[] = +{ + [DSMR_ENTRY_TYPE_DSM] = "segment", + [DSMR_ENTRY_TYPE_DSA] = "area", + [DSMR_ENTRY_TYPE_DSH] = "hash", +}; + +typedef struct DSMRegistryEntry +{ + char name[DSMR_NAME_LEN]; + DSMREntryType type; + union + { + NamedDSMState dsm; + NamedDSAState dsa; + NamedDSHState dsh; + } data; } DSMRegistryEntry; static const dshash_parameters dsh_params = { - offsetof(DSMRegistryEntry, handle), + offsetof(DSMRegistryEntry, type), sizeof(DSMRegistryEntry), dshash_strcmp, dshash_strhash, @@ -141,7 +203,7 @@ GetNamedDSMSegment(const char *name, size_t size, ereport(ERROR, (errmsg("DSM segment name cannot be empty"))); - if (strlen(name) >= offsetof(DSMRegistryEntry, handle)) + if (strlen(name) >= offsetof(DSMRegistryEntry, type)) ereport(ERROR, (errmsg("DSM segment name too long"))); @@ -158,32 +220,39 @@ GetNamedDSMSegment(const char *name, size_t size, entry = dshash_find_or_insert(dsm_registry_table, name, found); if (!(*found)) { + NamedDSMState *state = &entry->data.dsm; + dsm_segment *seg; + + entry->type = DSMR_ENTRY_TYPE_DSM; + /* Initialize the segment. */ - dsm_segment *seg = dsm_create(size, 0); + seg = dsm_create(size, 0); dsm_pin_segment(seg); dsm_pin_mapping(seg); - entry->handle = dsm_segment_handle(seg); - entry->size = size; + state->handle = dsm_segment_handle(seg); + state->size = size; ret = dsm_segment_address(seg); if (init_callback) (*init_callback) (ret); } - else if (entry->size != size) - { + else if (entry->type != DSMR_ENTRY_TYPE_DSM) ereport(ERROR, - (errmsg("requested DSM segment size does not match size of " - "existing segment"))); - } + (errmsg("requested DSM segment does not match type of existing entry"))); + else if (entry->data.dsm.size != size) + ereport(ERROR, + (errmsg("requested DSM segment size does not match size of existing segment"))); else { - dsm_segment *seg = dsm_find_mapping(entry->handle); + NamedDSMState *state = &entry->data.dsm; + dsm_segment *seg; /* If the existing segment is not already attached, attach it now. */ + seg = dsm_find_mapping(state->handle); if (seg == NULL) { - seg = dsm_attach(entry->handle); + seg = dsm_attach(state->handle); if (seg == NULL) elog(ERROR, "could not map dynamic shared memory segment"); @@ -198,3 +267,220 @@ GetNamedDSMSegment(const char *name, size_t size, return ret; } + +/* + * Initialize or attach a named DSA. + * + * This routine returns a pointer to the DSA. A new LWLock tranche ID will be + * generated if needed. Note that the lock tranche will be registered with the + * provided name. Also note that this should be called at most once for a + * given DSA in each backend. + */ +dsa_area * +GetNamedDSA(const char *name, bool *found) +{ + DSMRegistryEntry *entry; + MemoryContext oldcontext; + dsa_area *ret; + + Assert(found); + + if (!name || *name == '\0') + ereport(ERROR, + (errmsg("DSA name cannot be empty"))); + + if (strlen(name) >= offsetof(DSMRegistryEntry, type)) + ereport(ERROR, + (errmsg("DSA name too long"))); + + /* Be sure any local memory allocated by DSM/DSA routines is persistent. */ + oldcontext = MemoryContextSwitchTo(TopMemoryContext); + + /* Connect to the registry. */ + init_dsm_registry(); + + entry = dshash_find_or_insert(dsm_registry_table, name, found); + if (!(*found)) + { + NamedDSAState *state = &entry->data.dsa; + + entry->type = DSMR_ENTRY_TYPE_DSA; + + /* Initialize the LWLock tranche for the DSA. */ + state->tranche = LWLockNewTrancheId(); + strcpy(state->tranche_name, name); + LWLockRegisterTranche(state->tranche, state->tranche_name); + + /* Initialize the DSA. */ + ret = dsa_create(state->tranche); + dsa_pin(ret); + dsa_pin_mapping(ret); + + /* Store handle for other backends to use. */ + state->handle = dsa_get_handle(ret); + } + else if (entry->type != DSMR_ENTRY_TYPE_DSA) + ereport(ERROR, + (errmsg("requested DSA does not match type of existing entry"))); + else + { + NamedDSAState *state = &entry->data.dsa; + + if (dsa_is_attached(state->handle)) + ereport(ERROR, + (errmsg("requested DSA already attached to current process"))); + + /* Initialize existing LWLock tranche for the DSA. */ + LWLockRegisterTranche(state->tranche, state->tranche_name); + + /* Attach to existing DSA. */ + ret = dsa_attach(state->handle); + dsa_pin_mapping(ret); + } + + dshash_release_lock(dsm_registry_table, entry); + MemoryContextSwitchTo(oldcontext); + + return ret; +} + +/* + * Initialize or attach a named dshash table. + * + * This routine returns the address of the table. The tranche_id member of + * params is ignored; new tranche IDs will be generated if needed. Note that + * the DSA lock tranche will be registered with the provided name with " DSA" + * appended. The dshash lock tranche will be registered with the provided + * name. Also note that this should be called at most once for a given table + * in each backend. + */ +dshash_table * +GetNamedDSHash(const char *name, const dshash_parameters *params, bool *found) +{ + DSMRegistryEntry *entry; + MemoryContext oldcontext; + dshash_table *ret; + + Assert(params); + Assert(found); + + if (!name || *name == '\0') + ereport(ERROR, + (errmsg("DSHash name cannot be empty"))); + + if (strlen(name) >= offsetof(DSMRegistryEntry, type)) + ereport(ERROR, + (errmsg("DSHash name too long"))); + + /* Be sure any local memory allocated by DSM/DSA routines is persistent. */ + oldcontext = MemoryContextSwitchTo(TopMemoryContext); + + /* Connect to the registry. */ + init_dsm_registry(); + + entry = dshash_find_or_insert(dsm_registry_table, name, found); + if (!(*found)) + { + NamedDSAState *dsa_state = &entry->data.dsh.dsa; + NamedDSHState *dsh_state = &entry->data.dsh; + dshash_parameters params_copy; + dsa_area *dsa; + + entry->type = DSMR_ENTRY_TYPE_DSH; + + /* Initialize the LWLock tranche for the DSA. */ + dsa_state->tranche = LWLockNewTrancheId(); + sprintf(dsa_state->tranche_name, "%s%s", name, DSMR_DSA_TRANCHE_SUFFIX); + LWLockRegisterTranche(dsa_state->tranche, dsa_state->tranche_name); + + /* Initialize the LWLock tranche for the dshash table. */ + dsh_state->tranche = LWLockNewTrancheId(); + strcpy(dsh_state->tranche_name, name); + LWLockRegisterTranche(dsh_state->tranche, dsh_state->tranche_name); + + /* Initialize the DSA for the hash table. */ + dsa = dsa_create(dsa_state->tranche); + dsa_pin(dsa); + dsa_pin_mapping(dsa); + + /* Initialize the dshash table. */ + memcpy(¶ms_copy, params, sizeof(dshash_parameters)); + params_copy.tranche_id = dsh_state->tranche; + ret = dshash_create(dsa, ¶ms_copy, NULL); + + /* Store handles for other backends to use. */ + dsa_state->handle = dsa_get_handle(dsa); + dsh_state->handle = dshash_get_hash_table_handle(ret); + } + else if (entry->type != DSMR_ENTRY_TYPE_DSH) + ereport(ERROR, + (errmsg("requested DSHash does not match type of existing entry"))); + else + { + NamedDSAState *dsa_state = &entry->data.dsh.dsa; + NamedDSHState *dsh_state = &entry->data.dsh; + dsa_area *dsa; + + /* XXX: Should we verify params matches what table was created with? */ + + if (dsa_is_attached(dsa_state->handle)) + ereport(ERROR, + (errmsg("requested DSHash already attached to current process"))); + + /* Initialize existing LWLock tranches for the DSA and dshash table. */ + LWLockRegisterTranche(dsa_state->tranche, dsa_state->tranche_name); + LWLockRegisterTranche(dsh_state->tranche, dsh_state->tranche_name); + + /* Attach to existing DSA for the hash table. */ + dsa = dsa_attach(dsa_state->handle); + dsa_pin_mapping(dsa); + + /* Attach to existing dshash table. */ + ret = dshash_attach(dsa, params, dsh_state->handle, NULL); + } + + dshash_release_lock(dsm_registry_table, entry); + MemoryContextSwitchTo(oldcontext); + + return ret; +} + +Datum +pg_get_dsm_registry_allocations(PG_FUNCTION_ARGS) +{ + ReturnSetInfo *rsinfo = (ReturnSetInfo *) fcinfo->resultinfo; + DSMRegistryEntry *entry; + MemoryContext oldcontext; + dshash_seq_status status; + + InitMaterializedSRF(fcinfo, MAT_SRF_USE_EXPECTED_DESC); + + /* Be sure any local memory allocated by DSM/DSA routines is persistent. */ + oldcontext = MemoryContextSwitchTo(TopMemoryContext); + init_dsm_registry(); + MemoryContextSwitchTo(oldcontext); + + dshash_seq_init(&status, dsm_registry_table, false); + while ((entry = dshash_seq_next(&status)) != NULL) + { + Datum vals[3]; + bool nulls[3] = {0}; + + vals[0] = CStringGetTextDatum(entry->name); + vals[1] = CStringGetTextDatum(DSMREntryTypeNames[entry->type]); + + /* + * Since we can't know the size of DSA/dshash entries without first + * attaching to them, return NULL for those. + */ + if (entry->type == DSMR_ENTRY_TYPE_DSM) + vals[2] = Int64GetDatum(entry->data.dsm.size); + else + nulls[2] = true; + + tuplestore_putvalues(rsinfo->setResult, rsinfo->setDesc, vals, nulls); + } + dshash_seq_term(&status); + + return (Datum) 0; +} diff --git a/src/backend/storage/ipc/latch.c b/src/backend/storage/ipc/latch.c index c6aefd2f688..beadeb5e46a 100644 --- a/src/backend/storage/ipc/latch.c +++ b/src/backend/storage/ipc/latch.c @@ -187,9 +187,11 @@ WaitLatch(Latch *latch, int wakeEvents, long timeout, if (!(wakeEvents & WL_LATCH_SET)) latch = NULL; ModifyWaitEvent(LatchWaitSet, LatchWaitSetLatchPos, WL_LATCH_SET, latch); - ModifyWaitEvent(LatchWaitSet, LatchWaitSetPostmasterDeathPos, - (wakeEvents & (WL_EXIT_ON_PM_DEATH | WL_POSTMASTER_DEATH)), - NULL); + + if (IsUnderPostmaster) + ModifyWaitEvent(LatchWaitSet, LatchWaitSetPostmasterDeathPos, + (wakeEvents & (WL_EXIT_ON_PM_DEATH | WL_POSTMASTER_DEATH)), + NULL); if (WaitEventSetWait(LatchWaitSet, (wakeEvents & WL_TIMEOUT) ? timeout : -1, diff --git a/src/backend/storage/ipc/procarray.c b/src/backend/storage/ipc/procarray.c index e5b945a9ee3..bf987aed8d3 100644 --- a/src/backend/storage/ipc/procarray.c +++ b/src/backend/storage/ipc/procarray.c @@ -1622,58 +1622,6 @@ TransactionIdIsInProgress(TransactionId xid) return false; } -/* - * TransactionIdIsActive -- is xid the top-level XID of an active backend? - * - * This differs from TransactionIdIsInProgress in that it ignores prepared - * transactions, as well as transactions running on the primary if we're in - * hot standby. Also, we ignore subtransactions since that's not needed - * for current uses. - */ -bool -TransactionIdIsActive(TransactionId xid) -{ - bool result = false; - ProcArrayStruct *arrayP = procArray; - TransactionId *other_xids = ProcGlobal->xids; - int i; - - /* - * Don't bother checking a transaction older than RecentXmin; it could not - * possibly still be running. - */ - if (TransactionIdPrecedes(xid, RecentXmin)) - return false; - - LWLockAcquire(ProcArrayLock, LW_SHARED); - - for (i = 0; i < arrayP->numProcs; i++) - { - int pgprocno = arrayP->pgprocnos[i]; - PGPROC *proc = &allProcs[pgprocno]; - TransactionId pxid; - - /* Fetch xid just once - see GetNewTransactionId */ - pxid = UINT32_ACCESS_ONCE(other_xids[i]); - - if (!TransactionIdIsValid(pxid)) - continue; - - if (proc->pid == 0) - continue; /* ignore prepared transactions */ - - if (TransactionIdEquals(pxid, xid)) - { - result = true; - break; - } - } - - LWLockRelease(ProcArrayLock); - - return result; -} - /* * Determine XID horizons. @@ -2866,8 +2814,10 @@ GetRunningTransactionData(void) * * Similar to GetSnapshotData but returns just oldestActiveXid. We include * all PGPROCs with an assigned TransactionId, even VACUUM processes. - * We look at all databases, though there is no need to include WALSender - * since this has no effect on hot standby conflicts. + * + * If allDbs is true, we look at all databases, though there is no need to + * include WALSender since this has no effect on hot standby conflicts. If + * allDbs is false, skip processes attached to other databases. * * This is never executed during recovery so there is no need to look at * KnownAssignedXids. @@ -2875,9 +2825,12 @@ GetRunningTransactionData(void) * We don't worry about updating other counters, we want to keep this as * simple as possible and leave GetSnapshotData() as the primary code for * that bookkeeping. + * + * inCommitOnly indicates getting the oldestActiveXid among the transactions + * in the commit critical section. */ TransactionId -GetOldestActiveTransactionId(void) +GetOldestActiveTransactionId(bool inCommitOnly, bool allDbs) { ProcArrayStruct *arrayP = procArray; TransactionId *other_xids = ProcGlobal->xids; @@ -2904,6 +2857,8 @@ GetOldestActiveTransactionId(void) for (index = 0; index < arrayP->numProcs; index++) { TransactionId xid; + int pgprocno = arrayP->pgprocnos[index]; + PGPROC *proc = &allProcs[pgprocno]; /* Fetch xid just once - see GetNewTransactionId */ xid = UINT32_ACCESS_ONCE(other_xids[index]); @@ -2911,6 +2866,13 @@ GetOldestActiveTransactionId(void) if (!TransactionIdIsNormal(xid)) continue; + if (inCommitOnly && + (proc->delayChkptFlags & DELAY_CHKPT_IN_COMMIT) == 0) + continue; + + if (!allDbs && proc->databaseId != MyDatabaseId) + continue; + if (TransactionIdPrecedes(xid, oldestRunningXid)) oldestRunningXid = xid; diff --git a/src/backend/storage/ipc/procsignal.c b/src/backend/storage/ipc/procsignal.c index a9bb540b55a..087821311cc 100644 --- a/src/backend/storage/ipc/procsignal.c +++ b/src/backend/storage/ipc/procsignal.c @@ -728,7 +728,11 @@ procsignal_sigusr1_handler(SIGNAL_ARGS) void SendCancelRequest(int backendPID, const uint8 *cancel_key, int cancel_key_len) { - Assert(backendPID != 0); + if (backendPID == 0) + { + ereport(LOG, (errmsg("invalid cancel request with PID 0"))); + return; + } /* * See if we have a matching backend. Reading the pss_pid and diff --git a/src/backend/storage/ipc/shmem.c b/src/backend/storage/ipc/shmem.c index c9ae3b45b76..ca3656fc76f 100644 --- a/src/backend/storage/ipc/shmem.c +++ b/src/backend/storage/ipc/shmem.c @@ -679,12 +679,10 @@ pg_get_shmem_allocations_numa(PG_FUNCTION_ARGS) */ for (i = 0; i < shm_ent_page_count; i++) { - volatile uint64 touch pg_attribute_unused(); - page_ptrs[i] = startptr + (i * os_page_size); if (firstNumaTouch) - pg_numa_touch_mem_if_required(touch, page_ptrs[i]); + pg_numa_touch_mem_if_required(page_ptrs[i]); CHECK_FOR_INTERRUPTS(); } diff --git a/src/backend/storage/ipc/standby.c b/src/backend/storage/ipc/standby.c index 7fa8d9247e0..4222bdab078 100644 --- a/src/backend/storage/ipc/standby.c +++ b/src/backend/storage/ipc/standby.c @@ -1376,7 +1376,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) if (xlrec.subxid_overflow) elog(DEBUG2, - "snapshot of %d running transactions overflowed (lsn %X/%X oldest xid %u latest complete %u next xid %u)", + "snapshot of %d running transactions overflowed (lsn %X/%08X oldest xid %u latest complete %u next xid %u)", CurrRunningXacts->xcnt, LSN_FORMAT_ARGS(recptr), CurrRunningXacts->oldestRunningXid, @@ -1384,7 +1384,7 @@ LogCurrentRunningXacts(RunningTransactions CurrRunningXacts) CurrRunningXacts->nextXid); else elog(DEBUG2, - "snapshot of %d+%d running transaction ids (lsn %X/%X oldest xid %u latest complete %u next xid %u)", + "snapshot of %d+%d running transaction ids (lsn %X/%08X oldest xid %u latest complete %u next xid %u)", CurrRunningXacts->xcnt, CurrRunningXacts->subxcnt, LSN_FORMAT_ARGS(recptr), CurrRunningXacts->oldestRunningXid, diff --git a/src/backend/storage/lmgr/generate-lwlocknames.pl b/src/backend/storage/lmgr/generate-lwlocknames.pl index 4441b7cba0c..cd3e43c448a 100644 --- a/src/backend/storage/lmgr/generate-lwlocknames.pl +++ b/src/backend/storage/lmgr/generate-lwlocknames.pl @@ -10,7 +10,6 @@ use Getopt::Long; my $output_path = '.'; my $lastlockidx = -1; -my $continue = "\n"; GetOptions('outdir:s' => \$output_path); @@ -28,18 +27,24 @@ print $h "/* there is deliberately not an #ifndef LWLOCKNAMES_H here */\n\n"; # -# First, record the predefined LWLocks listed in wait_event_names.txt. We'll -# cross-check those with the ones in lwlocklist.h. +# First, record the predefined LWLocks and built-in tranches listed in +# wait_event_names.txt. We'll cross-check those with the ones in lwlocklist.h. # +my @wait_event_tranches; my @wait_event_lwlocks; my $record_lwlocks = 0; +my $in_tranches = 0; while (<$wait_event_names>) { chomp; # Check for end marker. - last if /^# END OF PREDEFINED LWLOCKS/; + if (/^# END OF PREDEFINED LWLOCKS/) + { + $in_tranches = 1; + next; + } # Skip comments and empty lines. next if /^#/; @@ -55,13 +60,29 @@ while (<$wait_event_names>) # Go to the next line if we are not yet recording LWLocks. next if not $record_lwlocks; + # Stop recording if we reach another section. + last if /^Section:/; + # Record the LWLock. (my $waiteventname, my $waitevendocsentence) = split(/\t/, $_); - push(@wait_event_lwlocks, $waiteventname); + + if ($in_tranches) + { + push(@wait_event_tranches, $waiteventname); + } + else + { + push(@wait_event_lwlocks, $waiteventname); + } } +# +# While gathering the list of predefined LWLocks, cross-check the lists in +# lwlocklist.h with the wait events we just recorded. +# my $in_comment = 0; -my $i = 0; +my $lwlock_count = 0; +my $tranche_count = 0; while (<$lwlocklist>) { chomp; @@ -82,40 +103,72 @@ while (<$lwlocklist>) next; } - die "unable to parse lwlocklist.h line \"$_\"" - unless /^PG_LWLOCK\((\d+),\s+(\w+)\)$/; + # + # Gather list of predefined LWLocks and cross-check with the wait events. + # + if (/^PG_LWLOCK\((\d+),\s+(\w+)\)$/) + { + my ($lockidx, $lockname) = ($1, $2); - (my $lockidx, my $lockname) = ($1, $2); + die "lwlocklist.h not in order" if $lockidx < $lastlockidx; + die "lwlocklist.h has duplicates" if $lockidx == $lastlockidx; - die "lwlocklist.h not in order" if $lockidx < $lastlockidx; - die "lwlocklist.h has duplicates" if $lockidx == $lastlockidx; + die "$lockname defined in lwlocklist.h but missing from " + . "wait_event_names.txt" + if $lwlock_count >= scalar @wait_event_lwlocks; + die "lists of predefined LWLocks do not match (first mismatch at " + . "$wait_event_lwlocks[$lwlock_count] in wait_event_names.txt and " + . "$lockname in lwlocklist.h)" + if $wait_event_lwlocks[$lwlock_count] ne $lockname; - die "$lockname defined in lwlocklist.h but missing from " - . "wait_event_names.txt" - if $i >= scalar @wait_event_lwlocks; - die "lists of predefined LWLocks do not match (first mismatch at " - . "$wait_event_lwlocks[$i] in wait_event_names.txt and $lockname in " - . "lwlocklist.h)" - if $wait_event_lwlocks[$i] ne $lockname; - $i++; + $lwlock_count++; - while ($lastlockidx < $lockidx - 1) + while ($lastlockidx < $lockidx - 1) + { + ++$lastlockidx; + } + $lastlockidx = $lockidx; + + # Add a "Lock" suffix to each lock name, as the C code depends on that. + printf $h "#define %-32s (&MainLWLockArray[$lockidx].lock)\n", + $lockname . "Lock"; + + next; + } + + # + # Cross-check the built-in LWLock tranches with the wait events. + # + if (/^PG_LWLOCKTRANCHE\((\w+),\s+(\w+)\)$/) { - ++$lastlockidx; - $continue = ",\n"; + my ($tranche_id, $tranche_name) = ($1, $2); + + die "$tranche_name defined in lwlocklist.h but missing from " + . "wait_event_names.txt" + if $tranche_count >= scalar @wait_event_tranches; + die + "lists of built-in LWLock tranches do not match (first mismatch at " + . "$wait_event_tranches[$tranche_count] in wait_event_names.txt and " + . "$tranche_name in lwlocklist.h)" + if $wait_event_tranches[$tranche_count] ne $tranche_name; + + $tranche_count++; + + next; } - $lastlockidx = $lockidx; - $continue = ",\n"; - # Add a "Lock" suffix to each lock name, as the C code depends on that - printf $h "#define %-32s (&MainLWLockArray[$lockidx].lock)\n", - $lockname . "Lock"; + die "unable to parse lwlocklist.h line \"$_\""; } die - "$wait_event_lwlocks[$i] defined in wait_event_names.txt but missing from " - . "lwlocklist.h" - if $i < scalar @wait_event_lwlocks; + "$wait_event_lwlocks[$lwlock_count] defined in wait_event_names.txt but " + . " missing from lwlocklist.h" + if $lwlock_count < scalar @wait_event_lwlocks; + +die + "$wait_event_tranches[$tranche_count] defined in wait_event_names.txt but " + . "missing from lwlocklist.h" + if $tranche_count < scalar @wait_event_tranches; print $h "\n"; printf $h "#define NUM_INDIVIDUAL_LWLOCKS %s\n", $lastlockidx + 1; diff --git a/src/backend/storage/lmgr/lock.c b/src/backend/storage/lmgr/lock.c index 86b06b9223f..62f3471448e 100644 --- a/src/backend/storage/lmgr/lock.c +++ b/src/backend/storage/lmgr/lock.c @@ -51,7 +51,7 @@ /* GUC variables */ int max_locks_per_xact; /* used to set the lock table size */ -bool log_lock_failure = false; +bool log_lock_failures = false; #define NLOCKENTS() \ mul_size(max_locks_per_xact, add_size(MaxBackends, max_prepared_xacts)) @@ -3539,9 +3539,9 @@ AtPrepare_Locks(void) * but that probably costs more cycles. */ void -PostPrepare_Locks(TransactionId xid) +PostPrepare_Locks(FullTransactionId fxid) { - PGPROC *newproc = TwoPhaseGetDummyProc(xid, false); + PGPROC *newproc = TwoPhaseGetDummyProc(fxid, false); HASH_SEQ_STATUS status; LOCALLOCK *locallock; LOCK *lock; @@ -4324,11 +4324,11 @@ DumpAllLocks(void) * and PANIC anyway. */ void -lock_twophase_recover(TransactionId xid, uint16 info, +lock_twophase_recover(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { TwoPhaseLockRecord *rec = (TwoPhaseLockRecord *) recdata; - PGPROC *proc = TwoPhaseGetDummyProc(xid, false); + PGPROC *proc = TwoPhaseGetDummyProc(fxid, false); LOCKTAG *locktag; LOCKMODE lockmode; LOCKMETHODID lockmethodid; @@ -4505,7 +4505,7 @@ lock_twophase_recover(TransactionId xid, uint16 info, * starting up into hot standby mode. */ void -lock_twophase_standby_recover(TransactionId xid, uint16 info, +lock_twophase_standby_recover(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { TwoPhaseLockRecord *rec = (TwoPhaseLockRecord *) recdata; @@ -4524,7 +4524,7 @@ lock_twophase_standby_recover(TransactionId xid, uint16 info, if (lockmode == AccessExclusiveLock && locktag->locktag_type == LOCKTAG_RELATION) { - StandbyAcquireAccessExclusiveLock(xid, + StandbyAcquireAccessExclusiveLock(XidFromFullTransactionId(fxid), locktag->locktag_field1 /* dboid */ , locktag->locktag_field2 /* reloid */ ); } @@ -4537,11 +4537,11 @@ lock_twophase_standby_recover(TransactionId xid, uint16 info, * Find and release the lock indicated by the 2PC record. */ void -lock_twophase_postcommit(TransactionId xid, uint16 info, +lock_twophase_postcommit(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { TwoPhaseLockRecord *rec = (TwoPhaseLockRecord *) recdata; - PGPROC *proc = TwoPhaseGetDummyProc(xid, true); + PGPROC *proc = TwoPhaseGetDummyProc(fxid, true); LOCKTAG *locktag; LOCKMETHODID lockmethodid; LockMethod lockMethodTable; @@ -4563,10 +4563,10 @@ lock_twophase_postcommit(TransactionId xid, uint16 info, * This is actually just the same as the COMMIT case. */ void -lock_twophase_postabort(TransactionId xid, uint16 info, +lock_twophase_postabort(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { - lock_twophase_postcommit(xid, info, recdata, len); + lock_twophase_postcommit(fxid, info, recdata, len); } /* diff --git a/src/backend/storage/lmgr/lwlock.c b/src/backend/storage/lmgr/lwlock.c index 46f44bc4511..ec9c345ffdf 100644 --- a/src/backend/storage/lmgr/lwlock.c +++ b/src/backend/storage/lmgr/lwlock.c @@ -122,9 +122,8 @@ StaticAssertDecl((LW_VAL_EXCLUSIVE & LW_FLAG_MASK) == 0, * own tranche. We absorb the names of these tranches from there into * BuiltinTrancheNames here. * - * 2. There are some predefined tranches for built-in groups of locks. - * These are listed in enum BuiltinTrancheIds in lwlock.h, and their names - * appear in BuiltinTrancheNames[] below. + * 2. There are some predefined tranches for built-in groups of locks defined + * in lwlocklist.h. We absorb the names of these tranches, too. * * 3. Extensions can create new tranches, via either RequestNamedLWLockTranche * or LWLockRegisterTranche. The names of these that are known in the current @@ -135,49 +134,10 @@ StaticAssertDecl((LW_VAL_EXCLUSIVE & LW_FLAG_MASK) == 0, */ static const char *const BuiltinTrancheNames[] = { #define PG_LWLOCK(id, lockname) [id] = CppAsString(lockname), +#define PG_LWLOCKTRANCHE(id, lockname) [LWTRANCHE_##id] = CppAsString(lockname), #include "storage/lwlocklist.h" #undef PG_LWLOCK - [LWTRANCHE_XACT_BUFFER] = "XactBuffer", - [LWTRANCHE_COMMITTS_BUFFER] = "CommitTsBuffer", - [LWTRANCHE_SUBTRANS_BUFFER] = "SubtransBuffer", - [LWTRANCHE_MULTIXACTOFFSET_BUFFER] = "MultiXactOffsetBuffer", - [LWTRANCHE_MULTIXACTMEMBER_BUFFER] = "MultiXactMemberBuffer", - [LWTRANCHE_NOTIFY_BUFFER] = "NotifyBuffer", - [LWTRANCHE_SERIAL_BUFFER] = "SerialBuffer", - [LWTRANCHE_WAL_INSERT] = "WALInsert", - [LWTRANCHE_BUFFER_CONTENT] = "BufferContent", - [LWTRANCHE_REPLICATION_ORIGIN_STATE] = "ReplicationOriginState", - [LWTRANCHE_REPLICATION_SLOT_IO] = "ReplicationSlotIO", - [LWTRANCHE_LOCK_FASTPATH] = "LockFastPath", - [LWTRANCHE_BUFFER_MAPPING] = "BufferMapping", - [LWTRANCHE_LOCK_MANAGER] = "LockManager", - [LWTRANCHE_PREDICATE_LOCK_MANAGER] = "PredicateLockManager", - [LWTRANCHE_PARALLEL_HASH_JOIN] = "ParallelHashJoin", - [LWTRANCHE_PARALLEL_BTREE_SCAN] = "ParallelBtreeScan", - [LWTRANCHE_PARALLEL_QUERY_DSA] = "ParallelQueryDSA", - [LWTRANCHE_PER_SESSION_DSA] = "PerSessionDSA", - [LWTRANCHE_PER_SESSION_RECORD_TYPE] = "PerSessionRecordType", - [LWTRANCHE_PER_SESSION_RECORD_TYPMOD] = "PerSessionRecordTypmod", - [LWTRANCHE_SHARED_TUPLESTORE] = "SharedTupleStore", - [LWTRANCHE_SHARED_TIDBITMAP] = "SharedTidBitmap", - [LWTRANCHE_PARALLEL_APPEND] = "ParallelAppend", - [LWTRANCHE_PER_XACT_PREDICATE_LIST] = "PerXactPredicateList", - [LWTRANCHE_PGSTATS_DSA] = "PgStatsDSA", - [LWTRANCHE_PGSTATS_HASH] = "PgStatsHash", - [LWTRANCHE_PGSTATS_DATA] = "PgStatsData", - [LWTRANCHE_LAUNCHER_DSA] = "LogicalRepLauncherDSA", - [LWTRANCHE_LAUNCHER_HASH] = "LogicalRepLauncherHash", - [LWTRANCHE_DSM_REGISTRY_DSA] = "DSMRegistryDSA", - [LWTRANCHE_DSM_REGISTRY_HASH] = "DSMRegistryHash", - [LWTRANCHE_COMMITTS_SLRU] = "CommitTsSLRU", - [LWTRANCHE_MULTIXACTOFFSET_SLRU] = "MultixactOffsetSLRU", - [LWTRANCHE_MULTIXACTMEMBER_SLRU] = "MultixactMemberSLRU", - [LWTRANCHE_NOTIFY_SLRU] = "NotifySLRU", - [LWTRANCHE_SERIAL_SLRU] = "SerialSLRU", - [LWTRANCHE_SUBTRANS_SLRU] = "SubtransSLRU", - [LWTRANCHE_XACT_SLRU] = "XactSLRU", - [LWTRANCHE_PARALLEL_VACUUM_DSA] = "ParallelVacuumDSA", - [LWTRANCHE_AIO_URING_COMPLETION] = "AioUringCompletion", +#undef PG_LWLOCKTRANCHE }; StaticAssertDecl(lengthof(BuiltinTrancheNames) == diff --git a/src/backend/storage/lmgr/predicate.c b/src/backend/storage/lmgr/predicate.c index d82114ffca1..c07fb588355 100644 --- a/src/backend/storage/lmgr/predicate.c +++ b/src/backend/storage/lmgr/predicate.c @@ -191,7 +191,7 @@ * AtPrepare_PredicateLocks(void); * PostPrepare_PredicateLocks(TransactionId xid); * PredicateLockTwoPhaseFinish(TransactionId xid, bool isCommit); - * predicatelock_twophase_recover(TransactionId xid, uint16 info, + * predicatelock_twophase_recover(FullTransactionId fxid, uint16 info, * void *recdata, uint32 len); */ @@ -4856,7 +4856,7 @@ AtPrepare_PredicateLocks(void) * anyway. We only need to clean up our local state. */ void -PostPrepare_PredicateLocks(TransactionId xid) +PostPrepare_PredicateLocks(FullTransactionId fxid) { if (MySerializableXact == InvalidSerializableXact) return; @@ -4879,12 +4879,12 @@ PostPrepare_PredicateLocks(TransactionId xid) * commits or aborts. */ void -PredicateLockTwoPhaseFinish(TransactionId xid, bool isCommit) +PredicateLockTwoPhaseFinish(FullTransactionId fxid, bool isCommit) { SERIALIZABLEXID *sxid; SERIALIZABLEXIDTAG sxidtag; - sxidtag.xid = xid; + sxidtag.xid = XidFromFullTransactionId(fxid); LWLockAcquire(SerializableXactHashLock, LW_SHARED); sxid = (SERIALIZABLEXID *) @@ -4906,10 +4906,11 @@ PredicateLockTwoPhaseFinish(TransactionId xid, bool isCommit) * Re-acquire a predicate lock belonging to a transaction that was prepared. */ void -predicatelock_twophase_recover(TransactionId xid, uint16 info, +predicatelock_twophase_recover(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { TwoPhasePredicateRecord *record; + TransactionId xid = XidFromFullTransactionId(fxid); Assert(len == sizeof(TwoPhasePredicateRecord)); diff --git a/src/backend/tcop/backend_startup.c b/src/backend/tcop/backend_startup.c index a7d1fec981f..14d5fc0b196 100644 --- a/src/backend/tcop/backend_startup.c +++ b/src/backend/tcop/backend_startup.c @@ -492,7 +492,7 @@ static int ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) { int32 len; - char *buf; + char *buf = NULL; ProtocolVersion proto; MemoryContext oldcontext; @@ -516,7 +516,7 @@ ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) * scanners, which may be less benign, but it's not really our job to * notice those.) */ - return STATUS_ERROR; + goto fail; } if (pq_getbytes(((char *) &len) + 1, 3) == EOF) @@ -526,7 +526,7 @@ ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) ereport(COMMERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("incomplete startup packet"))); - return STATUS_ERROR; + goto fail; } len = pg_ntoh32(len); @@ -538,7 +538,7 @@ ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) ereport(COMMERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("invalid length of startup packet"))); - return STATUS_ERROR; + goto fail; } /* @@ -554,7 +554,7 @@ ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) ereport(COMMERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), errmsg("incomplete startup packet"))); - return STATUS_ERROR; + goto fail; } pq_endmsgread(); @@ -568,7 +568,7 @@ ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) { ProcessCancelRequestPacket(port, buf, len); /* Not really an error, but we don't want to proceed further */ - return STATUS_ERROR; + goto fail; } if (proto == NEGOTIATE_SSL_CODE && !ssl_done) @@ -607,14 +607,16 @@ ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) ereport(COMMERROR, (errcode_for_socket_access(), errmsg("failed to send SSL negotiation response: %m"))); - return STATUS_ERROR; /* close the connection */ + goto fail; /* close the connection */ } #ifdef USE_SSL if (SSLok == 'S' && secure_open_server(port) == -1) - return STATUS_ERROR; + goto fail; #endif + pfree(buf); + /* * At this point we should have no data already buffered. If we do, * it was received before we performed the SSL handshake, so it wasn't @@ -661,14 +663,16 @@ ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) ereport(COMMERROR, (errcode_for_socket_access(), errmsg("failed to send GSSAPI negotiation response: %m"))); - return STATUS_ERROR; /* close the connection */ + goto fail; /* close the connection */ } #ifdef ENABLE_GSS if (GSSok == 'G' && secure_open_gssapi(port) == -1) - return STATUS_ERROR; + goto fail; #endif + pfree(buf); + /* * At this point we should have no data already buffered. If we do, * it was received before we performed the GSS handshake, so it wasn't @@ -863,7 +867,16 @@ ProcessStartupPacket(Port *port, bool ssl_done, bool gss_done) */ MemoryContextSwitchTo(oldcontext); + pfree(buf); + return STATUS_OK; + +fail: + /* be tidy, just to avoid Valgrind complaints */ + if (buf) + pfree(buf); + + return STATUS_ERROR; } /* @@ -881,7 +894,7 @@ ProcessCancelRequestPacket(Port *port, void *pkt, int pktlen) { ereport(COMMERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), - errmsg("invalid length of query cancel packet"))); + errmsg("invalid length of cancel request packet"))); return; } len = pktlen - offsetof(CancelRequestPacket, cancelAuthCode); @@ -889,7 +902,7 @@ ProcessCancelRequestPacket(Port *port, void *pkt, int pktlen) { ereport(COMMERROR, (errcode(ERRCODE_PROTOCOL_VIOLATION), - errmsg("invalid length of query cancel key"))); + errmsg("invalid length of cancel key in cancel request packet"))); return; } @@ -1077,7 +1090,7 @@ check_log_connections(char **newval, void **extra, GucSource source) if (!SplitIdentifierString(rawstring, ',', &elemlist)) { - GUC_check_errdetail("Invalid list syntax in parameter \"log_connections\"."); + GUC_check_errdetail("Invalid list syntax in parameter \"%s\".", "log_connections"); pfree(rawstring); list_free(elemlist); return false; diff --git a/src/backend/tcop/postgres.c b/src/backend/tcop/postgres.c index 4efd8120d2c..0cecd464902 100644 --- a/src/backend/tcop/postgres.c +++ b/src/backend/tcop/postgres.c @@ -988,6 +988,7 @@ pg_plan_queries(List *querytrees, const char *query_string, int cursorOptions, stmt->stmt_location = query->stmt_location; stmt->stmt_len = query->stmt_len; stmt->queryId = query->queryId; + stmt->planOrigin = PLAN_STMT_INTERNAL; } else { @@ -2034,7 +2035,7 @@ exec_bind_message(StringInfo input_message) { PlannedStmt *plan = lfirst_node(PlannedStmt, lc); - if (plan->planId != UINT64CONST(0)) + if (plan->planId != INT64CONST(0)) { pgstat_report_plan_id(plan->planId, false); break; @@ -2185,7 +2186,7 @@ exec_execute_message(const char *portal_name, long max_rows) { PlannedStmt *stmt = lfirst_node(PlannedStmt, lc); - if (stmt->planId != UINT64CONST(0)) + if (stmt->planId != INT64CONST(0)) { pgstat_report_plan_id(stmt->planId, false); break; diff --git a/src/backend/tcop/pquery.c b/src/backend/tcop/pquery.c index d1593f38b35..08791b8f75e 100644 --- a/src/backend/tcop/pquery.c +++ b/src/backend/tcop/pquery.c @@ -1350,24 +1350,15 @@ PortalRunMulti(Portal portal, PopActiveSnapshot(); /* - * If a query completion data was supplied, use it. Otherwise use the - * portal's query completion data. - * - * Exception: Clients expect INSERT/UPDATE/DELETE tags to have counts, so - * fake them with zeros. This can happen with DO INSTEAD rules if there - * is no replacement query of the same type as the original. We print "0 - * 0" here because technically there is no query of the matching tag type, - * and printing a non-zero count for a different query type seems wrong, - * e.g. an INSERT that does an UPDATE instead should not print "0 1" if - * one row was updated. See QueryRewrite(), step 3, for details. + * If a command tag was requested and we did not fill in a run-time- + * determined tag above, copy the parse-time tag from the Portal. (There + * might not be any tag there either, in edge cases such as empty prepared + * statements. That's OK.) */ - if (qc && qc->commandTag == CMDTAG_UNKNOWN) - { - if (portal->qc.commandTag != CMDTAG_UNKNOWN) - CopyQueryCompletion(qc, &portal->qc); - /* If the caller supplied a qc, we should have set it by now. */ - Assert(qc->commandTag != CMDTAG_UNKNOWN); - } + if (qc && + qc->commandTag == CMDTAG_UNKNOWN && + portal->qc.commandTag != CMDTAG_UNKNOWN) + CopyQueryCompletion(qc, &portal->qc); } /* diff --git a/src/backend/tcop/utility.c b/src/backend/tcop/utility.c index 25fe3d58016..4f4191b0ea6 100644 --- a/src/backend/tcop/utility.c +++ b/src/backend/tcop/utility.c @@ -943,17 +943,7 @@ standard_ProcessUtility(PlannedStmt *pstmt, break; case T_CheckPointStmt: - if (!has_privs_of_role(GetUserId(), ROLE_PG_CHECKPOINT)) - ereport(ERROR, - (errcode(ERRCODE_INSUFFICIENT_PRIVILEGE), - /* translator: %s is name of a SQL command, eg CHECKPOINT */ - errmsg("permission denied to execute %s command", - "CHECKPOINT"), - errdetail("Only roles with privileges of the \"%s\" role may execute this command.", - "pg_checkpoint"))); - - RequestCheckpoint(CHECKPOINT_IMMEDIATE | CHECKPOINT_WAIT | - (RecoveryInProgress() ? 0 : CHECKPOINT_FORCE)); + ExecCheckpoint(pstate, (CheckPointStmt *) parsetree); break; /* @@ -1244,6 +1234,7 @@ ProcessUtilitySlow(ParseState *pstate, wrapper->utilityStmt = stmt; wrapper->stmt_location = pstmt->stmt_location; wrapper->stmt_len = pstmt->stmt_len; + wrapper->planOrigin = PLAN_STMT_INTERNAL; ProcessUtility(wrapper, queryString, @@ -1343,7 +1334,7 @@ ProcessUtilitySlow(ParseState *pstate, */ switch (stmt->subtype) { - case 'T': /* ALTER DOMAIN DEFAULT */ + case AD_AlterDefault: /* * Recursively alter column default for table and, @@ -1353,30 +1344,30 @@ ProcessUtilitySlow(ParseState *pstate, AlterDomainDefault(stmt->typeName, stmt->def); break; - case 'N': /* ALTER DOMAIN DROP NOT NULL */ + case AD_DropNotNull: address = AlterDomainNotNull(stmt->typeName, false); break; - case 'O': /* ALTER DOMAIN SET NOT NULL */ + case AD_SetNotNull: address = AlterDomainNotNull(stmt->typeName, true); break; - case 'C': /* ADD CONSTRAINT */ + case AD_AddConstraint: address = AlterDomainAddConstraint(stmt->typeName, stmt->def, &secondaryObject); break; - case 'X': /* DROP CONSTRAINT */ + case AD_DropConstraint: address = AlterDomainDropConstraint(stmt->typeName, stmt->name, stmt->behavior, stmt->missing_ok); break; - case 'V': /* VALIDATE CONSTRAINT */ + case AD_ValidateConstraint: address = AlterDomainValidateConstraint(stmt->typeName, stmt->name); @@ -1974,6 +1965,7 @@ ProcessUtilityForAlterTable(Node *stmt, AlterTableUtilityContext *context) wrapper->utilityStmt = stmt; wrapper->stmt_location = context->pstmt->stmt_location; wrapper->stmt_len = context->pstmt->stmt_len; + wrapper->planOrigin = PLAN_STMT_INTERNAL; ProcessUtility(wrapper, context->queryString, diff --git a/src/backend/tsearch/dict_ispell.c b/src/backend/tsearch/dict_ispell.c index 63bd193a78a..debfbf956cc 100644 --- a/src/backend/tsearch/dict_ispell.c +++ b/src/backend/tsearch/dict_ispell.c @@ -47,24 +47,30 @@ dispell_init(PG_FUNCTION_ARGS) if (strcmp(defel->defname, "dictfile") == 0) { + char *filename; + if (dictloaded) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("multiple DictFile parameters"))); - NIImportDictionary(&(d->obj), - get_tsearch_config_filename(defGetString(defel), - "dict")); + filename = get_tsearch_config_filename(defGetString(defel), + "dict"); + NIImportDictionary(&(d->obj), filename); + pfree(filename); dictloaded = true; } else if (strcmp(defel->defname, "afffile") == 0) { + char *filename; + if (affloaded) ereport(ERROR, (errcode(ERRCODE_INVALID_PARAMETER_VALUE), errmsg("multiple AffFile parameters"))); - NIImportAffixes(&(d->obj), - get_tsearch_config_filename(defGetString(defel), - "affix")); + filename = get_tsearch_config_filename(defGetString(defel), + "affix"); + NIImportAffixes(&(d->obj), filename); + pfree(filename); affloaded = true; } else if (strcmp(defel->defname, "stopwords") == 0) diff --git a/src/backend/tsearch/dict_synonym.c b/src/backend/tsearch/dict_synonym.c index 0da5a9d6868..c2773eb01ad 100644 --- a/src/backend/tsearch/dict_synonym.c +++ b/src/backend/tsearch/dict_synonym.c @@ -199,6 +199,7 @@ skipline: } tsearch_readline_end(&trst); + pfree(filename); d->len = cur; qsort(d->syn, d->len, sizeof(Syn), compareSyn); diff --git a/src/backend/tsearch/dict_thesaurus.c b/src/backend/tsearch/dict_thesaurus.c index 1bebe36a691..1e6bbde1ca7 100644 --- a/src/backend/tsearch/dict_thesaurus.c +++ b/src/backend/tsearch/dict_thesaurus.c @@ -167,17 +167,17 @@ addWrd(DictThesaurus *d, char *b, char *e, uint32 idsubst, uint16 nwrd, uint16 p static void thesaurusRead(const char *filename, DictThesaurus *d) { + char *real_filename = get_tsearch_config_filename(filename, "ths"); tsearch_readline_state trst; uint32 idsubst = 0; bool useasis = false; char *line; - filename = get_tsearch_config_filename(filename, "ths"); - if (!tsearch_readline_begin(&trst, filename)) + if (!tsearch_readline_begin(&trst, real_filename)) ereport(ERROR, (errcode(ERRCODE_CONFIG_FILE_ERROR), errmsg("could not open thesaurus file \"%s\": %m", - filename))); + real_filename))); while ((line = tsearch_readline(&trst)) != NULL) { @@ -297,6 +297,7 @@ thesaurusRead(const char *filename, DictThesaurus *d) d->nsubst = idsubst; tsearch_readline_end(&trst); + pfree(real_filename); } static TheLexeme * diff --git a/src/backend/tsearch/ts_locale.c b/src/backend/tsearch/ts_locale.c index b77d8c23d36..4801fe90089 100644 --- a/src/backend/tsearch/ts_locale.c +++ b/src/backend/tsearch/ts_locale.c @@ -36,7 +36,7 @@ t_isalpha(const char *ptr) { int clen = pg_mblen(ptr); wchar_t character[WC_BUF_LEN]; - pg_locale_t mylocale = 0; /* TODO */ + locale_t mylocale = 0; /* TODO */ if (clen == 1 || database_ctype_is_c) return isalpha(TOUCHAR(ptr)); @@ -51,7 +51,7 @@ t_isalnum(const char *ptr) { int clen = pg_mblen(ptr); wchar_t character[WC_BUF_LEN]; - pg_locale_t mylocale = 0; /* TODO */ + locale_t mylocale = 0; /* TODO */ if (clen == 1 || database_ctype_is_c) return isalnum(TOUCHAR(ptr)); diff --git a/src/backend/tsearch/ts_selfuncs.c b/src/backend/tsearch/ts_selfuncs.c index 0c1d2bc1109..453a5e5c2ea 100644 --- a/src/backend/tsearch/ts_selfuncs.c +++ b/src/backend/tsearch/ts_selfuncs.c @@ -233,7 +233,7 @@ mcelem_tsquery_selec(TSQuery query, Datum *mcelem, int nmcelem, * The text Datums came from an array, so it cannot be compressed or * stored out-of-line -- it's safe to use VARSIZE_ANY*. */ - Assert(!VARATT_IS_COMPRESSED(mcelem[i]) && !VARATT_IS_EXTERNAL(mcelem[i])); + Assert(!VARATT_IS_COMPRESSED(DatumGetPointer(mcelem[i])) && !VARATT_IS_EXTERNAL(DatumGetPointer(mcelem[i]))); lookup[i].element = (text *) DatumGetPointer(mcelem[i]); lookup[i].frequency = numbers[i]; } diff --git a/src/backend/tsearch/wparser_def.c b/src/backend/tsearch/wparser_def.c index 79bcd32a063..e2dd3da3aa3 100644 --- a/src/backend/tsearch/wparser_def.c +++ b/src/backend/tsearch/wparser_def.c @@ -299,7 +299,7 @@ TParserInit(char *str, int len) */ if (prs->charmaxlen > 1) { - pg_locale_t mylocale = 0; /* TODO */ + locale_t mylocale = 0; /* TODO */ prs->usewide = true; if (database_ctype_is_c) diff --git a/src/backend/utils/activity/backend_status.c b/src/backend/utils/activity/backend_status.c index 9c2ed2cb9e0..a290cc4c975 100644 --- a/src/backend/utils/activity/backend_status.c +++ b/src/backend/utils/activity/backend_status.c @@ -321,7 +321,7 @@ pgstat_bestart_initial(void) lbeentry.st_progress_command = PROGRESS_COMMAND_INVALID; lbeentry.st_progress_command_target = InvalidOid; lbeentry.st_query_id = INT64CONST(0); - lbeentry.st_plan_id = UINT64CONST(0); + lbeentry.st_plan_id = INT64CONST(0); /* * we don't zero st_progress_param here to save cycles; nobody should @@ -600,7 +600,7 @@ pgstat_report_activity(BackendState state, const char *cmd_str) /* st_xact_start_timestamp and wait_event_info are also disabled */ beentry->st_xact_start_timestamp = 0; beentry->st_query_id = INT64CONST(0); - beentry->st_plan_id = UINT64CONST(0); + beentry->st_plan_id = INT64CONST(0); proc->wait_event_info = 0; PGSTAT_END_WRITE_ACTIVITY(beentry); } @@ -663,7 +663,7 @@ pgstat_report_activity(BackendState state, const char *cmd_str) if (state == STATE_RUNNING) { beentry->st_query_id = INT64CONST(0); - beentry->st_plan_id = UINT64CONST(0); + beentry->st_plan_id = INT64CONST(0); } if (cmd_str != NULL) @@ -722,7 +722,7 @@ pgstat_report_query_id(int64 query_id, bool force) * -------- */ void -pgstat_report_plan_id(uint64 plan_id, bool force) +pgstat_report_plan_id(int64 plan_id, bool force) { volatile PgBackendStatus *beentry = MyBEEntry; @@ -1154,7 +1154,7 @@ pgstat_get_my_query_id(void) * * Return current backend's plan identifier. */ -uint64 +int64 pgstat_get_my_plan_id(void) { if (!MyBEEntry) diff --git a/src/backend/utils/activity/pgstat.c b/src/backend/utils/activity/pgstat.c index 8b57845e870..6bc91ce0dad 100644 --- a/src/backend/utils/activity/pgstat.c +++ b/src/backend/utils/activity/pgstat.c @@ -212,6 +212,11 @@ int pgstat_fetch_consistency = PGSTAT_FETCH_CONSISTENCY_CACHE; PgStat_LocalState pgStatLocal; +/* + * Track pending reports for fixed-numbered stats, used by + * pgstat_report_stat(). + */ +bool pgstat_report_fixed = false; /* ---------- * Local data @@ -370,7 +375,6 @@ static const PgStat_KindInfo pgstat_kind_builtin_infos[PGSTAT_KIND_BUILTIN_SIZE] .shared_data_off = offsetof(PgStatShared_Backend, stats), .shared_data_len = sizeof(((PgStatShared_Backend *) 0)->stats), - .have_static_pending_cb = pgstat_backend_have_pending_cb, .flush_static_cb = pgstat_backend_flush_cb, .reset_timestamp_cb = pgstat_backend_reset_timestamp_cb, }, @@ -437,7 +441,6 @@ static const PgStat_KindInfo pgstat_kind_builtin_infos[PGSTAT_KIND_BUILTIN_SIZE] .shared_data_len = sizeof(((PgStatShared_IO *) 0)->stats), .flush_static_cb = pgstat_io_flush_cb, - .have_static_pending_cb = pgstat_io_have_pending_cb, .init_shmem_cb = pgstat_io_init_shmem_cb, .reset_all_cb = pgstat_io_reset_all_cb, .snapshot_cb = pgstat_io_snapshot_cb, @@ -455,7 +458,6 @@ static const PgStat_KindInfo pgstat_kind_builtin_infos[PGSTAT_KIND_BUILTIN_SIZE] .shared_data_len = sizeof(((PgStatShared_SLRU *) 0)->stats), .flush_static_cb = pgstat_slru_flush_cb, - .have_static_pending_cb = pgstat_slru_have_pending_cb, .init_shmem_cb = pgstat_slru_init_shmem_cb, .reset_all_cb = pgstat_slru_reset_all_cb, .snapshot_cb = pgstat_slru_snapshot_cb, @@ -474,7 +476,6 @@ static const PgStat_KindInfo pgstat_kind_builtin_infos[PGSTAT_KIND_BUILTIN_SIZE] .init_backend_cb = pgstat_wal_init_backend_cb, .flush_static_cb = pgstat_wal_flush_cb, - .have_static_pending_cb = pgstat_wal_have_pending_cb, .init_shmem_cb = pgstat_wal_init_shmem_cb, .reset_all_cb = pgstat_wal_reset_all_cb, .snapshot_cb = pgstat_wal_snapshot_cb, @@ -708,29 +709,10 @@ pgstat_report_stat(bool force) } /* Don't expend a clock check if nothing to do */ - if (dlist_is_empty(&pgStatPending)) + if (dlist_is_empty(&pgStatPending) && + !pgstat_report_fixed) { - bool do_flush = false; - - /* Check for pending stats */ - for (PgStat_Kind kind = PGSTAT_KIND_MIN; kind <= PGSTAT_KIND_MAX; kind++) - { - const PgStat_KindInfo *kind_info = pgstat_get_kind_info(kind); - - if (!kind_info) - continue; - if (!kind_info->have_static_pending_cb) - continue; - - if (kind_info->have_static_pending_cb()) - { - do_flush = true; - break; - } - } - - if (!do_flush) - return 0; + return 0; } /* @@ -784,16 +766,19 @@ pgstat_report_stat(bool force) partial_flush |= pgstat_flush_pending_entries(nowait); /* flush of other stats kinds */ - for (PgStat_Kind kind = PGSTAT_KIND_MIN; kind <= PGSTAT_KIND_MAX; kind++) + if (pgstat_report_fixed) { - const PgStat_KindInfo *kind_info = pgstat_get_kind_info(kind); + for (PgStat_Kind kind = PGSTAT_KIND_MIN; kind <= PGSTAT_KIND_MAX; kind++) + { + const PgStat_KindInfo *kind_info = pgstat_get_kind_info(kind); - if (!kind_info) - continue; - if (!kind_info->flush_static_cb) - continue; + if (!kind_info) + continue; + if (!kind_info->flush_static_cb) + continue; - partial_flush |= kind_info->flush_static_cb(nowait); + partial_flush |= kind_info->flush_static_cb(nowait); + } } last_flush = now; @@ -815,6 +800,7 @@ pgstat_report_stat(bool force) } pending_since = 0; + pgstat_report_fixed = false; return 0; } diff --git a/src/backend/utils/activity/pgstat_backend.c b/src/backend/utils/activity/pgstat_backend.c index 51256277e8d..8714a85e2d9 100644 --- a/src/backend/utils/activity/pgstat_backend.c +++ b/src/backend/utils/activity/pgstat_backend.c @@ -66,6 +66,7 @@ pgstat_count_backend_io_op_time(IOObject io_object, IOContext io_context, io_time); backend_has_iostats = true; + pgstat_report_fixed = true; } void @@ -81,6 +82,7 @@ pgstat_count_backend_io_op(IOObject io_object, IOContext io_context, PendingBackendStats.pending_io.bytes[io_object][io_context][io_op] += bytes; backend_has_iostats = true; + pgstat_report_fixed = true; } /* @@ -302,18 +304,6 @@ pgstat_flush_backend(bool nowait, bits32 flags) } /* - * Check if there are any backend stats waiting for flush. - */ -bool -pgstat_backend_have_pending_cb(void) -{ - if (!pgstat_tracks_backend_bktype(MyBackendType)) - return false; - - return (backend_has_iostats || pgstat_backend_wal_have_pending()); -} - -/* * Callback to flush out locally pending backend statistics. * * If some stats could not be flushed due to lock contention, return true. diff --git a/src/backend/utils/activity/pgstat_io.c b/src/backend/utils/activity/pgstat_io.c index d8d26379a57..13ae57ed649 100644 --- a/src/backend/utils/activity/pgstat_io.c +++ b/src/backend/utils/activity/pgstat_io.c @@ -80,6 +80,7 @@ pgstat_count_io_op(IOObject io_object, IOContext io_context, IOOp io_op, pgstat_count_backend_io_op(io_object, io_context, io_op, cnt, bytes); have_iostats = true; + pgstat_report_fixed = true; } /* @@ -168,15 +169,6 @@ pgstat_fetch_stat_io(void) } /* - * Check if there any IO stats waiting for flush. - */ -bool -pgstat_io_have_pending_cb(void) -{ - return have_iostats; -} - -/* * Simpler wrapper of pgstat_io_flush_cb() */ void diff --git a/src/backend/utils/activity/pgstat_relation.c b/src/backend/utils/activity/pgstat_relation.c index 28587e2916b..69df741cbf6 100644 --- a/src/backend/utils/activity/pgstat_relation.c +++ b/src/backend/utils/activity/pgstat_relation.c @@ -744,7 +744,7 @@ PostPrepare_PgStat_Relations(PgStat_SubXactStatus *xact_state) * Load the saved counts into our local pgstats state. */ void -pgstat_twophase_postcommit(TransactionId xid, uint16 info, +pgstat_twophase_postcommit(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; @@ -780,7 +780,7 @@ pgstat_twophase_postcommit(TransactionId xid, uint16 info, * as aborted. */ void -pgstat_twophase_postabort(TransactionId xid, uint16 info, +pgstat_twophase_postabort(FullTransactionId fxid, uint16 info, void *recdata, uint32 len) { TwoPhasePgStatRecord *rec = (TwoPhasePgStatRecord *) recdata; diff --git a/src/backend/utils/activity/pgstat_shmem.c b/src/backend/utils/activity/pgstat_shmem.c index 2e33293b000..53e7d534270 100644 --- a/src/backend/utils/activity/pgstat_shmem.c +++ b/src/backend/utils/activity/pgstat_shmem.c @@ -183,7 +183,7 @@ StatsShmemInit(void) p += MAXALIGN(pgstat_dsa_init_size()); dsa = dsa_create_in_place(ctl->raw_dsa_area, pgstat_dsa_init_size(), - LWTRANCHE_PGSTATS_DSA, 0); + LWTRANCHE_PGSTATS_DSA, NULL); dsa_pin(dsa); /* @@ -255,7 +255,8 @@ pgstat_attach_shmem(void) dsa_pin_mapping(pgStatLocal.dsa); pgStatLocal.shared_hash = dshash_attach(pgStatLocal.dsa, &dsh_params, - pgStatLocal.shmem->hash_handle, 0); + pgStatLocal.shmem->hash_handle, + NULL); MemoryContextSwitchTo(oldcontext); } diff --git a/src/backend/utils/activity/pgstat_slru.c b/src/backend/utils/activity/pgstat_slru.c index b9e940dde45..7bd8744accb 100644 --- a/src/backend/utils/activity/pgstat_slru.c +++ b/src/backend/utils/activity/pgstat_slru.c @@ -144,15 +144,6 @@ pgstat_get_slru_index(const char *name) } /* - * Check if there are any SLRU stats entries waiting for flush. - */ -bool -pgstat_slru_have_pending_cb(void) -{ - return have_slrustats; -} - -/* * Flush out locally pending SLRU stats entries * * If nowait is true, this function returns false on lock failure. Otherwise @@ -247,6 +238,7 @@ get_slru_entry(int slru_idx) Assert((slru_idx >= 0) && (slru_idx < SLRU_NUM_ELEMENTS)); have_slrustats = true; + pgstat_report_fixed = true; return &pending_SLRUStats[slru_idx]; } diff --git a/src/backend/utils/activity/pgstat_wal.c b/src/backend/utils/activity/pgstat_wal.c index 16a1ecb4d90..0d04480d2f6 100644 --- a/src/backend/utils/activity/pgstat_wal.c +++ b/src/backend/utils/activity/pgstat_wal.c @@ -72,6 +72,15 @@ pgstat_fetch_stat_wal(void) } /* + * To determine whether WAL usage happened. + */ +static inline bool +pgstat_wal_have_pending(void) +{ + return pgWalUsage.wal_records != prevWalUsage.wal_records; +} + +/* * Calculate how much WAL usage counters have increased by subtracting the * previous counters from the current ones. * @@ -92,7 +101,7 @@ pgstat_wal_flush_cb(bool nowait) * This function can be called even if nothing at all has happened. Avoid * taking lock for nothing in that case. */ - if (!pgstat_wal_have_pending_cb()) + if (!pgstat_wal_have_pending()) return false; /* @@ -136,15 +145,6 @@ pgstat_wal_init_backend_cb(void) prevWalUsage = pgWalUsage; } -/* - * To determine whether WAL usage happened. - */ -bool -pgstat_wal_have_pending_cb(void) -{ - return pgWalUsage.wal_records != prevWalUsage.wal_records; -} - void pgstat_wal_init_shmem_cb(void *stats) { diff --git a/src/backend/utils/activity/wait_event_names.txt b/src/backend/utils/activity/wait_event_names.txt index 4da68312b5f..0be307d2ca0 100644 --- a/src/backend/utils/activity/wait_event_names.txt +++ b/src/backend/utils/activity/wait_event_names.txt @@ -356,9 +356,13 @@ AioWorkerSubmissionQueue "Waiting to access AIO worker submission queue." # # END OF PREDEFINED LWLOCKS (DO NOT CHANGE THIS LINE) # -# Predefined LWLocks (i.e., those declared in lwlocknames.h) must be listed -# in the section above and must be listed in the same order as in -# lwlocknames.h. Other LWLocks must be listed in the section below. +# Predefined LWLocks (i.e., those declared at the top of lwlocknames.h) must be +# listed in the section above and must be listed in the same order as in +# lwlocknames.h. +# +# Likewise, the built-in LWLock tranches (i.e., those declared at the bottom of +# lwlocknames.h) must be listed in the section below and must be listed in the +# same order as in lwlocknames.h. # XactBuffer "Waiting for I/O on a transaction status SLRU buffer." diff --git a/src/backend/utils/adt/Makefile b/src/backend/utils/adt/Makefile index 4a233b63c32..ffeacf2b819 100644 --- a/src/backend/utils/adt/Makefile +++ b/src/backend/utils/adt/Makefile @@ -23,6 +23,7 @@ OBJS = \ arrayutils.o \ ascii.o \ bool.o \ + bytea.o \ cash.o \ char.o \ cryptohashfuncs.o \ diff --git a/src/backend/utils/adt/acl.c b/src/backend/utils/adt/acl.c index ca3c5ee3df3..1213f9106d5 100644 --- a/src/backend/utils/adt/acl.c +++ b/src/backend/utils/adt/acl.c @@ -135,6 +135,22 @@ static void RoleMembershipCacheCallback(Datum arg, int cacheid, uint32 hashvalue /* + * Test whether an identifier char can be left unquoted in ACLs. + * + * Formerly, we used isalnum() even on non-ASCII characters, resulting in + * unportable behavior. To ensure dump compatibility with old versions, + * we now treat high-bit-set characters as always requiring quoting during + * putid(), but getid() will always accept them without quotes. + */ +static inline bool +is_safe_acl_char(unsigned char c, bool is_getid) +{ + if (IS_HIGHBIT_SET(c)) + return is_getid; + return isalnum(c) || c == '_'; +} + +/* * getid * Consumes the first alphanumeric string (identifier) found in string * 's', ignoring any leading white space. If it finds a double quote @@ -159,21 +175,22 @@ getid(const char *s, char *n, Node *escontext) while (isspace((unsigned char) *s)) s++; - /* This code had better match what putid() does, below */ for (; *s != '\0' && - (isalnum((unsigned char) *s) || - *s == '_' || - *s == '"' || - in_quotes); + (in_quotes || *s == '"' || is_safe_acl_char(*s, true)); s++) { if (*s == '"') { + if (!in_quotes) + { + in_quotes = true; + continue; + } /* safe to look at next char (could be '\0' though) */ if (*(s + 1) != '"') { - in_quotes = !in_quotes; + in_quotes = false; continue; } /* it's an escaped double quote; skip the escaping char */ @@ -207,10 +224,10 @@ putid(char *p, const char *s) const char *src; bool safe = true; + /* Detect whether we need to use double quotes */ for (src = s; *src; src++) { - /* This test had better match what getid() does, above */ - if (!isalnum((unsigned char) *src) && *src != '_') + if (!is_safe_acl_char(*src, false)) { safe = false; break; diff --git a/src/backend/utils/adt/bytea.c b/src/backend/utils/adt/bytea.c new file mode 100644 index 00000000000..6e7b914c563 --- /dev/null +++ b/src/backend/utils/adt/bytea.c @@ -0,0 +1,1114 @@ +/*------------------------------------------------------------------------- + * + * bytea.c + * Functions for the bytea type. + * + * Portions Copyright (c) 2025, PostgreSQL Global Development Group + * + * + * IDENTIFICATION + * src/backend/utils/adt/bytea.c + * + *------------------------------------------------------------------------- + */ + +#include "postgres.h" + +#include "access/detoast.h" +#include "catalog/pg_collation_d.h" +#include "catalog/pg_type_d.h" +#include "common/int.h" +#include "fmgr.h" +#include "libpq/pqformat.h" +#include "port/pg_bitutils.h" +#include "utils/builtins.h" +#include "utils/bytea.h" +#include "utils/fmgrprotos.h" +#include "utils/memutils.h" +#include "utils/sortsupport.h" +#include "utils/varlena.h" +#include "varatt.h" + +/* GUC variable */ +int bytea_output = BYTEA_OUTPUT_HEX; + +static bytea *bytea_catenate(bytea *t1, bytea *t2); +static bytea *bytea_substring(Datum str, int S, int L, + bool length_not_specified); +static bytea *bytea_overlay(bytea *t1, bytea *t2, int sp, int sl); + +/* + * bytea_catenate + * Guts of byteacat(), broken out so it can be used by other functions + * + * Arguments can be in short-header form, but not compressed or out-of-line + */ +static bytea * +bytea_catenate(bytea *t1, bytea *t2) +{ + bytea *result; + int len1, + len2, + len; + char *ptr; + + len1 = VARSIZE_ANY_EXHDR(t1); + len2 = VARSIZE_ANY_EXHDR(t2); + + /* paranoia ... probably should throw error instead? */ + if (len1 < 0) + len1 = 0; + if (len2 < 0) + len2 = 0; + + len = len1 + len2 + VARHDRSZ; + result = (bytea *) palloc(len); + + /* Set size of result string... */ + SET_VARSIZE(result, len); + + /* Fill data field of result string... */ + ptr = VARDATA(result); + if (len1 > 0) + memcpy(ptr, VARDATA_ANY(t1), len1); + if (len2 > 0) + memcpy(ptr + len1, VARDATA_ANY(t2), len2); + + return result; +} + +#define PG_STR_GET_BYTEA(str_) \ + DatumGetByteaPP(DirectFunctionCall1(byteain, CStringGetDatum(str_))) + +static bytea * +bytea_substring(Datum str, + int S, + int L, + bool length_not_specified) +{ + int32 S1; /* adjusted start position */ + int32 L1; /* adjusted substring length */ + int32 E; /* end position */ + + /* + * The logic here should generally match text_substring(). + */ + S1 = Max(S, 1); + + if (length_not_specified) + { + /* + * Not passed a length - DatumGetByteaPSlice() grabs everything to the + * end of the string if we pass it a negative value for length. + */ + L1 = -1; + } + else if (L < 0) + { + /* SQL99 says to throw an error for E < S, i.e., negative length */ + ereport(ERROR, + (errcode(ERRCODE_SUBSTRING_ERROR), + errmsg("negative substring length not allowed"))); + L1 = -1; /* silence stupider compilers */ + } + else if (pg_add_s32_overflow(S, L, &E)) + { + /* + * L could be large enough for S + L to overflow, in which case the + * substring must run to end of string. + */ + L1 = -1; + } + else + { + /* + * A zero or negative value for the end position can happen if the + * start was negative or one. SQL99 says to return a zero-length + * string. + */ + if (E < 1) + return PG_STR_GET_BYTEA(""); + + L1 = E - S1; + } + + /* + * If the start position is past the end of the string, SQL99 says to + * return a zero-length string -- DatumGetByteaPSlice() will do that for + * us. We need only convert S1 to zero-based starting position. + */ + return DatumGetByteaPSlice(str, S1 - 1, L1); +} + +static bytea * +bytea_overlay(bytea *t1, bytea *t2, int sp, int sl) +{ + bytea *result; + bytea *s1; + bytea *s2; + int sp_pl_sl; + + /* + * Check for possible integer-overflow cases. For negative sp, throw a + * "substring length" error because that's what should be expected + * according to the spec's definition of OVERLAY(). + */ + if (sp <= 0) + ereport(ERROR, + (errcode(ERRCODE_SUBSTRING_ERROR), + errmsg("negative substring length not allowed"))); + if (pg_add_s32_overflow(sp, sl, &sp_pl_sl)) + ereport(ERROR, + (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), + errmsg("integer out of range"))); + + s1 = bytea_substring(PointerGetDatum(t1), 1, sp - 1, false); + s2 = bytea_substring(PointerGetDatum(t1), sp_pl_sl, -1, true); + result = bytea_catenate(s1, t2); + result = bytea_catenate(result, s2); + + return result; +} + +/***************************************************************************** + * USER I/O ROUTINES * + *****************************************************************************/ + +#define VAL(CH) ((CH) - '0') +#define DIG(VAL) ((VAL) + '0') + +/* + * byteain - converts from printable representation of byte array + * + * Non-printable characters must be passed as '\nnn' (octal) and are + * converted to internal form. '\' must be passed as '\\'. + */ +Datum +byteain(PG_FUNCTION_ARGS) +{ + char *inputText = PG_GETARG_CSTRING(0); + Node *escontext = fcinfo->context; + size_t len = strlen(inputText); + size_t bc; + char *tp; + char *rp; + bytea *result; + + /* Recognize hex input */ + if (inputText[0] == '\\' && inputText[1] == 'x') + { + bc = (len - 2) / 2 + VARHDRSZ; /* maximum possible length */ + result = palloc(bc); + bc = hex_decode_safe(inputText + 2, len - 2, VARDATA(result), + escontext); + SET_VARSIZE(result, bc + VARHDRSZ); /* actual length */ + + PG_RETURN_BYTEA_P(result); + } + + /* Else, it's the traditional escaped style */ + result = (bytea *) palloc(len + VARHDRSZ); /* maximum possible length */ + + tp = inputText; + rp = VARDATA(result); + while (*tp != '\0') + { + if (tp[0] != '\\') + *rp++ = *tp++; + else if ((tp[1] >= '0' && tp[1] <= '3') && + (tp[2] >= '0' && tp[2] <= '7') && + (tp[3] >= '0' && tp[3] <= '7')) + { + int v; + + v = VAL(tp[1]); + v <<= 3; + v += VAL(tp[2]); + v <<= 3; + *rp++ = v + VAL(tp[3]); + + tp += 4; + } + else if (tp[1] == '\\') + { + *rp++ = '\\'; + tp += 2; + } + else + { + /* + * one backslash, not followed by another or ### valid octal + */ + ereturn(escontext, (Datum) 0, + (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), + errmsg("invalid input syntax for type %s", "bytea"))); + } + } + + bc = rp - VARDATA(result); /* actual length */ + SET_VARSIZE(result, bc + VARHDRSZ); + + PG_RETURN_BYTEA_P(result); +} + +/* + * byteaout - converts to printable representation of byte array + * + * In the traditional escaped format, non-printable characters are + * printed as '\nnn' (octal) and '\' as '\\'. + */ +Datum +byteaout(PG_FUNCTION_ARGS) +{ + bytea *vlena = PG_GETARG_BYTEA_PP(0); + char *result; + char *rp; + + if (bytea_output == BYTEA_OUTPUT_HEX) + { + /* Print hex format */ + rp = result = palloc(VARSIZE_ANY_EXHDR(vlena) * 2 + 2 + 1); + *rp++ = '\\'; + *rp++ = 'x'; + rp += hex_encode(VARDATA_ANY(vlena), VARSIZE_ANY_EXHDR(vlena), rp); + } + else if (bytea_output == BYTEA_OUTPUT_ESCAPE) + { + /* Print traditional escaped format */ + char *vp; + uint64 len; + int i; + + len = 1; /* empty string has 1 char */ + vp = VARDATA_ANY(vlena); + for (i = VARSIZE_ANY_EXHDR(vlena); i != 0; i--, vp++) + { + if (*vp == '\\') + len += 2; + else if ((unsigned char) *vp < 0x20 || (unsigned char) *vp > 0x7e) + len += 4; + else + len++; + } + + /* + * In principle len can't overflow uint32 if the input fit in 1GB, but + * for safety let's check rather than relying on palloc's internal + * check. + */ + if (len > MaxAllocSize) + ereport(ERROR, + (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED), + errmsg_internal("result of bytea output conversion is too large"))); + rp = result = (char *) palloc(len); + + vp = VARDATA_ANY(vlena); + for (i = VARSIZE_ANY_EXHDR(vlena); i != 0; i--, vp++) + { + if (*vp == '\\') + { + *rp++ = '\\'; + *rp++ = '\\'; + } + else if ((unsigned char) *vp < 0x20 || (unsigned char) *vp > 0x7e) + { + int val; /* holds unprintable chars */ + + val = *vp; + rp[0] = '\\'; + rp[3] = DIG(val & 07); + val >>= 3; + rp[2] = DIG(val & 07); + val >>= 3; + rp[1] = DIG(val & 03); + rp += 4; + } + else + *rp++ = *vp; + } + } + else + { + elog(ERROR, "unrecognized \"bytea_output\" setting: %d", + bytea_output); + rp = result = NULL; /* keep compiler quiet */ + } + *rp = '\0'; + PG_RETURN_CSTRING(result); +} + +/* + * bytearecv - converts external binary format to bytea + */ +Datum +bytearecv(PG_FUNCTION_ARGS) +{ + StringInfo buf = (StringInfo) PG_GETARG_POINTER(0); + bytea *result; + int nbytes; + + nbytes = buf->len - buf->cursor; + result = (bytea *) palloc(nbytes + VARHDRSZ); + SET_VARSIZE(result, nbytes + VARHDRSZ); + pq_copymsgbytes(buf, VARDATA(result), nbytes); + PG_RETURN_BYTEA_P(result); +} + +/* + * byteasend - converts bytea to binary format + * + * This is a special case: just copy the input... + */ +Datum +byteasend(PG_FUNCTION_ARGS) +{ + bytea *vlena = PG_GETARG_BYTEA_P_COPY(0); + + PG_RETURN_BYTEA_P(vlena); +} + +Datum +bytea_string_agg_transfn(PG_FUNCTION_ARGS) +{ + StringInfo state; + + state = PG_ARGISNULL(0) ? NULL : (StringInfo) PG_GETARG_POINTER(0); + + /* Append the value unless null, preceding it with the delimiter. */ + if (!PG_ARGISNULL(1)) + { + bytea *value = PG_GETARG_BYTEA_PP(1); + bool isfirst = false; + + /* + * You might think we can just throw away the first delimiter, however + * we must keep it as we may be a parallel worker doing partial + * aggregation building a state to send to the main process. We need + * to keep the delimiter of every aggregation so that the combine + * function can properly join up the strings of two separately + * partially aggregated results. The first delimiter is only stripped + * off in the final function. To know how much to strip off the front + * of the string, we store the length of the first delimiter in the + * StringInfo's cursor field, which we don't otherwise need here. + */ + if (state == NULL) + { + MemoryContext aggcontext; + MemoryContext oldcontext; + + if (!AggCheckCallContext(fcinfo, &aggcontext)) + { + /* cannot be called directly because of internal-type argument */ + elog(ERROR, "bytea_string_agg_transfn called in non-aggregate context"); + } + + /* + * Create state in aggregate context. It'll stay there across + * subsequent calls. + */ + oldcontext = MemoryContextSwitchTo(aggcontext); + state = makeStringInfo(); + MemoryContextSwitchTo(oldcontext); + + isfirst = true; + } + + if (!PG_ARGISNULL(2)) + { + bytea *delim = PG_GETARG_BYTEA_PP(2); + + appendBinaryStringInfo(state, VARDATA_ANY(delim), + VARSIZE_ANY_EXHDR(delim)); + if (isfirst) + state->cursor = VARSIZE_ANY_EXHDR(delim); + } + + appendBinaryStringInfo(state, VARDATA_ANY(value), + VARSIZE_ANY_EXHDR(value)); + } + + /* + * The transition type for string_agg() is declared to be "internal", + * which is a pass-by-value type the same size as a pointer. + */ + if (state) + PG_RETURN_POINTER(state); + PG_RETURN_NULL(); +} + +Datum +bytea_string_agg_finalfn(PG_FUNCTION_ARGS) +{ + StringInfo state; + + /* cannot be called directly because of internal-type argument */ + Assert(AggCheckCallContext(fcinfo, NULL)); + + state = PG_ARGISNULL(0) ? NULL : (StringInfo) PG_GETARG_POINTER(0); + + if (state != NULL) + { + /* As per comment in transfn, strip data before the cursor position */ + bytea *result; + int strippedlen = state->len - state->cursor; + + result = (bytea *) palloc(strippedlen + VARHDRSZ); + SET_VARSIZE(result, strippedlen + VARHDRSZ); + memcpy(VARDATA(result), &state->data[state->cursor], strippedlen); + PG_RETURN_BYTEA_P(result); + } + else + PG_RETURN_NULL(); +} + +/*------------------------------------------------------------- + * byteaoctetlen + * + * get the number of bytes contained in an instance of type 'bytea' + *------------------------------------------------------------- + */ +Datum +byteaoctetlen(PG_FUNCTION_ARGS) +{ + Datum str = PG_GETARG_DATUM(0); + + /* We need not detoast the input at all */ + PG_RETURN_INT32(toast_raw_datum_size(str) - VARHDRSZ); +} + +/* + * byteacat - + * takes two bytea* and returns a bytea* that is the concatenation of + * the two. + * + * Cloned from textcat and modified as required. + */ +Datum +byteacat(PG_FUNCTION_ARGS) +{ + bytea *t1 = PG_GETARG_BYTEA_PP(0); + bytea *t2 = PG_GETARG_BYTEA_PP(1); + + PG_RETURN_BYTEA_P(bytea_catenate(t1, t2)); +} + +/* + * byteaoverlay + * Replace specified substring of first string with second + * + * The SQL standard defines OVERLAY() in terms of substring and concatenation. + * This code is a direct implementation of what the standard says. + */ +Datum +byteaoverlay(PG_FUNCTION_ARGS) +{ + bytea *t1 = PG_GETARG_BYTEA_PP(0); + bytea *t2 = PG_GETARG_BYTEA_PP(1); + int sp = PG_GETARG_INT32(2); /* substring start position */ + int sl = PG_GETARG_INT32(3); /* substring length */ + + PG_RETURN_BYTEA_P(bytea_overlay(t1, t2, sp, sl)); +} + +Datum +byteaoverlay_no_len(PG_FUNCTION_ARGS) +{ + bytea *t1 = PG_GETARG_BYTEA_PP(0); + bytea *t2 = PG_GETARG_BYTEA_PP(1); + int sp = PG_GETARG_INT32(2); /* substring start position */ + int sl; + + sl = VARSIZE_ANY_EXHDR(t2); /* defaults to length(t2) */ + PG_RETURN_BYTEA_P(bytea_overlay(t1, t2, sp, sl)); +} + +/* + * bytea_substr() + * Return a substring starting at the specified position. + * Cloned from text_substr and modified as required. + * + * Input: + * - string + * - starting position (is one-based) + * - string length (optional) + * + * If the starting position is zero or less, then return from the start of the string + * adjusting the length to be consistent with the "negative start" per SQL. + * If the length is less than zero, an ERROR is thrown. If no third argument + * (length) is provided, the length to the end of the string is assumed. + */ +Datum +bytea_substr(PG_FUNCTION_ARGS) +{ + PG_RETURN_BYTEA_P(bytea_substring(PG_GETARG_DATUM(0), + PG_GETARG_INT32(1), + PG_GETARG_INT32(2), + false)); +} + +/* + * bytea_substr_no_len - + * Wrapper to avoid opr_sanity failure due to + * one function accepting a different number of args. + */ +Datum +bytea_substr_no_len(PG_FUNCTION_ARGS) +{ + PG_RETURN_BYTEA_P(bytea_substring(PG_GETARG_DATUM(0), + PG_GETARG_INT32(1), + -1, + true)); +} + +/* + * bit_count + */ +Datum +bytea_bit_count(PG_FUNCTION_ARGS) +{ + bytea *t1 = PG_GETARG_BYTEA_PP(0); + + PG_RETURN_INT64(pg_popcount(VARDATA_ANY(t1), VARSIZE_ANY_EXHDR(t1))); +} + +/* + * byteapos - + * Return the position of the specified substring. + * Implements the SQL POSITION() function. + * Cloned from textpos and modified as required. + */ +Datum +byteapos(PG_FUNCTION_ARGS) +{ + bytea *t1 = PG_GETARG_BYTEA_PP(0); + bytea *t2 = PG_GETARG_BYTEA_PP(1); + int pos; + int px, + p; + int len1, + len2; + char *p1, + *p2; + + len1 = VARSIZE_ANY_EXHDR(t1); + len2 = VARSIZE_ANY_EXHDR(t2); + + if (len2 <= 0) + PG_RETURN_INT32(1); /* result for empty pattern */ + + p1 = VARDATA_ANY(t1); + p2 = VARDATA_ANY(t2); + + pos = 0; + px = (len1 - len2); + for (p = 0; p <= px; p++) + { + if ((*p2 == *p1) && (memcmp(p1, p2, len2) == 0)) + { + pos = p + 1; + break; + }; + p1++; + }; + + PG_RETURN_INT32(pos); +} + +/*------------------------------------------------------------- + * byteaGetByte + * + * this routine treats "bytea" as an array of bytes. + * It returns the Nth byte (a number between 0 and 255). + *------------------------------------------------------------- + */ +Datum +byteaGetByte(PG_FUNCTION_ARGS) +{ + bytea *v = PG_GETARG_BYTEA_PP(0); + int32 n = PG_GETARG_INT32(1); + int len; + int byte; + + len = VARSIZE_ANY_EXHDR(v); + + if (n < 0 || n >= len) + ereport(ERROR, + (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), + errmsg("index %d out of valid range, 0..%d", + n, len - 1))); + + byte = ((unsigned char *) VARDATA_ANY(v))[n]; + + PG_RETURN_INT32(byte); +} + +/*------------------------------------------------------------- + * byteaGetBit + * + * This routine treats a "bytea" type like an array of bits. + * It returns the value of the Nth bit (0 or 1). + * + *------------------------------------------------------------- + */ +Datum +byteaGetBit(PG_FUNCTION_ARGS) +{ + bytea *v = PG_GETARG_BYTEA_PP(0); + int64 n = PG_GETARG_INT64(1); + int byteNo, + bitNo; + int len; + int byte; + + len = VARSIZE_ANY_EXHDR(v); + + if (n < 0 || n >= (int64) len * 8) + ereport(ERROR, + (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), + errmsg("index %" PRId64 " out of valid range, 0..%" PRId64, + n, (int64) len * 8 - 1))); + + /* n/8 is now known < len, so safe to cast to int */ + byteNo = (int) (n / 8); + bitNo = (int) (n % 8); + + byte = ((unsigned char *) VARDATA_ANY(v))[byteNo]; + + if (byte & (1 << bitNo)) + PG_RETURN_INT32(1); + else + PG_RETURN_INT32(0); +} + +/*------------------------------------------------------------- + * byteaSetByte + * + * Given an instance of type 'bytea' creates a new one with + * the Nth byte set to the given value. + * + *------------------------------------------------------------- + */ +Datum +byteaSetByte(PG_FUNCTION_ARGS) +{ + bytea *res = PG_GETARG_BYTEA_P_COPY(0); + int32 n = PG_GETARG_INT32(1); + int32 newByte = PG_GETARG_INT32(2); + int len; + + len = VARSIZE(res) - VARHDRSZ; + + if (n < 0 || n >= len) + ereport(ERROR, + (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), + errmsg("index %d out of valid range, 0..%d", + n, len - 1))); + + /* + * Now set the byte. + */ + ((unsigned char *) VARDATA(res))[n] = newByte; + + PG_RETURN_BYTEA_P(res); +} + +/*------------------------------------------------------------- + * byteaSetBit + * + * Given an instance of type 'bytea' creates a new one with + * the Nth bit set to the given value. + * + *------------------------------------------------------------- + */ +Datum +byteaSetBit(PG_FUNCTION_ARGS) +{ + bytea *res = PG_GETARG_BYTEA_P_COPY(0); + int64 n = PG_GETARG_INT64(1); + int32 newBit = PG_GETARG_INT32(2); + int len; + int oldByte, + newByte; + int byteNo, + bitNo; + + len = VARSIZE(res) - VARHDRSZ; + + if (n < 0 || n >= (int64) len * 8) + ereport(ERROR, + (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), + errmsg("index %" PRId64 " out of valid range, 0..%" PRId64, + n, (int64) len * 8 - 1))); + + /* n/8 is now known < len, so safe to cast to int */ + byteNo = (int) (n / 8); + bitNo = (int) (n % 8); + + /* + * sanity check! + */ + if (newBit != 0 && newBit != 1) + ereport(ERROR, + (errcode(ERRCODE_INVALID_PARAMETER_VALUE), + errmsg("new bit must be 0 or 1"))); + + /* + * Update the byte. + */ + oldByte = ((unsigned char *) VARDATA(res))[byteNo]; + + if (newBit == 0) + newByte = oldByte & (~(1 << bitNo)); + else + newByte = oldByte | (1 << bitNo); + + ((unsigned char *) VARDATA(res))[byteNo] = newByte; + + PG_RETURN_BYTEA_P(res); +} + +/* + * Return reversed bytea + */ +Datum +bytea_reverse(PG_FUNCTION_ARGS) +{ + bytea *v = PG_GETARG_BYTEA_PP(0); + const char *p = VARDATA_ANY(v); + int len = VARSIZE_ANY_EXHDR(v); + const char *endp = p + len; + bytea *result = palloc(len + VARHDRSZ); + char *dst = (char *) VARDATA(result) + len; + + SET_VARSIZE(result, len + VARHDRSZ); + + while (p < endp) + *(--dst) = *p++; + + PG_RETURN_BYTEA_P(result); +} + + +/***************************************************************************** + * Comparison Functions used for bytea + * + * Note: btree indexes need these routines not to leak memory; therefore, + * be careful to free working copies of toasted datums. Most places don't + * need to be so careful. + *****************************************************************************/ + +Datum +byteaeq(PG_FUNCTION_ARGS) +{ + Datum arg1 = PG_GETARG_DATUM(0); + Datum arg2 = PG_GETARG_DATUM(1); + bool result; + Size len1, + len2; + + /* + * We can use a fast path for unequal lengths, which might save us from + * having to detoast one or both values. + */ + len1 = toast_raw_datum_size(arg1); + len2 = toast_raw_datum_size(arg2); + if (len1 != len2) + result = false; + else + { + bytea *barg1 = DatumGetByteaPP(arg1); + bytea *barg2 = DatumGetByteaPP(arg2); + + result = (memcmp(VARDATA_ANY(barg1), VARDATA_ANY(barg2), + len1 - VARHDRSZ) == 0); + + PG_FREE_IF_COPY(barg1, 0); + PG_FREE_IF_COPY(barg2, 1); + } + + PG_RETURN_BOOL(result); +} + +Datum +byteane(PG_FUNCTION_ARGS) +{ + Datum arg1 = PG_GETARG_DATUM(0); + Datum arg2 = PG_GETARG_DATUM(1); + bool result; + Size len1, + len2; + + /* + * We can use a fast path for unequal lengths, which might save us from + * having to detoast one or both values. + */ + len1 = toast_raw_datum_size(arg1); + len2 = toast_raw_datum_size(arg2); + if (len1 != len2) + result = true; + else + { + bytea *barg1 = DatumGetByteaPP(arg1); + bytea *barg2 = DatumGetByteaPP(arg2); + + result = (memcmp(VARDATA_ANY(barg1), VARDATA_ANY(barg2), + len1 - VARHDRSZ) != 0); + + PG_FREE_IF_COPY(barg1, 0); + PG_FREE_IF_COPY(barg2, 1); + } + + PG_RETURN_BOOL(result); +} + +Datum +bytealt(PG_FUNCTION_ARGS) +{ + bytea *arg1 = PG_GETARG_BYTEA_PP(0); + bytea *arg2 = PG_GETARG_BYTEA_PP(1); + int len1, + len2; + int cmp; + + len1 = VARSIZE_ANY_EXHDR(arg1); + len2 = VARSIZE_ANY_EXHDR(arg2); + + cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); + + PG_FREE_IF_COPY(arg1, 0); + PG_FREE_IF_COPY(arg2, 1); + + PG_RETURN_BOOL((cmp < 0) || ((cmp == 0) && (len1 < len2))); +} + +Datum +byteale(PG_FUNCTION_ARGS) +{ + bytea *arg1 = PG_GETARG_BYTEA_PP(0); + bytea *arg2 = PG_GETARG_BYTEA_PP(1); + int len1, + len2; + int cmp; + + len1 = VARSIZE_ANY_EXHDR(arg1); + len2 = VARSIZE_ANY_EXHDR(arg2); + + cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); + + PG_FREE_IF_COPY(arg1, 0); + PG_FREE_IF_COPY(arg2, 1); + + PG_RETURN_BOOL((cmp < 0) || ((cmp == 0) && (len1 <= len2))); +} + +Datum +byteagt(PG_FUNCTION_ARGS) +{ + bytea *arg1 = PG_GETARG_BYTEA_PP(0); + bytea *arg2 = PG_GETARG_BYTEA_PP(1); + int len1, + len2; + int cmp; + + len1 = VARSIZE_ANY_EXHDR(arg1); + len2 = VARSIZE_ANY_EXHDR(arg2); + + cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); + + PG_FREE_IF_COPY(arg1, 0); + PG_FREE_IF_COPY(arg2, 1); + + PG_RETURN_BOOL((cmp > 0) || ((cmp == 0) && (len1 > len2))); +} + +Datum +byteage(PG_FUNCTION_ARGS) +{ + bytea *arg1 = PG_GETARG_BYTEA_PP(0); + bytea *arg2 = PG_GETARG_BYTEA_PP(1); + int len1, + len2; + int cmp; + + len1 = VARSIZE_ANY_EXHDR(arg1); + len2 = VARSIZE_ANY_EXHDR(arg2); + + cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); + + PG_FREE_IF_COPY(arg1, 0); + PG_FREE_IF_COPY(arg2, 1); + + PG_RETURN_BOOL((cmp > 0) || ((cmp == 0) && (len1 >= len2))); +} + +Datum +byteacmp(PG_FUNCTION_ARGS) +{ + bytea *arg1 = PG_GETARG_BYTEA_PP(0); + bytea *arg2 = PG_GETARG_BYTEA_PP(1); + int len1, + len2; + int cmp; + + len1 = VARSIZE_ANY_EXHDR(arg1); + len2 = VARSIZE_ANY_EXHDR(arg2); + + cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); + if ((cmp == 0) && (len1 != len2)) + cmp = (len1 < len2) ? -1 : 1; + + PG_FREE_IF_COPY(arg1, 0); + PG_FREE_IF_COPY(arg2, 1); + + PG_RETURN_INT32(cmp); +} + +Datum +bytea_larger(PG_FUNCTION_ARGS) +{ + bytea *arg1 = PG_GETARG_BYTEA_PP(0); + bytea *arg2 = PG_GETARG_BYTEA_PP(1); + bytea *result; + int len1, + len2; + int cmp; + + len1 = VARSIZE_ANY_EXHDR(arg1); + len2 = VARSIZE_ANY_EXHDR(arg2); + + cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); + result = ((cmp > 0) || ((cmp == 0) && (len1 > len2)) ? arg1 : arg2); + + PG_RETURN_BYTEA_P(result); +} + +Datum +bytea_smaller(PG_FUNCTION_ARGS) +{ + bytea *arg1 = PG_GETARG_BYTEA_PP(0); + bytea *arg2 = PG_GETARG_BYTEA_PP(1); + bytea *result; + int len1, + len2; + int cmp; + + len1 = VARSIZE_ANY_EXHDR(arg1); + len2 = VARSIZE_ANY_EXHDR(arg2); + + cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); + result = ((cmp < 0) || ((cmp == 0) && (len1 < len2)) ? arg1 : arg2); + + PG_RETURN_BYTEA_P(result); +} + +Datum +bytea_sortsupport(PG_FUNCTION_ARGS) +{ + SortSupport ssup = (SortSupport) PG_GETARG_POINTER(0); + MemoryContext oldcontext; + + oldcontext = MemoryContextSwitchTo(ssup->ssup_cxt); + + /* Use generic string SortSupport, forcing "C" collation */ + varstr_sortsupport(ssup, BYTEAOID, C_COLLATION_OID); + + MemoryContextSwitchTo(oldcontext); + + PG_RETURN_VOID(); +} + +/* Cast bytea -> int2 */ +Datum +bytea_int2(PG_FUNCTION_ARGS) +{ + bytea *v = PG_GETARG_BYTEA_PP(0); + int len = VARSIZE_ANY_EXHDR(v); + uint16 result; + + /* Check that the byte array is not too long */ + if (len > sizeof(result)) + ereport(ERROR, + errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), + errmsg("smallint out of range")); + + /* Convert it to an integer; most significant bytes come first */ + result = 0; + for (int i = 0; i < len; i++) + { + result <<= BITS_PER_BYTE; + result |= ((unsigned char *) VARDATA_ANY(v))[i]; + } + + PG_RETURN_INT16(result); +} + +/* Cast bytea -> int4 */ +Datum +bytea_int4(PG_FUNCTION_ARGS) +{ + bytea *v = PG_GETARG_BYTEA_PP(0); + int len = VARSIZE_ANY_EXHDR(v); + uint32 result; + + /* Check that the byte array is not too long */ + if (len > sizeof(result)) + ereport(ERROR, + errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), + errmsg("integer out of range")); + + /* Convert it to an integer; most significant bytes come first */ + result = 0; + for (int i = 0; i < len; i++) + { + result <<= BITS_PER_BYTE; + result |= ((unsigned char *) VARDATA_ANY(v))[i]; + } + + PG_RETURN_INT32(result); +} + +/* Cast bytea -> int8 */ +Datum +bytea_int8(PG_FUNCTION_ARGS) +{ + bytea *v = PG_GETARG_BYTEA_PP(0); + int len = VARSIZE_ANY_EXHDR(v); + uint64 result; + + /* Check that the byte array is not too long */ + if (len > sizeof(result)) + ereport(ERROR, + errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), + errmsg("bigint out of range")); + + /* Convert it to an integer; most significant bytes come first */ + result = 0; + for (int i = 0; i < len; i++) + { + result <<= BITS_PER_BYTE; + result |= ((unsigned char *) VARDATA_ANY(v))[i]; + } + + PG_RETURN_INT64(result); +} + +/* Cast int2 -> bytea; can just use int2send() */ +Datum +int2_bytea(PG_FUNCTION_ARGS) +{ + return int2send(fcinfo); +} + +/* Cast int4 -> bytea; can just use int4send() */ +Datum +int4_bytea(PG_FUNCTION_ARGS) +{ + return int4send(fcinfo); +} + +/* Cast int8 -> bytea; can just use int8send() */ +Datum +int8_bytea(PG_FUNCTION_ARGS) +{ + return int8send(fcinfo); +} diff --git a/src/backend/utils/adt/date.c b/src/backend/utils/adt/date.c index 4227ab1a72b..344f58b92f7 100644 --- a/src/backend/utils/adt/date.c +++ b/src/backend/utils/adt/date.c @@ -1363,10 +1363,35 @@ timestamp_date(PG_FUNCTION_ARGS) { Timestamp timestamp = PG_GETARG_TIMESTAMP(0); DateADT result; + + result = timestamp2date_opt_overflow(timestamp, NULL); + PG_RETURN_DATEADT(result); +} + +/* + * Convert timestamp to date. + * + * On successful conversion, *overflow is set to zero if it's not NULL. + * + * If the timestamp is finite but out of the valid range for date, then: + * if overflow is NULL, we throw an out-of-range error. + * if overflow is not NULL, we store +1 or -1 there to indicate the sign + * of the overflow, and return the appropriate date infinity. + * + * Note: given the ranges of the types, overflow is only possible at + * the minimum end of the range, but we don't assume that in this code. + */ +DateADT +timestamp2date_opt_overflow(Timestamp timestamp, int *overflow) +{ + DateADT result; struct pg_tm tt, *tm = &tt; fsec_t fsec; + if (overflow) + *overflow = 0; + if (TIMESTAMP_IS_NOBEGIN(timestamp)) DATE_NOBEGIN(result); else if (TIMESTAMP_IS_NOEND(timestamp)) @@ -1374,14 +1399,30 @@ timestamp_date(PG_FUNCTION_ARGS) else { if (timestamp2tm(timestamp, NULL, tm, &fsec, NULL, NULL) != 0) + { + if (overflow) + { + if (timestamp < 0) + { + *overflow = -1; + DATE_NOBEGIN(result); + } + else + { + *overflow = 1; /* not actually reachable */ + DATE_NOEND(result); + } + return result; + } ereport(ERROR, (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE), errmsg("timestamp out of range"))); + } result = date2j(tm->tm_year, tm->tm_mon, tm->tm_mday) - POSTGRES_EPOCH_JDATE; } - PG_RETURN_DATEADT(result); + return result; } @@ -1408,11 +1449,36 @@ timestamptz_date(PG_FUNCTION_ARGS) { TimestampTz timestamp = PG_GETARG_TIMESTAMP(0); DateADT result; + + result = timestamptz2date_opt_overflow(timestamp, NULL); + PG_RETURN_DATEADT(result); +} + +/* + * Convert timestamptz to date. + * + * On successful conversion, *overflow is set to zero if it's not NULL. + * + * If the timestamptz is finite but out of the valid range for date, then: + * if overflow is NULL, we throw an out-of-range error. + * if overflow is not NULL, we store +1 or -1 there to indicate the sign + * of the overflow, and return the appropriate date infinity. + * + * Note: given the ranges of the types, overflow is only possible at + * the minimum end of the range, but we don't assume that in this code. + */ +DateADT +timestamptz2date_opt_overflow(TimestampTz timestamp, int *overflow) +{ + DateADT result; struct pg_tm tt, *tm = &tt; fsec_t fsec; int tz; + if (overflow) + *overflow = 0; + if (TIMESTAMP_IS_NOBEGIN(timestamp)) DATE_NOBEGIN(result); else if (TIMESTAMP_IS_NOEND(timestamp)) @@ -1420,14 +1486,30 @@ timestamptz_date(PG_FUNCTION_ARGS) else { if (timestamp2tm(timestamp, &tz, tm, &fsec, NULL, NULL) != 0) + { + if (overflow) + { + if (timestamp < 0) + { + *overflow = -1; + DATE_NOBEGIN(result); + } + else + { + *overflow = 1; /* not actually reachable */ + DATE_NOEND(result); + } + return result; + } ereport(ERROR, (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE), errmsg("timestamp out of range"))); + } result = date2j(tm->tm_year, tm->tm_mon, tm->tm_mday) - POSTGRES_EPOCH_JDATE; } - PG_RETURN_DATEADT(result); + return result; } diff --git a/src/backend/utils/adt/float.c b/src/backend/utils/adt/float.c index 6d20ae07ae7..7b97d2be6ca 100644 --- a/src/backend/utils/adt/float.c +++ b/src/backend/utils/adt/float.c @@ -4065,10 +4065,11 @@ float84ge(PG_FUNCTION_ARGS) * in the histogram. width_bucket() returns an integer indicating the * bucket number that 'operand' belongs to in an equiwidth histogram * with the specified characteristics. An operand smaller than the - * lower bound is assigned to bucket 0. An operand greater than the - * upper bound is assigned to an additional bucket (with number - * count+1). We don't allow "NaN" for any of the float8 inputs, and we - * don't allow either of the histogram bounds to be +/- infinity. + * lower bound is assigned to bucket 0. An operand greater than or equal + * to the upper bound is assigned to an additional bucket (with number + * count+1). We don't allow the histogram bounds to be NaN or +/- infinity, + * but we do allow those values for the operand (taking NaN to be larger + * than any other value, as we do in comparisons). */ Datum width_bucket_float8(PG_FUNCTION_ARGS) @@ -4084,12 +4085,11 @@ width_bucket_float8(PG_FUNCTION_ARGS) (errcode(ERRCODE_INVALID_ARGUMENT_FOR_WIDTH_BUCKET_FUNCTION), errmsg("count must be greater than zero"))); - if (isnan(operand) || isnan(bound1) || isnan(bound2)) + if (isnan(bound1) || isnan(bound2)) ereport(ERROR, (errcode(ERRCODE_INVALID_ARGUMENT_FOR_WIDTH_BUCKET_FUNCTION), - errmsg("operand, lower bound, and upper bound cannot be NaN"))); + errmsg("lower and upper bounds cannot be NaN"))); - /* Note that we allow "operand" to be infinite */ if (isinf(bound1) || isinf(bound2)) ereport(ERROR, (errcode(ERRCODE_INVALID_ARGUMENT_FOR_WIDTH_BUCKET_FUNCTION), @@ -4097,15 +4097,15 @@ width_bucket_float8(PG_FUNCTION_ARGS) if (bound1 < bound2) { - if (operand < bound1) - result = 0; - else if (operand >= bound2) + if (isnan(operand) || operand >= bound2) { if (pg_add_s32_overflow(count, 1, &result)) ereport(ERROR, (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), errmsg("integer out of range"))); } + else if (operand < bound1) + result = 0; else { if (!isinf(bound2 - bound1)) @@ -4135,7 +4135,7 @@ width_bucket_float8(PG_FUNCTION_ARGS) } else if (bound1 > bound2) { - if (operand > bound1) + if (isnan(operand) || operand > bound1) result = 0; else if (operand <= bound2) { diff --git a/src/backend/utils/adt/formatting.c b/src/backend/utils/adt/formatting.c index 5bd1e01f7e4..1d05481181d 100644 --- a/src/backend/utils/adt/formatting.c +++ b/src/backend/utils/adt/formatting.c @@ -3590,14 +3590,15 @@ DCH_from_char(FormatNode *node, const char *in, TmFromChar *out, if (matched < 2) ereturn(escontext,, (errcode(ERRCODE_INVALID_DATETIME_FORMAT), - errmsg("invalid input string for \"Y,YYY\""))); + errmsg("invalid value \"%s\" for \"%s\"", + s, "Y,YYY"))); /* years += (millennia * 1000); */ if (pg_mul_s32_overflow(millennia, 1000, &millennia) || pg_add_s32_overflow(years, millennia, &years)) ereturn(escontext,, (errcode(ERRCODE_DATETIME_FIELD_OVERFLOW), - errmsg("value for \"Y,YYY\" in source string is out of range"))); + errmsg("value for \"%s\" in source string is out of range", "Y,YYY"))); if (!from_char_set_int(&out->year, years, n, escontext)) return; diff --git a/src/backend/utils/adt/inet_net_pton.c b/src/backend/utils/adt/inet_net_pton.c index ef2236d9f04..3b0db2a3799 100644 --- a/src/backend/utils/adt/inet_net_pton.c +++ b/src/backend/utils/adt/inet_net_pton.c @@ -115,8 +115,7 @@ inet_cidr_pton_ipv4(const char *src, u_char *dst, size_t size) src++; /* skip x or X. */ while ((ch = *src++) != '\0' && isxdigit((unsigned char) ch)) { - if (isupper((unsigned char) ch)) - ch = tolower((unsigned char) ch); + ch = pg_ascii_tolower((unsigned char) ch); n = strchr(xdigits, ch) - xdigits; assert(n >= 0 && n <= 15); if (dirty == 0) diff --git a/src/backend/utils/adt/jsonb_gin.c b/src/backend/utils/adt/jsonb_gin.c index c1950792b5a..9b56248cf0b 100644 --- a/src/backend/utils/adt/jsonb_gin.c +++ b/src/backend/utils/adt/jsonb_gin.c @@ -896,8 +896,8 @@ gin_extract_jsonb_query(PG_FUNCTION_ARGS) continue; /* We rely on the array elements not being toasted */ entries[j++] = make_text_key(JGINFLAG_KEY, - VARDATA_ANY(key_datums[i]), - VARSIZE_ANY_EXHDR(key_datums[i])); + VARDATA_ANY(DatumGetPointer(key_datums[i])), + VARSIZE_ANY_EXHDR(DatumGetPointer(key_datums[i]))); } *nentries = j; diff --git a/src/backend/utils/adt/jsonb_op.c b/src/backend/utils/adt/jsonb_op.c index fa5603f26e1..51d38e321fb 100644 --- a/src/backend/utils/adt/jsonb_op.c +++ b/src/backend/utils/adt/jsonb_op.c @@ -63,8 +63,8 @@ jsonb_exists_any(PG_FUNCTION_ARGS) strVal.type = jbvString; /* We rely on the array elements not being toasted */ - strVal.val.string.val = VARDATA_ANY(key_datums[i]); - strVal.val.string.len = VARSIZE_ANY_EXHDR(key_datums[i]); + strVal.val.string.val = VARDATA_ANY(DatumGetPointer(key_datums[i])); + strVal.val.string.len = VARSIZE_ANY_EXHDR(DatumGetPointer(key_datums[i])); if (findJsonbValueFromContainer(&jb->root, JB_FOBJECT | JB_FARRAY, @@ -96,8 +96,8 @@ jsonb_exists_all(PG_FUNCTION_ARGS) strVal.type = jbvString; /* We rely on the array elements not being toasted */ - strVal.val.string.val = VARDATA_ANY(key_datums[i]); - strVal.val.string.len = VARSIZE_ANY_EXHDR(key_datums[i]); + strVal.val.string.val = VARDATA_ANY(DatumGetPointer(key_datums[i])); + strVal.val.string.len = VARSIZE_ANY_EXHDR(DatumGetPointer(key_datums[i])); if (findJsonbValueFromContainer(&jb->root, JB_FOBJECT | JB_FARRAY, diff --git a/src/backend/utils/adt/jsonb_util.c b/src/backend/utils/adt/jsonb_util.c index c8b6c15e059..82b807d067a 100644 --- a/src/backend/utils/adt/jsonb_util.c +++ b/src/backend/utils/adt/jsonb_util.c @@ -277,22 +277,16 @@ compareJsonbContainers(JsonbContainer *a, JsonbContainer *b) else { /* - * It's safe to assume that the types differed, and that the va - * and vb values passed were set. - * - * If the two values were of the same container type, then there'd - * have been a chance to observe the variation in the number of - * elements/pairs (when processing WJB_BEGIN_OBJECT, say). They're - * either two heterogeneously-typed containers, or a container and - * some scalar type. - * - * We don't have to consider the WJB_END_ARRAY and WJB_END_OBJECT - * cases here, because we would have seen the corresponding - * WJB_BEGIN_ARRAY and WJB_BEGIN_OBJECT tokens first, and - * concluded that they don't match. + * It's not possible for one iterator to report end of array or + * object while the other one reports something else, because we + * would have detected a length mismatch when we processed the + * container-start tokens above. Likewise we can't see WJB_DONE + * from one but not the other. So we have two different-type + * containers, or a container and some scalar type, or two + * different scalar types. Sort on the basis of the type code. */ - Assert(ra != WJB_END_ARRAY && ra != WJB_END_OBJECT); - Assert(rb != WJB_END_ARRAY && rb != WJB_END_OBJECT); + Assert(ra != WJB_DONE && ra != WJB_END_ARRAY && ra != WJB_END_OBJECT); + Assert(rb != WJB_DONE && rb != WJB_END_ARRAY && rb != WJB_END_OBJECT); Assert(va.type != vb.type); Assert(va.type != jbvBinary); @@ -852,15 +846,20 @@ JsonbIteratorInit(JsonbContainer *container) * It is our job to expand the jbvBinary representation without bothering them * with it. However, clients should not take it upon themselves to touch array * or Object element/pair buffers, since their element/pair pointers are - * garbage. Also, *val will not be set when returning WJB_END_ARRAY or - * WJB_END_OBJECT, on the assumption that it's only useful to access values - * when recursing in. + * garbage. + * + * *val is not meaningful when the result is WJB_DONE, WJB_END_ARRAY or + * WJB_END_OBJECT. However, we set val->type = jbvNull in those cases, + * so that callers may assume that val->type is always well-defined. */ JsonbIteratorToken JsonbIteratorNext(JsonbIterator **it, JsonbValue *val, bool skipNested) { if (*it == NULL) + { + val->type = jbvNull; return WJB_DONE; + } /* * When stepping into a nested container, we jump back here to start @@ -898,6 +897,7 @@ recurse: * nesting). */ *it = freeAndGetParent(*it); + val->type = jbvNull; return WJB_END_ARRAY; } @@ -951,6 +951,7 @@ recurse: * of nesting). */ *it = freeAndGetParent(*it); + val->type = jbvNull; return WJB_END_OBJECT; } else @@ -995,8 +996,10 @@ recurse: return WJB_VALUE; } - elog(ERROR, "invalid iterator state"); - return -1; + elog(ERROR, "invalid jsonb iterator state"); + /* satisfy compilers that don't know that elog(ERROR) doesn't return */ + val->type = jbvNull; + return WJB_DONE; } /* diff --git a/src/backend/utils/adt/jsonfuncs.c b/src/backend/utils/adt/jsonfuncs.c index bcb1720b6cd..370456408bf 100644 --- a/src/backend/utils/adt/jsonfuncs.c +++ b/src/backend/utils/adt/jsonfuncs.c @@ -4766,8 +4766,8 @@ jsonb_delete_array(PG_FUNCTION_ARGS) continue; /* We rely on the array elements not being toasted */ - keyptr = VARDATA_ANY(keys_elems[i]); - keylen = VARSIZE_ANY_EXHDR(keys_elems[i]); + keyptr = VARDATA_ANY(DatumGetPointer(keys_elems[i])); + keylen = VARSIZE_ANY_EXHDR(DatumGetPointer(keys_elems[i])); if (keylen == v.val.string.len && memcmp(keyptr, v.val.string.val, keylen) == 0) { diff --git a/src/backend/utils/adt/jsonpath_exec.c b/src/backend/utils/adt/jsonpath_exec.c index dbab24737ef..407041b14a1 100644 --- a/src/backend/utils/adt/jsonpath_exec.c +++ b/src/backend/utils/adt/jsonpath_exec.c @@ -3074,8 +3074,8 @@ JsonItemFromDatum(Datum val, Oid typid, int32 typmod, JsonbValue *res) case TEXTOID: case VARCHAROID: res->type = jbvString; - res->val.string.val = VARDATA_ANY(val); - res->val.string.len = VARSIZE_ANY_EXHDR(val); + res->val.string.val = VARDATA_ANY(DatumGetPointer(val)); + res->val.string.len = VARSIZE_ANY_EXHDR(DatumGetPointer(val)); break; case DATEOID: case TIMEOID: diff --git a/src/backend/utils/adt/like.c b/src/backend/utils/adt/like.c index 7f4cf614585..4216ac17f43 100644 --- a/src/backend/utils/adt/like.c +++ b/src/backend/utils/adt/like.c @@ -98,7 +98,7 @@ SB_lower_char(unsigned char c, pg_locale_t locale) else if (locale->is_default) return pg_tolower(c); else - return tolower_l(c, locale->info.lt); + return char_tolower(c, locale); } @@ -209,7 +209,17 @@ Generic_Text_IC_like(text *str, text *pat, Oid collation) * way. */ - if (pg_database_encoding_max_length() > 1 || (locale->provider == COLLPROVIDER_ICU)) + if (locale->ctype_is_c || + (char_tolower_enabled(locale) && + pg_database_encoding_max_length() == 1)) + { + p = VARDATA_ANY(pat); + plen = VARSIZE_ANY_EXHDR(pat); + s = VARDATA_ANY(str); + slen = VARSIZE_ANY_EXHDR(str); + return SB_IMatchText(s, slen, p, plen, locale); + } + else { pat = DatumGetTextPP(DirectFunctionCall1Coll(lower, collation, PointerGetDatum(pat))); @@ -224,14 +234,6 @@ Generic_Text_IC_like(text *str, text *pat, Oid collation) else return MB_MatchText(s, slen, p, plen, 0); } - else - { - p = VARDATA_ANY(pat); - plen = VARSIZE_ANY_EXHDR(pat); - s = VARDATA_ANY(str); - slen = VARSIZE_ANY_EXHDR(str); - return SB_IMatchText(s, slen, p, plen, locale); - } } /* diff --git a/src/backend/utils/adt/like_support.c b/src/backend/utils/adt/like_support.c index 8fdc677371f..999f23f86d5 100644 --- a/src/backend/utils/adt/like_support.c +++ b/src/backend/utils/adt/like_support.c @@ -1495,13 +1495,8 @@ pattern_char_isalpha(char c, bool is_multibyte, { if (locale->ctype_is_c) return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z'); - else if (is_multibyte && IS_HIGHBIT_SET(c)) - return true; - else if (locale->provider != COLLPROVIDER_LIBC) - return IS_HIGHBIT_SET(c) || - (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z'); else - return isalpha_l((unsigned char) c, locale->info.lt); + return char_is_cased(c, locale); } diff --git a/src/backend/utils/adt/mcxtfuncs.c b/src/backend/utils/adt/mcxtfuncs.c index 396c2f223b4..fe6dce9cba3 100644 --- a/src/backend/utils/adt/mcxtfuncs.c +++ b/src/backend/utils/adt/mcxtfuncs.c @@ -38,7 +38,7 @@ typedef struct MemoryContextId { MemoryContext context; int context_id; -} MemoryContextId; +} MemoryContextId; /* * int_list_to_array diff --git a/src/backend/utils/adt/meson.build b/src/backend/utils/adt/meson.build index 244f48f4fd7..ed9bbd7b926 100644 --- a/src/backend/utils/adt/meson.build +++ b/src/backend/utils/adt/meson.build @@ -12,6 +12,7 @@ backend_sources += files( 'arrayutils.c', 'ascii.c', 'bool.c', + 'bytea.c', 'cash.c', 'char.c', 'cryptohashfuncs.c', diff --git a/src/backend/utils/adt/multirangetypes.c b/src/backend/utils/adt/multirangetypes.c index cd84ced5b48..46f2ec0c29f 100644 --- a/src/backend/utils/adt/multirangetypes.c +++ b/src/backend/utils/adt/multirangetypes.c @@ -394,12 +394,13 @@ multirange_send(PG_FUNCTION_ARGS) for (int i = 0; i < range_count; i++) { Datum range; + bytea *outputbytes; range = RangeTypePGetDatum(ranges[i]); - range = PointerGetDatum(SendFunctionCall(&cache->typioproc, range)); + outputbytes = SendFunctionCall(&cache->typioproc, range); - pq_sendint32(buf, VARSIZE(range) - VARHDRSZ); - pq_sendbytes(buf, VARDATA(range), VARSIZE(range) - VARHDRSZ); + pq_sendint32(buf, VARSIZE(outputbytes) - VARHDRSZ); + pq_sendbytes(buf, VARDATA(outputbytes), VARSIZE(outputbytes) - VARHDRSZ); } PG_RETURN_BYTEA_P(pq_endtypsend(buf)); @@ -2833,7 +2834,7 @@ hash_multirange(PG_FUNCTION_ARGS) upper_hash = 0; /* Merge hashes of flags and bounds */ - range_hash = hash_uint32((uint32) flags); + range_hash = hash_bytes_uint32((uint32) flags); range_hash ^= lower_hash; range_hash = pg_rotate_left32(range_hash, 1); range_hash ^= upper_hash; diff --git a/src/backend/utils/adt/network.c b/src/backend/utils/adt/network.c index f03fcc1147b..9fd211b2d45 100644 --- a/src/backend/utils/adt/network.c +++ b/src/backend/utils/adt/network.c @@ -12,8 +12,6 @@ #include <netinet/in.h> #include <arpa/inet.h> -#include "access/stratnum.h" -#include "catalog/pg_opfamily.h" #include "catalog/pg_type.h" #include "common/hashfn.h" #include "common/ip.h" diff --git a/src/backend/utils/adt/network_spgist.c b/src/backend/utils/adt/network_spgist.c index a84747d9275..602276a35c3 100644 --- a/src/backend/utils/adt/network_spgist.c +++ b/src/backend/utils/adt/network_spgist.c @@ -37,7 +37,6 @@ #include "catalog/pg_type.h" #include "utils/fmgrprotos.h" #include "utils/inet.h" -#include "varatt.h" static int inet_spg_node_number(const inet *val, int commonbits); diff --git a/src/backend/utils/adt/numeric.c b/src/backend/utils/adt/numeric.c index 40dcbc7b671..c9233565d57 100644 --- a/src/backend/utils/adt/numeric.c +++ b/src/backend/utils/adt/numeric.c @@ -1958,9 +1958,11 @@ generate_series_numeric_support(PG_FUNCTION_ARGS) * in the histogram. width_bucket() returns an integer indicating the * bucket number that 'operand' belongs to in an equiwidth histogram * with the specified characteristics. An operand smaller than the - * lower bound is assigned to bucket 0. An operand greater than the - * upper bound is assigned to an additional bucket (with number - * count+1). We don't allow "NaN" for any of the numeric arguments. + * lower bound is assigned to bucket 0. An operand greater than or equal + * to the upper bound is assigned to an additional bucket (with number + * count+1). We don't allow the histogram bounds to be NaN or +/- infinity, + * but we do allow those values for the operand (taking NaN to be larger + * than any other value, as we do in comparisons). */ Datum width_bucket_numeric(PG_FUNCTION_ARGS) @@ -1978,17 +1980,13 @@ width_bucket_numeric(PG_FUNCTION_ARGS) (errcode(ERRCODE_INVALID_ARGUMENT_FOR_WIDTH_BUCKET_FUNCTION), errmsg("count must be greater than zero"))); - if (NUMERIC_IS_SPECIAL(operand) || - NUMERIC_IS_SPECIAL(bound1) || - NUMERIC_IS_SPECIAL(bound2)) + if (NUMERIC_IS_SPECIAL(bound1) || NUMERIC_IS_SPECIAL(bound2)) { - if (NUMERIC_IS_NAN(operand) || - NUMERIC_IS_NAN(bound1) || - NUMERIC_IS_NAN(bound2)) + if (NUMERIC_IS_NAN(bound1) || NUMERIC_IS_NAN(bound2)) ereport(ERROR, (errcode(ERRCODE_INVALID_ARGUMENT_FOR_WIDTH_BUCKET_FUNCTION), - errmsg("operand, lower bound, and upper bound cannot be NaN"))); - /* We allow "operand" to be infinite; cmp_numerics will cope */ + errmsg("lower and upper bounds cannot be NaN"))); + if (NUMERIC_IS_INF(bound1) || NUMERIC_IS_INF(bound2)) ereport(ERROR, (errcode(ERRCODE_INVALID_ARGUMENT_FOR_WIDTH_BUCKET_FUNCTION), diff --git a/src/backend/utils/adt/pg_locale.c b/src/backend/utils/adt/pg_locale.c index f5e31c433a0..97c2ac1faf9 100644 --- a/src/backend/utils/adt/pg_locale.c +++ b/src/backend/utils/adt/pg_locale.c @@ -41,7 +41,6 @@ #include "mb/pg_wchar.h" #include "miscadmin.h" #include "utils/builtins.h" -#include "utils/formatting.h" #include "utils/guc_hooks.h" #include "utils/lsyscache.h" #include "utils/memutils.h" @@ -80,31 +79,6 @@ extern pg_locale_t create_pg_locale_icu(Oid collid, MemoryContext context); extern pg_locale_t create_pg_locale_libc(Oid collid, MemoryContext context); extern char *get_collation_actual_version_libc(const char *collcollate); -extern size_t strlower_builtin(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strtitle_builtin(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strupper_builtin(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strfold_builtin(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); - -extern size_t strlower_icu(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strtitle_icu(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strupper_icu(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strfold_icu(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); - -extern size_t strlower_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strtitle_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strupper_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); - /* GUC settings */ char *locale_messages; char *locale_monetary; @@ -1093,6 +1067,9 @@ create_pg_locale(Oid collid, MemoryContext context) Assert((result->collate_is_c && result->collate == NULL) || (!result->collate_is_c && result->collate != NULL)); + Assert((result->ctype_is_c && result->ctype == NULL) || + (!result->ctype_is_c && result->ctype != NULL)); + datum = SysCacheGetAttr(COLLOID, tp, Anum_pg_collation_collversion, &isnull); if (!isnull) @@ -1257,77 +1234,31 @@ size_t pg_strlower(char *dst, size_t dstsize, const char *src, ssize_t srclen, pg_locale_t locale) { - if (locale->provider == COLLPROVIDER_BUILTIN) - return strlower_builtin(dst, dstsize, src, srclen, locale); -#ifdef USE_ICU - else if (locale->provider == COLLPROVIDER_ICU) - return strlower_icu(dst, dstsize, src, srclen, locale); -#endif - else if (locale->provider == COLLPROVIDER_LIBC) - return strlower_libc(dst, dstsize, src, srclen, locale); - else - /* shouldn't happen */ - PGLOCALE_SUPPORT_ERROR(locale->provider); - - return 0; /* keep compiler quiet */ + return locale->ctype->strlower(dst, dstsize, src, srclen, locale); } size_t pg_strtitle(char *dst, size_t dstsize, const char *src, ssize_t srclen, pg_locale_t locale) { - if (locale->provider == COLLPROVIDER_BUILTIN) - return strtitle_builtin(dst, dstsize, src, srclen, locale); -#ifdef USE_ICU - else if (locale->provider == COLLPROVIDER_ICU) - return strtitle_icu(dst, dstsize, src, srclen, locale); -#endif - else if (locale->provider == COLLPROVIDER_LIBC) - return strtitle_libc(dst, dstsize, src, srclen, locale); - else - /* shouldn't happen */ - PGLOCALE_SUPPORT_ERROR(locale->provider); - - return 0; /* keep compiler quiet */ + return locale->ctype->strtitle(dst, dstsize, src, srclen, locale); } size_t pg_strupper(char *dst, size_t dstsize, const char *src, ssize_t srclen, pg_locale_t locale) { - if (locale->provider == COLLPROVIDER_BUILTIN) - return strupper_builtin(dst, dstsize, src, srclen, locale); -#ifdef USE_ICU - else if (locale->provider == COLLPROVIDER_ICU) - return strupper_icu(dst, dstsize, src, srclen, locale); -#endif - else if (locale->provider == COLLPROVIDER_LIBC) - return strupper_libc(dst, dstsize, src, srclen, locale); - else - /* shouldn't happen */ - PGLOCALE_SUPPORT_ERROR(locale->provider); - - return 0; /* keep compiler quiet */ + return locale->ctype->strupper(dst, dstsize, src, srclen, locale); } size_t pg_strfold(char *dst, size_t dstsize, const char *src, ssize_t srclen, pg_locale_t locale) { - if (locale->provider == COLLPROVIDER_BUILTIN) - return strfold_builtin(dst, dstsize, src, srclen, locale); -#ifdef USE_ICU - else if (locale->provider == COLLPROVIDER_ICU) - return strfold_icu(dst, dstsize, src, srclen, locale); -#endif - /* for libc, just use strlower */ - else if (locale->provider == COLLPROVIDER_LIBC) - return strlower_libc(dst, dstsize, src, srclen, locale); + if (locale->ctype->strfold) + return locale->ctype->strfold(dst, dstsize, src, srclen, locale); else - /* shouldn't happen */ - PGLOCALE_SUPPORT_ERROR(locale->provider); - - return 0; /* keep compiler quiet */ + return locale->ctype->strlower(dst, dstsize, src, srclen, locale); } /* @@ -1465,6 +1396,41 @@ pg_strnxfrm_prefix(char *dest, size_t destsize, const char *src, } /* + * char_is_cased() + * + * Fuzzy test of whether the given char is case-varying or not. The argument + * is a single byte, so in a multibyte encoding, just assume any non-ASCII + * char is case-varying. + */ +bool +char_is_cased(char ch, pg_locale_t locale) +{ + return locale->ctype->char_is_cased(ch, locale); +} + +/* + * char_tolower_enabled() + * + * Does the provider support char_tolower()? + */ +bool +char_tolower_enabled(pg_locale_t locale) +{ + return (locale->ctype->char_tolower != NULL); +} + +/* + * char_tolower() + * + * Convert char (single-byte encoding) to lowercase. + */ +char +char_tolower(unsigned char ch, pg_locale_t locale) +{ + return locale->ctype->char_tolower(ch, locale); +} + +/* * Return required encoding ID for the given locale, or -1 if any encoding is * valid for the locale. */ diff --git a/src/backend/utils/adt/pg_locale_builtin.c b/src/backend/utils/adt/pg_locale_builtin.c index f51768830cd..0c9fbdb40f2 100644 --- a/src/backend/utils/adt/pg_locale_builtin.c +++ b/src/backend/utils/adt/pg_locale_builtin.c @@ -18,22 +18,12 @@ #include "mb/pg_wchar.h" #include "miscadmin.h" #include "utils/builtins.h" -#include "utils/memutils.h" #include "utils/pg_locale.h" #include "utils/syscache.h" extern pg_locale_t create_pg_locale_builtin(Oid collid, MemoryContext context); extern char *get_collation_actual_version_builtin(const char *collcollate); -extern size_t strlower_builtin(char *dest, size_t destsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strtitle_builtin(char *dest, size_t destsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strupper_builtin(char *dest, size_t destsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strfold_builtin(char *dest, size_t destsize, const char *src, - ssize_t srclen, pg_locale_t locale); - struct WordBoundaryState { @@ -77,7 +67,7 @@ initcap_wbnext(void *state) return wbstate->len; } -size_t +static size_t strlower_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) { @@ -85,7 +75,7 @@ strlower_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen, locale->info.builtin.casemap_full); } -size_t +static size_t strtitle_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) { @@ -103,7 +93,7 @@ strtitle_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen, initcap_wbnext, &wbstate); } -size_t +static size_t strupper_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) { @@ -111,7 +101,7 @@ strupper_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen, locale->info.builtin.casemap_full); } -size_t +static size_t strfold_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) { @@ -119,6 +109,98 @@ strfold_builtin(char *dest, size_t destsize, const char *src, ssize_t srclen, locale->info.builtin.casemap_full); } +static bool +wc_isdigit_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_isdigit(wc, !locale->info.builtin.casemap_full); +} + +static bool +wc_isalpha_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_isalpha(wc); +} + +static bool +wc_isalnum_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_isalnum(wc, !locale->info.builtin.casemap_full); +} + +static bool +wc_isupper_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_isupper(wc); +} + +static bool +wc_islower_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_islower(wc); +} + +static bool +wc_isgraph_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_isgraph(wc); +} + +static bool +wc_isprint_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_isprint(wc); +} + +static bool +wc_ispunct_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_ispunct(wc, !locale->info.builtin.casemap_full); +} + +static bool +wc_isspace_builtin(pg_wchar wc, pg_locale_t locale) +{ + return pg_u_isspace(wc); +} + +static bool +char_is_cased_builtin(char ch, pg_locale_t locale) +{ + return IS_HIGHBIT_SET(ch) || + (ch >= 'A' && ch <= 'Z') || (ch >= 'a' && ch <= 'z'); +} + +static pg_wchar +wc_toupper_builtin(pg_wchar wc, pg_locale_t locale) +{ + return unicode_uppercase_simple(wc); +} + +static pg_wchar +wc_tolower_builtin(pg_wchar wc, pg_locale_t locale) +{ + return unicode_lowercase_simple(wc); +} + +static const struct ctype_methods ctype_methods_builtin = { + .strlower = strlower_builtin, + .strtitle = strtitle_builtin, + .strupper = strupper_builtin, + .strfold = strfold_builtin, + .wc_isdigit = wc_isdigit_builtin, + .wc_isalpha = wc_isalpha_builtin, + .wc_isalnum = wc_isalnum_builtin, + .wc_isupper = wc_isupper_builtin, + .wc_islower = wc_islower_builtin, + .wc_isgraph = wc_isgraph_builtin, + .wc_isprint = wc_isprint_builtin, + .wc_ispunct = wc_ispunct_builtin, + .wc_isspace = wc_isspace_builtin, + .char_is_cased = char_is_cased_builtin, + .wc_tolower = wc_tolower_builtin, + .wc_toupper = wc_toupper_builtin, +}; + pg_locale_t create_pg_locale_builtin(Oid collid, MemoryContext context) { @@ -158,10 +240,11 @@ create_pg_locale_builtin(Oid collid, MemoryContext context) result->info.builtin.locale = MemoryContextStrdup(context, locstr); result->info.builtin.casemap_full = (strcmp(locstr, "PG_UNICODE_FAST") == 0); - result->provider = COLLPROVIDER_BUILTIN; result->deterministic = true; result->collate_is_c = true; result->ctype_is_c = (strcmp(locstr, "C") == 0); + if (!result->ctype_is_c) + result->ctype = &ctype_methods_builtin; return result; } diff --git a/src/backend/utils/adt/pg_locale_icu.c b/src/backend/utils/adt/pg_locale_icu.c index a32c32a0744..96741e08269 100644 --- a/src/backend/utils/adt/pg_locale_icu.c +++ b/src/backend/utils/adt/pg_locale_icu.c @@ -48,19 +48,22 @@ #define TEXTBUFLEN 1024 extern pg_locale_t create_pg_locale_icu(Oid collid, MemoryContext context); -extern size_t strlower_icu(char *dest, size_t destsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strtitle_icu(char *dest, size_t destsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strupper_icu(char *dest, size_t destsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strfold_icu(char *dest, size_t destsize, const char *src, - ssize_t srclen, pg_locale_t locale); #ifdef USE_ICU extern UCollator *pg_ucol_open(const char *loc_str); +static size_t strlower_icu(char *dest, size_t destsize, const char *src, + ssize_t srclen, pg_locale_t locale); +static size_t strtitle_icu(char *dest, size_t destsize, const char *src, + ssize_t srclen, pg_locale_t locale); +static size_t strupper_icu(char *dest, size_t destsize, const char *src, + ssize_t srclen, pg_locale_t locale); +static size_t strfold_icu(char *dest, size_t destsize, const char *src, + ssize_t srclen, pg_locale_t locale); +static int strncoll_icu(const char *arg1, ssize_t len1, + const char *arg2, ssize_t len2, + pg_locale_t locale); static size_t strnxfrm_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale); @@ -118,6 +121,25 @@ static int32_t u_strFoldCase_default(UChar *dest, int32_t destCapacity, const char *locale, UErrorCode *pErrorCode); +static bool +char_is_cased_icu(char ch, pg_locale_t locale) +{ + return IS_HIGHBIT_SET(ch) || + (ch >= 'A' && ch <= 'Z') || (ch >= 'a' && ch <= 'z'); +} + +static pg_wchar +toupper_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_toupper(wc); +} + +static pg_wchar +tolower_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_tolower(wc); +} + static const struct collate_methods collate_methods_icu = { .strncoll = strncoll_icu, .strnxfrm = strnxfrm_icu, @@ -136,6 +158,78 @@ static const struct collate_methods collate_methods_icu_utf8 = { .strxfrm_is_safe = true, }; +static bool +wc_isdigit_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_isdigit(wc); +} + +static bool +wc_isalpha_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_isalpha(wc); +} + +static bool +wc_isalnum_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_isalnum(wc); +} + +static bool +wc_isupper_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_isupper(wc); +} + +static bool +wc_islower_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_islower(wc); +} + +static bool +wc_isgraph_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_isgraph(wc); +} + +static bool +wc_isprint_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_isprint(wc); +} + +static bool +wc_ispunct_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_ispunct(wc); +} + +static bool +wc_isspace_icu(pg_wchar wc, pg_locale_t locale) +{ + return u_isspace(wc); +} + +static const struct ctype_methods ctype_methods_icu = { + .strlower = strlower_icu, + .strtitle = strtitle_icu, + .strupper = strupper_icu, + .strfold = strfold_icu, + .wc_isdigit = wc_isdigit_icu, + .wc_isalpha = wc_isalpha_icu, + .wc_isalnum = wc_isalnum_icu, + .wc_isupper = wc_isupper_icu, + .wc_islower = wc_islower_icu, + .wc_isgraph = wc_isgraph_icu, + .wc_isprint = wc_isprint_icu, + .wc_ispunct = wc_ispunct_icu, + .wc_isspace = wc_isspace_icu, + .char_is_cased = char_is_cased_icu, + .wc_toupper = toupper_icu, + .wc_tolower = tolower_icu, +}; #endif pg_locale_t @@ -198,7 +292,6 @@ create_pg_locale_icu(Oid collid, MemoryContext context) result = MemoryContextAllocZero(context, sizeof(struct pg_locale_struct)); result->info.icu.locale = MemoryContextStrdup(context, iculocstr); result->info.icu.ucol = collator; - result->provider = COLLPROVIDER_ICU; result->deterministic = deterministic; result->collate_is_c = false; result->ctype_is_c = false; @@ -206,6 +299,7 @@ create_pg_locale_icu(Oid collid, MemoryContext context) result->collate = &collate_methods_icu_utf8; else result->collate = &collate_methods_icu; + result->ctype = &ctype_methods_icu; return result; #else @@ -379,7 +473,7 @@ make_icu_collator(const char *iculocstr, const char *icurules) } } -size_t +static size_t strlower_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) { @@ -399,7 +493,7 @@ strlower_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, return result_len; } -size_t +static size_t strtitle_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) { @@ -419,7 +513,7 @@ strtitle_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, return result_len; } -size_t +static size_t strupper_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) { @@ -439,7 +533,7 @@ strupper_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, return result_len; } -size_t +static size_t strfold_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) { @@ -474,8 +568,6 @@ strncoll_icu_utf8(const char *arg1, ssize_t len1, const char *arg2, ssize_t len2 int result; UErrorCode status; - Assert(locale->provider == COLLPROVIDER_ICU); - Assert(GetDatabaseEncoding() == PG_UTF8); status = U_ZERO_ERROR; @@ -503,8 +595,6 @@ strnxfrm_icu(char *dest, size_t destsize, const char *src, ssize_t srclen, size_t uchar_bsize; Size result_bsize; - Assert(locale->provider == COLLPROVIDER_ICU); - init_icu_converter(); ulen = uchar_length(icu_converter, src, srclen); @@ -549,8 +639,6 @@ strnxfrm_prefix_icu_utf8(char *dest, size_t destsize, uint32_t state[2]; UErrorCode status; - Assert(locale->provider == COLLPROVIDER_ICU); - Assert(GetDatabaseEncoding() == PG_UTF8); uiter_setUTF8(&iter, src, srclen); @@ -749,8 +837,6 @@ strncoll_icu(const char *arg1, ssize_t len1, *uchar2; int result; - Assert(locale->provider == COLLPROVIDER_ICU); - /* if encoding is UTF8, use more efficient strncoll_icu_utf8 */ #ifdef HAVE_UCOL_STRCOLLUTF8 Assert(GetDatabaseEncoding() != PG_UTF8); @@ -799,8 +885,6 @@ strnxfrm_prefix_icu(char *dest, size_t destsize, size_t uchar_bsize; Size result_bsize; - Assert(locale->provider == COLLPROVIDER_ICU); - /* if encoding is UTF8, use more efficient strnxfrm_prefix_icu_utf8 */ Assert(GetDatabaseEncoding() != PG_UTF8); diff --git a/src/backend/utils/adt/pg_locale_libc.c b/src/backend/utils/adt/pg_locale_libc.c index 199857e22db..8d88b53c375 100644 --- a/src/backend/utils/adt/pg_locale_libc.c +++ b/src/backend/utils/adt/pg_locale_libc.c @@ -34,6 +34,46 @@ #endif /* + * For the libc provider, to provide as much functionality as possible on a + * variety of platforms without going so far as to implement everything from + * scratch, we use several implementation strategies depending on the + * situation: + * + * 1. In C/POSIX collations, we use hard-wired code. We can't depend on + * the <ctype.h> functions since those will obey LC_CTYPE. Note that these + * collations don't give a fig about multibyte characters. + * + * 2. When working in UTF8 encoding, we use the <wctype.h> functions. + * This assumes that every platform uses Unicode codepoints directly + * as the wchar_t representation of Unicode. (XXX: ICU makes this assumption + * even for non-UTF8 encodings, which may be a problem.) On some platforms + * wchar_t is only 16 bits wide, so we have to punt for codepoints > 0xFFFF. + * + * 3. In all other encodings, we use the <ctype.h> functions for pg_wchar + * values up to 255, and punt for values above that. This is 100% correct + * only in single-byte encodings such as LATINn. However, non-Unicode + * multibyte encodings are mostly Far Eastern character sets for which the + * properties being tested here aren't very relevant for higher code values + * anyway. The difficulty with using the <wctype.h> functions with + * non-Unicode multibyte encodings is that we can have no certainty that + * the platform's wchar_t representation matches what we do in pg_wchar + * conversions. + * + * As a special case, in the "default" collation, (2) and (3) force ASCII + * letters to follow ASCII upcase/downcase rules, while in a non-default + * collation we just let the library functions do what they will. The case + * where this matters is treatment of I/i in Turkish, and the behavior is + * meant to match the upper()/lower() SQL functions. + * + * We store the active collation setting in static variables. In principle + * it could be passed down to here via the regex library's "struct vars" data + * structure; but that would require somewhat invasive changes in the regex + * library, and right now there's no real benefit to be gained from that. + * + * NB: the coding here assumes pg_wchar is an unsigned type. + */ + +/* * Size of stack buffer to use for string transformations, used to avoid heap * allocations in typical cases. This should be large enough that most strings * will fit, but small enough that we feel comfortable putting it on the @@ -43,13 +83,6 @@ extern pg_locale_t create_pg_locale_libc(Oid collid, MemoryContext context); -extern size_t strlower_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strtitle_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); -extern size_t strupper_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale); - static int strncoll_libc(const char *arg1, ssize_t len1, const char *arg2, ssize_t len2, pg_locale_t locale); @@ -85,6 +118,251 @@ static size_t strupper_libc_mb(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale); +static bool +wc_isdigit_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return isdigit_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_isalpha_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return isalpha_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_isalnum_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return isalnum_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_isupper_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return isupper_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_islower_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return islower_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_isgraph_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return isgraph_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_isprint_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return isprint_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_ispunct_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return ispunct_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_isspace_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + return isspace_l((unsigned char) wc, locale->info.lt); +} + +static bool +wc_isdigit_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswdigit_l((wint_t) wc, locale->info.lt); +} + +static bool +wc_isalpha_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswalpha_l((wint_t) wc, locale->info.lt); +} + +static bool +wc_isalnum_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswalnum_l((wint_t) wc, locale->info.lt); +} + +static bool +wc_isupper_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswupper_l((wint_t) wc, locale->info.lt); +} + +static bool +wc_islower_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswlower_l((wint_t) wc, locale->info.lt); +} + +static bool +wc_isgraph_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswgraph_l((wint_t) wc, locale->info.lt); +} + +static bool +wc_isprint_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswprint_l((wint_t) wc, locale->info.lt); +} + +static bool +wc_ispunct_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswpunct_l((wint_t) wc, locale->info.lt); +} + +static bool +wc_isspace_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + return iswspace_l((wint_t) wc, locale->info.lt); +} + +static char +char_tolower_libc(unsigned char ch, pg_locale_t locale) +{ + Assert(pg_database_encoding_max_length() == 1); + return tolower_l(ch, locale->info.lt); +} + +static bool +char_is_cased_libc(char ch, pg_locale_t locale) +{ + bool is_multibyte = pg_database_encoding_max_length() > 1; + + if (is_multibyte && IS_HIGHBIT_SET(ch)) + return true; + else + return isalpha_l((unsigned char) ch, locale->info.lt); +} + +static pg_wchar +toupper_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + Assert(GetDatabaseEncoding() != PG_UTF8); + + /* force C behavior for ASCII characters, per comments above */ + if (locale->is_default && wc <= (pg_wchar) 127) + return pg_ascii_toupper((unsigned char) wc); + if (wc <= (pg_wchar) UCHAR_MAX) + return toupper_l((unsigned char) wc, locale->info.lt); + else + return wc; +} + +static pg_wchar +toupper_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + Assert(GetDatabaseEncoding() == PG_UTF8); + + /* force C behavior for ASCII characters, per comments above */ + if (locale->is_default && wc <= (pg_wchar) 127) + return pg_ascii_toupper((unsigned char) wc); + if (sizeof(wchar_t) >= 4 || wc <= (pg_wchar) 0xFFFF) + return towupper_l((wint_t) wc, locale->info.lt); + else + return wc; +} + +static pg_wchar +tolower_libc_sb(pg_wchar wc, pg_locale_t locale) +{ + Assert(GetDatabaseEncoding() != PG_UTF8); + + /* force C behavior for ASCII characters, per comments above */ + if (locale->is_default && wc <= (pg_wchar) 127) + return pg_ascii_tolower((unsigned char) wc); + if (wc <= (pg_wchar) UCHAR_MAX) + return tolower_l((unsigned char) wc, locale->info.lt); + else + return wc; +} + +static pg_wchar +tolower_libc_mb(pg_wchar wc, pg_locale_t locale) +{ + Assert(GetDatabaseEncoding() == PG_UTF8); + + /* force C behavior for ASCII characters, per comments above */ + if (locale->is_default && wc <= (pg_wchar) 127) + return pg_ascii_tolower((unsigned char) wc); + if (sizeof(wchar_t) >= 4 || wc <= (pg_wchar) 0xFFFF) + return towlower_l((wint_t) wc, locale->info.lt); + else + return wc; +} + +static const struct ctype_methods ctype_methods_libc_sb = { + .strlower = strlower_libc_sb, + .strtitle = strtitle_libc_sb, + .strupper = strupper_libc_sb, + .wc_isdigit = wc_isdigit_libc_sb, + .wc_isalpha = wc_isalpha_libc_sb, + .wc_isalnum = wc_isalnum_libc_sb, + .wc_isupper = wc_isupper_libc_sb, + .wc_islower = wc_islower_libc_sb, + .wc_isgraph = wc_isgraph_libc_sb, + .wc_isprint = wc_isprint_libc_sb, + .wc_ispunct = wc_ispunct_libc_sb, + .wc_isspace = wc_isspace_libc_sb, + .char_is_cased = char_is_cased_libc, + .char_tolower = char_tolower_libc, + .wc_toupper = toupper_libc_sb, + .wc_tolower = tolower_libc_sb, + .max_chr = UCHAR_MAX, +}; + +/* + * Non-UTF8 multibyte encodings use multibyte semantics for case mapping, but + * single-byte semantics for pattern matching. + */ +static const struct ctype_methods ctype_methods_libc_other_mb = { + .strlower = strlower_libc_mb, + .strtitle = strtitle_libc_mb, + .strupper = strupper_libc_mb, + .wc_isdigit = wc_isdigit_libc_sb, + .wc_isalpha = wc_isalpha_libc_sb, + .wc_isalnum = wc_isalnum_libc_sb, + .wc_isupper = wc_isupper_libc_sb, + .wc_islower = wc_islower_libc_sb, + .wc_isgraph = wc_isgraph_libc_sb, + .wc_isprint = wc_isprint_libc_sb, + .wc_ispunct = wc_ispunct_libc_sb, + .wc_isspace = wc_isspace_libc_sb, + .char_is_cased = char_is_cased_libc, + .char_tolower = char_tolower_libc, + .wc_toupper = toupper_libc_sb, + .wc_tolower = tolower_libc_sb, + .max_chr = UCHAR_MAX, +}; + +static const struct ctype_methods ctype_methods_libc_utf8 = { + .strlower = strlower_libc_mb, + .strtitle = strtitle_libc_mb, + .strupper = strupper_libc_mb, + .wc_isdigit = wc_isdigit_libc_mb, + .wc_isalpha = wc_isalpha_libc_mb, + .wc_isalnum = wc_isalnum_libc_mb, + .wc_isupper = wc_isupper_libc_mb, + .wc_islower = wc_islower_libc_mb, + .wc_isgraph = wc_isgraph_libc_mb, + .wc_isprint = wc_isprint_libc_mb, + .wc_ispunct = wc_ispunct_libc_mb, + .wc_isspace = wc_isspace_libc_mb, + .char_is_cased = char_is_cased_libc, + .char_tolower = char_tolower_libc, + .wc_toupper = toupper_libc_mb, + .wc_tolower = tolower_libc_mb, +}; + static const struct collate_methods collate_methods_libc = { .strncoll = strncoll_libc, .strnxfrm = strnxfrm_libc, @@ -119,36 +397,6 @@ static const struct collate_methods collate_methods_libc_win32_utf8 = { }; #endif -size_t -strlower_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale) -{ - if (pg_database_encoding_max_length() > 1) - return strlower_libc_mb(dst, dstsize, src, srclen, locale); - else - return strlower_libc_sb(dst, dstsize, src, srclen, locale); -} - -size_t -strtitle_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale) -{ - if (pg_database_encoding_max_length() > 1) - return strtitle_libc_mb(dst, dstsize, src, srclen, locale); - else - return strtitle_libc_sb(dst, dstsize, src, srclen, locale); -} - -size_t -strupper_libc(char *dst, size_t dstsize, const char *src, - ssize_t srclen, pg_locale_t locale) -{ - if (pg_database_encoding_max_length() > 1) - return strupper_libc_mb(dst, dstsize, src, srclen, locale); - else - return strupper_libc_sb(dst, dstsize, src, srclen, locale); -} - static size_t strlower_libc_sb(char *dest, size_t destsize, const char *src, ssize_t srclen, pg_locale_t locale) @@ -209,7 +457,7 @@ strlower_libc_mb(char *dest, size_t destsize, const char *src, ssize_t srclen, /* Output workspace cannot have more codes than input bytes */ workspace = (wchar_t *) palloc((srclen + 1) * sizeof(wchar_t)); - char2wchar(workspace, srclen + 1, src, srclen, locale); + char2wchar(workspace, srclen + 1, src, srclen, loc); for (curr_char = 0; workspace[curr_char] != 0; curr_char++) workspace[curr_char] = towlower_l(workspace[curr_char], loc); @@ -220,7 +468,7 @@ strlower_libc_mb(char *dest, size_t destsize, const char *src, ssize_t srclen, max_size = curr_char * pg_database_encoding_max_length(); result = palloc(max_size + 1); - result_size = wchar2char(result, workspace, max_size + 1, locale); + result_size = wchar2char(result, workspace, max_size + 1, loc); if (result_size + 1 > destsize) return result_size; @@ -304,7 +552,7 @@ strtitle_libc_mb(char *dest, size_t destsize, const char *src, ssize_t srclen, /* Output workspace cannot have more codes than input bytes */ workspace = (wchar_t *) palloc((srclen + 1) * sizeof(wchar_t)); - char2wchar(workspace, srclen + 1, src, srclen, locale); + char2wchar(workspace, srclen + 1, src, srclen, loc); for (curr_char = 0; workspace[curr_char] != 0; curr_char++) { @@ -321,7 +569,7 @@ strtitle_libc_mb(char *dest, size_t destsize, const char *src, ssize_t srclen, max_size = curr_char * pg_database_encoding_max_length(); result = palloc(max_size + 1); - result_size = wchar2char(result, workspace, max_size + 1, locale); + result_size = wchar2char(result, workspace, max_size + 1, loc); if (result_size + 1 > destsize) return result_size; @@ -392,7 +640,7 @@ strupper_libc_mb(char *dest, size_t destsize, const char *src, ssize_t srclen, /* Output workspace cannot have more codes than input bytes */ workspace = (wchar_t *) palloc((srclen + 1) * sizeof(wchar_t)); - char2wchar(workspace, srclen + 1, src, srclen, locale); + char2wchar(workspace, srclen + 1, src, srclen, loc); for (curr_char = 0; workspace[curr_char] != 0; curr_char++) workspace[curr_char] = towupper_l(workspace[curr_char], loc); @@ -403,7 +651,7 @@ strupper_libc_mb(char *dest, size_t destsize, const char *src, ssize_t srclen, max_size = curr_char * pg_database_encoding_max_length(); result = palloc(max_size + 1); - result_size = wchar2char(result, workspace, max_size + 1, locale); + result_size = wchar2char(result, workspace, max_size + 1, loc); if (result_size + 1 > destsize) return result_size; @@ -465,7 +713,6 @@ create_pg_locale_libc(Oid collid, MemoryContext context) loc = make_libc_collator(collate, ctype); result = MemoryContextAllocZero(context, sizeof(struct pg_locale_struct)); - result->provider = COLLPROVIDER_LIBC; result->deterministic = true; result->collate_is_c = (strcmp(collate, "C") == 0) || (strcmp(collate, "POSIX") == 0); @@ -481,6 +728,15 @@ create_pg_locale_libc(Oid collid, MemoryContext context) #endif result->collate = &collate_methods_libc; } + if (!result->ctype_is_c) + { + if (GetDatabaseEncoding() == PG_UTF8) + result->ctype = &ctype_methods_libc_utf8; + else if (pg_database_encoding_max_length() > 1) + result->ctype = &ctype_methods_libc_other_mb; + else + result->ctype = &ctype_methods_libc_sb; + } return result; } @@ -576,8 +832,6 @@ strncoll_libc(const char *arg1, ssize_t len1, const char *arg2, ssize_t len2, const char *arg2n; int result; - Assert(locale->provider == COLLPROVIDER_LIBC); - if (bufsize1 + bufsize2 > TEXTBUFLEN) buf = palloc(bufsize1 + bufsize2); @@ -632,8 +886,6 @@ strnxfrm_libc(char *dest, size_t destsize, const char *src, ssize_t srclen, size_t bufsize = srclen + 1; size_t result; - Assert(locale->provider == COLLPROVIDER_LIBC); - if (srclen == -1) return strxfrm_l(dest, src, destsize, locale->info.lt); @@ -742,7 +994,6 @@ strncoll_libc_win32_utf8(const char *arg1, ssize_t len1, const char *arg2, int r; int result; - Assert(locale->provider == COLLPROVIDER_LIBC); Assert(GetDatabaseEncoding() == PG_UTF8); if (len1 == -1) @@ -879,7 +1130,7 @@ wcstombs_l(char *dest, const wchar_t *src, size_t n, locale_t loc) * zero-terminated. The output will be zero-terminated iff there is room. */ size_t -wchar2char(char *to, const wchar_t *from, size_t tolen, pg_locale_t locale) +wchar2char(char *to, const wchar_t *from, size_t tolen, locale_t loc) { size_t result; @@ -909,7 +1160,7 @@ wchar2char(char *to, const wchar_t *from, size_t tolen, pg_locale_t locale) } else #endif /* WIN32 */ - if (locale == (pg_locale_t) 0) + if (loc == (locale_t) 0) { /* Use wcstombs directly for the default locale */ result = wcstombs(to, from, tolen); @@ -917,7 +1168,7 @@ wchar2char(char *to, const wchar_t *from, size_t tolen, pg_locale_t locale) else { /* Use wcstombs_l for nondefault locales */ - result = wcstombs_l(to, from, tolen, locale->info.lt); + result = wcstombs_l(to, from, tolen, loc); } return result; @@ -934,7 +1185,7 @@ wchar2char(char *to, const wchar_t *from, size_t tolen, pg_locale_t locale) */ size_t char2wchar(wchar_t *to, size_t tolen, const char *from, size_t fromlen, - pg_locale_t locale) + locale_t loc) { size_t result; @@ -969,7 +1220,7 @@ char2wchar(wchar_t *to, size_t tolen, const char *from, size_t fromlen, /* mbstowcs requires ending '\0' */ char *str = pnstrdup(from, fromlen); - if (locale == (pg_locale_t) 0) + if (loc == (locale_t) 0) { /* Use mbstowcs directly for the default locale */ result = mbstowcs(to, str, tolen); @@ -977,7 +1228,7 @@ char2wchar(wchar_t *to, size_t tolen, const char *from, size_t fromlen, else { /* Use mbstowcs_l for nondefault locales */ - result = mbstowcs_l(to, str, tolen, locale->info.lt); + result = mbstowcs_l(to, str, tolen, loc); } pfree(str); diff --git a/src/backend/utils/adt/pg_lsn.c b/src/backend/utils/adt/pg_lsn.c index 16311590a14..12de2446f5b 100644 --- a/src/backend/utils/adt/pg_lsn.c +++ b/src/backend/utils/adt/pg_lsn.c @@ -83,7 +83,7 @@ pg_lsn_out(PG_FUNCTION_ARGS) char buf[MAXPG_LSNLEN + 1]; char *result; - snprintf(buf, sizeof buf, "%X/%X", LSN_FORMAT_ARGS(lsn)); + snprintf(buf, sizeof buf, "%X/%08X", LSN_FORMAT_ARGS(lsn)); result = pstrdup(buf); PG_RETURN_CSTRING(result); } diff --git a/src/backend/utils/adt/pg_upgrade_support.c b/src/backend/utils/adt/pg_upgrade_support.c index d44f8c262ba..a4f8b4faa90 100644 --- a/src/backend/utils/adt/pg_upgrade_support.c +++ b/src/backend/utils/adt/pg_upgrade_support.c @@ -21,6 +21,7 @@ #include "commands/extension.h" #include "miscadmin.h" #include "replication/logical.h" +#include "replication/logicallauncher.h" #include "replication/origin.h" #include "replication/worker_internal.h" #include "storage/lmgr.h" @@ -410,3 +411,21 @@ binary_upgrade_replorigin_advance(PG_FUNCTION_ARGS) PG_RETURN_VOID(); } + +/* + * binary_upgrade_create_conflict_detection_slot + * + * Create a replication slot to retain information necessary for conflict + * detection such as dead tuples, commit timestamps, and origins. + */ +Datum +binary_upgrade_create_conflict_detection_slot(PG_FUNCTION_ARGS) +{ + CHECK_IS_BINARY_UPGRADE; + + CreateConflictDetectionSlot(); + + ReplicationSlotRelease(); + + PG_RETURN_VOID(); +} diff --git a/src/backend/utils/adt/pgstatfuncs.c b/src/backend/utils/adt/pgstatfuncs.c index e980109f245..c756c2bebaa 100644 --- a/src/backend/utils/adt/pgstatfuncs.c +++ b/src/backend/utils/adt/pgstatfuncs.c @@ -1510,7 +1510,7 @@ pg_stat_io_build_tuples(ReturnSetInfo *rsinfo, bktype_stats->bytes[io_obj][io_context][io_op]; /* Convert to numeric */ - snprintf(buf, sizeof buf, UINT64_FORMAT, byte); + snprintf(buf, sizeof buf, INT64_FORMAT, byte); values[byte_idx] = DirectFunctionCall3(numeric_in, CStringGetDatum(buf), ObjectIdGetDatum(0), @@ -2171,7 +2171,7 @@ pg_stat_get_replication_slot(PG_FUNCTION_ARGS) Datum pg_stat_get_subscription_stats(PG_FUNCTION_ARGS) { -#define PG_STAT_GET_SUBSCRIPTION_STATS_COLS 11 +#define PG_STAT_GET_SUBSCRIPTION_STATS_COLS 12 Oid subid = PG_GETARG_OID(0); TupleDesc tupdesc; Datum values[PG_STAT_GET_SUBSCRIPTION_STATS_COLS] = {0}; @@ -2197,15 +2197,17 @@ pg_stat_get_subscription_stats(PG_FUNCTION_ARGS) INT8OID, -1, 0); TupleDescInitEntry(tupdesc, (AttrNumber) 6, "confl_update_exists", INT8OID, -1, 0); - TupleDescInitEntry(tupdesc, (AttrNumber) 7, "confl_update_missing", + TupleDescInitEntry(tupdesc, (AttrNumber) 7, "confl_update_deleted", INT8OID, -1, 0); - TupleDescInitEntry(tupdesc, (AttrNumber) 8, "confl_delete_origin_differs", + TupleDescInitEntry(tupdesc, (AttrNumber) 8, "confl_update_missing", INT8OID, -1, 0); - TupleDescInitEntry(tupdesc, (AttrNumber) 9, "confl_delete_missing", + TupleDescInitEntry(tupdesc, (AttrNumber) 9, "confl_delete_origin_differs", INT8OID, -1, 0); - TupleDescInitEntry(tupdesc, (AttrNumber) 10, "confl_multiple_unique_conflicts", + TupleDescInitEntry(tupdesc, (AttrNumber) 10, "confl_delete_missing", INT8OID, -1, 0); - TupleDescInitEntry(tupdesc, (AttrNumber) 11, "stats_reset", + TupleDescInitEntry(tupdesc, (AttrNumber) 11, "confl_multiple_unique_conflicts", + INT8OID, -1, 0); + TupleDescInitEntry(tupdesc, (AttrNumber) 12, "stats_reset", TIMESTAMPTZOID, -1, 0); BlessTupleDesc(tupdesc); diff --git a/src/backend/utils/adt/rangetypes.c b/src/backend/utils/adt/rangetypes.c index 66cc0acf4a7..c83b239b3bb 100644 --- a/src/backend/utils/adt/rangetypes.c +++ b/src/backend/utils/adt/rangetypes.c @@ -285,8 +285,7 @@ range_send(PG_FUNCTION_ARGS) if (RANGE_HAS_LBOUND(flags)) { - Datum bound = PointerGetDatum(SendFunctionCall(&cache->typioproc, - lower.val)); + bytea *bound = SendFunctionCall(&cache->typioproc, lower.val); uint32 bound_len = VARSIZE(bound) - VARHDRSZ; char *bound_data = VARDATA(bound); @@ -296,8 +295,7 @@ range_send(PG_FUNCTION_ARGS) if (RANGE_HAS_UBOUND(flags)) { - Datum bound = PointerGetDatum(SendFunctionCall(&cache->typioproc, - upper.val)); + bytea *bound = SendFunctionCall(&cache->typioproc, upper.val); uint32 bound_len = VARSIZE(bound) - VARHDRSZ; char *bound_data = VARDATA(bound); @@ -1444,7 +1442,7 @@ hash_range(PG_FUNCTION_ARGS) upper_hash = 0; /* Merge hashes of flags and bounds */ - result = hash_uint32((uint32) flags); + result = hash_bytes_uint32((uint32) flags); result ^= lower_hash; result = pg_rotate_left32(result, 1); result ^= upper_hash; diff --git a/src/backend/utils/adt/regproc.c b/src/backend/utils/adt/regproc.c index 5ee608a2b39..b8bbe95e82e 100644 --- a/src/backend/utils/adt/regproc.c +++ b/src/backend/utils/adt/regproc.c @@ -30,6 +30,7 @@ #include "catalog/pg_ts_config.h" #include "catalog/pg_ts_dict.h" #include "catalog/pg_type.h" +#include "commands/dbcommands.h" #include "lib/stringinfo.h" #include "mb/pg_wchar.h" #include "miscadmin.h" @@ -1764,6 +1765,123 @@ regnamespacesend(PG_FUNCTION_ARGS) } /* + * regdatabasein - converts database name to database OID + * + * We also accept a numeric OID, for symmetry with the output routine. + * + * '-' signifies unknown (OID 0). In all other cases, the input must + * match an existing pg_database entry. + */ +Datum +regdatabasein(PG_FUNCTION_ARGS) +{ + char *db_name_or_oid = PG_GETARG_CSTRING(0); + Node *escontext = fcinfo->context; + Oid result; + List *names; + + /* Handle "-" or numeric OID */ + if (parseDashOrOid(db_name_or_oid, &result, escontext)) + PG_RETURN_OID(result); + + /* The rest of this wouldn't work in bootstrap mode */ + if (IsBootstrapProcessingMode()) + elog(ERROR, "regdatabase values must be OIDs in bootstrap mode"); + + /* Normal case: see if the name matches any pg_database entry. */ + names = stringToQualifiedNameList(db_name_or_oid, escontext); + if (names == NIL) + PG_RETURN_NULL(); + + if (list_length(names) != 1) + ereturn(escontext, (Datum) 0, + (errcode(ERRCODE_INVALID_NAME), + errmsg("invalid name syntax"))); + + result = get_database_oid(strVal(linitial(names)), true); + + if (!OidIsValid(result)) + ereturn(escontext, (Datum) 0, + (errcode(ERRCODE_UNDEFINED_OBJECT), + errmsg("database \"%s\" does not exist", + strVal(linitial(names))))); + + PG_RETURN_OID(result); +} + +/* + * to_regdatabase - converts database name to database OID + * + * If the name is not found, we return NULL. + */ +Datum +to_regdatabase(PG_FUNCTION_ARGS) +{ + char *db_name = text_to_cstring(PG_GETARG_TEXT_PP(0)); + Datum result; + ErrorSaveContext escontext = {T_ErrorSaveContext}; + + if (!DirectInputFunctionCallSafe(regdatabasein, db_name, + InvalidOid, -1, + (Node *) &escontext, + &result)) + PG_RETURN_NULL(); + PG_RETURN_DATUM(result); +} + +/* + * regdatabaseout - converts database OID to database name + */ +Datum +regdatabaseout(PG_FUNCTION_ARGS) +{ + Oid dboid = PG_GETARG_OID(0); + char *result; + + if (dboid == InvalidOid) + { + result = pstrdup("-"); + PG_RETURN_CSTRING(result); + } + + result = get_database_name(dboid); + + if (result) + { + /* pstrdup is not really necessary, but it avoids a compiler warning */ + result = pstrdup(quote_identifier(result)); + } + else + { + /* If OID doesn't match any database, return it numerically */ + result = (char *) palloc(NAMEDATALEN); + snprintf(result, NAMEDATALEN, "%u", dboid); + } + + PG_RETURN_CSTRING(result); +} + +/* + * regdatabaserecv - converts external binary format to regdatabase + */ +Datum +regdatabaserecv(PG_FUNCTION_ARGS) +{ + /* Exactly the same as oidrecv, so share code */ + return oidrecv(fcinfo); +} + +/* + * regdatabasesend - converts regdatabase to binary format + */ +Datum +regdatabasesend(PG_FUNCTION_ARGS) +{ + /* Exactly the same as oidsend, so share code */ + return oidsend(fcinfo); +} + +/* * text_regclass: convert text to regclass * * This could be replaced by CoerceViaIO, except that we need to treat diff --git a/src/backend/utils/adt/ri_triggers.c b/src/backend/utils/adt/ri_triggers.c index 6239900fa28..059fc5ebf60 100644 --- a/src/backend/utils/adt/ri_triggers.c +++ b/src/backend/utils/adt/ri_triggers.c @@ -30,7 +30,6 @@ #include "access/xact.h" #include "catalog/pg_collation.h" #include "catalog/pg_constraint.h" -#include "catalog/pg_proc.h" #include "commands/trigger.h" #include "executor/executor.h" #include "executor/spi.h" @@ -46,7 +45,6 @@ #include "utils/inval.h" #include "utils/lsyscache.h" #include "utils/memutils.h" -#include "utils/rangetypes.h" #include "utils/rel.h" #include "utils/rls.h" #include "utils/ruleutils.h" diff --git a/src/backend/utils/adt/selfuncs.c b/src/backend/utils/adt/selfuncs.c index a96b1b9c0bc..17fbfa9b410 100644 --- a/src/backend/utils/adt/selfuncs.c +++ b/src/backend/utils/adt/selfuncs.c @@ -103,7 +103,6 @@ #include "access/table.h" #include "access/tableam.h" #include "access/visibilitymap.h" -#include "catalog/pg_am.h" #include "catalog/pg_collation.h" #include "catalog/pg_operator.h" #include "catalog/pg_statistic.h" @@ -3799,18 +3798,25 @@ estimate_multivariate_bucketsize(PlannerInfo *root, RelOptInfo *inner, List *hashclauses, Selectivity *innerbucketsize) { - List *clauses = list_copy(hashclauses); - List *otherclauses = NIL; - double ndistinct = 1.0; + List *clauses; + List *otherclauses; + double ndistinct; if (list_length(hashclauses) <= 1) - + { /* * Nothing to do for a single clause. Could we employ univariate * extended stat here? */ return hashclauses; + } + /* "clauses" is the list of hashclauses we've not dealt with yet */ + clauses = list_copy(hashclauses); + /* "otherclauses" holds clauses we are going to return to caller */ + otherclauses = NIL; + /* current estimate of ndistinct */ + ndistinct = 1.0; while (clauses != NIL) { ListCell *lc; @@ -3875,12 +3881,13 @@ estimate_multivariate_bucketsize(PlannerInfo *root, RelOptInfo *inner, group_rel = root->simple_rel_array[relid]; } else if (group_relid != relid) - + { /* * Being in the group forming state we don't need other * clauses. */ continue; + } /* * We're going to add the new clause to the varinfos list. We @@ -4620,6 +4627,7 @@ convert_to_scalar(Datum value, Oid valuetypid, Oid collid, double *scaledvalue, case REGDICTIONARYOID: case REGROLEOID: case REGNAMESPACEOID: + case REGDATABASEOID: *scaledvalue = convert_numeric_to_scalar(value, valuetypid, &failure); *scaledlobound = convert_numeric_to_scalar(lobound, boundstypid, @@ -4752,6 +4760,7 @@ convert_numeric_to_scalar(Datum value, Oid typid, bool *failure) case REGDICTIONARYOID: case REGROLEOID: case REGNAMESPACEOID: + case REGDATABASEOID: /* we can treat OIDs as integers... */ return (double) DatumGetObjectId(value); } diff --git a/src/backend/utils/adt/tid.c b/src/backend/utils/adt/tid.c index 1b0df111717..39dab3e42df 100644 --- a/src/backend/utils/adt/tid.c +++ b/src/backend/utils/adt/tid.c @@ -84,7 +84,7 @@ tidin(PG_FUNCTION_ARGS) /* * Cope with possibility that unsigned long is wider than BlockNumber, in * which case strtoul will not raise an error for some values that are out - * of the range of BlockNumber. (See similar code in oidin().) + * of the range of BlockNumber. (See similar code in uint32in_subr().) */ #if SIZEOF_LONG > 4 if (cvt != (unsigned long) blockNumber && diff --git a/src/backend/utils/adt/timestamp.c b/src/backend/utils/adt/timestamp.c index 347089b7626..25cff56c3d0 100644 --- a/src/backend/utils/adt/timestamp.c +++ b/src/backend/utils/adt/timestamp.c @@ -5312,10 +5312,10 @@ isoweekdate2date(int isoweek, int wday, int *year, int *mon, int *mday) int date2isoweek(int year, int mon, int mday) { - float8 result; int day0, day4, - dayn; + dayn, + week; /* current day */ dayn = date2j(year, mon, mday); @@ -5338,13 +5338,13 @@ date2isoweek(int year, int mon, int mday) day0 = j2day(day4 - 1); } - result = (dayn - (day4 - day0)) / 7 + 1; + week = (dayn - (day4 - day0)) / 7 + 1; /* * Sometimes the last few days in a year will fall into the first week of * the next year, so check for this. */ - if (result >= 52) + if (week >= 52) { day4 = date2j(year + 1, 1, 4); @@ -5352,10 +5352,10 @@ date2isoweek(int year, int mon, int mday) day0 = j2day(day4 - 1); if (dayn >= day4 - day0) - result = (dayn - (day4 - day0)) / 7 + 1; + week = (dayn - (day4 - day0)) / 7 + 1; } - return (int) result; + return week; } @@ -5367,10 +5367,10 @@ date2isoweek(int year, int mon, int mday) int date2isoyear(int year, int mon, int mday) { - float8 result; int day0, day4, - dayn; + dayn, + week; /* current day */ dayn = date2j(year, mon, mday); @@ -5395,13 +5395,13 @@ date2isoyear(int year, int mon, int mday) year--; } - result = (dayn - (day4 - day0)) / 7 + 1; + week = (dayn - (day4 - day0)) / 7 + 1; /* * Sometimes the last few days in a year will fall into the first week of * the next year, so check for this. */ - if (result >= 52) + if (week >= 52) { day4 = date2j(year + 1, 1, 4); @@ -6477,7 +6477,7 @@ timestamp2timestamptz_opt_overflow(Timestamp timestamp, int *overflow) if (TIMESTAMP_NOT_FINITE(timestamp)) return timestamp; - /* We don't expect this to fail, but check it pro forma */ + /* timestamp2tm should not fail on valid timestamps, but cope */ if (timestamp2tm(timestamp, NULL, tm, &fsec, NULL, NULL) == 0) { tz = DetermineTimeZoneOffset(tm, session_timezone); @@ -6485,23 +6485,22 @@ timestamp2timestamptz_opt_overflow(Timestamp timestamp, int *overflow) result = dt2local(timestamp, -tz); if (IS_VALID_TIMESTAMP(result)) - { return result; + } + + if (overflow) + { + if (timestamp < 0) + { + *overflow = -1; + TIMESTAMP_NOBEGIN(result); } - else if (overflow) + else { - if (result < MIN_TIMESTAMP) - { - *overflow = -1; - TIMESTAMP_NOBEGIN(result); - } - else - { - *overflow = 1; - TIMESTAMP_NOEND(result); - } - return result; + *overflow = 1; + TIMESTAMP_NOEND(result); } + return result; } ereport(ERROR, @@ -6531,27 +6530,81 @@ timestamptz_timestamp(PG_FUNCTION_ARGS) PG_RETURN_TIMESTAMP(timestamptz2timestamp(timestamp)); } +/* + * Convert timestamptz to timestamp, throwing error for overflow. + */ static Timestamp timestamptz2timestamp(TimestampTz timestamp) { + return timestamptz2timestamp_opt_overflow(timestamp, NULL); +} + +/* + * Convert timestamp with time zone to timestamp. + * + * On successful conversion, *overflow is set to zero if it's not NULL. + * + * If the timestamptz is finite but out of the valid range for timestamp, then: + * if overflow is NULL, we throw an out-of-range error. + * if overflow is not NULL, we store +1 or -1 there to indicate the sign + * of the overflow, and return the appropriate timestamp infinity. + */ +Timestamp +timestamptz2timestamp_opt_overflow(TimestampTz timestamp, int *overflow) +{ Timestamp result; struct pg_tm tt, *tm = &tt; fsec_t fsec; int tz; + if (overflow) + *overflow = 0; + if (TIMESTAMP_NOT_FINITE(timestamp)) result = timestamp; else { if (timestamp2tm(timestamp, &tz, tm, &fsec, NULL, NULL) != 0) + { + if (overflow) + { + if (timestamp < 0) + { + *overflow = -1; + TIMESTAMP_NOBEGIN(result); + } + else + { + *overflow = 1; + TIMESTAMP_NOEND(result); + } + return result; + } ereport(ERROR, (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE), errmsg("timestamp out of range"))); + } if (tm2timestamp(tm, fsec, NULL, &result) != 0) + { + if (overflow) + { + if (timestamp < 0) + { + *overflow = -1; + TIMESTAMP_NOBEGIN(result); + } + else + { + *overflow = 1; + TIMESTAMP_NOEND(result); + } + return result; + } ereport(ERROR, (errcode(ERRCODE_DATETIME_VALUE_OUT_OF_RANGE), errmsg("timestamp out of range"))); + } } return result; } diff --git a/src/backend/utils/adt/tsvector_op.c b/src/backend/utils/adt/tsvector_op.c index 1fa1275ca63..0625da9532f 100644 --- a/src/backend/utils/adt/tsvector_op.c +++ b/src/backend/utils/adt/tsvector_op.c @@ -329,8 +329,8 @@ tsvector_setweight_by_filter(PG_FUNCTION_ARGS) if (nulls[i]) continue; - lex = VARDATA(dlexemes[i]); - lex_len = VARSIZE(dlexemes[i]) - VARHDRSZ; + lex = VARDATA(DatumGetPointer(dlexemes[i])); + lex_len = VARSIZE(DatumGetPointer(dlexemes[i])) - VARHDRSZ; lex_pos = tsvector_bsearch(tsout, lex, lex_len); if (lex_pos >= 0 && (j = POSDATALEN(tsout, entry + lex_pos)) != 0) @@ -443,10 +443,10 @@ compare_text_lexemes(const void *va, const void *vb) { Datum a = *((const Datum *) va); Datum b = *((const Datum *) vb); - char *alex = VARDATA_ANY(a); - int alex_len = VARSIZE_ANY_EXHDR(a); - char *blex = VARDATA_ANY(b); - int blex_len = VARSIZE_ANY_EXHDR(b); + char *alex = VARDATA_ANY(DatumGetPointer(a)); + int alex_len = VARSIZE_ANY_EXHDR(DatumGetPointer(a)); + char *blex = VARDATA_ANY(DatumGetPointer(b)); + int blex_len = VARSIZE_ANY_EXHDR(DatumGetPointer(b)); return tsCompareString(alex, alex_len, blex, blex_len, false); } @@ -605,8 +605,8 @@ tsvector_delete_arr(PG_FUNCTION_ARGS) if (nulls[i]) continue; - lex = VARDATA(dlexemes[i]); - lex_len = VARSIZE(dlexemes[i]) - VARHDRSZ; + lex = VARDATA(DatumGetPointer(dlexemes[i])); + lex_len = VARSIZE(DatumGetPointer(dlexemes[i])) - VARHDRSZ; lex_pos = tsvector_bsearch(tsin, lex, lex_len); if (lex_pos >= 0) @@ -770,7 +770,7 @@ array_to_tsvector(PG_FUNCTION_ARGS) (errcode(ERRCODE_NULL_VALUE_NOT_ALLOWED), errmsg("lexeme array may not contain nulls"))); - if (VARSIZE(dlexemes[i]) - VARHDRSZ == 0) + if (VARSIZE(DatumGetPointer(dlexemes[i])) - VARHDRSZ == 0) ereport(ERROR, (errcode(ERRCODE_ZERO_LENGTH_CHARACTER_STRING), errmsg("lexeme array may not contain empty strings"))); @@ -786,7 +786,7 @@ array_to_tsvector(PG_FUNCTION_ARGS) /* Calculate space needed for surviving lexemes. */ for (i = 0; i < nitems; i++) - datalen += VARSIZE(dlexemes[i]) - VARHDRSZ; + datalen += VARSIZE(DatumGetPointer(dlexemes[i])) - VARHDRSZ; tslen = CALCDATASIZE(nitems, datalen); /* Allocate and fill tsvector. */ @@ -798,8 +798,8 @@ array_to_tsvector(PG_FUNCTION_ARGS) cur = STRPTR(tsout); for (i = 0; i < nitems; i++) { - char *lex = VARDATA(dlexemes[i]); - int lex_len = VARSIZE(dlexemes[i]) - VARHDRSZ; + char *lex = VARDATA(DatumGetPointer(dlexemes[i])); + int lex_len = VARSIZE(DatumGetPointer(dlexemes[i])) - VARHDRSZ; memcpy(cur, lex, lex_len); arrout[i].haspos = 0; diff --git a/src/backend/utils/adt/varlena.c b/src/backend/utils/adt/varlena.c index 3e4d5568bde..ffae8c23abf 100644 --- a/src/backend/utils/adt/varlena.c +++ b/src/backend/utils/adt/varlena.c @@ -35,7 +35,6 @@ #include "port/pg_bswap.h" #include "regex/regex.h" #include "utils/builtins.h" -#include "utils/bytea.h" #include "utils/guc.h" #include "utils/lsyscache.h" #include "utils/memutils.h" @@ -43,10 +42,6 @@ #include "utils/sortsupport.h" #include "utils/varlena.h" - -/* GUC variable */ -int bytea_output = BYTEA_OUTPUT_HEX; - typedef struct varlena VarString; /* @@ -148,12 +143,6 @@ static int text_position_get_match_pos(TextPositionState *state); static void text_position_cleanup(TextPositionState *state); static void check_collation_set(Oid collid); static int text_cmp(text *arg1, text *arg2, Oid collid); -static bytea *bytea_catenate(bytea *t1, bytea *t2); -static bytea *bytea_substring(Datum str, - int S, - int L, - bool length_not_specified); -static bytea *bytea_overlay(bytea *t1, bytea *t2, int sp, int sl); static void appendStringInfoText(StringInfo str, const text *t); static bool split_text(FunctionCallInfo fcinfo, SplitTextOutputData *tstate); static void split_text_accum_result(SplitTextOutputData *tstate, @@ -279,307 +268,6 @@ text_to_cstring_buffer(const text *src, char *dst, size_t dst_len) * USER I/O ROUTINES * *****************************************************************************/ - -#define VAL(CH) ((CH) - '0') -#define DIG(VAL) ((VAL) + '0') - -/* - * byteain - converts from printable representation of byte array - * - * Non-printable characters must be passed as '\nnn' (octal) and are - * converted to internal form. '\' must be passed as '\\'. - * ereport(ERROR, ...) if bad form. - * - * BUGS: - * The input is scanned twice. - * The error checking of input is minimal. - */ -Datum -byteain(PG_FUNCTION_ARGS) -{ - char *inputText = PG_GETARG_CSTRING(0); - Node *escontext = fcinfo->context; - char *tp; - char *rp; - int bc; - bytea *result; - - /* Recognize hex input */ - if (inputText[0] == '\\' && inputText[1] == 'x') - { - size_t len = strlen(inputText); - - bc = (len - 2) / 2 + VARHDRSZ; /* maximum possible length */ - result = palloc(bc); - bc = hex_decode_safe(inputText + 2, len - 2, VARDATA(result), - escontext); - SET_VARSIZE(result, bc + VARHDRSZ); /* actual length */ - - PG_RETURN_BYTEA_P(result); - } - - /* Else, it's the traditional escaped style */ - for (bc = 0, tp = inputText; *tp != '\0'; bc++) - { - if (tp[0] != '\\') - tp++; - else if ((tp[0] == '\\') && - (tp[1] >= '0' && tp[1] <= '3') && - (tp[2] >= '0' && tp[2] <= '7') && - (tp[3] >= '0' && tp[3] <= '7')) - tp += 4; - else if ((tp[0] == '\\') && - (tp[1] == '\\')) - tp += 2; - else - { - /* - * one backslash, not followed by another or ### valid octal - */ - ereturn(escontext, (Datum) 0, - (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), - errmsg("invalid input syntax for type %s", "bytea"))); - } - } - - bc += VARHDRSZ; - - result = (bytea *) palloc(bc); - SET_VARSIZE(result, bc); - - tp = inputText; - rp = VARDATA(result); - while (*tp != '\0') - { - if (tp[0] != '\\') - *rp++ = *tp++; - else if ((tp[0] == '\\') && - (tp[1] >= '0' && tp[1] <= '3') && - (tp[2] >= '0' && tp[2] <= '7') && - (tp[3] >= '0' && tp[3] <= '7')) - { - bc = VAL(tp[1]); - bc <<= 3; - bc += VAL(tp[2]); - bc <<= 3; - *rp++ = bc + VAL(tp[3]); - - tp += 4; - } - else if ((tp[0] == '\\') && - (tp[1] == '\\')) - { - *rp++ = '\\'; - tp += 2; - } - else - { - /* - * We should never get here. The first pass should not allow it. - */ - ereturn(escontext, (Datum) 0, - (errcode(ERRCODE_INVALID_TEXT_REPRESENTATION), - errmsg("invalid input syntax for type %s", "bytea"))); - } - } - - PG_RETURN_BYTEA_P(result); -} - -/* - * byteaout - converts to printable representation of byte array - * - * In the traditional escaped format, non-printable characters are - * printed as '\nnn' (octal) and '\' as '\\'. - */ -Datum -byteaout(PG_FUNCTION_ARGS) -{ - bytea *vlena = PG_GETARG_BYTEA_PP(0); - char *result; - char *rp; - - if (bytea_output == BYTEA_OUTPUT_HEX) - { - /* Print hex format */ - rp = result = palloc(VARSIZE_ANY_EXHDR(vlena) * 2 + 2 + 1); - *rp++ = '\\'; - *rp++ = 'x'; - rp += hex_encode(VARDATA_ANY(vlena), VARSIZE_ANY_EXHDR(vlena), rp); - } - else if (bytea_output == BYTEA_OUTPUT_ESCAPE) - { - /* Print traditional escaped format */ - char *vp; - uint64 len; - int i; - - len = 1; /* empty string has 1 char */ - vp = VARDATA_ANY(vlena); - for (i = VARSIZE_ANY_EXHDR(vlena); i != 0; i--, vp++) - { - if (*vp == '\\') - len += 2; - else if ((unsigned char) *vp < 0x20 || (unsigned char) *vp > 0x7e) - len += 4; - else - len++; - } - - /* - * In principle len can't overflow uint32 if the input fit in 1GB, but - * for safety let's check rather than relying on palloc's internal - * check. - */ - if (len > MaxAllocSize) - ereport(ERROR, - (errcode(ERRCODE_PROGRAM_LIMIT_EXCEEDED), - errmsg_internal("result of bytea output conversion is too large"))); - rp = result = (char *) palloc(len); - - vp = VARDATA_ANY(vlena); - for (i = VARSIZE_ANY_EXHDR(vlena); i != 0; i--, vp++) - { - if (*vp == '\\') - { - *rp++ = '\\'; - *rp++ = '\\'; - } - else if ((unsigned char) *vp < 0x20 || (unsigned char) *vp > 0x7e) - { - int val; /* holds unprintable chars */ - - val = *vp; - rp[0] = '\\'; - rp[3] = DIG(val & 07); - val >>= 3; - rp[2] = DIG(val & 07); - val >>= 3; - rp[1] = DIG(val & 03); - rp += 4; - } - else - *rp++ = *vp; - } - } - else - { - elog(ERROR, "unrecognized \"bytea_output\" setting: %d", - bytea_output); - rp = result = NULL; /* keep compiler quiet */ - } - *rp = '\0'; - PG_RETURN_CSTRING(result); -} - -/* - * bytearecv - converts external binary format to bytea - */ -Datum -bytearecv(PG_FUNCTION_ARGS) -{ - StringInfo buf = (StringInfo) PG_GETARG_POINTER(0); - bytea *result; - int nbytes; - - nbytes = buf->len - buf->cursor; - result = (bytea *) palloc(nbytes + VARHDRSZ); - SET_VARSIZE(result, nbytes + VARHDRSZ); - pq_copymsgbytes(buf, VARDATA(result), nbytes); - PG_RETURN_BYTEA_P(result); -} - -/* - * byteasend - converts bytea to binary format - * - * This is a special case: just copy the input... - */ -Datum -byteasend(PG_FUNCTION_ARGS) -{ - bytea *vlena = PG_GETARG_BYTEA_P_COPY(0); - - PG_RETURN_BYTEA_P(vlena); -} - -Datum -bytea_string_agg_transfn(PG_FUNCTION_ARGS) -{ - StringInfo state; - - state = PG_ARGISNULL(0) ? NULL : (StringInfo) PG_GETARG_POINTER(0); - - /* Append the value unless null, preceding it with the delimiter. */ - if (!PG_ARGISNULL(1)) - { - bytea *value = PG_GETARG_BYTEA_PP(1); - bool isfirst = false; - - /* - * You might think we can just throw away the first delimiter, however - * we must keep it as we may be a parallel worker doing partial - * aggregation building a state to send to the main process. We need - * to keep the delimiter of every aggregation so that the combine - * function can properly join up the strings of two separately - * partially aggregated results. The first delimiter is only stripped - * off in the final function. To know how much to strip off the front - * of the string, we store the length of the first delimiter in the - * StringInfo's cursor field, which we don't otherwise need here. - */ - if (state == NULL) - { - state = makeStringAggState(fcinfo); - isfirst = true; - } - - if (!PG_ARGISNULL(2)) - { - bytea *delim = PG_GETARG_BYTEA_PP(2); - - appendBinaryStringInfo(state, VARDATA_ANY(delim), - VARSIZE_ANY_EXHDR(delim)); - if (isfirst) - state->cursor = VARSIZE_ANY_EXHDR(delim); - } - - appendBinaryStringInfo(state, VARDATA_ANY(value), - VARSIZE_ANY_EXHDR(value)); - } - - /* - * The transition type for string_agg() is declared to be "internal", - * which is a pass-by-value type the same size as a pointer. - */ - if (state) - PG_RETURN_POINTER(state); - PG_RETURN_NULL(); -} - -Datum -bytea_string_agg_finalfn(PG_FUNCTION_ARGS) -{ - StringInfo state; - - /* cannot be called directly because of internal-type argument */ - Assert(AggCheckCallContext(fcinfo, NULL)); - - state = PG_ARGISNULL(0) ? NULL : (StringInfo) PG_GETARG_POINTER(0); - - if (state != NULL) - { - /* As per comment in transfn, strip data before the cursor position */ - bytea *result; - int strippedlen = state->len - state->cursor; - - result = (bytea *) palloc(strippedlen + VARHDRSZ); - SET_VARSIZE(result, strippedlen + VARHDRSZ); - memcpy(VARDATA(result), &state->data[state->cursor], strippedlen); - PG_RETURN_BYTEA_P(result); - } - else - PG_RETURN_NULL(); -} - /* * textin - converts cstring to internal representation */ @@ -2959,467 +2647,6 @@ bttext_pattern_sortsupport(PG_FUNCTION_ARGS) } -/*------------------------------------------------------------- - * byteaoctetlen - * - * get the number of bytes contained in an instance of type 'bytea' - *------------------------------------------------------------- - */ -Datum -byteaoctetlen(PG_FUNCTION_ARGS) -{ - Datum str = PG_GETARG_DATUM(0); - - /* We need not detoast the input at all */ - PG_RETURN_INT32(toast_raw_datum_size(str) - VARHDRSZ); -} - -/* - * byteacat - - * takes two bytea* and returns a bytea* that is the concatenation of - * the two. - * - * Cloned from textcat and modified as required. - */ -Datum -byteacat(PG_FUNCTION_ARGS) -{ - bytea *t1 = PG_GETARG_BYTEA_PP(0); - bytea *t2 = PG_GETARG_BYTEA_PP(1); - - PG_RETURN_BYTEA_P(bytea_catenate(t1, t2)); -} - -/* - * bytea_catenate - * Guts of byteacat(), broken out so it can be used by other functions - * - * Arguments can be in short-header form, but not compressed or out-of-line - */ -static bytea * -bytea_catenate(bytea *t1, bytea *t2) -{ - bytea *result; - int len1, - len2, - len; - char *ptr; - - len1 = VARSIZE_ANY_EXHDR(t1); - len2 = VARSIZE_ANY_EXHDR(t2); - - /* paranoia ... probably should throw error instead? */ - if (len1 < 0) - len1 = 0; - if (len2 < 0) - len2 = 0; - - len = len1 + len2 + VARHDRSZ; - result = (bytea *) palloc(len); - - /* Set size of result string... */ - SET_VARSIZE(result, len); - - /* Fill data field of result string... */ - ptr = VARDATA(result); - if (len1 > 0) - memcpy(ptr, VARDATA_ANY(t1), len1); - if (len2 > 0) - memcpy(ptr + len1, VARDATA_ANY(t2), len2); - - return result; -} - -#define PG_STR_GET_BYTEA(str_) \ - DatumGetByteaPP(DirectFunctionCall1(byteain, CStringGetDatum(str_))) - -/* - * bytea_substr() - * Return a substring starting at the specified position. - * Cloned from text_substr and modified as required. - * - * Input: - * - string - * - starting position (is one-based) - * - string length (optional) - * - * If the starting position is zero or less, then return from the start of the string - * adjusting the length to be consistent with the "negative start" per SQL. - * If the length is less than zero, an ERROR is thrown. If no third argument - * (length) is provided, the length to the end of the string is assumed. - */ -Datum -bytea_substr(PG_FUNCTION_ARGS) -{ - PG_RETURN_BYTEA_P(bytea_substring(PG_GETARG_DATUM(0), - PG_GETARG_INT32(1), - PG_GETARG_INT32(2), - false)); -} - -/* - * bytea_substr_no_len - - * Wrapper to avoid opr_sanity failure due to - * one function accepting a different number of args. - */ -Datum -bytea_substr_no_len(PG_FUNCTION_ARGS) -{ - PG_RETURN_BYTEA_P(bytea_substring(PG_GETARG_DATUM(0), - PG_GETARG_INT32(1), - -1, - true)); -} - -static bytea * -bytea_substring(Datum str, - int S, - int L, - bool length_not_specified) -{ - int32 S1; /* adjusted start position */ - int32 L1; /* adjusted substring length */ - int32 E; /* end position */ - - /* - * The logic here should generally match text_substring(). - */ - S1 = Max(S, 1); - - if (length_not_specified) - { - /* - * Not passed a length - DatumGetByteaPSlice() grabs everything to the - * end of the string if we pass it a negative value for length. - */ - L1 = -1; - } - else if (L < 0) - { - /* SQL99 says to throw an error for E < S, i.e., negative length */ - ereport(ERROR, - (errcode(ERRCODE_SUBSTRING_ERROR), - errmsg("negative substring length not allowed"))); - L1 = -1; /* silence stupider compilers */ - } - else if (pg_add_s32_overflow(S, L, &E)) - { - /* - * L could be large enough for S + L to overflow, in which case the - * substring must run to end of string. - */ - L1 = -1; - } - else - { - /* - * A zero or negative value for the end position can happen if the - * start was negative or one. SQL99 says to return a zero-length - * string. - */ - if (E < 1) - return PG_STR_GET_BYTEA(""); - - L1 = E - S1; - } - - /* - * If the start position is past the end of the string, SQL99 says to - * return a zero-length string -- DatumGetByteaPSlice() will do that for - * us. We need only convert S1 to zero-based starting position. - */ - return DatumGetByteaPSlice(str, S1 - 1, L1); -} - -/* - * byteaoverlay - * Replace specified substring of first string with second - * - * The SQL standard defines OVERLAY() in terms of substring and concatenation. - * This code is a direct implementation of what the standard says. - */ -Datum -byteaoverlay(PG_FUNCTION_ARGS) -{ - bytea *t1 = PG_GETARG_BYTEA_PP(0); - bytea *t2 = PG_GETARG_BYTEA_PP(1); - int sp = PG_GETARG_INT32(2); /* substring start position */ - int sl = PG_GETARG_INT32(3); /* substring length */ - - PG_RETURN_BYTEA_P(bytea_overlay(t1, t2, sp, sl)); -} - -Datum -byteaoverlay_no_len(PG_FUNCTION_ARGS) -{ - bytea *t1 = PG_GETARG_BYTEA_PP(0); - bytea *t2 = PG_GETARG_BYTEA_PP(1); - int sp = PG_GETARG_INT32(2); /* substring start position */ - int sl; - - sl = VARSIZE_ANY_EXHDR(t2); /* defaults to length(t2) */ - PG_RETURN_BYTEA_P(bytea_overlay(t1, t2, sp, sl)); -} - -static bytea * -bytea_overlay(bytea *t1, bytea *t2, int sp, int sl) -{ - bytea *result; - bytea *s1; - bytea *s2; - int sp_pl_sl; - - /* - * Check for possible integer-overflow cases. For negative sp, throw a - * "substring length" error because that's what should be expected - * according to the spec's definition of OVERLAY(). - */ - if (sp <= 0) - ereport(ERROR, - (errcode(ERRCODE_SUBSTRING_ERROR), - errmsg("negative substring length not allowed"))); - if (pg_add_s32_overflow(sp, sl, &sp_pl_sl)) - ereport(ERROR, - (errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), - errmsg("integer out of range"))); - - s1 = bytea_substring(PointerGetDatum(t1), 1, sp - 1, false); - s2 = bytea_substring(PointerGetDatum(t1), sp_pl_sl, -1, true); - result = bytea_catenate(s1, t2); - result = bytea_catenate(result, s2); - - return result; -} - -/* - * bit_count - */ -Datum -bytea_bit_count(PG_FUNCTION_ARGS) -{ - bytea *t1 = PG_GETARG_BYTEA_PP(0); - - PG_RETURN_INT64(pg_popcount(VARDATA_ANY(t1), VARSIZE_ANY_EXHDR(t1))); -} - -/* - * byteapos - - * Return the position of the specified substring. - * Implements the SQL POSITION() function. - * Cloned from textpos and modified as required. - */ -Datum -byteapos(PG_FUNCTION_ARGS) -{ - bytea *t1 = PG_GETARG_BYTEA_PP(0); - bytea *t2 = PG_GETARG_BYTEA_PP(1); - int pos; - int px, - p; - int len1, - len2; - char *p1, - *p2; - - len1 = VARSIZE_ANY_EXHDR(t1); - len2 = VARSIZE_ANY_EXHDR(t2); - - if (len2 <= 0) - PG_RETURN_INT32(1); /* result for empty pattern */ - - p1 = VARDATA_ANY(t1); - p2 = VARDATA_ANY(t2); - - pos = 0; - px = (len1 - len2); - for (p = 0; p <= px; p++) - { - if ((*p2 == *p1) && (memcmp(p1, p2, len2) == 0)) - { - pos = p + 1; - break; - }; - p1++; - }; - - PG_RETURN_INT32(pos); -} - -/*------------------------------------------------------------- - * byteaGetByte - * - * this routine treats "bytea" as an array of bytes. - * It returns the Nth byte (a number between 0 and 255). - *------------------------------------------------------------- - */ -Datum -byteaGetByte(PG_FUNCTION_ARGS) -{ - bytea *v = PG_GETARG_BYTEA_PP(0); - int32 n = PG_GETARG_INT32(1); - int len; - int byte; - - len = VARSIZE_ANY_EXHDR(v); - - if (n < 0 || n >= len) - ereport(ERROR, - (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), - errmsg("index %d out of valid range, 0..%d", - n, len - 1))); - - byte = ((unsigned char *) VARDATA_ANY(v))[n]; - - PG_RETURN_INT32(byte); -} - -/*------------------------------------------------------------- - * byteaGetBit - * - * This routine treats a "bytea" type like an array of bits. - * It returns the value of the Nth bit (0 or 1). - * - *------------------------------------------------------------- - */ -Datum -byteaGetBit(PG_FUNCTION_ARGS) -{ - bytea *v = PG_GETARG_BYTEA_PP(0); - int64 n = PG_GETARG_INT64(1); - int byteNo, - bitNo; - int len; - int byte; - - len = VARSIZE_ANY_EXHDR(v); - - if (n < 0 || n >= (int64) len * 8) - ereport(ERROR, - (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), - errmsg("index %" PRId64 " out of valid range, 0..%" PRId64, - n, (int64) len * 8 - 1))); - - /* n/8 is now known < len, so safe to cast to int */ - byteNo = (int) (n / 8); - bitNo = (int) (n % 8); - - byte = ((unsigned char *) VARDATA_ANY(v))[byteNo]; - - if (byte & (1 << bitNo)) - PG_RETURN_INT32(1); - else - PG_RETURN_INT32(0); -} - -/*------------------------------------------------------------- - * byteaSetByte - * - * Given an instance of type 'bytea' creates a new one with - * the Nth byte set to the given value. - * - *------------------------------------------------------------- - */ -Datum -byteaSetByte(PG_FUNCTION_ARGS) -{ - bytea *res = PG_GETARG_BYTEA_P_COPY(0); - int32 n = PG_GETARG_INT32(1); - int32 newByte = PG_GETARG_INT32(2); - int len; - - len = VARSIZE(res) - VARHDRSZ; - - if (n < 0 || n >= len) - ereport(ERROR, - (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), - errmsg("index %d out of valid range, 0..%d", - n, len - 1))); - - /* - * Now set the byte. - */ - ((unsigned char *) VARDATA(res))[n] = newByte; - - PG_RETURN_BYTEA_P(res); -} - -/*------------------------------------------------------------- - * byteaSetBit - * - * Given an instance of type 'bytea' creates a new one with - * the Nth bit set to the given value. - * - *------------------------------------------------------------- - */ -Datum -byteaSetBit(PG_FUNCTION_ARGS) -{ - bytea *res = PG_GETARG_BYTEA_P_COPY(0); - int64 n = PG_GETARG_INT64(1); - int32 newBit = PG_GETARG_INT32(2); - int len; - int oldByte, - newByte; - int byteNo, - bitNo; - - len = VARSIZE(res) - VARHDRSZ; - - if (n < 0 || n >= (int64) len * 8) - ereport(ERROR, - (errcode(ERRCODE_ARRAY_SUBSCRIPT_ERROR), - errmsg("index %" PRId64 " out of valid range, 0..%" PRId64, - n, (int64) len * 8 - 1))); - - /* n/8 is now known < len, so safe to cast to int */ - byteNo = (int) (n / 8); - bitNo = (int) (n % 8); - - /* - * sanity check! - */ - if (newBit != 0 && newBit != 1) - ereport(ERROR, - (errcode(ERRCODE_INVALID_PARAMETER_VALUE), - errmsg("new bit must be 0 or 1"))); - - /* - * Update the byte. - */ - oldByte = ((unsigned char *) VARDATA(res))[byteNo]; - - if (newBit == 0) - newByte = oldByte & (~(1 << bitNo)); - else - newByte = oldByte | (1 << bitNo); - - ((unsigned char *) VARDATA(res))[byteNo] = newByte; - - PG_RETURN_BYTEA_P(res); -} - -/* - * Return reversed bytea - */ -Datum -bytea_reverse(PG_FUNCTION_ARGS) -{ - bytea *v = PG_GETARG_BYTEA_PP(0); - const char *p = VARDATA_ANY(v); - int len = VARSIZE_ANY_EXHDR(v); - const char *endp = p + len; - bytea *result = palloc(len + VARHDRSZ); - char *dst = (char *) VARDATA(result) + len; - - SET_VARSIZE(result, len + VARHDRSZ); - - while (p < endp) - *(--dst) = *p++; - - PG_RETURN_BYTEA_P(result); -} - - /* text_name() * Converts a text type to a Name type. */ @@ -3849,331 +3076,6 @@ SplitGUCList(char *rawstring, char separator, return true; } - -/***************************************************************************** - * Comparison Functions used for bytea - * - * Note: btree indexes need these routines not to leak memory; therefore, - * be careful to free working copies of toasted datums. Most places don't - * need to be so careful. - *****************************************************************************/ - -Datum -byteaeq(PG_FUNCTION_ARGS) -{ - Datum arg1 = PG_GETARG_DATUM(0); - Datum arg2 = PG_GETARG_DATUM(1); - bool result; - Size len1, - len2; - - /* - * We can use a fast path for unequal lengths, which might save us from - * having to detoast one or both values. - */ - len1 = toast_raw_datum_size(arg1); - len2 = toast_raw_datum_size(arg2); - if (len1 != len2) - result = false; - else - { - bytea *barg1 = DatumGetByteaPP(arg1); - bytea *barg2 = DatumGetByteaPP(arg2); - - result = (memcmp(VARDATA_ANY(barg1), VARDATA_ANY(barg2), - len1 - VARHDRSZ) == 0); - - PG_FREE_IF_COPY(barg1, 0); - PG_FREE_IF_COPY(barg2, 1); - } - - PG_RETURN_BOOL(result); -} - -Datum -byteane(PG_FUNCTION_ARGS) -{ - Datum arg1 = PG_GETARG_DATUM(0); - Datum arg2 = PG_GETARG_DATUM(1); - bool result; - Size len1, - len2; - - /* - * We can use a fast path for unequal lengths, which might save us from - * having to detoast one or both values. - */ - len1 = toast_raw_datum_size(arg1); - len2 = toast_raw_datum_size(arg2); - if (len1 != len2) - result = true; - else - { - bytea *barg1 = DatumGetByteaPP(arg1); - bytea *barg2 = DatumGetByteaPP(arg2); - - result = (memcmp(VARDATA_ANY(barg1), VARDATA_ANY(barg2), - len1 - VARHDRSZ) != 0); - - PG_FREE_IF_COPY(barg1, 0); - PG_FREE_IF_COPY(barg2, 1); - } - - PG_RETURN_BOOL(result); -} - -Datum -bytealt(PG_FUNCTION_ARGS) -{ - bytea *arg1 = PG_GETARG_BYTEA_PP(0); - bytea *arg2 = PG_GETARG_BYTEA_PP(1); - int len1, - len2; - int cmp; - - len1 = VARSIZE_ANY_EXHDR(arg1); - len2 = VARSIZE_ANY_EXHDR(arg2); - - cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); - - PG_FREE_IF_COPY(arg1, 0); - PG_FREE_IF_COPY(arg2, 1); - - PG_RETURN_BOOL((cmp < 0) || ((cmp == 0) && (len1 < len2))); -} - -Datum -byteale(PG_FUNCTION_ARGS) -{ - bytea *arg1 = PG_GETARG_BYTEA_PP(0); - bytea *arg2 = PG_GETARG_BYTEA_PP(1); - int len1, - len2; - int cmp; - - len1 = VARSIZE_ANY_EXHDR(arg1); - len2 = VARSIZE_ANY_EXHDR(arg2); - - cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); - - PG_FREE_IF_COPY(arg1, 0); - PG_FREE_IF_COPY(arg2, 1); - - PG_RETURN_BOOL((cmp < 0) || ((cmp == 0) && (len1 <= len2))); -} - -Datum -byteagt(PG_FUNCTION_ARGS) -{ - bytea *arg1 = PG_GETARG_BYTEA_PP(0); - bytea *arg2 = PG_GETARG_BYTEA_PP(1); - int len1, - len2; - int cmp; - - len1 = VARSIZE_ANY_EXHDR(arg1); - len2 = VARSIZE_ANY_EXHDR(arg2); - - cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); - - PG_FREE_IF_COPY(arg1, 0); - PG_FREE_IF_COPY(arg2, 1); - - PG_RETURN_BOOL((cmp > 0) || ((cmp == 0) && (len1 > len2))); -} - -Datum -byteage(PG_FUNCTION_ARGS) -{ - bytea *arg1 = PG_GETARG_BYTEA_PP(0); - bytea *arg2 = PG_GETARG_BYTEA_PP(1); - int len1, - len2; - int cmp; - - len1 = VARSIZE_ANY_EXHDR(arg1); - len2 = VARSIZE_ANY_EXHDR(arg2); - - cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); - - PG_FREE_IF_COPY(arg1, 0); - PG_FREE_IF_COPY(arg2, 1); - - PG_RETURN_BOOL((cmp > 0) || ((cmp == 0) && (len1 >= len2))); -} - -Datum -byteacmp(PG_FUNCTION_ARGS) -{ - bytea *arg1 = PG_GETARG_BYTEA_PP(0); - bytea *arg2 = PG_GETARG_BYTEA_PP(1); - int len1, - len2; - int cmp; - - len1 = VARSIZE_ANY_EXHDR(arg1); - len2 = VARSIZE_ANY_EXHDR(arg2); - - cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); - if ((cmp == 0) && (len1 != len2)) - cmp = (len1 < len2) ? -1 : 1; - - PG_FREE_IF_COPY(arg1, 0); - PG_FREE_IF_COPY(arg2, 1); - - PG_RETURN_INT32(cmp); -} - -Datum -bytea_larger(PG_FUNCTION_ARGS) -{ - bytea *arg1 = PG_GETARG_BYTEA_PP(0); - bytea *arg2 = PG_GETARG_BYTEA_PP(1); - bytea *result; - int len1, - len2; - int cmp; - - len1 = VARSIZE_ANY_EXHDR(arg1); - len2 = VARSIZE_ANY_EXHDR(arg2); - - cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); - result = ((cmp > 0) || ((cmp == 0) && (len1 > len2)) ? arg1 : arg2); - - PG_RETURN_BYTEA_P(result); -} - -Datum -bytea_smaller(PG_FUNCTION_ARGS) -{ - bytea *arg1 = PG_GETARG_BYTEA_PP(0); - bytea *arg2 = PG_GETARG_BYTEA_PP(1); - bytea *result; - int len1, - len2; - int cmp; - - len1 = VARSIZE_ANY_EXHDR(arg1); - len2 = VARSIZE_ANY_EXHDR(arg2); - - cmp = memcmp(VARDATA_ANY(arg1), VARDATA_ANY(arg2), Min(len1, len2)); - result = ((cmp < 0) || ((cmp == 0) && (len1 < len2)) ? arg1 : arg2); - - PG_RETURN_BYTEA_P(result); -} - -Datum -bytea_sortsupport(PG_FUNCTION_ARGS) -{ - SortSupport ssup = (SortSupport) PG_GETARG_POINTER(0); - MemoryContext oldcontext; - - oldcontext = MemoryContextSwitchTo(ssup->ssup_cxt); - - /* Use generic string SortSupport, forcing "C" collation */ - varstr_sortsupport(ssup, BYTEAOID, C_COLLATION_OID); - - MemoryContextSwitchTo(oldcontext); - - PG_RETURN_VOID(); -} - -/* Cast bytea -> int2 */ -Datum -bytea_int2(PG_FUNCTION_ARGS) -{ - bytea *v = PG_GETARG_BYTEA_PP(0); - int len = VARSIZE_ANY_EXHDR(v); - uint16 result; - - /* Check that the byte array is not too long */ - if (len > sizeof(result)) - ereport(ERROR, - errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), - errmsg("smallint out of range")); - - /* Convert it to an integer; most significant bytes come first */ - result = 0; - for (int i = 0; i < len; i++) - { - result <<= BITS_PER_BYTE; - result |= ((unsigned char *) VARDATA_ANY(v))[i]; - } - - PG_RETURN_INT16(result); -} - -/* Cast bytea -> int4 */ -Datum -bytea_int4(PG_FUNCTION_ARGS) -{ - bytea *v = PG_GETARG_BYTEA_PP(0); - int len = VARSIZE_ANY_EXHDR(v); - uint32 result; - - /* Check that the byte array is not too long */ - if (len > sizeof(result)) - ereport(ERROR, - errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), - errmsg("integer out of range")); - - /* Convert it to an integer; most significant bytes come first */ - result = 0; - for (int i = 0; i < len; i++) - { - result <<= BITS_PER_BYTE; - result |= ((unsigned char *) VARDATA_ANY(v))[i]; - } - - PG_RETURN_INT32(result); -} - -/* Cast bytea -> int8 */ -Datum -bytea_int8(PG_FUNCTION_ARGS) -{ - bytea *v = PG_GETARG_BYTEA_PP(0); - int len = VARSIZE_ANY_EXHDR(v); - uint64 result; - - /* Check that the byte array is not too long */ - if (len > sizeof(result)) - ereport(ERROR, - errcode(ERRCODE_NUMERIC_VALUE_OUT_OF_RANGE), - errmsg("bigint out of range")); - - /* Convert it to an integer; most significant bytes come first */ - result = 0; - for (int i = 0; i < len; i++) - { - result <<= BITS_PER_BYTE; - result |= ((unsigned char *) VARDATA_ANY(v))[i]; - } - - PG_RETURN_INT64(result); -} - -/* Cast int2 -> bytea; can just use int2send() */ -Datum -int2_bytea(PG_FUNCTION_ARGS) -{ - return int2send(fcinfo); -} - -/* Cast int4 -> bytea; can just use int4send() */ -Datum -int4_bytea(PG_FUNCTION_ARGS) -{ - return int4send(fcinfo); -} - -/* Cast int8 -> bytea; can just use int8send() */ -Datum -int8_bytea(PG_FUNCTION_ARGS) -{ - return int8send(fcinfo); -} - /* * appendStringInfoText * diff --git a/src/backend/utils/adt/xml.c b/src/backend/utils/adt/xml.c index a4150bff2ea..182e8f75db7 100644 --- a/src/backend/utils/adt/xml.c +++ b/src/backend/utils/adt/xml.c @@ -529,14 +529,36 @@ xmltext(PG_FUNCTION_ARGS) #ifdef USE_LIBXML text *arg = PG_GETARG_TEXT_PP(0); text *result; - xmlChar *xmlbuf = NULL; + volatile xmlChar *xmlbuf = NULL; + PgXmlErrorContext *xmlerrcxt; + + /* First we gotta spin up some error handling. */ + xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL); + + PG_TRY(); + { + xmlbuf = xmlEncodeSpecialChars(NULL, xml_text2xmlChar(arg)); + + if (xmlbuf == NULL || xmlerrcxt->err_occurred) + xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY, + "could not allocate xmlChar"); - xmlbuf = xmlEncodeSpecialChars(NULL, xml_text2xmlChar(arg)); + result = cstring_to_text_with_len((const char *) xmlbuf, + xmlStrlen((const xmlChar *) xmlbuf)); + } + PG_CATCH(); + { + if (xmlbuf) + xmlFree((xmlChar *) xmlbuf); - Assert(xmlbuf); + pg_xml_done(xmlerrcxt, true); + PG_RE_THROW(); + } + PG_END_TRY(); + + xmlFree((xmlChar *) xmlbuf); + pg_xml_done(xmlerrcxt, false); - result = cstring_to_text_with_len((const char *) xmlbuf, xmlStrlen(xmlbuf)); - xmlFree(xmlbuf); PG_RETURN_XML_P(result); #else NO_XML_SUPPORT(); @@ -663,7 +685,7 @@ xmltotext_with_options(xmltype *data, XmlOptionType xmloption_arg, bool indent) volatile xmlBufferPtr buf = NULL; volatile xmlSaveCtxtPtr ctxt = NULL; ErrorSaveContext escontext = {T_ErrorSaveContext}; - PgXmlErrorContext *xmlerrcxt; + PgXmlErrorContext *volatile xmlerrcxt = NULL; #endif if (xmloption_arg != XMLOPTION_DOCUMENT && !indent) @@ -704,13 +726,18 @@ xmltotext_with_options(xmltype *data, XmlOptionType xmloption_arg, bool indent) return (text *) data; } - /* Otherwise, we gotta spin up some error handling. */ - xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL); - + /* + * Otherwise, we gotta spin up some error handling. Unlike most other + * routines in this module, we already have a libxml "doc" structure to + * free, so we need to call pg_xml_init() inside the PG_TRY and be + * prepared for it to fail (typically due to palloc OOM). + */ PG_TRY(); { size_t decl_len = 0; + xmlerrcxt = pg_xml_init(PG_XML_STRICTNESS_ALL); + /* The serialized data will go into this buffer. */ buf = xmlBufferCreate(); @@ -770,7 +797,10 @@ xmltotext_with_options(xmltype *data, XmlOptionType xmloption_arg, bool indent) if (oldroot != NULL) xmlFreeNode(oldroot); - xmlAddChildList(root, content_nodes); + if (xmlAddChildList(root, content_nodes) == NULL || + xmlerrcxt->err_occurred) + xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR, + "could not append xml node list"); /* * We use this node to insert newlines in the dump. Note: in at @@ -838,10 +868,10 @@ xmltotext_with_options(xmltype *data, XmlOptionType xmloption_arg, bool indent) xmlSaveClose(ctxt); if (buf) xmlBufferFree(buf); - if (doc) - xmlFreeDoc(doc); + xmlFreeDoc(doc); - pg_xml_done(xmlerrcxt, true); + if (xmlerrcxt) + pg_xml_done(xmlerrcxt, true); PG_RE_THROW(); } @@ -931,7 +961,10 @@ xmlelement(XmlExpr *xexpr, xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY, "could not allocate xmlTextWriter"); - xmlTextWriterStartElement(writer, (xmlChar *) xexpr->name); + if (xmlTextWriterStartElement(writer, (xmlChar *) xexpr->name) < 0 || + xmlerrcxt->err_occurred) + xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR, + "could not start xml element"); forboth(arg, named_arg_strings, narg, xexpr->arg_names) { @@ -939,19 +972,30 @@ xmlelement(XmlExpr *xexpr, char *argname = strVal(lfirst(narg)); if (str) - xmlTextWriterWriteAttribute(writer, - (xmlChar *) argname, - (xmlChar *) str); + { + if (xmlTextWriterWriteAttribute(writer, + (xmlChar *) argname, + (xmlChar *) str) < 0 || + xmlerrcxt->err_occurred) + xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR, + "could not write xml attribute"); + } } foreach(arg, arg_strings) { char *str = (char *) lfirst(arg); - xmlTextWriterWriteRaw(writer, (xmlChar *) str); + if (xmlTextWriterWriteRaw(writer, (xmlChar *) str) < 0 || + xmlerrcxt->err_occurred) + xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR, + "could not write raw xml text"); } - xmlTextWriterEndElement(writer); + if (xmlTextWriterEndElement(writer) < 0 || + xmlerrcxt->err_occurred) + xml_ereport(xmlerrcxt, ERROR, ERRCODE_INTERNAL_ERROR, + "could not end xml element"); /* we MUST do this now to flush data out to the buffer ... */ xmlFreeTextWriter(writer); @@ -1725,7 +1769,7 @@ xml_doctype_in_content(const xmlChar *str) * xmloption_arg, but a DOCTYPE node in the input can force DOCUMENT mode). * * If parsed_nodes isn't NULL and we parse in CONTENT mode, the list - * of parsed nodes from the xmlParseInNodeContext call will be returned + * of parsed nodes from the xmlParseBalancedChunkMemory call will be returned * to *parsed_nodes. (It is caller's responsibility to free that.) * * Errors normally result in ereport(ERROR), but if escontext is an @@ -1751,6 +1795,7 @@ xml_parse(text *data, XmlOptionType xmloption_arg, PgXmlErrorContext *xmlerrcxt; volatile xmlParserCtxtPtr ctxt = NULL; volatile xmlDocPtr doc = NULL; + volatile int save_keep_blanks = -1; /* * This step looks annoyingly redundant, but we must do it to have a @@ -1778,7 +1823,6 @@ xml_parse(text *data, XmlOptionType xmloption_arg, PG_TRY(); { bool parse_as_document = false; - int options; int res_code; size_t count = 0; xmlChar *version = NULL; @@ -1809,18 +1853,6 @@ xml_parse(text *data, XmlOptionType xmloption_arg, parse_as_document = true; } - /* - * Select parse options. - * - * Note that here we try to apply DTD defaults (XML_PARSE_DTDATTR) - * according to SQL/XML:2008 GR 10.16.7.d: 'Default values defined by - * internal DTD are applied'. As for external DTDs, we try to support - * them too (see SQL/XML:2008 GR 10.16.7.e), but that doesn't really - * happen because xmlPgEntityLoader prevents it. - */ - options = XML_PARSE_NOENT | XML_PARSE_DTDATTR - | (preserve_whitespace ? 0 : XML_PARSE_NOBLANKS); - /* initialize output parameters */ if (parsed_xmloptiontype != NULL) *parsed_xmloptiontype = parse_as_document ? XMLOPTION_DOCUMENT : @@ -1830,11 +1862,26 @@ xml_parse(text *data, XmlOptionType xmloption_arg, if (parse_as_document) { + int options; + + /* set up parser context used by xmlCtxtReadDoc */ ctxt = xmlNewParserCtxt(); if (ctxt == NULL || xmlerrcxt->err_occurred) xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY, "could not allocate parser context"); + /* + * Select parse options. + * + * Note that here we try to apply DTD defaults (XML_PARSE_DTDATTR) + * according to SQL/XML:2008 GR 10.16.7.d: 'Default values defined + * by internal DTD are applied'. As for external DTDs, we try to + * support them too (see SQL/XML:2008 GR 10.16.7.e), but that + * doesn't really happen because xmlPgEntityLoader prevents it. + */ + options = XML_PARSE_NOENT | XML_PARSE_DTDATTR + | (preserve_whitespace ? 0 : XML_PARSE_NOBLANKS); + doc = xmlCtxtReadDoc(ctxt, utf8string, NULL, /* no URL */ "UTF-8", @@ -1856,10 +1903,7 @@ xml_parse(text *data, XmlOptionType xmloption_arg, } else { - xmlNodePtr root; - xmlNodePtr oldroot PG_USED_FOR_ASSERTS_ONLY; - - /* set up document with empty root node to be the context node */ + /* set up document that xmlParseBalancedChunkMemory will add to */ doc = xmlNewDoc(version); if (doc == NULL || xmlerrcxt->err_occurred) xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY, @@ -1872,43 +1916,22 @@ xml_parse(text *data, XmlOptionType xmloption_arg, "could not allocate XML document"); doc->standalone = standalone; - root = xmlNewNode(NULL, (const xmlChar *) "content-root"); - if (root == NULL || xmlerrcxt->err_occurred) - xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY, - "could not allocate xml node"); - - /* - * This attaches root to doc, so we need not free it separately; - * and there can't yet be any old root to free. - */ - oldroot = xmlDocSetRootElement(doc, root); - Assert(oldroot == NULL); + /* set parse options --- have to do this the ugly way */ + save_keep_blanks = xmlKeepBlanksDefault(preserve_whitespace ? 1 : 0); /* allow empty content */ if (*(utf8string + count)) { - xmlNodePtr node_list = NULL; - xmlParserErrors res; - - res = xmlParseInNodeContext(root, - (char *) utf8string + count, - strlen((char *) utf8string + count), - options, - &node_list); - - if (res != XML_ERR_OK || xmlerrcxt->err_occurred) + res_code = xmlParseBalancedChunkMemory(doc, NULL, NULL, 0, + utf8string + count, + parsed_nodes); + if (res_code != 0 || xmlerrcxt->err_occurred) { - xmlFreeNodeList(node_list); xml_errsave(escontext, xmlerrcxt, ERRCODE_INVALID_XML_CONTENT, "invalid XML content"); goto fail; } - - if (parsed_nodes != NULL) - *parsed_nodes = node_list; - else - xmlFreeNodeList(node_list); } } @@ -1917,6 +1940,8 @@ fail: } PG_CATCH(); { + if (save_keep_blanks != -1) + xmlKeepBlanksDefault(save_keep_blanks); if (doc != NULL) xmlFreeDoc(doc); if (ctxt != NULL) @@ -1928,6 +1953,9 @@ fail: } PG_END_TRY(); + if (save_keep_blanks != -1) + xmlKeepBlanksDefault(save_keep_blanks); + if (ctxt != NULL) xmlFreeParserCtxt(ctxt); @@ -4220,20 +4248,27 @@ xml_xmlnodetoxmltype(xmlNodePtr cur, PgXmlErrorContext *xmlerrcxt) } else { - xmlChar *str; + volatile xmlChar *str = NULL; - str = xmlXPathCastNodeToString(cur); PG_TRY(); { + char *escaped; + + str = xmlXPathCastNodeToString(cur); + if (str == NULL || xmlerrcxt->err_occurred) + xml_ereport(xmlerrcxt, ERROR, ERRCODE_OUT_OF_MEMORY, + "could not allocate xmlChar"); + /* Here we rely on XML having the same representation as TEXT */ - char *escaped = escape_xml((char *) str); + escaped = escape_xml((char *) str); result = (xmltype *) cstring_to_text(escaped); pfree(escaped); } PG_FINALLY(); { - xmlFree(str); + if (str) + xmlFree((xmlChar *) str); } PG_END_TRY(); } diff --git a/src/backend/utils/cache/catcache.c b/src/backend/utils/cache/catcache.c index 657648996c2..e2cd3feaf81 100644 --- a/src/backend/utils/cache/catcache.c +++ b/src/backend/utils/cache/catcache.c @@ -213,7 +213,7 @@ namehashfast(Datum datum) { char *key = NameStr(*DatumGetName(datum)); - return hash_any((unsigned char *) key, strlen(key)); + return hash_bytes((unsigned char *) key, strlen(key)); } static bool @@ -317,6 +317,7 @@ GetCCHashEqFuncs(Oid keytype, CCHashFN *hashfunc, RegProcedure *eqfunc, CCFastEq case REGDICTIONARYOID: case REGROLEOID: case REGNAMESPACEOID: + case REGDATABASEOID: *hashfunc = int4hashfast; *fasteqfunc = int4eqfast; *eqfunc = F_OIDEQ; diff --git a/src/backend/utils/cache/evtcache.c b/src/backend/utils/cache/evtcache.c index ce596bf5638..b9d5a5998be 100644 --- a/src/backend/utils/cache/evtcache.c +++ b/src/backend/utils/cache/evtcache.c @@ -78,7 +78,6 @@ BuildEventTriggerCache(void) { HASHCTL ctl; HTAB *cache; - MemoryContext oldcontext; Relation rel; Relation irel; SysScanDesc scan; @@ -110,9 +109,6 @@ BuildEventTriggerCache(void) (Datum) 0); } - /* Switch to correct memory context. */ - oldcontext = MemoryContextSwitchTo(EventTriggerCacheContext); - /* Prevent the memory context from being nuked while we're rebuilding. */ EventTriggerCacheState = ETCS_REBUILD_STARTED; @@ -145,6 +141,7 @@ BuildEventTriggerCache(void) bool evttags_isnull; EventTriggerCacheEntry *entry; bool found; + MemoryContext oldcontext; /* Get next tuple. */ tup = systable_getnext_ordered(scan, ForwardScanDirection); @@ -171,6 +168,9 @@ BuildEventTriggerCache(void) else continue; + /* Switch to correct memory context. */ + oldcontext = MemoryContextSwitchTo(EventTriggerCacheContext); + /* Allocate new cache item. */ item = palloc0(sizeof(EventTriggerCacheItem)); item->fnoid = form->evtfoid; @@ -188,6 +188,9 @@ BuildEventTriggerCache(void) entry->triggerlist = lappend(entry->triggerlist, item); else entry->triggerlist = list_make1(item); + + /* Restore previous memory context. */ + MemoryContextSwitchTo(oldcontext); } /* Done with pg_event_trigger scan. */ @@ -195,9 +198,6 @@ BuildEventTriggerCache(void) index_close(irel, AccessShareLock); relation_close(rel, AccessShareLock); - /* Restore previous memory context. */ - MemoryContextSwitchTo(oldcontext); - /* Install new cache. */ EventTriggerCache = cache; @@ -240,6 +240,8 @@ DecodeTextArrayToBitmapset(Datum array) } pfree(elems); + if ((Pointer) arr != DatumGetPointer(array)) + pfree(arr); return bms; } diff --git a/src/backend/utils/cache/plancache.c b/src/backend/utils/cache/plancache.c index 89a1c79e984..6661d2c6b73 100644 --- a/src/backend/utils/cache/plancache.c +++ b/src/backend/utils/cache/plancache.c @@ -463,8 +463,7 @@ CompleteCachedPlan(CachedPlanSource *plansource, /* * Save the final parameter types (or other parameter specification data) - * into the source_context, as well as our other parameters. Also save - * the result tuple descriptor. + * into the source_context, as well as our other parameters. */ MemoryContextSwitchTo(source_context); @@ -480,9 +479,25 @@ CompleteCachedPlan(CachedPlanSource *plansource, plansource->parserSetupArg = parserSetupArg; plansource->cursor_options = cursor_options; plansource->fixed_result = fixed_result; - plansource->resultDesc = PlanCacheComputeResultDesc(querytree_list); + /* + * Also save the result tuple descriptor. PlanCacheComputeResultDesc may + * leak some cruft; normally we just accept that to save a copy step, but + * in USE_VALGRIND mode be tidy by running it in the caller's context. + */ +#ifdef USE_VALGRIND + MemoryContextSwitchTo(oldcxt); + plansource->resultDesc = PlanCacheComputeResultDesc(querytree_list); + if (plansource->resultDesc) + { + MemoryContextSwitchTo(source_context); + plansource->resultDesc = CreateTupleDescCopy(plansource->resultDesc); + MemoryContextSwitchTo(oldcxt); + } +#else + plansource->resultDesc = PlanCacheComputeResultDesc(querytree_list); MemoryContextSwitchTo(oldcxt); +#endif plansource->is_complete = true; plansource->is_valid = true; @@ -1283,6 +1298,7 @@ GetCachedPlan(CachedPlanSource *plansource, ParamListInfo boundParams, CachedPlan *plan = NULL; List *qlist; bool customplan; + ListCell *lc; /* Assert caller is doing things in a sane order */ Assert(plansource->magic == CACHEDPLANSOURCE_MAGIC); @@ -1385,6 +1401,13 @@ GetCachedPlan(CachedPlanSource *plansource, ParamListInfo boundParams, plan->is_saved = true; } + foreach(lc, plan->stmt_list) + { + PlannedStmt *pstmt = (PlannedStmt *) lfirst(lc); + + pstmt->planOrigin = customplan ? PLAN_STMT_CACHE_CUSTOM : PLAN_STMT_CACHE_GENERIC; + } + return plan; } diff --git a/src/backend/utils/cache/ts_cache.c b/src/backend/utils/cache/ts_cache.c index 18cccd778fd..e8ae53238d0 100644 --- a/src/backend/utils/cache/ts_cache.c +++ b/src/backend/utils/cache/ts_cache.c @@ -321,7 +321,9 @@ lookup_ts_dictionary_cache(Oid dictId) /* * Init method runs in dictionary's private memory context, and we - * make sure the options are stored there too + * make sure the options are stored there too. This typically + * results in a small amount of memory leakage, but it's not worth + * complicating the API for tmplinit functions to avoid it. */ oldcontext = MemoryContextSwitchTo(entry->dictCtx); diff --git a/src/backend/utils/cache/typcache.c b/src/backend/utils/cache/typcache.c index f9aec38a11f..6a347698edf 100644 --- a/src/backend/utils/cache/typcache.c +++ b/src/backend/utils/cache/typcache.c @@ -1171,9 +1171,6 @@ load_domaintype_info(TypeCacheEntry *typentry) elog(ERROR, "domain \"%s\" constraint \"%s\" has NULL conbin", NameStr(typTup->typname), NameStr(c->conname)); - /* Convert conbin to C string in caller context */ - constring = TextDatumGetCString(val); - /* Create the DomainConstraintCache object and context if needed */ if (dcc == NULL) { @@ -1189,9 +1186,8 @@ load_domaintype_info(TypeCacheEntry *typentry) dcc->dccRefCount = 0; } - /* Create node trees in DomainConstraintCache's context */ - oldcxt = MemoryContextSwitchTo(dcc->dccContext); - + /* Convert conbin to a node tree, still in caller's context */ + constring = TextDatumGetCString(val); check_expr = (Expr *) stringToNode(constring); /* @@ -1206,10 +1202,13 @@ load_domaintype_info(TypeCacheEntry *typentry) */ check_expr = expression_planner(check_expr); + /* Create only the minimally needed stuff in dccContext */ + oldcxt = MemoryContextSwitchTo(dcc->dccContext); + r = makeNode(DomainConstraintState); r->constrainttype = DOM_CONSTRAINT_CHECK; r->name = pstrdup(NameStr(c->conname)); - r->check_expr = check_expr; + r->check_expr = copyObject(check_expr); r->check_exprstate = NULL; MemoryContextSwitchTo(oldcxt); diff --git a/src/backend/utils/error/elog.c b/src/backend/utils/error/elog.c index 47af743990f..afce1a8e1f0 100644 --- a/src/backend/utils/error/elog.c +++ b/src/backend/utils/error/elog.c @@ -1128,12 +1128,15 @@ set_backtrace(ErrorData *edata, int num_skip) nframes = backtrace(buf, lengthof(buf)); strfrms = backtrace_symbols(buf, nframes); - if (strfrms == NULL) - return; - - for (int i = num_skip; i < nframes; i++) - appendStringInfo(&errtrace, "\n%s", strfrms[i]); - free(strfrms); + if (strfrms != NULL) + { + for (int i = num_skip; i < nframes; i++) + appendStringInfo(&errtrace, "\n%s", strfrms[i]); + free(strfrms); + } + else + appendStringInfoString(&errtrace, + "insufficient memory for backtrace generation"); } #else appendStringInfoString(&errtrace, diff --git a/src/backend/utils/fmgr/dfmgr.c b/src/backend/utils/fmgr/dfmgr.c index 603632581d0..4bb84ff7087 100644 --- a/src/backend/utils/fmgr/dfmgr.c +++ b/src/backend/utils/fmgr/dfmgr.c @@ -99,6 +99,14 @@ load_external_function(const char *filename, const char *funcname, void *lib_handle; void *retval; + /* + * If the value starts with "$libdir/", strip that. This is because many + * extensions have hardcoded '$libdir/foo' as their library name, which + * prevents using the path. + */ + if (strncmp(filename, "$libdir/", 8) == 0) + filename += 8; + /* Expand the possibly-abbreviated filename to an exact path name */ fullname = expand_dynamic_library_name(filename); @@ -456,14 +464,6 @@ expand_dynamic_library_name(const char *name) Assert(name); - /* - * If the value starts with "$libdir/", strip that. This is because many - * extensions have hardcoded '$libdir/foo' as their library name, which - * prevents using the path. - */ - if (strncmp(name, "$libdir/", 8) == 0) - name += 8; - have_slash = (first_dir_separator(name) != NULL); if (!have_slash) diff --git a/src/backend/utils/hash/dynahash.c b/src/backend/utils/hash/dynahash.c index 1ad155d446e..81da03629f0 100644 --- a/src/backend/utils/hash/dynahash.c +++ b/src/backend/utils/hash/dynahash.c @@ -22,10 +22,11 @@ * lookup key's hash value as a partition number --- this will work because * of the way calc_bucket() maps hash values to bucket numbers. * - * For hash tables in shared memory, the memory allocator function should - * match malloc's semantics of returning NULL on failure. For hash tables - * in local memory, we typically use palloc() which will throw error on - * failure. The code in this file has to cope with both cases. + * The memory allocator function should match malloc's semantics of returning + * NULL on failure. (This is essential for hash tables in shared memory. + * For hash tables in local memory, we used to use palloc() which will throw + * error on failure; but we no longer do, so it's untested whether this + * module will still cope with that behavior.) * * dynahash.c provides support for these types of lookup keys: * @@ -98,6 +99,7 @@ #include "access/xact.h" #include "common/hashfn.h" +#include "lib/ilist.h" #include "port/pg_bitutils.h" #include "storage/shmem.h" #include "storage/spin.h" @@ -195,6 +197,7 @@ struct HASHHDR long ssize; /* segment size --- must be power of 2 */ int sshift; /* segment shift = log2(ssize) */ int nelem_alloc; /* number of entries to allocate at once */ + bool isfixed; /* if true, don't enlarge */ #ifdef HASH_STATISTICS @@ -227,7 +230,6 @@ struct HTAB MemoryContext hcxt; /* memory context if default allocator used */ char *tabname; /* table name (for error messages) */ bool isshared; /* true if table is in shared memory */ - bool isfixed; /* if true, don't enlarge */ /* freezing a shared table isn't allowed, so we can keep state here */ bool frozen; /* true = no more inserts allowed */ @@ -236,6 +238,16 @@ struct HTAB Size keysize; /* hash key length in bytes */ long ssize; /* segment size --- must be power of 2 */ int sshift; /* segment shift = log2(ssize) */ + + /* + * In a USE_VALGRIND build, non-shared hashtables keep an slist chain of + * all the element blocks they have allocated. This pacifies Valgrind, + * which would otherwise often claim that the element blocks are "possibly + * lost" for lack of any non-interior pointers to their starts. + */ +#ifdef USE_VALGRIND + slist_head element_blocks; +#endif }; /* @@ -618,8 +630,10 @@ hash_create(const char *tabname, long nelem, const HASHCTL *info, int flags) } } + /* Set isfixed if requested, but not till after we build initial entries */ if (flags & HASH_FIXED_SIZE) - hashp->isfixed = true; + hctl->isfixed = true; + return hashp; } @@ -644,6 +658,8 @@ hdefault(HTAB *hashp) hctl->ssize = DEF_SEGSIZE; hctl->sshift = DEF_SEGSIZE_SHIFT; + hctl->isfixed = false; /* can be enlarged */ + #ifdef HASH_STATISTICS hctl->accesses = hctl->collisions = 0; #endif @@ -1708,23 +1724,51 @@ element_alloc(HTAB *hashp, int nelem, int freelist_idx) { HASHHDR *hctl = hashp->hctl; Size elementSize; + Size requestSize; + char *allocedBlock; HASHELEMENT *firstElement; HASHELEMENT *tmpElement; HASHELEMENT *prevElement; int i; - if (hashp->isfixed) + if (hctl->isfixed) return false; /* Each element has a HASHELEMENT header plus user data. */ elementSize = MAXALIGN(sizeof(HASHELEMENT)) + MAXALIGN(hctl->entrysize); + requestSize = nelem * elementSize; + + /* Add space for slist_node list link if we need one. */ +#ifdef USE_VALGRIND + if (!hashp->isshared) + requestSize += MAXALIGN(sizeof(slist_node)); +#endif + + /* Allocate the memory. */ CurrentDynaHashCxt = hashp->hcxt; - firstElement = (HASHELEMENT *) hashp->alloc(nelem * elementSize); + allocedBlock = hashp->alloc(requestSize); - if (!firstElement) + if (!allocedBlock) return false; + /* + * If USE_VALGRIND, each allocated block of elements of a non-shared + * hashtable is chained into a list, so that Valgrind won't think it's + * been leaked. + */ +#ifdef USE_VALGRIND + if (hashp->isshared) + firstElement = (HASHELEMENT *) allocedBlock; + else + { + slist_push_head(&hashp->element_blocks, (slist_node *) allocedBlock); + firstElement = (HASHELEMENT *) (allocedBlock + MAXALIGN(sizeof(slist_node))); + } +#else + firstElement = (HASHELEMENT *) allocedBlock; +#endif + /* prepare to link all the new entries into the freelist */ prevElement = NULL; tmpElement = firstElement; diff --git a/src/backend/utils/init/miscinit.c b/src/backend/utils/init/miscinit.c index 43b4dbccc3d..65d8cbfaed5 100644 --- a/src/backend/utils/init/miscinit.c +++ b/src/backend/utils/init/miscinit.c @@ -1183,7 +1183,6 @@ UnlinkLockFiles(int status, Datum arg) /* Should we complain if the unlink fails? */ } /* Since we're about to exit, no need to reclaim storage */ - lock_files = NIL; /* * Lock file removal should always be the last externally visible action diff --git a/src/backend/utils/init/postinit.c b/src/backend/utils/init/postinit.c index c86ceefda94..641e535a73c 100644 --- a/src/backend/utils/init/postinit.c +++ b/src/backend/utils/init/postinit.c @@ -417,12 +417,11 @@ CheckMyDatabase(const char *name, bool am_superuser, bool override_allow_connect datum = SysCacheGetAttrNotNull(DATABASEOID, tup, Anum_pg_database_datctype); ctype = TextDatumGetCString(datum); - if (pg_perm_setlocale(LC_COLLATE, collate) == NULL) - ereport(FATAL, - (errmsg("database locale is incompatible with operating system"), - errdetail("The database was initialized with LC_COLLATE \"%s\", " - " which is not recognized by setlocale().", collate), - errhint("Recreate the database with another locale or install the missing locale."))); + /* + * Historcally, we set LC_COLLATE from datcollate, as well. That's no + * longer necessary because all collation behavior is handled through + * pg_locale_t. + */ if (pg_perm_setlocale(LC_CTYPE, ctype) == NULL) ereport(FATAL, diff --git a/src/backend/utils/misc/guc.c b/src/backend/utils/misc/guc.c index 667df448732..e404c345e6e 100644 --- a/src/backend/utils/misc/guc.c +++ b/src/backend/utils/misc/guc.c @@ -249,6 +249,7 @@ static void reapply_stacked_values(struct config_generic *variable, const char *curvalue, GucContext curscontext, GucSource cursource, Oid cursrole); +static void free_placeholder(struct config_string *pHolder); static bool validate_option_array_item(const char *name, const char *value, bool skipIfNoPermissions); static void write_auto_conf_file(int fd, const char *filename, ConfigVariable *head); @@ -4722,8 +4723,13 @@ AlterSystemSetConfigFile(AlterSystemStmt *altersysstmt) * the config file cannot cause postmaster start to fail, so we * don't have to be too tense about possibly installing a bad * value.) + * + * As an exception, we skip this check if this is a RESET command + * for an unknown custom GUC, else there'd be no way for users to + * remove such settings with reserved prefixes. */ - (void) assignable_custom_variable_name(name, false, ERROR); + if (value || !valid_custom_variable_name(name)) + (void) assignable_custom_variable_name(name, false, ERROR); } /* @@ -5018,16 +5024,8 @@ define_custom_variable(struct config_generic *variable) set_config_sourcefile(name, pHolder->gen.sourcefile, pHolder->gen.sourceline); - /* - * Free up as much as we conveniently can of the placeholder structure. - * (This neglects any stack items, so it's possible for some memory to be - * leaked. Since this can only happen once per session per variable, it - * doesn't seem worth spending much code on.) - */ - set_string_field(pHolder, pHolder->variable, NULL); - set_string_field(pHolder, &pHolder->reset_val, NULL); - - guc_free(pHolder); + /* Now we can free the no-longer-referenced placeholder variable */ + free_placeholder(pHolder); } /* @@ -5127,6 +5125,25 @@ reapply_stacked_values(struct config_generic *variable, } /* + * Free up a no-longer-referenced placeholder GUC variable. + * + * This neglects any stack items, so it's possible for some memory to be + * leaked. Since this can only happen once per session per variable, it + * doesn't seem worth spending much code on. + */ +static void +free_placeholder(struct config_string *pHolder) +{ + /* Placeholders are always STRING type, so free their values */ + Assert(pHolder->gen.vartype == PGC_STRING); + set_string_field(pHolder, pHolder->variable, NULL); + set_string_field(pHolder, &pHolder->reset_val, NULL); + + guc_free(unconstify(char *, pHolder->gen.name)); + guc_free(pHolder); +} + +/* * Functions for extensions to call to define their custom GUC variables. */ void @@ -5286,9 +5303,7 @@ MarkGUCPrefixReserved(const char *className) /* * Check for existing placeholders. We must actually remove invalid - * placeholders, else future parallel worker startups will fail. (We - * don't bother trying to free associated memory, since this shouldn't - * happen often.) + * placeholders, else future parallel worker startups will fail. */ hash_seq_init(&status, guc_hashtab); while ((hentry = (GUCHashEntry *) hash_seq_search(&status)) != NULL) @@ -5312,6 +5327,8 @@ MarkGUCPrefixReserved(const char *className) NULL); /* Remove it from any lists it's in, too */ RemoveGUCFromLists(var); + /* And free it */ + free_placeholder((struct config_string *) var); } } @@ -6711,6 +6728,7 @@ validate_option_array_item(const char *name, const char *value, { struct config_generic *gconf; + bool reset_custom; /* * There are three cases to consider: @@ -6729,16 +6747,21 @@ validate_option_array_item(const char *name, const char *value, * it's assumed to be fully validated.) * * name is not known and can't be created as a placeholder. Throw error, - * unless skipIfNoPermissions is true, in which case return false. + * unless skipIfNoPermissions or reset_custom is true. If reset_custom is + * true, this is a RESET or RESET ALL operation for an unknown custom GUC + * with a reserved prefix, in which case we want to fall through to the + * placeholder case described in the preceding paragraph (else there'd be + * no way for users to remove them). Otherwise, return false. */ - gconf = find_option(name, true, skipIfNoPermissions, ERROR); - if (!gconf) + reset_custom = (!value && valid_custom_variable_name(name)); + gconf = find_option(name, true, skipIfNoPermissions || reset_custom, ERROR); + if (!gconf && !reset_custom) { /* not known, failed to make a placeholder */ return false; } - if (gconf->flags & GUC_CUSTOM_PLACEHOLDER) + if (!gconf || gconf->flags & GUC_CUSTOM_PLACEHOLDER) { /* * We cannot do any meaningful check on the value, so only permissions diff --git a/src/backend/utils/misc/guc_tables.c b/src/backend/utils/misc/guc_tables.c index 2f8cbd86759..d14b1678e7f 100644 --- a/src/backend/utils/misc/guc_tables.c +++ b/src/backend/utils/misc/guc_tables.c @@ -1028,7 +1028,7 @@ struct config_bool ConfigureNamesBool[] = }, { {"enable_distinct_reordering", PGC_USERSET, QUERY_TUNING_METHOD, - gettext_noop("Enables reordering of DISTINCT pathkeys."), + gettext_noop("Enables reordering of DISTINCT keys."), NULL, GUC_EXPLAIN }, @@ -1602,11 +1602,11 @@ struct config_bool ConfigureNamesBool[] = NULL, NULL, NULL }, { - {"log_lock_failure", PGC_SUSET, LOGGING_WHAT, + {"log_lock_failures", PGC_SUSET, LOGGING_WHAT, gettext_noop("Logs lock failures."), NULL }, - &log_lock_failure, + &log_lock_failures, false, NULL, NULL, NULL }, @@ -3081,7 +3081,7 @@ struct config_int ConfigureNamesInt[] = }, &max_slot_wal_keep_size_mb, -1, -1, MAX_KILOBYTES, - check_max_slot_wal_keep_size, NULL, NULL + NULL, NULL, NULL }, { @@ -3100,11 +3100,11 @@ struct config_int ConfigureNamesInt[] = gettext_noop("Sets the duration a replication slot can remain idle before " "it is invalidated."), NULL, - GUC_UNIT_MIN + GUC_UNIT_S }, - &idle_replication_slot_timeout_mins, - 0, 0, INT_MAX / SECS_PER_MINUTE, - check_idle_replication_slot_timeout, NULL, NULL + &idle_replication_slot_timeout_secs, + 0, 0, INT_MAX, + NULL, NULL, NULL }, { @@ -4837,7 +4837,7 @@ struct config_string ConfigureNamesString[] = { {"ssl_groups", PGC_SIGHUP, CONN_AUTH_SSL, gettext_noop("Sets the group(s) to use for Diffie-Hellman key exchange."), - gettext_noop("Multiple groups can be specified using colon-separated list."), + gettext_noop("Multiple groups can be specified using a colon-separated list."), GUC_SUPERUSER_ONLY }, &SSLECDHCurve, diff --git a/src/backend/utils/misc/injection_point.c b/src/backend/utils/misc/injection_point.c index f58ebc8ee52..83b887b6978 100644 --- a/src/backend/utils/misc/injection_point.c +++ b/src/backend/utils/misc/injection_point.c @@ -584,3 +584,49 @@ IsInjectionPointAttached(const char *name) return false; /* silence compiler */ #endif } + +/* + * Retrieve a list of all the injection points currently attached. + * + * This list is palloc'd in the current memory context. + */ +List * +InjectionPointList(void) +{ +#ifdef USE_INJECTION_POINTS + List *inj_points = NIL; + uint32 max_inuse; + + LWLockAcquire(InjectionPointLock, LW_SHARED); + + max_inuse = pg_atomic_read_u32(&ActiveInjectionPoints->max_inuse); + + for (uint32 idx = 0; idx < max_inuse; idx++) + { + InjectionPointEntry *entry; + InjectionPointData *inj_point; + uint64 generation; + + entry = &ActiveInjectionPoints->entries[idx]; + generation = pg_atomic_read_u64(&entry->generation); + + /* skip free slots */ + if (generation % 2 == 0) + continue; + + inj_point = (InjectionPointData *) palloc0(sizeof(InjectionPointData)); + inj_point->name = pstrdup(entry->name); + inj_point->library = pstrdup(entry->library); + inj_point->function = pstrdup(entry->function); + inj_points = lappend(inj_points, inj_point); + } + + LWLockRelease(InjectionPointLock); + + return inj_points; + +#else + elog(ERROR, "Injection points are not supported by this build"); + return NIL; /* keep compiler quiet */ +#endif +} diff --git a/src/backend/utils/misc/postgresql.conf.sample b/src/backend/utils/misc/postgresql.conf.sample index 87ce76b18f4..a9d8293474a 100644 --- a/src/backend/utils/misc/postgresql.conf.sample +++ b/src/backend/utils/misc/postgresql.conf.sample @@ -342,7 +342,7 @@ # (change requires restart) #wal_keep_size = 0 # in megabytes; 0 disables #max_slot_wal_keep_size = -1 # in megabytes; -1 disables -#idle_replication_slot_timeout = 0 # in minutes; 0 disables +#idle_replication_slot_timeout = 0 # in seconds; 0 disables #wal_sender_timeout = 60s # in milliseconds; 0 disables #track_commit_timestamp = off # collect timestamp of transaction commit # (change requires restart) @@ -624,7 +624,7 @@ # %% = '%' # e.g. '<%u%%%d> ' #log_lock_waits = off # log lock waits >= deadlock_timeout -#log_lock_failure = off # log lock failures +#log_lock_failures = off # log lock failures #log_recovery_conflict_waits = off # log standby recovery conflict waits # >= deadlock_timeout #log_parameter_max_length = -1 # when logging statements, limit logged diff --git a/src/backend/utils/misc/ps_status.c b/src/backend/utils/misc/ps_status.c index e08b26e8c14..4df25944deb 100644 --- a/src/backend/utils/misc/ps_status.c +++ b/src/backend/utils/misc/ps_status.c @@ -100,6 +100,17 @@ static void flush_ps_display(void); static int save_argc; static char **save_argv; +/* + * Valgrind seems not to consider the global "environ" variable as a valid + * root pointer; so when we allocate a new environment array, it claims that + * data is leaked. To fix that, keep our own statically-allocated copy of the + * pointer. (Oddly, this doesn't seem to be a problem for "argv".) + */ +#if defined(PS_USE_CLOBBER_ARGV) && defined(USE_VALGRIND) +extern char **ps_status_new_environ; +char **ps_status_new_environ; +#endif + /* * Call this early in startup to save the original argc/argv values. @@ -206,6 +217,11 @@ save_ps_display_args(int argc, char **argv) } new_environ[i] = NULL; environ = new_environ; + + /* See notes about Valgrind above. */ +#ifdef USE_VALGRIND + ps_status_new_environ = new_environ; +#endif } /* diff --git a/src/backend/utils/mmgr/alignedalloc.c b/src/backend/utils/mmgr/alignedalloc.c index 7eea695de62..b1be7426914 100644 --- a/src/backend/utils/mmgr/alignedalloc.c +++ b/src/backend/utils/mmgr/alignedalloc.c @@ -45,6 +45,15 @@ AlignedAllocFree(void *pointer) GetMemoryChunkContext(unaligned)->name, chunk); #endif + /* + * Create a dummy vchunk covering the start of the unaligned chunk, but + * not overlapping the aligned chunk. This will be freed while pfree'ing + * the unaligned chunk, keeping Valgrind happy. Then when we return to + * the outer pfree, that will clean up the vchunk for the aligned chunk. + */ + VALGRIND_MEMPOOL_ALLOC(GetMemoryChunkContext(unaligned), unaligned, + (char *) pointer - (char *) unaligned); + /* Recursively pfree the unaligned chunk */ pfree(unaligned); } @@ -123,6 +132,15 @@ AlignedAllocRealloc(void *pointer, Size size, int flags) VALGRIND_MAKE_MEM_DEFINED(pointer, old_size); memcpy(newptr, pointer, Min(size, old_size)); + /* + * Create a dummy vchunk covering the start of the old unaligned chunk, + * but not overlapping the aligned chunk. This will be freed while + * pfree'ing the old unaligned chunk, keeping Valgrind happy. Then when + * we return to repalloc, it will move the vchunk for the aligned chunk. + */ + VALGRIND_MEMPOOL_ALLOC(ctx, unaligned, + (char *) pointer - (char *) unaligned); + pfree(unaligned); return newptr; diff --git a/src/backend/utils/mmgr/aset.c b/src/backend/utils/mmgr/aset.c index 666ecd8f78d..9ef109ca586 100644 --- a/src/backend/utils/mmgr/aset.c +++ b/src/backend/utils/mmgr/aset.c @@ -103,6 +103,8 @@ #define ALLOC_BLOCKHDRSZ MAXALIGN(sizeof(AllocBlockData)) #define ALLOC_CHUNKHDRSZ sizeof(MemoryChunk) +#define FIRST_BLOCKHDRSZ (MAXALIGN(sizeof(AllocSetContext)) + \ + ALLOC_BLOCKHDRSZ) typedef struct AllocBlockData *AllocBlock; /* forward reference */ @@ -458,6 +460,21 @@ AllocSetContextCreateInternal(MemoryContext parent, * we'd leak the header/initial block if we ereport in this stretch. */ + /* Create a vpool associated with the context */ + VALGRIND_CREATE_MEMPOOL(set, 0, false); + + /* + * Create a vchunk covering both the AllocSetContext struct and the keeper + * block's header. (Perhaps it would be more sensible for these to be two + * separate vchunks, but doing that seems to tickle bugs in some versions + * of Valgrind.) We must have these vchunks, and also a vchunk for each + * subsequently-added block header, so that Valgrind considers the + * pointers within them while checking for leaked memory. Note that + * Valgrind doesn't distinguish between these vchunks and those created by + * mcxt.c for the user-accessible-data chunks we allocate. + */ + VALGRIND_MEMPOOL_ALLOC(set, set, FIRST_BLOCKHDRSZ); + /* Fill in the initial block's block header */ block = KeeperBlock(set); block->aset = set; @@ -585,6 +602,14 @@ AllocSetReset(MemoryContext context) #ifdef CLOBBER_FREED_MEMORY wipe_mem(block, block->freeptr - ((char *) block)); #endif + + /* + * We need to free the block header's vchunk explicitly, although + * the user-data vchunks within will go away in the TRIM below. + * Otherwise Valgrind complains about leaked allocations. + */ + VALGRIND_MEMPOOL_FREE(set, block); + free(block); } block = next; @@ -592,6 +617,14 @@ AllocSetReset(MemoryContext context) Assert(context->mem_allocated == keepersize); + /* + * Instruct Valgrind to throw away all the vchunks associated with this + * context, except for the one covering the AllocSetContext and + * keeper-block header. This gets rid of the vchunks for whatever user + * data is getting discarded by the context reset. + */ + VALGRIND_MEMPOOL_TRIM(set, set, FIRST_BLOCKHDRSZ); + /* Reset block size allocation sequence, too */ set->nextBlockSize = set->initBlockSize; } @@ -648,6 +681,9 @@ AllocSetDelete(MemoryContext context) freelist->first_free = (AllocSetContext *) oldset->header.nextchild; freelist->num_free--; + /* Destroy the context's vpool --- see notes below */ + VALGRIND_DESTROY_MEMPOOL(oldset); + /* All that remains is to free the header/initial block */ free(oldset); } @@ -675,13 +711,24 @@ AllocSetDelete(MemoryContext context) #endif if (!IsKeeperBlock(set, block)) + { + /* As in AllocSetReset, free block-header vchunks explicitly */ + VALGRIND_MEMPOOL_FREE(set, block); free(block); + } block = next; } Assert(context->mem_allocated == keepersize); + /* + * Destroy the vpool. We don't seem to need to explicitly free the + * initial block's header vchunk, nor any user-data vchunks that Valgrind + * still knows about; they'll all go away automatically. + */ + VALGRIND_DESTROY_MEMPOOL(set); + /* Finally, free the context header, including the keeper block */ free(set); } @@ -716,6 +763,9 @@ AllocSetAllocLarge(MemoryContext context, Size size, int flags) if (block == NULL) return MemoryContextAllocationFailure(context, size, flags); + /* Make a vchunk covering the new block's header */ + VALGRIND_MEMPOOL_ALLOC(set, block, ALLOC_BLOCKHDRSZ); + context->mem_allocated += blksize; block->aset = set; @@ -922,6 +972,9 @@ AllocSetAllocFromNewBlock(MemoryContext context, Size size, int flags, if (block == NULL) return MemoryContextAllocationFailure(context, size, flags); + /* Make a vchunk covering the new block's header */ + VALGRIND_MEMPOOL_ALLOC(set, block, ALLOC_BLOCKHDRSZ); + context->mem_allocated += blksize; block->aset = set; @@ -1104,6 +1157,10 @@ AllocSetFree(void *pointer) #ifdef CLOBBER_FREED_MEMORY wipe_mem(block, block->freeptr - ((char *) block)); #endif + + /* As in AllocSetReset, free block-header vchunks explicitly */ + VALGRIND_MEMPOOL_FREE(set, block); + free(block); } else @@ -1184,6 +1241,7 @@ AllocSetRealloc(void *pointer, Size size, int flags) * realloc() to make the containing block bigger, or smaller, with * minimum space wastage. */ + AllocBlock newblock; Size chksize; Size blksize; Size oldblksize; @@ -1223,14 +1281,21 @@ AllocSetRealloc(void *pointer, Size size, int flags) blksize = chksize + ALLOC_BLOCKHDRSZ + ALLOC_CHUNKHDRSZ; oldblksize = block->endptr - ((char *) block); - block = (AllocBlock) realloc(block, blksize); - if (block == NULL) + newblock = (AllocBlock) realloc(block, blksize); + if (newblock == NULL) { /* Disallow access to the chunk header. */ VALGRIND_MAKE_MEM_NOACCESS(chunk, ALLOC_CHUNKHDRSZ); return MemoryContextAllocationFailure(&set->header, size, flags); } + /* + * Move the block-header vchunk explicitly. (mcxt.c will take care of + * moving the vchunk for the user data.) + */ + VALGRIND_MEMPOOL_CHANGE(set, block, newblock, ALLOC_BLOCKHDRSZ); + block = newblock; + /* updated separately, not to underflow when (oldblksize > blksize) */ set->header.mem_allocated -= oldblksize; set->header.mem_allocated += blksize; @@ -1294,7 +1359,7 @@ AllocSetRealloc(void *pointer, Size size, int flags) /* Ensure any padding bytes are marked NOACCESS. */ VALGRIND_MAKE_MEM_NOACCESS((char *) pointer + size, chksize - size); - /* Disallow access to the chunk header . */ + /* Disallow access to the chunk header. */ VALGRIND_MAKE_MEM_NOACCESS(chunk, ALLOC_CHUNKHDRSZ); return pointer; diff --git a/src/backend/utils/mmgr/bump.c b/src/backend/utils/mmgr/bump.c index f7a37d1b3e8..2805d55a2ec 100644 --- a/src/backend/utils/mmgr/bump.c +++ b/src/backend/utils/mmgr/bump.c @@ -45,7 +45,9 @@ #include "utils/memutils_memorychunk.h" #include "utils/memutils_internal.h" -#define Bump_BLOCKHDRSZ MAXALIGN(sizeof(BumpBlock)) +#define Bump_BLOCKHDRSZ MAXALIGN(sizeof(BumpBlock)) +#define FIRST_BLOCKHDRSZ (MAXALIGN(sizeof(BumpContext)) + \ + Bump_BLOCKHDRSZ) /* No chunk header unless built with MEMORY_CONTEXT_CHECKING */ #ifdef MEMORY_CONTEXT_CHECKING @@ -189,6 +191,12 @@ BumpContextCreate(MemoryContext parent, const char *name, Size minContextSize, * Avoid writing code that can fail between here and MemoryContextCreate; * we'd leak the header and initial block if we ereport in this stretch. */ + + /* See comments about Valgrind interactions in aset.c */ + VALGRIND_CREATE_MEMPOOL(set, 0, false); + /* This vchunk covers the BumpContext and the keeper block header */ + VALGRIND_MEMPOOL_ALLOC(set, set, FIRST_BLOCKHDRSZ); + dlist_init(&set->blocks); /* Fill in the initial block's block header */ @@ -262,6 +270,14 @@ BumpReset(MemoryContext context) BumpBlockFree(set, block); } + /* + * Instruct Valgrind to throw away all the vchunks associated with this + * context, except for the one covering the BumpContext and keeper-block + * header. This gets rid of the vchunks for whatever user data is getting + * discarded by the context reset. + */ + VALGRIND_MEMPOOL_TRIM(set, set, FIRST_BLOCKHDRSZ); + /* Reset block size allocation sequence, too */ set->nextBlockSize = set->initBlockSize; @@ -279,6 +295,10 @@ BumpDelete(MemoryContext context) { /* Reset to release all releasable BumpBlocks */ BumpReset(context); + + /* Destroy the vpool -- see notes in aset.c */ + VALGRIND_DESTROY_MEMPOOL(context); + /* And free the context header and keeper block */ free(context); } @@ -318,6 +338,9 @@ BumpAllocLarge(MemoryContext context, Size size, int flags) if (block == NULL) return MemoryContextAllocationFailure(context, size, flags); + /* Make a vchunk covering the new block's header */ + VALGRIND_MEMPOOL_ALLOC(set, block, Bump_BLOCKHDRSZ); + context->mem_allocated += blksize; /* the block is completely full */ @@ -455,6 +478,9 @@ BumpAllocFromNewBlock(MemoryContext context, Size size, int flags, if (block == NULL) return MemoryContextAllocationFailure(context, size, flags); + /* Make a vchunk covering the new block's header */ + VALGRIND_MEMPOOL_ALLOC(set, block, Bump_BLOCKHDRSZ); + context->mem_allocated += blksize; /* initialize the new block */ @@ -606,6 +632,9 @@ BumpBlockFree(BumpContext *set, BumpBlock *block) wipe_mem(block, ((char *) block->endptr - (char *) block)); #endif + /* As in aset.c, free block-header vchunks explicitly */ + VALGRIND_MEMPOOL_FREE(set, block); + free(block); } diff --git a/src/backend/utils/mmgr/dsa.c b/src/backend/utils/mmgr/dsa.c index 17d4f7a7a06..be43e9351c3 100644 --- a/src/backend/utils/mmgr/dsa.c +++ b/src/backend/utils/mmgr/dsa.c @@ -532,6 +532,21 @@ dsa_attach(dsa_handle handle) } /* + * Returns whether the area with the given handle was already attached by the + * current process. The area must have been created with dsa_create (not + * dsa_create_in_place). + */ +bool +dsa_is_attached(dsa_handle handle) +{ + /* + * An area handle is really a DSM segment handle for the first segment, so + * we can just search for that. + */ + return dsm_find_mapping(handle) != NULL; +} + +/* * Attach to an area that was created with dsa_create_in_place. The caller * must somehow know the location in memory that was used when the area was * created, though it may be mapped at a different virtual address in this diff --git a/src/backend/utils/mmgr/generation.c b/src/backend/utils/mmgr/generation.c index 18679ad4f1e..cfafc9bf082 100644 --- a/src/backend/utils/mmgr/generation.c +++ b/src/backend/utils/mmgr/generation.c @@ -45,6 +45,8 @@ #define Generation_BLOCKHDRSZ MAXALIGN(sizeof(GenerationBlock)) #define Generation_CHUNKHDRSZ sizeof(MemoryChunk) +#define FIRST_BLOCKHDRSZ (MAXALIGN(sizeof(GenerationContext)) + \ + Generation_BLOCKHDRSZ) #define Generation_CHUNK_FRACTION 8 @@ -221,6 +223,12 @@ GenerationContextCreate(MemoryContext parent, * Avoid writing code that can fail between here and MemoryContextCreate; * we'd leak the header if we ereport in this stretch. */ + + /* See comments about Valgrind interactions in aset.c */ + VALGRIND_CREATE_MEMPOOL(set, 0, false); + /* This vchunk covers the GenerationContext and the keeper block header */ + VALGRIND_MEMPOOL_ALLOC(set, set, FIRST_BLOCKHDRSZ); + dlist_init(&set->blocks); /* Fill in the initial block's block header */ @@ -309,6 +317,14 @@ GenerationReset(MemoryContext context) GenerationBlockFree(set, block); } + /* + * Instruct Valgrind to throw away all the vchunks associated with this + * context, except for the one covering the GenerationContext and + * keeper-block header. This gets rid of the vchunks for whatever user + * data is getting discarded by the context reset. + */ + VALGRIND_MEMPOOL_TRIM(set, set, FIRST_BLOCKHDRSZ); + /* set it so new allocations to make use of the keeper block */ set->block = KeeperBlock(set); @@ -329,6 +345,10 @@ GenerationDelete(MemoryContext context) { /* Reset to release all releasable GenerationBlocks */ GenerationReset(context); + + /* Destroy the vpool -- see notes in aset.c */ + VALGRIND_DESTROY_MEMPOOL(context); + /* And free the context header and keeper block */ free(context); } @@ -365,6 +385,9 @@ GenerationAllocLarge(MemoryContext context, Size size, int flags) if (block == NULL) return MemoryContextAllocationFailure(context, size, flags); + /* Make a vchunk covering the new block's header */ + VALGRIND_MEMPOOL_ALLOC(set, block, Generation_BLOCKHDRSZ); + context->mem_allocated += blksize; /* block with a single (used) chunk */ @@ -487,6 +510,9 @@ GenerationAllocFromNewBlock(MemoryContext context, Size size, int flags, if (block == NULL) return MemoryContextAllocationFailure(context, size, flags); + /* Make a vchunk covering the new block's header */ + VALGRIND_MEMPOOL_ALLOC(set, block, Generation_BLOCKHDRSZ); + context->mem_allocated += blksize; /* initialize the new block */ @@ -677,6 +703,9 @@ GenerationBlockFree(GenerationContext *set, GenerationBlock *block) wipe_mem(block, block->blksize); #endif + /* As in aset.c, free block-header vchunks explicitly */ + VALGRIND_MEMPOOL_FREE(set, block); + free(block); } diff --git a/src/backend/utils/mmgr/mcxt.c b/src/backend/utils/mmgr/mcxt.c index 15fa4d0a55e..47fd774c7d2 100644 --- a/src/backend/utils/mmgr/mcxt.c +++ b/src/backend/utils/mmgr/mcxt.c @@ -8,6 +8,23 @@ * context-type-specific operations via the function pointers in a * context's MemoryContextMethods struct. * + * A note about Valgrind support: when USE_VALGRIND is defined, we provide + * support for memory leak tracking at the allocation-unit level. Valgrind + * does leak detection by tracking allocated "chunks", which can be grouped + * into "pools". The "chunk" terminology is overloaded, since we use that + * word for our allocation units, and it's sometimes important to distinguish + * those from the Valgrind objects that describe them. To reduce confusion, + * let's use the terms "vchunk" and "vpool" for the Valgrind objects. + * + * We use a separate vpool for each memory context. The context-type-specific + * code is responsible for creating and deleting the vpools, and also for + * creating vchunks to cover its management data structures such as block + * headers. (There must be a vchunk that includes every pointer we want + * Valgrind to consider for leak-tracking purposes.) This module creates + * and deletes the vchunks that cover the caller-visible allocated chunks. + * However, the context-type-specific code must handle cleaning up those + * vchunks too during memory context reset operations. + * * * Portions Copyright (c) 1996-2025, PostgreSQL Global Development Group * Portions Copyright (c) 1994, Regents of the University of California @@ -418,8 +435,6 @@ MemoryContextResetOnly(MemoryContext context) context->methods->reset(context); context->isReset = true; - VALGRIND_DESTROY_MEMPOOL(context); - VALGRIND_CREATE_MEMPOOL(context, 0, false); } } @@ -526,8 +541,6 @@ MemoryContextDeleteOnly(MemoryContext context) context->ident = NULL; context->methods->delete_context(context); - - VALGRIND_DESTROY_MEMPOOL(context); } /* @@ -560,9 +573,7 @@ MemoryContextDeleteChildren(MemoryContext context) * the specified context, since that means it will automatically be freed * when no longer needed. * - * There is no API for deregistering a callback once registered. If you - * want it to not do anything anymore, adjust the state pointed to by its - * "arg" to indicate that. + * Note that callers can assume this cannot fail. */ void MemoryContextRegisterResetCallback(MemoryContext context, @@ -578,6 +589,41 @@ MemoryContextRegisterResetCallback(MemoryContext context, } /* + * MemoryContextUnregisterResetCallback + * Undo the effects of MemoryContextRegisterResetCallback. + * + * This can be used if a callback's effects are no longer required + * at some point before the context has been reset/deleted. It is the + * caller's responsibility to pfree the callback struct (if needed). + * + * An assertion failure occurs if the callback was not registered. + * We could alternatively define that case as a no-op, but that seems too + * likely to mask programming errors such as passing the wrong context. + */ +void +MemoryContextUnregisterResetCallback(MemoryContext context, + MemoryContextCallback *cb) +{ + MemoryContextCallback *prev, + *cur; + + Assert(MemoryContextIsValid(context)); + + for (prev = NULL, cur = context->reset_cbs; cur != NULL; + prev = cur, cur = cur->next) + { + if (cur != cb) + continue; + if (prev) + prev->next = cur->next; + else + context->reset_cbs = cur->next; + return; + } + Assert(false); +} + +/* * MemoryContextCallResetCallbacks * Internal function to call all registered callbacks for context. */ @@ -1137,8 +1183,6 @@ MemoryContextCreate(MemoryContext node, node->nextchild = NULL; node->allowInCritSection = false; } - - VALGRIND_CREATE_MEMPOOL(node, 0, false); } /* @@ -1421,7 +1465,13 @@ MemoryContextAllocAligned(MemoryContext context, void *unaligned; void *aligned; - /* wouldn't make much sense to waste that much space */ + /* + * Restrict alignto to ensure that it can fit into the "value" field of + * the redirection MemoryChunk, and that the distance back to the start of + * the unaligned chunk will fit into the space available for that. This + * isn't a limitation in practice, since it wouldn't make much sense to + * waste that much space. + */ Assert(alignto < (128 * 1024 * 1024)); /* ensure alignto is a power of 2 */ @@ -1458,10 +1508,15 @@ MemoryContextAllocAligned(MemoryContext context, alloc_size += 1; #endif - /* perform the actual allocation */ - unaligned = MemoryContextAllocExtended(context, alloc_size, flags); + /* + * Perform the actual allocation, but do not pass down MCXT_ALLOC_ZERO. + * This ensures that wasted bytes beyond the aligned chunk do not become + * DEFINED. + */ + unaligned = MemoryContextAllocExtended(context, alloc_size, + flags & ~MCXT_ALLOC_ZERO); - /* set the aligned pointer */ + /* compute the aligned pointer */ aligned = (void *) TYPEALIGN(alignto, (char *) unaligned + sizeof(MemoryChunk)); @@ -1489,12 +1544,23 @@ MemoryContextAllocAligned(MemoryContext context, set_sentinel(aligned, size); #endif - /* Mark the bytes before the redirection header as noaccess */ - VALGRIND_MAKE_MEM_NOACCESS(unaligned, - (char *) alignedchunk - (char *) unaligned); + /* + * MemoryContextAllocExtended marked the whole unaligned chunk as a + * vchunk. Undo that, instead making just the aligned chunk be a vchunk. + * This prevents Valgrind from complaining that the vchunk is possibly + * leaked, since only pointers to the aligned chunk will exist. + * + * After these calls, the aligned chunk will be marked UNDEFINED, and all + * the rest of the unaligned chunk (the redirection chunk header, the + * padding bytes before it, and any wasted trailing bytes) will be marked + * NOACCESS, which is what we want. + */ + VALGRIND_MEMPOOL_FREE(context, unaligned); + VALGRIND_MEMPOOL_ALLOC(context, aligned, size); - /* Disallow access to the redirection chunk header. */ - VALGRIND_MAKE_MEM_NOACCESS(alignedchunk, sizeof(MemoryChunk)); + /* Now zero (and make DEFINED) just the aligned chunk, if requested */ + if ((flags & MCXT_ALLOC_ZERO) != 0) + MemSetAligned(aligned, 0, size); return aligned; } @@ -1528,16 +1594,12 @@ void pfree(void *pointer) { #ifdef USE_VALGRIND - MemoryContextMethodID method = GetMemoryChunkMethodID(pointer); MemoryContext context = GetMemoryChunkContext(pointer); #endif MCXT_METHOD(pointer, free_p) (pointer); -#ifdef USE_VALGRIND - if (method != MCTX_ALIGNED_REDIRECT_ID) - VALGRIND_MEMPOOL_FREE(context, pointer); -#endif + VALGRIND_MEMPOOL_FREE(context, pointer); } /* @@ -1547,9 +1609,6 @@ pfree(void *pointer) void * repalloc(void *pointer, Size size) { -#ifdef USE_VALGRIND - MemoryContextMethodID method = GetMemoryChunkMethodID(pointer); -#endif #if defined(USE_ASSERT_CHECKING) || defined(USE_VALGRIND) MemoryContext context = GetMemoryChunkContext(pointer); #endif @@ -1572,10 +1631,7 @@ repalloc(void *pointer, Size size) */ ret = MCXT_METHOD(pointer, realloc) (pointer, size, 0); -#ifdef USE_VALGRIND - if (method != MCTX_ALIGNED_REDIRECT_ID) - VALGRIND_MEMPOOL_CHANGE(context, pointer, ret, size); -#endif + VALGRIND_MEMPOOL_CHANGE(context, pointer, ret, size); return ret; } diff --git a/src/backend/utils/mmgr/slab.c b/src/backend/utils/mmgr/slab.c index d32c0d318fb..0e35abcf5a0 100644 --- a/src/backend/utils/mmgr/slab.c +++ b/src/backend/utils/mmgr/slab.c @@ -377,6 +377,11 @@ SlabContextCreate(MemoryContext parent, * we'd leak the header if we ereport in this stretch. */ + /* See comments about Valgrind interactions in aset.c */ + VALGRIND_CREATE_MEMPOOL(slab, 0, false); + /* This vchunk covers the SlabContext only */ + VALGRIND_MEMPOOL_ALLOC(slab, slab, sizeof(SlabContext)); + /* Fill in SlabContext-specific header fields */ slab->chunkSize = (uint32) chunkSize; slab->fullChunkSize = (uint32) fullChunkSize; @@ -451,6 +456,10 @@ SlabReset(MemoryContext context) #ifdef CLOBBER_FREED_MEMORY wipe_mem(block, slab->blockSize); #endif + + /* As in aset.c, free block-header vchunks explicitly */ + VALGRIND_MEMPOOL_FREE(slab, block); + free(block); context->mem_allocated -= slab->blockSize; } @@ -467,11 +476,23 @@ SlabReset(MemoryContext context) #ifdef CLOBBER_FREED_MEMORY wipe_mem(block, slab->blockSize); #endif + + /* As in aset.c, free block-header vchunks explicitly */ + VALGRIND_MEMPOOL_FREE(slab, block); + free(block); context->mem_allocated -= slab->blockSize; } } + /* + * Instruct Valgrind to throw away all the vchunks associated with this + * context, except for the one covering the SlabContext. This gets rid of + * the vchunks for whatever user data is getting discarded by the context + * reset. + */ + VALGRIND_MEMPOOL_TRIM(slab, slab, sizeof(SlabContext)); + slab->curBlocklistIndex = 0; Assert(context->mem_allocated == 0); @@ -486,6 +507,10 @@ SlabDelete(MemoryContext context) { /* Reset to release all the SlabBlocks */ SlabReset(context); + + /* Destroy the vpool -- see notes in aset.c */ + VALGRIND_DESTROY_MEMPOOL(context); + /* And free the context header */ free(context); } @@ -567,6 +592,9 @@ SlabAllocFromNewBlock(MemoryContext context, Size size, int flags) if (unlikely(block == NULL)) return MemoryContextAllocationFailure(context, size, flags); + /* Make a vchunk covering the new block's header */ + VALGRIND_MEMPOOL_ALLOC(slab, block, Slab_BLOCKHDRSZ); + block->slab = slab; context->mem_allocated += slab->blockSize; @@ -795,6 +823,10 @@ SlabFree(void *pointer) #ifdef CLOBBER_FREED_MEMORY wipe_mem(block, slab->blockSize); #endif + + /* As in aset.c, free block-header vchunks explicitly */ + VALGRIND_MEMPOOL_FREE(slab, block); + free(block); slab->header.mem_allocated -= slab->blockSize; } |