/* ---------------------------------------------------------------- * ExecUpdate * * note: we can't run UPDATE queries with transactions * off because UPDATEs are actually INSERTs and our * scan will mistakenly loop forever, updating the tuple * it just inserted.. This should be fixed but until it * is, we don't want to get stuck in an infinite loop * which corrupts your database.. * ---------------------------------------------------------------- */ void ExecUpdate(TupleTableSlot *slot, ItemPointer tupleid, TupleTableSlot *planSlot, DestReceiver *dest, EState *estate) { HeapTuple tuple; ResultRelInfo *resultRelInfo; Relation resultRelationDesc; HTSU_Result result; ItemPointerData update_ctid; TransactionId update_xmax; /* * abort the operation if not running transactions */ if (IsBootstrapProcessingMode()) elog(ERROR, "cannot UPDATE during bootstrap"); /* * get the heap tuple out of the tuple table slot, making sure we have a * writable copy */ tuple = ExecFetchSlotHeapTuple(slot); /* * get information on the (current) result relation */ resultRelInfo = estate->es_result_relation_info; resultRelationDesc = resultRelInfo->ri_RelationDesc; /* see if this update would move the tuple to a different partition */ if (estate->es_result_partitions) { AttrNumber max_attr; Datum *values; bool *nulls; Oid targetid; Assert(estate->es_partition_state != NULL && estate->es_partition_state->accessMethods != NULL); if (!estate->es_partition_state->accessMethods->part_cxt) estate->es_partition_state->accessMethods->part_cxt = GetPerTupleExprContext(estate)->ecxt_per_tuple_memory; Assert(PointerIsValid(estate->es_result_partitions)); max_attr = estate->es_partition_state->max_partition_attr; slot_getsomeattrs(slot, max_attr); values = slot_get_values(slot); nulls = slot_get_isnull(slot); targetid = selectPartition(estate->es_result_partitions, values, nulls, slot->tts_tupleDescriptor, estate->es_partition_state->accessMethods); if (!OidIsValid(targetid)) ereport(ERROR, (errcode(ERRCODE_NO_PARTITION_FOR_PARTITIONING_KEY), errmsg("no partition for partitioning key"))); if (RelationGetRelid(resultRelationDesc) != targetid) { ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), errmsg("moving tuple from partition \"%s\" to " "partition \"%s\" not supported", get_rel_name(RelationGetRelid(resultRelationDesc)), get_rel_name(targetid)), errOmitLocation(true))); } } /* BEFORE ROW UPDATE Triggers */ if (resultRelInfo->ri_TrigDesc && resultRelInfo->ri_TrigDesc->n_before_row[TRIGGER_EVENT_UPDATE] > 0) { HeapTuple newtuple; newtuple = ExecBRUpdateTriggers(estate, resultRelInfo, tupleid, tuple, estate->es_snapshot->curcid); if (newtuple == NULL) /* "do nothing" */ return; if (newtuple != tuple) /* modified by Trigger(s) */ { /* * Put the modified tuple into a slot for convenience of routines * below. We assume the tuple was allocated in per-tuple memory * context, and therefore will go away by itself. The tuple table * slot should not try to clear it. */ TupleTableSlot *newslot = estate->es_trig_tuple_slot; if (newslot->tts_tupleDescriptor != slot->tts_tupleDescriptor) ExecSetSlotDescriptor(newslot, slot->tts_tupleDescriptor); ExecStoreGenericTuple(newtuple, newslot, false); newslot->tts_tableOid = slot->tts_tableOid; /* for constraints */ slot = newslot; tuple = newtuple; } } /* * Check the constraints of the tuple * * If we generate a new candidate tuple after EvalPlanQual testing, we * must loop back here and recheck constraints. (We don't need to redo * triggers, however. If there are any BEFORE triggers then trigger.c * will have done heap_lock_tuple to lock the correct tuple, so there's no * need to do them again.) */ lreplace:; if (resultRelationDesc->rd_att->constr) ExecConstraints(resultRelInfo, slot, estate); if (!GpPersistent_IsPersistentRelation(resultRelationDesc->rd_id)) { /* * Normal UPDATE path. */ /* * replace the heap tuple * * Note: if es_crosscheck_snapshot isn't InvalidSnapshot, we check that * the row to be updated is visible to that snapshot, and throw a can't- * serialize error if not. This is a special-case behavior needed for * referential integrity updates in serializable transactions. */ result = heap_update(resultRelationDesc, tupleid, tuple, &update_ctid, &update_xmax, estate->es_snapshot->curcid, estate->es_crosscheck_snapshot, true /* wait for commit */ ); switch (result) { case HeapTupleSelfUpdated: /* already deleted by self; nothing to do */ return; case HeapTupleMayBeUpdated: break; case HeapTupleUpdated: if (IsXactIsoLevelSerializable) ereport(ERROR, (errcode(ERRCODE_T_R_SERIALIZATION_FAILURE), errmsg("could not serialize access due to concurrent update"))); else if (!ItemPointerEquals(tupleid, &update_ctid)) { TupleTableSlot *epqslot; epqslot = EvalPlanQual(estate, resultRelInfo->ri_RangeTableIndex, &update_ctid, update_xmax, estate->es_snapshot->curcid); if (!TupIsNull(epqslot)) { *tupleid = update_ctid; slot = ExecFilterJunk(estate->es_junkFilter, epqslot); tuple = ExecFetchSlotHeapTuple(slot); goto lreplace; } } /* tuple already deleted; nothing to do */ return; default: elog(ERROR, "unrecognized heap_update status: %u", result); return; } } else { HeapTuple persistentTuple; /* * Persistent metadata path. */ persistentTuple = heap_copytuple(tuple); persistentTuple->t_self = *tupleid; frozen_heap_inplace_update(resultRelationDesc, persistentTuple); heap_freetuple(persistentTuple); } IncrReplaced(); (estate->es_processed)++; /* * Note: instead of having to update the old index tuples associated with * the heap tuple, all we do is form and insert new index tuples. This is * because UPDATEs are actually DELETEs and INSERTs, and index tuple * deletion is done later by VACUUM (see notes in ExecDelete). All we do * here is insert new index tuples. -cim 9/27/89 */ /* * insert index entries for tuple * * Note: heap_update returns the tid (location) of the new tuple in the * t_self field. */ if (resultRelInfo->ri_NumIndices > 0) ExecInsertIndexTuples(slot, &(tuple->t_self), estate, false); /* AFTER ROW UPDATE Triggers */ ExecARUpdateTriggers(estate, resultRelInfo, tupleid, tuple); }
/* * lazy_truncate_heap - try to truncate off any empty pages at the end */ static void lazy_truncate_heap(Relation onerel, LVRelStats *vacrelstats) { BlockNumber old_rel_pages = vacrelstats->rel_pages; BlockNumber new_rel_pages; PageFreeSpaceInfo *pageSpaces; int n; int i, j; PGRUsage ru0; /* * Persistent table TIDs are stored in other locations like gp_relation_node * and changeTracking logs, which continue to have references to CTID even * if PT tuple is marked deleted. This TID is used to read tuple during * crash recovery or segment resyncs. Hence need to avoid truncating * persistent tables to avoid error / crash in heap_fetch using the TID * on lazy vacuum. */ if (GpPersistent_IsPersistentRelation(RelationGetRelid(onerel))) return; pg_rusage_init(&ru0); /* * We need full exclusive lock on the relation in order to do truncation. * If we can't get it, give up rather than waiting --- we don't want to * block other backends, and we don't want to deadlock (which is quite * possible considering we already hold a lower-grade lock). */ if (!ConditionalLockRelation(onerel, AccessExclusiveLock)) return; /* * Now that we have exclusive lock, look to see if the rel has grown * whilst we were vacuuming with non-exclusive lock. If so, give up; the * newly added pages presumably contain non-deletable tuples. */ new_rel_pages = RelationGetNumberOfBlocks(onerel); if (new_rel_pages != old_rel_pages) { /* might as well use the latest news when we update pg_class stats */ vacrelstats->rel_pages = new_rel_pages; UnlockRelation(onerel, AccessExclusiveLock); return; } /* * Scan backwards from the end to verify that the end pages actually * contain no tuples. This is *necessary*, not optional, because other * backends could have added tuples to these pages whilst we were * vacuuming. */ new_rel_pages = count_nondeletable_pages(onerel, vacrelstats); if (new_rel_pages >= old_rel_pages) { /* can't do anything after all */ UnlockRelation(onerel, AccessExclusiveLock); return; } /* * Okay to truncate. */ RelationTruncate( onerel, new_rel_pages, /* markPersistentAsPhysicallyTruncated */ true); /* * Note: once we have truncated, we *must* keep the exclusive lock until * commit. The sinval message that will be sent at commit (as a result of * vac_update_relstats()) must be received by other backends, to cause * them to reset their rd_targblock values, before they can safely access * the table again. */ /* * Drop free-space info for removed blocks; these must not get entered * into the FSM! */ pageSpaces = vacrelstats->free_pages; n = vacrelstats->num_free_pages; j = 0; for (i = 0; i < n; i++) { if (pageSpaces[i].blkno < new_rel_pages) { pageSpaces[j] = pageSpaces[i]; j++; } } vacrelstats->num_free_pages = j; /* * If tot_free_pages was more than num_free_pages, we can't tell for sure * what its correct value is now, because we don't know which of the * forgotten pages are getting truncated. Conservatively set it equal to * num_free_pages. */ vacrelstats->tot_free_pages = j; /* We destroyed the heap ordering, so mark array unordered */ vacrelstats->fs_is_heap = false; /* update statistics */ vacrelstats->rel_pages = new_rel_pages; vacrelstats->pages_removed = old_rel_pages - new_rel_pages; ereport(elevel, (errmsg("\"%s\": truncated %u to %u pages", RelationGetRelationName(onerel), old_rel_pages, new_rel_pages), errdetail("%s.", pg_rusage_show(&ru0)))); }
/* ---------------------------------------------------------------- * ExecUpdate * * note: we can't run UPDATE queries with transactions * off because UPDATEs are actually INSERTs and our * scan will mistakenly loop forever, updating the tuple * it just inserted.. This should be fixed but until it * is, we don't want to get stuck in an infinite loop * which corrupts your database.. * ---------------------------------------------------------------- */ void ExecUpdate(TupleTableSlot *slot, ItemPointer tupleid, TupleTableSlot *planSlot, DestReceiver *dest, EState *estate) { void* tuple; ResultRelInfo *resultRelInfo; Relation resultRelationDesc; HTSU_Result result; ItemPointerData update_ctid; TransactionId update_xmax; AOTupleId aoTupleId = AOTUPLEID_INIT; TupleTableSlot *partslot = NULL; /* * abort the operation if not running transactions */ if (IsBootstrapProcessingMode()) elog(ERROR, "cannot UPDATE during bootstrap"); /* * get information on the (current) result relation */ resultRelInfo = estate->es_result_relation_info; resultRelationDesc = resultRelInfo->ri_RelationDesc; bool rel_is_heap = RelationIsHeap(resultRelationDesc); bool rel_is_aorows = RelationIsAoRows(resultRelationDesc); bool rel_is_aocols = RelationIsAoCols(resultRelationDesc); bool rel_is_external = RelationIsExternal(resultRelationDesc); /* * get the heap tuple out of the tuple table slot, making sure we have a * writable copy */ if (rel_is_heap) { partslot = slot; tuple = ExecFetchSlotHeapTuple(partslot); } else if (rel_is_aorows || rel_is_aocols) { /* * It is necessary to reconstruct a logically compatible tuple to * a phyiscally compatible tuple. The slot's tuple descriptor comes * from the projection target list, which doesn't indicate dropped * columns, and MemTuple cannot deal with cases without converting * the target list back into the original relation's tuple desc. */ partslot = reconstructMatchingTupleSlot(slot, resultRelInfo); /* * We directly inline toasted columns here as update with toasted columns * would create two references to the same toasted value. */ tuple = ExecFetchSlotMemTuple(partslot, true); } else if (rel_is_external) { if (estate->es_result_partitions && estate->es_result_partitions->part->parrelid != 0) { ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), errmsg("Update external partitions not supported."))); return; } else { partslot = slot; tuple = ExecFetchSlotHeapTuple(partslot); } } else { Insist(false); } /* see if this update would move the tuple to a different partition */ if (estate->es_result_partitions) checkPartitionUpdate(estate, partslot, resultRelInfo); /* BEFORE ROW UPDATE Triggers */ if (resultRelInfo->ri_TrigDesc && resultRelInfo->ri_TrigDesc->n_before_row[TRIGGER_EVENT_UPDATE] > 0) { HeapTuple newtuple; newtuple = ExecBRUpdateTriggers(estate, resultRelInfo, tupleid, tuple, estate->es_snapshot->curcid); if (newtuple == NULL) /* "do nothing" */ return; if (newtuple != tuple) /* modified by Trigger(s) */ { /* * Put the modified tuple into a slot for convenience of routines * below. We assume the tuple was allocated in per-tuple memory * context, and therefore will go away by itself. The tuple table * slot should not try to clear it. */ TupleTableSlot *newslot = estate->es_trig_tuple_slot; if (newslot->tts_tupleDescriptor != partslot->tts_tupleDescriptor) ExecSetSlotDescriptor(newslot, partslot->tts_tupleDescriptor); ExecStoreGenericTuple(newtuple, newslot, false); newslot->tts_tableOid = partslot->tts_tableOid; /* for constraints */ partslot = newslot; tuple = newtuple; } } /* * Check the constraints of the tuple * * If we generate a new candidate tuple after EvalPlanQual testing, we * must loop back here and recheck constraints. (We don't need to redo * triggers, however. If there are any BEFORE triggers then trigger.c * will have done heap_lock_tuple to lock the correct tuple, so there's no * need to do them again.) */ lreplace:; if (resultRelationDesc->rd_att->constr) ExecConstraints(resultRelInfo, partslot, estate); if (!GpPersistent_IsPersistentRelation(resultRelationDesc->rd_id)) { /* * Normal UPDATE path. */ /* * replace the heap tuple * * Note: if es_crosscheck_snapshot isn't InvalidSnapshot, we check that * the row to be updated is visible to that snapshot, and throw a can't- * serialize error if not. This is a special-case behavior needed for * referential integrity updates in serializable transactions. */ if (rel_is_heap) { result = heap_update(resultRelationDesc, tupleid, tuple, &update_ctid, &update_xmax, estate->es_snapshot->curcid, estate->es_crosscheck_snapshot, true /* wait for commit */ ); } else if (rel_is_aorows) { if (IsXactIsoLevelSerializable) { ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), errmsg("Updates on append-only tables are not supported in serializable transactions."))); } if (resultRelInfo->ri_updateDesc == NULL) { ResultRelInfoSetSegno(resultRelInfo, estate->es_result_aosegnos); resultRelInfo->ri_updateDesc = (AppendOnlyUpdateDesc) appendonly_update_init(resultRelationDesc, ActiveSnapshot, resultRelInfo->ri_aosegno); } result = appendonly_update(resultRelInfo->ri_updateDesc, tuple, (AOTupleId *) tupleid, &aoTupleId); } else if (rel_is_aocols) { if (IsXactIsoLevelSerializable) { ereport(ERROR, (errcode(ERRCODE_FEATURE_NOT_SUPPORTED), errmsg("Updates on append-only tables are not supported in serializable transactions."))); } if (resultRelInfo->ri_updateDesc == NULL) { ResultRelInfoSetSegno(resultRelInfo, estate->es_result_aosegnos); resultRelInfo->ri_updateDesc = (AppendOnlyUpdateDesc) aocs_update_init(resultRelationDesc, resultRelInfo->ri_aosegno); } result = aocs_update(resultRelInfo->ri_updateDesc, partslot, (AOTupleId *) tupleid, &aoTupleId); } else { Assert(!"We should not be here"); } switch (result) { case HeapTupleSelfUpdated: /* already deleted by self; nothing to do */ return; case HeapTupleMayBeUpdated: break; case HeapTupleUpdated: if (IsXactIsoLevelSerializable) ereport(ERROR, (errcode(ERRCODE_T_R_SERIALIZATION_FAILURE), errmsg("could not serialize access due to concurrent update"))); else if (!ItemPointerEquals(tupleid, &update_ctid)) { TupleTableSlot *epqslot; epqslot = EvalPlanQual(estate, resultRelInfo->ri_RangeTableIndex, &update_ctid, update_xmax, estate->es_snapshot->curcid); if (!TupIsNull(epqslot)) { *tupleid = update_ctid; partslot = ExecFilterJunk(estate->es_junkFilter, epqslot); tuple = ExecFetchSlotHeapTuple(partslot); goto lreplace; } } /* tuple already deleted; nothing to do */ return; default: elog(ERROR, "unrecognized heap_update status: %u", result); return; } } else { HeapTuple persistentTuple; /* * Persistent metadata path. */ persistentTuple = heap_copytuple(tuple); persistentTuple->t_self = *tupleid; frozen_heap_inplace_update(resultRelationDesc, persistentTuple); heap_freetuple(persistentTuple); } IncrReplaced(); (estate->es_processed)++; (resultRelInfo->ri_aoprocessed)++; /* * Note: instead of having to update the old index tuples associated with * the heap tuple, all we do is form and insert new index tuples. This is * because UPDATEs are actually DELETEs and INSERTs, and index tuple * deletion is done later by VACUUM (see notes in ExecDelete). All we do * here is insert new index tuples. -cim 9/27/89 */ /* * insert index entries for tuple * * Note: heap_update returns the tid (location) of the new tuple in the * t_self field. */ if (rel_is_aorows || rel_is_aocols) { if (resultRelInfo->ri_NumIndices > 0) ExecInsertIndexTuples(partslot, (ItemPointer)&aoTupleId, estate, false); } else { if (resultRelInfo->ri_NumIndices > 0) ExecInsertIndexTuples(partslot, &(((HeapTuple) tuple)->t_self), estate, false); } /* AFTER ROW UPDATE Triggers */ ExecARUpdateTriggers(estate, resultRelInfo, tupleid, tuple); }