/* --------------------------------
* ExecStoreMinimalTuple
*
* Like ExecStoreTuple, but insert a "minimal" tuple into the slot.
*
* No 'buffer' parameter since minimal tuples are never stored in relations.
* --------------------------------
*/
TupleTableSlot *
ExecStoreMinimalTuple(MemTuple mtup,
TupleTableSlot *slot,
bool shouldFree)
{
/*
* sanity checks
*/
Assert(mtup != NULL);
Assert(slot != NULL);
Assert(slot->tts_tupleDescriptor != NULL);
Assert(slot->tts_mt_bind != NULL);
/* Acctually we are storing a HeapTuple! */
if(!is_heaptuple_memtuple((HeapTuple) mtup))
return ExecStoreHeapTuple((HeapTuple) mtup, slot, InvalidBuffer, shouldFree);
/*
* Free any old physical tuple belonging to the slot.
*/
free_heaptuple_memtuple(slot);
/*
* Store the new tuple into the specified slot.
*/
/* Clear tts_flags, here isempty set to false */
if(shouldFree)
TupSetShouldFree(slot);
else
TupClearShouldFree(slot);
slot->PRIVATE_tts_memtuple = mtup;
TupClearIsEmpty(slot);
slot->PRIVATE_tts_nvalid = 0;
/*
* Drop the pin on the referenced buffer, if there is one.
*/
if (BufferIsValid(slot->tts_buffer))
ReleaseBuffer(slot->tts_buffer);
slot->tts_buffer = InvalidBuffer;
return slot;
}
/* ----------------------------------------------------------------
* IndexNext
*
* Retrieve a tuple from the IndexScan node's currentRelation
* using the index specified in the IndexScanState information.
* ----------------------------------------------------------------
*/
TupleTableSlot *
IndexNext(IndexScanState *node)
{
EState *estate;
ExprContext *econtext;
ScanDirection direction;
IndexScanDesc scandesc;
Index scanrelid;
HeapTuple tuple;
TupleTableSlot *slot;
/*
* extract necessary information from index scan node
*/
estate = node->ss.ps.state;
direction = estate->es_direction;
initScanDesc(node);
/* flip direction if this is an overall backward scan */
if (ScanDirectionIsBackward(((IndexScan *) node->ss.ps.plan)->indexorderdir))
{
if (ScanDirectionIsForward(direction))
direction = BackwardScanDirection;
else if (ScanDirectionIsBackward(direction))
direction = ForwardScanDirection;
}
scandesc = node->iss_ScanDesc;
econtext = node->ss.ps.ps_ExprContext;
slot = node->ss.ss_ScanTupleSlot;
scanrelid = ((IndexScan *) node->ss.ps.plan)->scan.scanrelid;
/*
* Check if we are evaluating PlanQual for tuple of this relation.
* Additional checking is not good, but no other way for now. We could
* introduce new nodes for this case and handle IndexScan --> NewNode
* switching in Init/ReScan plan...
*/
if (estate->es_evTuple != NULL &&
estate->es_evTuple[scanrelid - 1] != NULL)
{
if (estate->es_evTupleNull[scanrelid - 1])
{
if (!node->ss.ps.delayEagerFree)
{
ExecEagerFreeIndexScan(node);
}
return ExecClearTuple(slot);
}
ExecStoreGenericTuple(estate->es_evTuple[scanrelid - 1], slot, false);
/* Does the tuple meet the indexqual condition? */
econtext->ecxt_scantuple = slot;
ResetExprContext(econtext);
if (!ExecQual(node->indexqualorig, econtext, false))
{
if (!node->ss.ps.delayEagerFree)
{
ExecEagerFreeIndexScan(node);
}
ExecClearTuple(slot); /* would not be returned by scan */
}
/* Flag for the next call that no more tuples */
estate->es_evTupleNull[scanrelid - 1] = true;
Gpmon_M_Incr_Rows_Out(GpmonPktFromIndexScanState(node));
CheckSendPlanStateGpmonPkt(&node->ss.ps);
return slot;
}
/*
* ok, now that we have what we need, fetch the next tuple.
*/
if ((tuple = index_getnext(scandesc, direction)) != NULL)
{
/*
* Store the scanned tuple in the scan tuple slot of the scan state.
* Note: we pass 'false' because tuples returned by amgetnext are
* pointers onto disk pages and must not be pfree()'d.
*/
ExecStoreHeapTuple(tuple, /* tuple to store */
slot, /* slot to store in */
scandesc->xs_cbuf, /* buffer containing tuple */
false); /* don't pfree */
Gpmon_M_Incr_Rows_Out(GpmonPktFromIndexScanState(node));
CheckSendPlanStateGpmonPkt(&node->ss.ps);
return slot;
}
if (!node->ss.ps.delayEagerFree)
{
ExecEagerFreeIndexScan(node);
}
/*
* if we get here it means the index scan failed so we are at the end of
* the scan..
*/
return ExecClearTuple(slot);
}
/*
* Handle UPDATE message.
*
* TODO: FDW support
*/
static void
apply_handle_update(StringInfo s)
{
LogicalRepRelMapEntry *rel;
LogicalRepRelId relid;
Oid idxoid;
EState *estate;
EPQState epqstate;
LogicalRepTupleData oldtup;
LogicalRepTupleData newtup;
bool has_oldtup;
TupleTableSlot *localslot;
TupleTableSlot *remoteslot;
bool found;
MemoryContext oldctx;
ensure_transaction();
relid = logicalrep_read_update(s, &has_oldtup, &oldtup,
&newtup);
rel = logicalrep_rel_open(relid, RowExclusiveLock);
if (!should_apply_changes_for_rel(rel))
{
/*
* The relation can't become interesting in the middle of the
* transaction so it's safe to unlock it.
*/
logicalrep_rel_close(rel, RowExclusiveLock);
return;
}
/* Check if we can do the update. */
check_relation_updatable(rel);
/* Initialize the executor state. */
estate = create_estate_for_relation(rel);
remoteslot = ExecInitExtraTupleSlot(estate,
RelationGetDescr(rel->localrel));
localslot = ExecInitExtraTupleSlot(estate,
RelationGetDescr(rel->localrel));
EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
PushActiveSnapshot(GetTransactionSnapshot());
ExecOpenIndices(estate->es_result_relation_info, false);
/* Build the search tuple. */
oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
slot_store_cstrings(remoteslot, rel,
has_oldtup ? oldtup.values : newtup.values);
MemoryContextSwitchTo(oldctx);
/*
* Try to find tuple using either replica identity index, primary key or
* if needed, sequential scan.
*/
idxoid = GetRelationIdentityOrPK(rel->localrel);
Assert(OidIsValid(idxoid) ||
(rel->remoterel.replident == REPLICA_IDENTITY_FULL && has_oldtup));
if (OidIsValid(idxoid))
found = RelationFindReplTupleByIndex(rel->localrel, idxoid,
LockTupleExclusive,
remoteslot, localslot);
else
found = RelationFindReplTupleSeq(rel->localrel, LockTupleExclusive,
remoteslot, localslot);
ExecClearTuple(remoteslot);
/*
* Tuple found.
*
* Note this will fail if there are other conflicting unique indexes.
*/
if (found)
{
/* Process and store remote tuple in the slot */
oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
ExecStoreHeapTuple(localslot->tts_tuple, remoteslot, false);
slot_modify_cstrings(remoteslot, rel, newtup.values, newtup.changed);
MemoryContextSwitchTo(oldctx);
EvalPlanQualSetSlot(&epqstate, remoteslot);
/* Do the actual update. */
ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
}
else
{
/*
* The tuple to be updated could not be found.
*
* TODO what to do here, change the log level to LOG perhaps?
*/
elog(DEBUG1,
"logical replication did not find row for update "
"in replication target relation \"%s\"",
RelationGetRelationName(rel->localrel));
}
/* Cleanup. */
ExecCloseIndices(estate->es_result_relation_info);
PopActiveSnapshot();
/* Handle queued AFTER triggers. */
AfterTriggerEndQuery(estate);
EvalPlanQualEnd(&epqstate);
ExecResetTupleTable(estate->es_tupleTable, false);
FreeExecutorState(estate);
logicalrep_rel_close(rel, NoLock);
CommandCounterIncrement();
}
/* ----------------------------------------------------------------
* IndexOnlyNext
*
* Retrieve a tuple from the IndexOnlyScan node's index.
* ----------------------------------------------------------------
*/
static TupleTableSlot *
IndexOnlyNext(IndexOnlyScanState *node)
{
EState *estate;
ExprContext *econtext;
ScanDirection direction;
IndexScanDesc scandesc;
TupleTableSlot *slot;
ItemPointer tid;
/*
* extract necessary information from index scan node
*/
estate = node->ss.ps.state;
direction = estate->es_direction;
/* flip direction if this is an overall backward scan */
if (ScanDirectionIsBackward(((IndexOnlyScan *) node->ss.ps.plan)->indexorderdir))
{
if (ScanDirectionIsForward(direction))
direction = BackwardScanDirection;
else if (ScanDirectionIsBackward(direction))
direction = ForwardScanDirection;
}
scandesc = node->ioss_ScanDesc;
econtext = node->ss.ps.ps_ExprContext;
slot = node->ss.ss_ScanTupleSlot;
if (scandesc == NULL)
{
/*
* We reach here if the index only scan is not parallel, or if we're
* serially executing an index only scan that was planned to be
* parallel.
*/
scandesc = index_beginscan(node->ss.ss_currentRelation,
node->ioss_RelationDesc,
estate->es_snapshot,
node->ioss_NumScanKeys,
node->ioss_NumOrderByKeys);
node->ioss_ScanDesc = scandesc;
/* Set it up for index-only scan */
node->ioss_ScanDesc->xs_want_itup = true;
node->ioss_VMBuffer = InvalidBuffer;
/*
* If no run-time keys to calculate or they are ready, go ahead and
* pass the scankeys to the index AM.
*/
if (node->ioss_NumRuntimeKeys == 0 || node->ioss_RuntimeKeysReady)
index_rescan(scandesc,
node->ioss_ScanKeys,
node->ioss_NumScanKeys,
node->ioss_OrderByKeys,
node->ioss_NumOrderByKeys);
}
/*
* OK, now that we have what we need, fetch the next tuple.
*/
while ((tid = index_getnext_tid(scandesc, direction)) != NULL)
{
HeapTuple tuple = NULL;
CHECK_FOR_INTERRUPTS();
/*
* We can skip the heap fetch if the TID references a heap page on
* which all tuples are known visible to everybody. In any case,
* we'll use the index tuple not the heap tuple as the data source.
*
* Note on Memory Ordering Effects: visibilitymap_get_status does not
* lock the visibility map buffer, and therefore the result we read
* here could be slightly stale. However, it can't be stale enough to
* matter.
*
* We need to detect clearing a VM bit due to an insert right away,
* because the tuple is present in the index page but not visible. The
* reading of the TID by this scan (using a shared lock on the index
* buffer) is serialized with the insert of the TID into the index
* (using an exclusive lock on the index buffer). Because the VM bit
* is cleared before updating the index, and locking/unlocking of the
* index page acts as a full memory barrier, we are sure to see the
* cleared bit if we see a recently-inserted TID.
*
* Deletes do not update the index page (only VACUUM will clear out
* the TID), so the clearing of the VM bit by a delete is not
* serialized with this test below, and we may see a value that is
* significantly stale. However, we don't care about the delete right
* away, because the tuple is still visible until the deleting
* transaction commits or the statement ends (if it's our
* transaction). In either case, the lock on the VM buffer will have
* been released (acting as a write barrier) after clearing the bit.
* And for us to have a snapshot that includes the deleting
* transaction (making the tuple invisible), we must have acquired
* ProcArrayLock after that time, acting as a read barrier.
*
* It's worth going through this complexity to avoid needing to lock
* the VM buffer, which could cause significant contention.
*/
if (!VM_ALL_VISIBLE(scandesc->heapRelation,
ItemPointerGetBlockNumber(tid),
&node->ioss_VMBuffer))
{
/*
* Rats, we have to visit the heap to check visibility.
*/
InstrCountTuples2(node, 1);
tuple = index_fetch_heap(scandesc);
if (tuple == NULL)
continue; /* no visible tuple, try next index entry */
/*
* Only MVCC snapshots are supported here, so there should be no
* need to keep following the HOT chain once a visible entry has
* been found. If we did want to allow that, we'd need to keep
* more state to remember not to call index_getnext_tid next time.
*/
if (scandesc->xs_continue_hot)
elog(ERROR, "non-MVCC snapshots are not supported in index-only scans");
/*
* Note: at this point we are holding a pin on the heap page, as
* recorded in scandesc->xs_cbuf. We could release that pin now,
* but it's not clear whether it's a win to do so. The next index
* entry might require a visit to the same heap page.
*/
}
/*
* Fill the scan tuple slot with data from the index. This might be
* provided in either HeapTuple or IndexTuple format. Conceivably an
* index AM might fill both fields, in which case we prefer the heap
* format, since it's probably a bit cheaper to fill a slot from.
*/
if (scandesc->xs_hitup)
{
/*
* We don't take the trouble to verify that the provided tuple has
* exactly the slot's format, but it seems worth doing a quick
* check on the number of fields.
*/
Assert(slot->tts_tupleDescriptor->natts ==
scandesc->xs_hitupdesc->natts);
ExecStoreHeapTuple(scandesc->xs_hitup, slot, false);
}
else if (scandesc->xs_itup)
StoreIndexTuple(slot, scandesc->xs_itup, scandesc->xs_itupdesc);
else
elog(ERROR, "no data returned for index-only scan");
/*
* If the index was lossy, we have to recheck the index quals.
* (Currently, this can never happen, but we should support the case
* for possible future use, eg with GiST indexes.)
*/
if (scandesc->xs_recheck)
{
econtext->ecxt_scantuple = slot;
if (!ExecQualAndReset(node->indexqual, econtext))
{
/* Fails recheck, so drop it and loop back for another */
InstrCountFiltered2(node, 1);
continue;
}
}
/*
* We don't currently support rechecking ORDER BY distances. (In
* principle, if the index can support retrieval of the originally
* indexed value, it should be able to produce an exact distance
* calculation too. So it's not clear that adding code here for
* recheck/re-sort would be worth the trouble. But we should at least
* throw an error if someone tries it.)
*/
if (scandesc->numberOfOrderBys > 0 && scandesc->xs_recheckorderby)
ereport(ERROR,
(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
errmsg("lossy distance functions are not supported in index-only scans")));
/*
* Predicate locks for index-only scans must be acquired at the page
* level when the heap is not accessed, since tuple-level predicate
* locks need the tuple's xmin value. If we had to visit the tuple
* anyway, then we already have the tuple-level lock and can skip the
* page lock.
*/
if (tuple == NULL)
PredicateLockPage(scandesc->heapRelation,
ItemPointerGetBlockNumber(tid),
estate->es_snapshot);
return slot;
}
/*
* if we get here it means the index scan failed so we are at the end of
* the scan..
*/
return ExecClearTuple(slot);
}
TupleTableSlot *
HeapScanNext(ScanState *scanState)
{
Assert(IsA(scanState, TableScanState) ||
IsA(scanState, DynamicTableScanState));
SeqScanState *node = (SeqScanState *)scanState;
Assert(node->opaque != NULL);
HeapTuple tuple;
HeapScanDesc scandesc;
Index scanrelid;
EState *estate;
ScanDirection direction;
TupleTableSlot *slot;
Assert((node->ss.scan_state & SCAN_SCAN) != 0);
/*
* get information from the estate and scan state
*/
estate = node->ss.ps.state;
scandesc = node->opaque->ss_currentScanDesc;
scanrelid = ((SeqScan *) node->ss.ps.plan)->scanrelid;
direction = estate->es_direction;
slot = node->ss.ss_ScanTupleSlot;
/*
* Check if we are evaluating PlanQual for tuple of this relation.
* Additional checking is not good, but no other way for now. We could
* introduce new nodes for this case and handle SeqScan --> NewNode
* switching in Init/ReScan plan...
*/
if (estate->es_evTuple != NULL &&
estate->es_evTuple[scanrelid - 1] != NULL)
{
if (estate->es_evTupleNull[scanrelid - 1])
{
return ExecClearTuple(slot);
}
ExecStoreGenericTuple(estate->es_evTuple[scanrelid - 1], slot, false);
/*
* Note that unlike IndexScan, SeqScan never uses keys in
* heap_beginscan (and this is very bad) - so, here we do not check
* the keys.
*/
/* Flag for the next call that no more tuples */
estate->es_evTupleNull[scanrelid - 1] = true;
return slot;
}
/*
* get the next tuple from the access methods
*/
if (node->opaque->ss_heapTupleData.bot == node->opaque->ss_heapTupleData.top &&
!node->opaque->ss_heapTupleData.seen_EOS)
{
node->opaque->ss_heapTupleData.last = NULL;
node->opaque->ss_heapTupleData.bot = 0;
node->opaque->ss_heapTupleData.top = lengthof(node->opaque->ss_heapTupleData.item);
heap_getnextx(scandesc, direction, node->opaque->ss_heapTupleData.item,
&node->opaque->ss_heapTupleData.top,
&node->opaque->ss_heapTupleData.seen_EOS);
if (scandesc->rs_pageatatime &&
IsA(scanState, TableScanState))
{
Gpmon_M_Incr(GpmonPktFromTableScanState((TableScanState *)scanState), GPMON_TABLESCAN_PAGE);
CheckSendPlanStateGpmonPkt(&node->ss.ps);
}
}
node->opaque->ss_heapTupleData.last = NULL;
if (node->opaque->ss_heapTupleData.bot < node->opaque->ss_heapTupleData.top)
{
node->opaque->ss_heapTupleData.last =
&node->opaque->ss_heapTupleData.item[node->opaque->ss_heapTupleData.bot++];
}
tuple = node->opaque->ss_heapTupleData.last;
/*
* save the tuple and the buffer returned to us by the access methods in
* our scan tuple slot and return the slot. Note: we pass 'false' because
* tuples returned by heap_getnext() are pointers onto disk pages and were
* not created with palloc() and so should not be pfree()'d. Note also
* that ExecStoreTuple will increment the refcount of the buffer; the
* refcount will not be dropped until the tuple table slot is cleared.
*/
if (tuple)
{
ExecStoreHeapTuple(tuple,
slot,
scandesc->rs_cbuf,
false);
}
else
{
ExecClearTuple(slot);
}
return slot;
}
/*
* unique_key_recheck - trigger function to do a deferred uniqueness check.
*
* This now also does deferred exclusion-constraint checks, so the name is
* somewhat historical.
*
* This is invoked as an AFTER ROW trigger for both INSERT and UPDATE,
* for any rows recorded as potentially violating a deferrable unique
* or exclusion constraint.
*
* This may be an end-of-statement check, a commit-time check, or a
* check triggered by a SET CONSTRAINTS command.
*/
Datum
unique_key_recheck(PG_FUNCTION_ARGS)
{
TriggerData *trigdata = (TriggerData *) fcinfo->context;
const char *funcname = "unique_key_recheck";
HeapTuple new_row;
ItemPointerData tmptid;
Relation indexRel;
IndexInfo *indexInfo;
EState *estate;
ExprContext *econtext;
TupleTableSlot *slot;
Datum values[INDEX_MAX_KEYS];
bool isnull[INDEX_MAX_KEYS];
/*
* Make sure this is being called as an AFTER ROW trigger. Note:
* translatable error strings are shared with ri_triggers.c, so resist the
* temptation to fold the function name into them.
*/
if (!CALLED_AS_TRIGGER(fcinfo))
ereport(ERROR,
(errcode(ERRCODE_E_R_I_E_TRIGGER_PROTOCOL_VIOLATED),
errmsg("function \"%s\" was not called by trigger manager",
funcname)));
if (!TRIGGER_FIRED_AFTER(trigdata->tg_event) ||
!TRIGGER_FIRED_FOR_ROW(trigdata->tg_event))
ereport(ERROR,
(errcode(ERRCODE_E_R_I_E_TRIGGER_PROTOCOL_VIOLATED),
errmsg("function \"%s\" must be fired AFTER ROW",
funcname)));
/*
* Get the new data that was inserted/updated.
*/
if (TRIGGER_FIRED_BY_INSERT(trigdata->tg_event))
new_row = trigdata->tg_trigtuple;
else if (TRIGGER_FIRED_BY_UPDATE(trigdata->tg_event))
new_row = trigdata->tg_newtuple;
else
{
ereport(ERROR,
(errcode(ERRCODE_E_R_I_E_TRIGGER_PROTOCOL_VIOLATED),
errmsg("function \"%s\" must be fired for INSERT or UPDATE",
funcname)));
new_row = NULL; /* keep compiler quiet */
}
/*
* If the new_row is now dead (ie, inserted and then deleted within our
* transaction), we can skip the check. However, we have to be careful,
* because this trigger gets queued only in response to index insertions;
* which means it does not get queued for HOT updates. The row we are
* called for might now be dead, but have a live HOT child, in which case
* we still need to make the check --- effectively, we're applying the
* check against the live child row, although we can use the values from
* this row since by definition all columns of interest to us are the
* same.
*
* This might look like just an optimization, because the index AM will
* make this identical test before throwing an error. But it's actually
* needed for correctness, because the index AM will also throw an error
* if it doesn't find the index entry for the row. If the row's dead then
* it's possible the index entry has also been marked dead, and even
* removed.
*/
tmptid = new_row->t_self;
if (!heap_hot_search(&tmptid, trigdata->tg_relation, SnapshotSelf, NULL))
{
/*
* All rows in the HOT chain are dead, so skip the check.
*/
return PointerGetDatum(NULL);
}
/*
* Open the index, acquiring a RowExclusiveLock, just as if we were going
* to update it. (This protects against possible changes of the index
* schema, not against concurrent updates.)
*/
indexRel = index_open(trigdata->tg_trigger->tgconstrindid,
RowExclusiveLock);
indexInfo = BuildIndexInfo(indexRel);
/*
* The heap tuple must be put into a slot for FormIndexDatum.
*/
slot = MakeSingleTupleTableSlot(RelationGetDescr(trigdata->tg_relation));
ExecStoreHeapTuple(new_row, slot, InvalidBuffer, false);
/*
* Typically the index won't have expressions, but if it does we need an
* EState to evaluate them. We need it for exclusion constraints too,
* even if they are just on simple columns.
*/
if (indexInfo->ii_Expressions != NIL ||
indexInfo->ii_ExclusionOps != NULL)
{
estate = CreateExecutorState();
econtext = GetPerTupleExprContext(estate);
econtext->ecxt_scantuple = slot;
}
else
estate = NULL;
/*
* Form the index values and isnull flags for the index entry that we need
* to check.
*
* Note: if the index uses functions that are not as immutable as they are
* supposed to be, this could produce an index tuple different from the
* original. The index AM can catch such errors by verifying that it
* finds a matching index entry with the tuple's TID. For exclusion
* constraints we check this in check_exclusion_constraint().
*/
FormIndexDatum(indexInfo, slot, estate, values, isnull);
/*
* Now do the appropriate check.
*/
if (indexInfo->ii_ExclusionOps == NULL)
{
/*
* Note: this is not a real insert; it is a check that the index entry
* that has already been inserted is unique. Passing t_self is
* correct even if t_self is now dead, because that is the TID the
* index will know about.
*/
index_insert(indexRel, values, isnull, &(new_row->t_self),
trigdata->tg_relation, UNIQUE_CHECK_EXISTING);
}
else
{
/*
* For exclusion constraints we just do the normal check, but now it's
* okay to throw error. In the HOT-update case, we must use the live
* HOT child's TID here, else check_exclusion_constraint will think
* the child is a conflict.
*/
check_exclusion_constraint(trigdata->tg_relation, indexRel, indexInfo,
&tmptid, values, isnull,
estate, false, false);
}
/*
* If that worked, then this index entry is unique or non-excluded, and we
* are done.
*/
if (estate != NULL)
FreeExecutorState(estate);
ExecDropSingleTupleTableSlot(slot);
index_close(indexRel, RowExclusiveLock);
return PointerGetDatum(NULL);
}
/*
* ExecSetOp for non-hashed case
*/
static TupleTableSlot *
setop_retrieve_direct(SetOpState *setopstate)
{
PlanState *outerPlan;
SetOpStatePerGroup pergroup;
TupleTableSlot *outerslot;
TupleTableSlot *resultTupleSlot;
ExprContext *econtext = setopstate->ps.ps_ExprContext;
/*
* get state info from node
*/
outerPlan = outerPlanState(setopstate);
pergroup = (SetOpStatePerGroup) setopstate->pergroup;
resultTupleSlot = setopstate->ps.ps_ResultTupleSlot;
/*
* We loop retrieving groups until we find one we should return
*/
while (!setopstate->setop_done)
{
/*
* If we don't already have the first tuple of the new group, fetch it
* from the outer plan.
*/
if (setopstate->grp_firstTuple == NULL)
{
outerslot = ExecProcNode(outerPlan);
if (!TupIsNull(outerslot))
{
/* Make a copy of the first input tuple */
setopstate->grp_firstTuple = ExecCopySlotHeapTuple(outerslot);
}
else
{
/* outer plan produced no tuples at all */
setopstate->setop_done = true;
return NULL;
}
}
/*
* Store the copied first input tuple in the tuple table slot reserved
* for it. The tuple will be deleted when it is cleared from the
* slot.
*/
ExecStoreHeapTuple(setopstate->grp_firstTuple,
resultTupleSlot,
true);
setopstate->grp_firstTuple = NULL; /* don't keep two pointers */
/* Initialize working state for a new input tuple group */
initialize_counts(pergroup);
/* Count the first input tuple */
advance_counts(pergroup,
fetch_tuple_flag(setopstate, resultTupleSlot));
/*
* Scan the outer plan until we exhaust it or cross a group boundary.
*/
for (;;)
{
outerslot = ExecProcNode(outerPlan);
if (TupIsNull(outerslot))
{
/* no more outer-plan tuples available */
setopstate->setop_done = true;
break;
}
/*
* Check whether we've crossed a group boundary.
*/
econtext->ecxt_outertuple = resultTupleSlot;
econtext->ecxt_innertuple = outerslot;
if (!ExecQualAndReset(setopstate->eqfunction, econtext))
{
/*
* Save the first input tuple of the next group.
*/
setopstate->grp_firstTuple = ExecCopySlotHeapTuple(outerslot);
break;
}
/* Still in same group, so count this tuple */
advance_counts(pergroup,
fetch_tuple_flag(setopstate, outerslot));
}
/*
* Done scanning input tuple group. See if we should emit any copies
* of result tuple, and if so return the first copy.
*/
set_output_count(setopstate, pergroup);
if (setopstate->numOutput > 0)
{
setopstate->numOutput--;
return resultTupleSlot;
}
}
/* No more groups */
ExecClearTuple(resultTupleSlot);
return NULL;
}
/*
* Search the relation 'rel' for tuple using the sequential scan.
*
* If a matching tuple is found, lock it with lockmode, fill the slot with its
* contents, and return true. Return false otherwise.
*
* Note that this stops on the first matching tuple.
*
* This can obviously be quite slow on tables that have more than few rows.
*/
bool
RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
TupleTableSlot *searchslot, TupleTableSlot *outslot)
{
HeapTuple scantuple;
HeapScanDesc scan;
SnapshotData snap;
TransactionId xwait;
bool found;
TupleDesc desc = RelationGetDescr(rel);
Assert(equalTupleDescs(desc, outslot->tts_tupleDescriptor));
/* Start a heap scan. */
InitDirtySnapshot(snap);
scan = heap_beginscan(rel, &snap, 0, NULL);
retry:
found = false;
heap_rescan(scan, NULL);
/* Try to find the tuple */
while ((scantuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
{
if (!tuple_equals_slot(desc, scantuple, searchslot))
continue;
found = true;
ExecStoreHeapTuple(scantuple, outslot, false);
ExecMaterializeSlot(outslot);
xwait = TransactionIdIsValid(snap.xmin) ?
snap.xmin : snap.xmax;
/*
* If the tuple is locked, wait for locking transaction to finish and
* retry.
*/
if (TransactionIdIsValid(xwait))
{
XactLockTableWait(xwait, NULL, NULL, XLTW_None);
goto retry;
}
}
/* Found tuple, try to lock it in the lockmode. */
if (found)
{
Buffer buf;
HeapUpdateFailureData hufd;
HTSU_Result res;
HeapTupleData locktup;
HeapTupleTableSlot *hslot = (HeapTupleTableSlot *)outslot;
/* Only a heap tuple has item pointers. */
Assert(TTS_IS_HEAPTUPLE(outslot) || TTS_IS_BUFFERTUPLE(outslot));
ItemPointerCopy(&hslot->tuple->t_self, &locktup.t_self);
PushActiveSnapshot(GetLatestSnapshot());
res = heap_lock_tuple(rel, &locktup, GetCurrentCommandId(false),
lockmode,
LockWaitBlock,
false /* don't follow updates */ ,
&buf, &hufd);
/* the tuple slot already has the buffer pinned */
ReleaseBuffer(buf);
PopActiveSnapshot();
switch (res)
{
case HeapTupleMayBeUpdated:
break;
case HeapTupleUpdated:
/* XXX: Improve handling here */
if (ItemPointerIndicatesMovedPartitions(&hufd.ctid))
ereport(LOG,
(errcode(ERRCODE_T_R_SERIALIZATION_FAILURE),
errmsg("tuple to be locked was already moved to another partition due to concurrent update, retrying")));
else
ereport(LOG,
(errcode(ERRCODE_T_R_SERIALIZATION_FAILURE),
errmsg("concurrent update, retrying")));
goto retry;
case HeapTupleInvisible:
elog(ERROR, "attempted to lock invisible tuple");
break;
default:
elog(ERROR, "unexpected heap_lock_tuple status: %u", res);
break;
}
}
heap_endscan(scan);
return found;
}
/*
* Search the relation 'rel' for tuple using the index.
*
* If a matching tuple is found, lock it with lockmode, fill the slot with its
* contents, and return true. Return false otherwise.
*/
bool
RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
LockTupleMode lockmode,
TupleTableSlot *searchslot,
TupleTableSlot *outslot)
{
HeapTuple scantuple;
ScanKeyData skey[INDEX_MAX_KEYS];
IndexScanDesc scan;
SnapshotData snap;
TransactionId xwait;
Relation idxrel;
bool found;
/* Open the index. */
idxrel = index_open(idxoid, RowExclusiveLock);
/* Start an index scan. */
InitDirtySnapshot(snap);
scan = index_beginscan(rel, idxrel, &snap,
IndexRelationGetNumberOfKeyAttributes(idxrel),
0);
/* Build scan key. */
build_replindex_scan_key(skey, rel, idxrel, searchslot);
retry:
found = false;
index_rescan(scan, skey, IndexRelationGetNumberOfKeyAttributes(idxrel), NULL, 0);
/* Try to find the tuple */
if ((scantuple = index_getnext(scan, ForwardScanDirection)) != NULL)
{
found = true;
ExecStoreHeapTuple(scantuple, outslot, false);
ExecMaterializeSlot(outslot);
xwait = TransactionIdIsValid(snap.xmin) ?
snap.xmin : snap.xmax;
/*
* If the tuple is locked, wait for locking transaction to finish and
* retry.
*/
if (TransactionIdIsValid(xwait))
{
XactLockTableWait(xwait, NULL, NULL, XLTW_None);
goto retry;
}
}
/* Found tuple, try to lock it in the lockmode. */
if (found)
{
Buffer buf;
HeapUpdateFailureData hufd;
HTSU_Result res;
HeapTupleData locktup;
HeapTupleTableSlot *hslot = (HeapTupleTableSlot *)outslot;
/* Only a heap tuple has item pointers. */
Assert(TTS_IS_HEAPTUPLE(outslot) || TTS_IS_BUFFERTUPLE(outslot));
ItemPointerCopy(&hslot->tuple->t_self, &locktup.t_self);
PushActiveSnapshot(GetLatestSnapshot());
res = heap_lock_tuple(rel, &locktup, GetCurrentCommandId(false),
lockmode,
LockWaitBlock,
false /* don't follow updates */ ,
&buf, &hufd);
/* the tuple slot already has the buffer pinned */
ReleaseBuffer(buf);
PopActiveSnapshot();
switch (res)
{
case HeapTupleMayBeUpdated:
break;
case HeapTupleUpdated:
/* XXX: Improve handling here */
if (ItemPointerIndicatesMovedPartitions(&hufd.ctid))
ereport(LOG,
(errcode(ERRCODE_T_R_SERIALIZATION_FAILURE),
errmsg("tuple to be locked was already moved to another partition due to concurrent update, retrying")));
else
ereport(LOG,
(errcode(ERRCODE_T_R_SERIALIZATION_FAILURE),
errmsg("concurrent update, retrying")));
goto retry;
case HeapTupleInvisible:
elog(ERROR, "attempted to lock invisible tuple");
break;
default:
elog(ERROR, "unexpected heap_lock_tuple status: %u", res);
break;
}
}
index_endscan(scan);
/* Don't release lock until commit. */
index_close(idxrel, NoLock);
return found;
}
/*
* CatalogIndexInsert - insert index entries for one catalog tuple
*
* This should be called for each inserted or updated catalog tuple.
*
* This is effectively a cut-down version of ExecInsertIndexTuples.
*/
static void
CatalogIndexInsert(CatalogIndexState indstate, HeapTuple heapTuple)
{
int i;
int numIndexes;
RelationPtr relationDescs;
Relation heapRelation;
TupleTableSlot *slot;
IndexInfo **indexInfoArray;
Datum values[INDEX_MAX_KEYS];
bool isnull[INDEX_MAX_KEYS];
/* HOT update does not require index inserts */
if (HeapTupleIsHeapOnly(heapTuple))
return;
/*
* Get information from the state structure. Fall out if nothing to do.
*/
numIndexes = indstate->ri_NumIndices;
if (numIndexes == 0)
return;
relationDescs = indstate->ri_IndexRelationDescs;
indexInfoArray = indstate->ri_IndexRelationInfo;
heapRelation = indstate->ri_RelationDesc;
/* Need a slot to hold the tuple being examined */
slot = MakeSingleTupleTableSlot(RelationGetDescr(heapRelation),
&TTSOpsHeapTuple);
ExecStoreHeapTuple(heapTuple, slot, false);
/*
* for each index, form and insert the index tuple
*/
for (i = 0; i < numIndexes; i++)
{
IndexInfo *indexInfo;
indexInfo = indexInfoArray[i];
/* If the index is marked as read-only, ignore it */
if (!indexInfo->ii_ReadyForInserts)
continue;
/*
* Expressional and partial indexes on system catalogs are not
* supported, nor exclusion constraints, nor deferred uniqueness
*/
Assert(indexInfo->ii_Expressions == NIL);
Assert(indexInfo->ii_Predicate == NIL);
Assert(indexInfo->ii_ExclusionOps == NULL);
Assert(relationDescs[i]->rd_index->indimmediate);
Assert(indexInfo->ii_NumIndexKeyAttrs != 0);
/*
* FormIndexDatum fills in its values and isnull parameters with the
* appropriate values for the column(s) of the index.
*/
FormIndexDatum(indexInfo,
slot,
NULL, /* no expression eval to do */
values,
isnull);
/*
* The index AM does the rest.
*/
index_insert(relationDescs[i], /* index relation */
values, /* array of index Datums */
isnull, /* is-null flags */
&(heapTuple->t_self), /* tid of heap tuple */
heapRelation,
relationDescs[i]->rd_index->indisunique ?
UNIQUE_CHECK_YES : UNIQUE_CHECK_NO,
indexInfo);
}
ExecDropSingleTupleTableSlot(slot);
}
/*
* Store the next tuple for a given reader into the appropriate slot.
*
* Returns true if successful, false if not (either reader is exhausted,
* or we didn't want to wait for a tuple). Sets done flag if reader
* is found to be exhausted.
*/
static bool
gather_merge_readnext(GatherMergeState *gm_state, int reader, bool nowait)
{
GMReaderTupleBuffer *tuple_buffer;
HeapTuple tup;
/*
* If we're being asked to generate a tuple from the leader, then we just
* call ExecProcNode as normal to produce one.
*/
if (reader == 0)
{
if (gm_state->need_to_scan_locally)
{
PlanState *outerPlan = outerPlanState(gm_state);
TupleTableSlot *outerTupleSlot;
EState *estate = gm_state->ps.state;
/* Install our DSA area while executing the plan. */
estate->es_query_dsa = gm_state->pei ? gm_state->pei->area : NULL;
outerTupleSlot = ExecProcNode(outerPlan);
estate->es_query_dsa = NULL;
if (!TupIsNull(outerTupleSlot))
{
gm_state->gm_slots[0] = outerTupleSlot;
return true;
}
/* need_to_scan_locally serves as "done" flag for leader */
gm_state->need_to_scan_locally = false;
}
return false;
}
/* Otherwise, check the state of the relevant tuple buffer. */
tuple_buffer = &gm_state->gm_tuple_buffers[reader - 1];
if (tuple_buffer->nTuples > tuple_buffer->readCounter)
{
/* Return any tuple previously read that is still buffered. */
tup = tuple_buffer->tuple[tuple_buffer->readCounter++];
}
else if (tuple_buffer->done)
{
/* Reader is known to be exhausted. */
return false;
}
else
{
/* Read and buffer next tuple. */
tup = gm_readnext_tuple(gm_state,
reader,
nowait,
&tuple_buffer->done);
if (!HeapTupleIsValid(tup))
return false;
/*
* Attempt to read more tuples in nowait mode and store them in the
* pending-tuple array for the reader.
*/
load_tuple_array(gm_state, reader);
}
Assert(HeapTupleIsValid(tup));
/* Build the TupleTableSlot for the given tuple */
ExecStoreHeapTuple(tup, /* tuple to store */
gm_state->gm_slots[reader], /* slot in which to store
* the tuple */
true); /* pfree tuple when done with it */
return true;
}
