On Sat, 2021-06-05 at 09:47 -0700, Zhihong Yu wrote:
> On Fri, Jun 4, 2021 at 4:14 PM Jacob Champion <pchamp...@vmware.com> wrote:
> > Agreed. I'm going to double-check with Deep that the new calls
> > to table_tuple_fetch_row_version() should be projecting the full row,
> > then post an updated patch some time next week.
(The discussions over the fallout of the inheritance_planner fallout
are still going, but in the meantime here's an updated v4 that builds
and passes `make check`.)
> + return
> relation->rd_tableam->scan_begin_with_column_projection(relation, snapshot,
> 0, NULL,
> + parallel_scan, flags, proj);
>
> scan_begin_with_column_projection() adds a parameter to scan_begin().
> Can scan_begin() be enhanced with this projection parameter ?
> Otherwise in the future we may have
> scan_begin_with_column_projection_with_x_y ...
Maybe; I agree that would match the current "extension" APIs a little
better. I'll let Deep and/or Ashwin chime in on why this design was
chosen.
> + /* Make sure the the new slot is not dependent on the original tuple */
>
> Double 'the' in the comment. More than one place with duplicate 'the'
> in the patch.
Fixed.
> +typedef struct neededColumnContext
> +{
> + Bitmapset **mask;
> + int n;
>
> Should field n be named ncol ? 'n' seems too general.
Agreed; changed to ncol.
> + * TODO: Remove this hack!! This should be done once at the start of
> the tid scan.
>
> Would the above be addressed in the next patch ?
I have not had time to get to this in v4, sorry.
> Toward the end of extract_scan_columns():
>
> + bms_free(rte->scanCols);
> + rte->scanCols = bms_make_singleton(0);
> + break;
>
> Should 'goto outer;' be in place of 'break;' (since rte->scanCols has
> been assigned for whole-row) ?
Agreed and fixed. Thank you!
--Jacob
From 7ac00847f17e2a0448ae06370598db0f034fcc7c Mon Sep 17 00:00:00 2001
From: Jacob Champion <pchamp...@vmware.com>
Date: Tue, 2 Mar 2021 08:59:45 -0800
Subject: [PATCH v4] tableam: accept column projection list
TODO: check the additional bms_make_singleton(0) calls in
src/backend/executor/nodeModifyTable.c
Co-authored-by: Soumyadeep Chakraborty <soumyadeep2...@gmail.com>
Co-authored-by: Melanie Plageman <melanieplage...@gmail.com>
Co-authored-by: Ashwin Agrawal <aagra...@pivotal.io>
Co-authored-by: Jacob Champion <pchamp...@vmware.com>
---
src/backend/access/heap/heapam_handler.c | 5 +-
src/backend/access/nbtree/nbtsort.c | 3 +-
src/backend/access/table/tableam.c | 5 +-
src/backend/commands/trigger.c | 19 +++-
src/backend/executor/execMain.c | 3 +-
src/backend/executor/execPartition.c | 2 +
src/backend/executor/execReplication.c | 6 +-
src/backend/executor/execScan.c | 108 +++++++++++++++++++++
src/backend/executor/nodeIndexscan.c | 10 ++
src/backend/executor/nodeLockRows.c | 7 +-
src/backend/executor/nodeModifyTable.c | 47 +++++++--
src/backend/executor/nodeSeqscan.c | 43 ++++++++-
src/backend/executor/nodeTidscan.c | 11 ++-
src/backend/nodes/copyfuncs.c | 2 +
src/backend/nodes/equalfuncs.c | 2 +
src/backend/nodes/outfuncs.c | 2 +
src/backend/nodes/readfuncs.c | 2 +
src/backend/optimizer/path/allpaths.c | 85 +++++++++++++++-
src/backend/optimizer/prep/preptlist.c | 13 ++-
src/backend/optimizer/util/inherit.c | 37 ++++++-
src/backend/parser/analyze.c | 40 ++++++--
src/backend/parser/parse_relation.c | 18 ++++
src/backend/partitioning/partbounds.c | 14 ++-
src/backend/rewrite/rewriteHandler.c | 8 ++
src/include/access/tableam.h | 118 +++++++++++++++++++++--
src/include/executor/executor.h | 6 ++
src/include/nodes/execnodes.h | 1 +
src/include/nodes/parsenodes.h | 14 +++
src/include/utils/rel.h | 14 +++
29 files changed, 598 insertions(+), 47 deletions(-)
diff --git a/src/backend/access/heap/heapam_handler.c b/src/backend/access/heap/heapam_handler.c
index e2cd79ec54..e440ae4a69 100644
--- a/src/backend/access/heap/heapam_handler.c
+++ b/src/backend/access/heap/heapam_handler.c
@@ -182,7 +182,8 @@ static bool
heapam_fetch_row_version(Relation relation,
ItemPointer tid,
Snapshot snapshot,
- TupleTableSlot *slot)
+ TupleTableSlot *slot,
+ Bitmapset *project_cols)
{
BufferHeapTupleTableSlot *bslot = (BufferHeapTupleTableSlot *) slot;
Buffer buffer;
@@ -350,7 +351,7 @@ static TM_Result
heapam_tuple_lock(Relation relation, ItemPointer tid, Snapshot snapshot,
TupleTableSlot *slot, CommandId cid, LockTupleMode mode,
LockWaitPolicy wait_policy, uint8 flags,
- TM_FailureData *tmfd)
+ TM_FailureData *tmfd, Bitmapset *project_cols)
{
BufferHeapTupleTableSlot *bslot = (BufferHeapTupleTableSlot *) slot;
TM_Result result;
diff --git a/src/backend/access/nbtree/nbtsort.c b/src/backend/access/nbtree/nbtsort.c
index 2c4d7f6e25..57462eb9c7 100644
--- a/src/backend/access/nbtree/nbtsort.c
+++ b/src/backend/access/nbtree/nbtsort.c
@@ -1966,7 +1966,8 @@ _bt_parallel_scan_and_sort(BTSpool *btspool, BTSpool *btspool2,
indexInfo = BuildIndexInfo(btspool->index);
indexInfo->ii_Concurrent = btshared->isconcurrent;
scan = table_beginscan_parallel(btspool->heap,
- ParallelTableScanFromBTShared(btshared));
+ ParallelTableScanFromBTShared(btshared),
+ NULL);
reltuples = table_index_build_scan(btspool->heap, btspool->index, indexInfo,
true, progress, _bt_build_callback,
(void *) &buildstate, scan);
diff --git a/src/backend/access/table/tableam.c b/src/backend/access/table/tableam.c
index 5ea5bdd810..9bda57247a 100644
--- a/src/backend/access/table/tableam.c
+++ b/src/backend/access/table/tableam.c
@@ -172,7 +172,7 @@ table_parallelscan_initialize(Relation rel, ParallelTableScanDesc pscan,
}
TableScanDesc
-table_beginscan_parallel(Relation relation, ParallelTableScanDesc parallel_scan)
+table_beginscan_parallel(Relation relation, ParallelTableScanDesc parallel_scan, Bitmapset *proj)
{
Snapshot snapshot;
uint32 flags = SO_TYPE_SEQSCAN |
@@ -194,6 +194,9 @@ table_beginscan_parallel(Relation relation, ParallelTableScanDesc parallel_scan)
snapshot = SnapshotAny;
}
+ if (proj)
+ return relation->rd_tableam->scan_begin_with_column_projection(relation, snapshot, 0, NULL,
+ parallel_scan, flags, proj);
return relation->rd_tableam->scan_begin(relation, snapshot, 0, NULL,
parallel_scan, flags);
}
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 07c73f39de..5dd2a12cb5 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -2532,6 +2532,10 @@ ExecBRDeleteTriggers(EState *estate, EPQState *epqstate,
if (newtuple != trigtuple)
heap_freetuple(newtuple);
}
+
+ /* Make sure the new slot is not dependent on the original tuple */
+ ExecMaterializeSlot(slot);
+
if (should_free)
heap_freetuple(trigtuple);
@@ -2816,6 +2820,10 @@ ExecBRUpdateTriggers(EState *estate, EPQState *epqstate,
newtuple = NULL;
}
}
+
+ /* Make sure the new slot is not dependent on the original tuple */
+ ExecMaterializeSlot(newslot);
+
if (should_free_trig)
heap_freetuple(trigtuple);
@@ -3023,7 +3031,8 @@ GetTupleForTrigger(EState *estate,
estate->es_output_cid,
lockmode, LockWaitBlock,
lockflags,
- &tmfd);
+ &tmfd,
+ bms_make_singleton(0));
switch (test)
{
@@ -3098,7 +3107,7 @@ GetTupleForTrigger(EState *estate,
* suffices.
*/
if (!table_tuple_fetch_row_version(relation, tid, SnapshotAny,
- oldslot))
+ oldslot, bms_make_singleton(0)))
elog(ERROR, "failed to fetch tuple for trigger");
}
@@ -3962,7 +3971,8 @@ AfterTriggerExecute(EState *estate,
if (!table_tuple_fetch_row_version(rel, &(event->ate_ctid1),
SnapshotAny,
- LocTriggerData.tg_trigslot))
+ LocTriggerData.tg_trigslot,
+ bms_make_singleton(0)))
elog(ERROR, "failed to fetch tuple1 for AFTER trigger");
LocTriggerData.tg_trigtuple =
ExecFetchSlotHeapTuple(LocTriggerData.tg_trigslot, false, &should_free_trig);
@@ -3981,7 +3991,8 @@ AfterTriggerExecute(EState *estate,
if (!table_tuple_fetch_row_version(rel, &(event->ate_ctid2),
SnapshotAny,
- LocTriggerData.tg_newslot))
+ LocTriggerData.tg_newslot,
+ bms_make_singleton(0)))
elog(ERROR, "failed to fetch tuple2 for AFTER trigger");
LocTriggerData.tg_newtuple =
ExecFetchSlotHeapTuple(LocTriggerData.tg_newslot, false, &should_free_new);
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index b3ce4bae53..0197fbd0da 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -2575,7 +2575,8 @@ EvalPlanQualFetchRowMark(EPQState *epqstate, Index rti, TupleTableSlot *slot)
/* ordinary table, fetch the tuple */
if (!table_tuple_fetch_row_version(erm->relation,
(ItemPointer) DatumGetPointer(datum),
- SnapshotAny, slot))
+ SnapshotAny, slot,
+ bms_make_singleton(0)))
elog(ERROR, "failed to fetch tuple for EvalPlanQual recheck");
return true;
}
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 606c920b06..b82650ef32 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -837,6 +837,8 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
ExecInitQual((List *) clause, &mtstate->ps);
}
}
+
+ PopulateNeededColumnsForOnConflictUpdate(leaf_part_rri);
}
}
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 1e285e0349..3b4a65fa11 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -180,7 +180,8 @@ retry:
lockmode,
LockWaitBlock,
0 /* don't follow updates */ ,
- &tmfd);
+ &tmfd,
+ bms_make_singleton(0));
PopActiveSnapshot();
@@ -357,7 +358,8 @@ retry:
lockmode,
LockWaitBlock,
0 /* don't follow updates */ ,
- &tmfd);
+ &tmfd,
+ bms_make_singleton(0));
PopActiveSnapshot();
diff --git a/src/backend/executor/execScan.c b/src/backend/executor/execScan.c
index 69ab34573e..e25c999b4d 100644
--- a/src/backend/executor/execScan.c
+++ b/src/backend/executor/execScan.c
@@ -20,7 +20,9 @@
#include "executor/executor.h"
#include "miscadmin.h"
+#include "nodes/nodeFuncs.h"
#include "utils/memutils.h"
+#include "utils/rel.h"
@@ -340,3 +342,109 @@ ExecScanReScan(ScanState *node)
}
}
}
+
+typedef struct neededColumnContext
+{
+ Bitmapset **mask;
+ int ncol;
+} neededColumnContext;
+
+static bool
+neededColumnContextWalker(Node *node, neededColumnContext *c)
+{
+ if (node == NULL || contains_whole_row_col(*c->mask))
+ return false;
+
+ if (IsA(node, Var))
+ {
+ Var *var = (Var *)node;
+
+ if (var->varattno > 0)
+ {
+ Assert(var->varattno <= c->ncol);
+ *(c->mask) = bms_add_member(*(c->mask), var->varattno);
+ }
+ else if(var->varattno == 0) {
+ bms_free(*(c->mask));
+ *(c->mask) = bms_make_singleton(0);
+ }
+
+ return false;
+ }
+ return expression_tree_walker(node, neededColumnContextWalker, (void * )c);
+}
+
+/*
+ * n specifies the number of allowed entries in mask: we use
+ * it for bounds-checking in the walker above.
+ */
+void
+PopulateNeededColumnsForNode(Node *expr, int ncol, Bitmapset **scanCols)
+{
+ neededColumnContext c;
+
+ c.mask = scanCols;
+ c.ncol = ncol;
+
+ neededColumnContextWalker(expr, &c);
+}
+
+Bitmapset *
+PopulateNeededColumnsForScan(ScanState *scanstate, int ncol)
+{
+ Bitmapset *result = NULL;
+ Plan *plan = scanstate->ps.plan;
+
+ PopulateNeededColumnsForNode((Node *) plan->targetlist, ncol, &result);
+ PopulateNeededColumnsForNode((Node *) plan->qual, ncol, &result);
+
+ if (IsA(plan, IndexScan))
+ {
+ PopulateNeededColumnsForNode((Node *) ((IndexScan *) plan)->indexqualorig, ncol, &result);
+ PopulateNeededColumnsForNode((Node *) ((IndexScan *) plan)->indexorderbyorig, ncol, &result);
+ }
+ else if (IsA(plan, BitmapHeapScan))
+ PopulateNeededColumnsForNode((Node *) ((BitmapHeapScan *) plan)->bitmapqualorig, ncol, &result);
+
+ return result;
+}
+
+Bitmapset *
+PopulateNeededColumnsForEPQ(EPQState *epqstate, int ncol)
+{
+ Bitmapset *epqCols = NULL;
+ Assert(epqstate && epqstate->plan);
+ PopulateNeededColumnsForNode((Node *) epqstate->plan->qual,
+ ncol,
+ &epqCols);
+ return epqCols;
+}
+
+void
+PopulateNeededColumnsForOnConflictUpdate(ResultRelInfo *resultRelInfo)
+{
+ ExprState *onConflictSetWhere = resultRelInfo->ri_onConflict->oc_WhereClause;
+ ProjectionInfo *oc_ProjInfo = resultRelInfo->ri_onConflict->oc_ProjInfo;
+ Relation relation = resultRelInfo->ri_RelationDesc;
+ Bitmapset *proj_cols = NULL;
+ ListCell *lc;
+
+ if (onConflictSetWhere && onConflictSetWhere->expr)
+ PopulateNeededColumnsForNode((Node *) onConflictSetWhere->expr,
+ RelationGetDescr(relation)->natts,
+ &proj_cols);
+
+ if (oc_ProjInfo)
+ PopulateNeededColumnsForNode((Node *) oc_ProjInfo->pi_state.expr,
+ RelationGetDescr(relation)->natts,
+ &proj_cols);
+
+ foreach(lc, resultRelInfo->ri_WithCheckOptionExprs)
+ {
+ ExprState *wcoExpr = (ExprState *) lfirst(lc);
+ PopulateNeededColumnsForNode((Node *) wcoExpr->expr,
+ RelationGetDescr(relation)->natts,
+ &proj_cols);
+ }
+ resultRelInfo->ri_onConflict->proj_cols = proj_cols;
+}
diff --git a/src/backend/executor/nodeIndexscan.c b/src/backend/executor/nodeIndexscan.c
index 2fffb1b437..4923bb258c 100644
--- a/src/backend/executor/nodeIndexscan.c
+++ b/src/backend/executor/nodeIndexscan.c
@@ -115,6 +115,16 @@ IndexNext(IndexScanState *node)
node->iss_NumScanKeys,
node->iss_NumOrderByKeys);
+ if (table_scans_leverage_column_projection(node->ss.ss_currentRelation))
+ {
+ Bitmapset *proj = NULL;
+ Scan *planNode = (Scan *)node->ss.ps.plan;
+ int rti = planNode->scanrelid;
+ RangeTblEntry *rte = list_nth(estate->es_plannedstmt->rtable, rti - 1);
+ proj = rte->scanCols;
+ table_index_fetch_set_column_projection(scandesc->xs_heapfetch, proj);
+ }
+
node->iss_ScanDesc = scandesc;
/*
diff --git a/src/backend/executor/nodeLockRows.c b/src/backend/executor/nodeLockRows.c
index b2e5c30079..bbff96a85f 100644
--- a/src/backend/executor/nodeLockRows.c
+++ b/src/backend/executor/nodeLockRows.c
@@ -80,6 +80,7 @@ lnext:
int lockflags = 0;
TM_Result test;
TupleTableSlot *markSlot;
+ Bitmapset *epqCols = NULL;
/* clear any leftover test tuple for this rel */
markSlot = EvalPlanQualSlot(&node->lr_epqstate, erm->relation, erm->rti);
@@ -179,11 +180,15 @@ lnext:
if (!IsolationUsesXactSnapshot())
lockflags |= TUPLE_LOCK_FLAG_FIND_LAST_VERSION;
+ epqCols = PopulateNeededColumnsForEPQ(&node->lr_epqstate,
+ RelationGetDescr(erm->relation)->natts);
+
test = table_tuple_lock(erm->relation, &tid, estate->es_snapshot,
markSlot, estate->es_output_cid,
lockmode, erm->waitPolicy,
lockflags,
- &tmfd);
+ &tmfd,
+ epqCols);
switch (test)
{
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 379b056310..c9fec21449 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -244,7 +244,7 @@ ExecCheckTIDVisible(EState *estate,
if (!IsolationUsesXactSnapshot())
return;
- if (!table_tuple_fetch_row_version(rel, tid, SnapshotAny, tempSlot))
+ if (!table_tuple_fetch_row_version(rel, tid, SnapshotAny, tempSlot, NULL))
elog(ERROR, "failed to fetch conflicting tuple for ON CONFLICT");
ExecCheckTupleVisible(estate, rel, tempSlot);
ExecClearTuple(tempSlot);
@@ -1202,6 +1202,7 @@ ldelete:;
{
TupleTableSlot *inputslot;
TupleTableSlot *epqslot;
+ Bitmapset *epqCols = NULL;
if (IsolationUsesXactSnapshot())
ereport(ERROR,
@@ -1216,12 +1217,15 @@ ldelete:;
inputslot = EvalPlanQualSlot(epqstate, resultRelationDesc,
resultRelInfo->ri_RangeTableIndex);
+ epqCols = PopulateNeededColumnsForEPQ(epqstate,
+ RelationGetDescr(resultRelationDesc)->natts);
+
result = table_tuple_lock(resultRelationDesc, tupleid,
estate->es_snapshot,
inputslot, estate->es_output_cid,
LockTupleExclusive, LockWaitBlock,
TUPLE_LOCK_FLAG_FIND_LAST_VERSION,
- &tmfd);
+ &tmfd, epqCols);
switch (result)
{
@@ -1374,8 +1378,23 @@ ldelete:;
}
else
{
+ RangeTblEntry *resultrte = exec_rt_fetch(resultRelInfo->ri_RangeTableIndex, estate);
+ Bitmapset *project_cols = resultrte->returningCols;
+ /*
+ * XXX returningCols should never be empty if we have a RETURNING
+ * clause. Right now, if we have a view, we fail to populate the
+ * returningCols of its base table's RTE.
+ * If we encounter such a situation now, for correctness, ensure
+ * that we fetch all the columns.
+ */
+ if(bms_is_empty(resultrte->returningCols))
+ {
+ bms_free(resultrte->returningCols);
+ project_cols = bms_make_singleton(0);
+ }
if (!table_tuple_fetch_row_version(resultRelationDesc, tupleid,
- SnapshotAny, slot))
+ SnapshotAny, slot,
+ project_cols))
elog(ERROR, "failed to fetch deleted tuple for DELETE RETURNING");
}
}
@@ -1523,7 +1542,8 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
if (!table_tuple_fetch_row_version(resultRelInfo->ri_RelationDesc,
tupleid,
SnapshotAny,
- oldSlot))
+ oldSlot,
+ bms_make_singleton(0) /* TODO */))
elog(ERROR, "failed to fetch tuple being updated");
*retry_slot = ExecGetUpdateNewTuple(resultRelInfo, epqslot,
oldSlot);
@@ -1822,6 +1842,7 @@ lreplace:;
TupleTableSlot *inputslot;
TupleTableSlot *epqslot;
TupleTableSlot *oldSlot;
+ Bitmapset *epqCols = NULL;
if (IsolationUsesXactSnapshot())
ereport(ERROR,
@@ -1835,12 +1856,14 @@ lreplace:;
inputslot = EvalPlanQualSlot(epqstate, resultRelationDesc,
resultRelInfo->ri_RangeTableIndex);
+ epqCols = PopulateNeededColumnsForEPQ(epqstate,
+ RelationGetDescr(resultRelationDesc)->natts);
result = table_tuple_lock(resultRelationDesc, tupleid,
estate->es_snapshot,
inputslot, estate->es_output_cid,
lockmode, LockWaitBlock,
TUPLE_LOCK_FLAG_FIND_LAST_VERSION,
- &tmfd);
+ &tmfd, epqCols);
switch (result)
{
@@ -1864,7 +1887,8 @@ lreplace:;
if (!table_tuple_fetch_row_version(resultRelationDesc,
tupleid,
SnapshotAny,
- oldSlot))
+ oldSlot,
+ bms_make_singleton(0) /* TODO */))
elog(ERROR, "failed to fetch tuple being updated");
slot = ExecGetUpdateNewTuple(resultRelInfo,
epqslot, oldSlot);
@@ -1981,6 +2005,8 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
Relation relation = resultRelInfo->ri_RelationDesc;
ExprState *onConflictSetWhere = resultRelInfo->ri_onConflict->oc_WhereClause;
TupleTableSlot *existing = resultRelInfo->ri_onConflict->oc_Existing;
+ ProjectionInfo *oc_ProjInfo = resultRelInfo->ri_onConflict->oc_ProjInfo;
+ Bitmapset *proj_cols = resultRelInfo->ri_onConflict->proj_cols;
TM_FailureData tmfd;
LockTupleMode lockmode;
TM_Result test;
@@ -2001,7 +2027,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
estate->es_snapshot,
existing, estate->es_output_cid,
lockmode, LockWaitBlock, 0,
- &tmfd);
+ &tmfd, proj_cols);
switch (test)
{
case TM_Ok:
@@ -2149,7 +2175,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
}
/* Project the new tuple version */
- ExecProject(resultRelInfo->ri_onConflict->oc_ProjInfo);
+ ExecProject(oc_ProjInfo);
/*
* Note that it is possible that the target tuple has been modified in
@@ -2571,7 +2597,8 @@ ExecModifyTable(PlanState *pstate)
Assert(tupleid != NULL);
if (!table_tuple_fetch_row_version(relation, tupleid,
SnapshotAny,
- oldSlot))
+ oldSlot,
+ bms_make_singleton(0) /* TODO */))
elog(ERROR, "failed to fetch tuple being updated");
}
slot = ExecGetUpdateNewTuple(resultRelInfo, planSlot,
@@ -3033,6 +3060,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
&mtstate->ps);
onconfl->oc_WhereClause = qualexpr;
}
+
+ PopulateNeededColumnsForOnConflictUpdate(resultRelInfo);
}
/*
diff --git a/src/backend/executor/nodeSeqscan.c b/src/backend/executor/nodeSeqscan.c
index 066f9ae37e..2535d1c7f4 100644
--- a/src/backend/executor/nodeSeqscan.c
+++ b/src/backend/executor/nodeSeqscan.c
@@ -31,6 +31,7 @@
#include "access/tableam.h"
#include "executor/execdebug.h"
#include "executor/nodeSeqscan.h"
+#include "nodes/nodeFuncs.h"
#include "utils/rel.h"
static TupleTableSlot *SeqNext(SeqScanState *node);
@@ -68,9 +69,22 @@ SeqNext(SeqScanState *node)
* We reach here if the scan is not parallel, or if we're serially
* executing a scan that was planned to be parallel.
*/
- scandesc = table_beginscan(node->ss.ss_currentRelation,
- estate->es_snapshot,
- 0, NULL);
+ if (table_scans_leverage_column_projection(node->ss.ss_currentRelation))
+ {
+ Scan *planNode = (Scan *)node->ss.ps.plan;
+ int rti = planNode->scanrelid;
+ RangeTblEntry *rte = list_nth(estate->es_plannedstmt->rtable, rti - 1);
+ scandesc = table_beginscan_with_column_projection(node->ss.ss_currentRelation,
+ estate->es_snapshot,
+ 0, NULL,
+ rte->scanCols);
+ }
+ else
+ {
+ scandesc = table_beginscan(node->ss.ss_currentRelation,
+ estate->es_snapshot,
+ 0, NULL);
+ }
node->ss.ss_currentScanDesc = scandesc;
}
@@ -270,14 +284,22 @@ ExecSeqScanInitializeDSM(SeqScanState *node,
{
EState *estate = node->ss.ps.state;
ParallelTableScanDesc pscan;
+ Bitmapset *proj = NULL;
pscan = shm_toc_allocate(pcxt->toc, node->pscan_len);
+
+ if (table_scans_leverage_column_projection(node->ss.ss_currentRelation))
+ {
+ proj = PopulateNeededColumnsForScan(&node->ss,
+ node->ss.ss_currentRelation->rd_att->natts);
+ }
+
table_parallelscan_initialize(node->ss.ss_currentRelation,
pscan,
estate->es_snapshot);
shm_toc_insert(pcxt->toc, node->ss.ps.plan->plan_node_id, pscan);
node->ss.ss_currentScanDesc =
- table_beginscan_parallel(node->ss.ss_currentRelation, pscan);
+ table_beginscan_parallel(node->ss.ss_currentRelation, pscan, proj);
}
/* ----------------------------------------------------------------
@@ -307,8 +329,19 @@ ExecSeqScanInitializeWorker(SeqScanState *node,
ParallelWorkerContext *pwcxt)
{
ParallelTableScanDesc pscan;
+ Bitmapset *proj = NULL;
+
+ /*
+ * FIXME: this is duplicate work with ExecSeqScanInitializeDSM. In future
+ * plan will have the we have projection list, then this overhead will not exist.
+ */
+ if (table_scans_leverage_column_projection(node->ss.ss_currentRelation))
+ {
+ proj = PopulateNeededColumnsForScan(&node->ss,
+ node->ss.ss_currentRelation->rd_att->natts);
+ }
pscan = shm_toc_lookup(pwcxt->toc, node->ss.ps.plan->plan_node_id, false);
node->ss.ss_currentScanDesc =
- table_beginscan_parallel(node->ss.ss_currentRelation, pscan);
+ table_beginscan_parallel(node->ss.ss_currentRelation, pscan, proj);
}
diff --git a/src/backend/executor/nodeTidscan.c b/src/backend/executor/nodeTidscan.c
index 48c3737da2..ffb00c5f03 100644
--- a/src/backend/executor/nodeTidscan.c
+++ b/src/backend/executor/nodeTidscan.c
@@ -366,6 +366,7 @@ TidNext(TidScanState *node)
while (node->tss_TidPtr >= 0 && node->tss_TidPtr < numTids)
{
ItemPointerData tid = tidList[node->tss_TidPtr];
+ Bitmapset *project_cols = NULL;
/*
* For WHERE CURRENT OF, the tuple retrieved from the cursor might
@@ -375,7 +376,15 @@ TidNext(TidScanState *node)
if (node->tss_isCurrentOf)
table_tuple_get_latest_tid(scan, &tid);
- if (table_tuple_fetch_row_version(heapRelation, &tid, snapshot, slot))
+ /*
+ * TODO: Remove this hack!! This should be done once at the start of the tid scan.
+ * Ideally we should probably set the list of projection cols in the
+ * generic scan desc, perhaps in TableScanDesc.
+ */
+ project_cols = PopulateNeededColumnsForScan((ScanState *) node,
+ RelationGetDescr(heapRelation)->natts);
+
+ if (table_tuple_fetch_row_version(heapRelation, &tid, snapshot, slot, project_cols))
return slot;
/* Bad TID or failed snapshot qual; try next */
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 90770a89b0..7dfdce02bf 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -2495,8 +2495,10 @@ _copyRangeTblEntry(const RangeTblEntry *from)
COPY_SCALAR_FIELD(checkAsUser);
COPY_BITMAPSET_FIELD(selectedCols);
COPY_BITMAPSET_FIELD(insertedCols);
+ COPY_BITMAPSET_FIELD(returningCols);
COPY_BITMAPSET_FIELD(updatedCols);
COPY_BITMAPSET_FIELD(extraUpdatedCols);
+ COPY_BITMAPSET_FIELD(scanCols);
COPY_NODE_FIELD(securityQuals);
return newnode;
diff --git a/src/backend/nodes/equalfuncs.c b/src/backend/nodes/equalfuncs.c
index ce76d093dd..dedfbd9a9b 100644
--- a/src/backend/nodes/equalfuncs.c
+++ b/src/backend/nodes/equalfuncs.c
@@ -2750,8 +2750,10 @@ _equalRangeTblEntry(const RangeTblEntry *a, const RangeTblEntry *b)
COMPARE_SCALAR_FIELD(checkAsUser);
COMPARE_BITMAPSET_FIELD(selectedCols);
COMPARE_BITMAPSET_FIELD(insertedCols);
+ COMPARE_BITMAPSET_FIELD(returningCols);
COMPARE_BITMAPSET_FIELD(updatedCols);
COMPARE_BITMAPSET_FIELD(extraUpdatedCols);
+ COMPARE_BITMAPSET_FIELD(scanCols);
COMPARE_NODE_FIELD(securityQuals);
return true;
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index 8da8b14f0e..2928f6e5d0 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -3296,8 +3296,10 @@ _outRangeTblEntry(StringInfo str, const RangeTblEntry *node)
WRITE_OID_FIELD(checkAsUser);
WRITE_BITMAPSET_FIELD(selectedCols);
WRITE_BITMAPSET_FIELD(insertedCols);
+ WRITE_BITMAPSET_FIELD(returningCols);
WRITE_BITMAPSET_FIELD(updatedCols);
WRITE_BITMAPSET_FIELD(extraUpdatedCols);
+ WRITE_BITMAPSET_FIELD(scanCols);
WRITE_NODE_FIELD(securityQuals);
}
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 3772ea07df..e7b8625713 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1508,8 +1508,10 @@ _readRangeTblEntry(void)
READ_OID_FIELD(checkAsUser);
READ_BITMAPSET_FIELD(selectedCols);
READ_BITMAPSET_FIELD(insertedCols);
+ READ_BITMAPSET_FIELD(returningCols);
READ_BITMAPSET_FIELD(updatedCols);
READ_BITMAPSET_FIELD(extraUpdatedCols);
+ READ_BITMAPSET_FIELD(scanCols);
READ_NODE_FIELD(securityQuals);
READ_DONE();
diff --git a/src/backend/optimizer/path/allpaths.c b/src/backend/optimizer/path/allpaths.c
index 353454b183..caf113fda8 100644
--- a/src/backend/optimizer/path/allpaths.c
+++ b/src/backend/optimizer/path/allpaths.c
@@ -27,6 +27,7 @@
#include "miscadmin.h"
#include "nodes/makefuncs.h"
#include "nodes/nodeFuncs.h"
+#include "utils/rel.h"
#ifdef OPTIMIZER_DEBUG
#include "nodes/print.h"
#endif
@@ -142,7 +143,7 @@ static void subquery_push_qual(Query *subquery,
static void recurse_push_qual(Node *setOp, Query *topquery,
RangeTblEntry *rte, Index rti, Node *qual);
static void remove_unused_subquery_outputs(Query *subquery, RelOptInfo *rel);
-
+static void extract_scan_columns(PlannerInfo *root);
/*
* make_one_rel
@@ -185,6 +186,8 @@ make_one_rel(PlannerInfo *root, List *joinlist)
*/
set_base_rel_sizes(root);
+ extract_scan_columns(root);
+
/*
* We should now have size estimates for every actual table involved in
* the query, and we also know which if any have been deleted from the
@@ -235,6 +238,86 @@ make_one_rel(PlannerInfo *root, List *joinlist)
return rel;
}
+static void
+extract_scan_columns(PlannerInfo *root)
+{
+ for (int i = 1; i < root->simple_rel_array_size; i++)
+ {
+ ListCell *lc;
+ RangeTblEntry *rte = root->simple_rte_array[i];
+ RelOptInfo *rel = root->simple_rel_array[i];
+ if (rte == NULL)
+ continue;
+ if (rel == NULL)
+ continue;
+ if (IS_DUMMY_REL(rel))
+ continue;
+ rte->scanCols = NULL;
+ foreach(lc, rel->reltarget->exprs)
+ {
+ Node *node;
+ List *vars;
+ ListCell *lc1;
+ node = lfirst(lc);
+ /*
+ * TODO: suggest a default for vars_only to make maintenance less burdensome
+ */
+ vars = pull_var_clause(node,
+ PVC_RECURSE_AGGREGATES |
+ PVC_RECURSE_WINDOWFUNCS |
+ PVC_RECURSE_PLACEHOLDERS);
+ foreach(lc1, vars)
+ {
+ Var *var = lfirst(lc1);
+ if (var->varno == i)
+ {
+ if (var->varattno > 0)
+ rte->scanCols = bms_add_member(rte->scanCols, var->varattno);
+ else if (var->varattno == 0)
+ {
+ /*
+ * If there is a whole-row var, we have to fetch the whole row.
+ */
+ bms_free(rte->scanCols);
+ rte->scanCols = bms_make_singleton(0);
+ goto outer;
+ }
+ }
+ }
+ }
+ foreach(lc, rel->baserestrictinfo)
+ {
+ RestrictInfo *rinfo = (RestrictInfo *) lfirst(lc);
+ List *vars = pull_var_clause((Node *)rinfo->clause,
+ PVC_RECURSE_AGGREGATES |
+ PVC_RECURSE_WINDOWFUNCS |
+ PVC_RECURSE_PLACEHOLDERS);
+ ListCell *lc1;
+ if (contains_whole_row_col(rte->scanCols))
+ break;
+ foreach(lc1, vars)
+ {
+ Var *var = lfirst(lc1);
+ if (var->varno == i)
+ {
+ if (var->varattno > 0)
+ rte->scanCols = bms_add_member(rte->scanCols, var->varattno);
+ else if (var->varattno == 0)
+ {
+ /*
+ * If there is a whole-row var, we have to fetch the whole row.
+ */
+ bms_free(rte->scanCols);
+ rte->scanCols = bms_make_singleton(0);
+ goto outer;
+ }
+ }
+ }
+ }
+ outer:;
+ }
+}
+
/*
* set_base_rel_consider_startup
* Set the consider_[param_]startup flags for each base-relation entry.
diff --git a/src/backend/optimizer/prep/preptlist.c b/src/backend/optimizer/prep/preptlist.c
index e9434580d6..ca1490a951 100644
--- a/src/backend/optimizer/prep/preptlist.c
+++ b/src/backend/optimizer/prep/preptlist.c
@@ -205,8 +205,19 @@ preprocess_targetlist(PlannerInfo *root)
* to make these Vars available for the RETURNING calculation. Vars that
* belong to the result rel don't need to be added, because they will be
* made to refer to the actual heap tuple.
+ *
+ * XXX: Avoid adding cols from the returningList to avoid overestimation
+ * of scanCols from RelOptInfo->reltarget exprs. This is done to avoid
+ * additional cols from the RETURNING clause making its way into scanCols
+ * for queries such as:
+ * delete from base_tbl using other_tbl t where base_tbl.col1 = t.col1 returning *;
+ * where base_tbl is the root table of an inheritance hierarchy
+ * TODO: Delete the result_relation guard below if and when
+ * inheritance_planner() is refactored to not fake a round of planning
+ * pretending we have a SELECT query (which causes result_relation to be 0
+ * in the first place)
*/
- if (parse->returningList && list_length(parse->rtable) > 1)
+ if (result_relation && parse->returningList && list_length(parse->rtable) > 1)
{
List *vars;
ListCell *l;
diff --git a/src/backend/optimizer/util/inherit.c b/src/backend/optimizer/util/inherit.c
index 992ef87b9d..eea59f758a 100644
--- a/src/backend/optimizer/util/inherit.c
+++ b/src/backend/optimizer/util/inherit.c
@@ -50,7 +50,9 @@ static Bitmapset *translate_col_privs(const Bitmapset *parent_privs,
static void expand_appendrel_subquery(PlannerInfo *root, RelOptInfo *rel,
RangeTblEntry *rte, Index rti);
-
+static Bitmapset *
+translate_parent_cols(Bitmapset *parent_cols, List *translated_vars,
+ Relation parent_rel);
/*
* expand_inherited_rtentry
* Expand a rangetable entry that has the "inh" bit set.
@@ -539,6 +541,13 @@ expand_single_inheritance_child(PlannerInfo *root, RangeTblEntry *parentrte,
childrte->alias = childrte->eref = makeAlias(parentrte->eref->aliasname,
child_colnames);
+ if (childOID != parentOID)
+ childrte->returningCols =
+ translate_parent_cols(parentrte->returningCols,
+ appinfo->translated_vars, parentrel);
+ else
+ childrte->returningCols = bms_copy(parentrte->returningCols);
+
/*
* Translate the column permissions bitmaps to the child's attnums (we
* have to build the translated_vars list before we can do this). But if
@@ -648,6 +657,32 @@ expand_single_inheritance_child(PlannerInfo *root, RangeTblEntry *parentrte,
}
}
+/*
+ * We need to translate the list of ordinal attnos from a parent table's
+ * RangeTblEntry to the ordinal attribute numbers for the child's entry.
+ */
+
+static Bitmapset *
+translate_parent_cols(Bitmapset *parent_cols, List *translated_vars,
+ Relation parent_rel)
+{
+ int col = -1;
+ Bitmapset *result = NULL;
+ /*
+ * Enumerate the set of parent columns for translation if there is a whole
+ * row var
+ */
+ if(contains_whole_row_col(parent_cols))
+ parent_cols = get_ordinal_attnos(parent_rel);
+ while ((col = bms_next_member(parent_cols, col)) >= 0)
+ {
+ Var *var = (Var *) list_nth(translated_vars, col - 1);
+ if (var)
+ result = bms_add_member(result, var->varattno);
+ }
+ return result;
+}
+
/*
* translate_col_privs
* Translate a bitmapset representing per-column privileges from the
diff --git a/src/backend/parser/analyze.c b/src/backend/parser/analyze.c
index 9cede29d6a..4d045e44ae 100644
--- a/src/backend/parser/analyze.c
+++ b/src/backend/parser/analyze.c
@@ -73,7 +73,7 @@ static void determineRecursiveColTypes(ParseState *pstate,
Node *larg, List *nrtargetlist);
static Query *transformReturnStmt(ParseState *pstate, ReturnStmt *stmt);
static Query *transformUpdateStmt(ParseState *pstate, UpdateStmt *stmt);
-static List *transformReturningList(ParseState *pstate, List *returningList);
+static void transformReturningList(ParseState *pstate, Query *qry, List *returningList);
static List *transformUpdateTargetList(ParseState *pstate,
List *targetList);
static Query *transformPLAssignStmt(ParseState *pstate,
@@ -469,7 +469,7 @@ transformDeleteStmt(ParseState *pstate, DeleteStmt *stmt)
qual = transformWhereClause(pstate, stmt->whereClause,
EXPR_KIND_WHERE, "WHERE");
- qry->returningList = transformReturningList(pstate, stmt->returningList);
+ transformReturningList(pstate, qry, stmt->returningList);
/* done building the range table and jointree */
qry->rtable = pstate->p_rtable;
@@ -888,8 +888,7 @@ transformInsertStmt(ParseState *pstate, InsertStmt *stmt)
/* Process RETURNING, if any. */
if (stmt->returningList)
- qry->returningList = transformReturningList(pstate,
- stmt->returningList);
+ transformReturningList(pstate, qry, stmt->returningList);
/* done building the range table and jointree */
qry->rtable = pstate->p_rtable;
@@ -2341,7 +2340,7 @@ transformUpdateStmt(ParseState *pstate, UpdateStmt *stmt)
qual = transformWhereClause(pstate, stmt->whereClause,
EXPR_KIND_WHERE, "WHERE");
- qry->returningList = transformReturningList(pstate, stmt->returningList);
+ transformReturningList(pstate, qry, stmt->returningList);
/*
* Now we are done with SELECT-like processing, and can get on with
@@ -2436,14 +2435,16 @@ transformUpdateTargetList(ParseState *pstate, List *origTlist)
* transformReturningList -
* handle a RETURNING clause in INSERT/UPDATE/DELETE
*/
-static List *
-transformReturningList(ParseState *pstate, List *returningList)
+static void
+transformReturningList(ParseState *pstate, Query *qry, List *returningList)
{
List *rlist;
int save_next_resno;
+ List *vars;
+ ListCell *l;
if (returningList == NIL)
- return NIL; /* nothing to do */
+ return;
/*
* We need to assign resnos starting at one in the RETURNING list. Save
@@ -2456,6 +2457,27 @@ transformReturningList(ParseState *pstate, List *returningList)
/* transform RETURNING identically to a SELECT targetlist */
rlist = transformTargetList(pstate, returningList, EXPR_KIND_RETURNING);
+ vars = pull_var_clause((Node *) rlist,
+ PVC_RECURSE_AGGREGATES |
+ PVC_RECURSE_WINDOWFUNCS |
+ PVC_INCLUDE_PLACEHOLDERS);
+ foreach (l, vars)
+ {
+ Var *var = (Var *) lfirst(l);
+ RangeTblEntry *rte = (RangeTblEntry *) list_nth(pstate->p_rtable, var->varno - 1);
+ if (var->varattno > 0)
+ rte->returningCols = bms_add_member(rte->returningCols, var->varattno);
+ else if (var->varattno == 0)
+ {
+ /*
+ * If there is a whole-row var, we have to fetch the whole row.
+ */
+ bms_free(rte->returningCols);
+ rte->returningCols = bms_make_singleton(0);
+ break;
+ }
+ }
+
/*
* Complain if the nonempty tlist expanded to nothing (which is possible
* if it contains only a star-expansion of a zero-column table). If we
@@ -2479,7 +2501,7 @@ transformReturningList(ParseState *pstate, List *returningList)
/* restore state */
pstate->p_next_resno = save_next_resno;
- return rlist;
+ qry->returningList = rlist;
}
diff --git a/src/backend/parser/parse_relation.c b/src/backend/parser/parse_relation.c
index 7465919044..68723562ee 100644
--- a/src/backend/parser/parse_relation.c
+++ b/src/backend/parser/parse_relation.c
@@ -1456,7 +1456,9 @@ addRangeTableEntry(ParseState *pstate,
rte->selectedCols = NULL;
rte->insertedCols = NULL;
rte->updatedCols = NULL;
+ rte->returningCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
/*
* Add completed RTE to pstate's range table list, so that we know its
@@ -1543,7 +1545,9 @@ addRangeTableEntryForRelation(ParseState *pstate,
rte->checkAsUser = InvalidOid; /* not set-uid by default, either */
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
+ rte->scanCols = NULL;
rte->extraUpdatedCols = NULL;
/*
@@ -1640,8 +1644,10 @@ addRangeTableEntryForSubquery(ParseState *pstate,
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
/*
* Add completed RTE to pstate's range table list, so that we know its
@@ -1946,8 +1952,10 @@ addRangeTableEntryForFunction(ParseState *pstate,
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
/*
* Add completed RTE to pstate's range table list, so that we know its
@@ -2017,8 +2025,10 @@ addRangeTableEntryForTableFunc(ParseState *pstate,
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
/*
* Add completed RTE to pstate's range table list, so that we know its
@@ -2104,8 +2114,10 @@ addRangeTableEntryForValues(ParseState *pstate,
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
/*
* Add completed RTE to pstate's range table list, so that we know its
@@ -2195,8 +2207,10 @@ addRangeTableEntryForJoin(ParseState *pstate,
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
/*
* Add completed RTE to pstate's range table list, so that we know its
@@ -2345,8 +2359,10 @@ addRangeTableEntryForCTE(ParseState *pstate,
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
/*
* Add completed RTE to pstate's range table list, so that we know its
@@ -2470,6 +2486,8 @@ addRangeTableEntryForENR(ParseState *pstate,
rte->requiredPerms = 0;
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
+ rte->returningCols = NULL;
+ rte->scanCols = NULL;
/*
* Add completed RTE to pstate's range table list, so that we know its
diff --git a/src/backend/partitioning/partbounds.c b/src/backend/partitioning/partbounds.c
index 7925fcce3b..a379458c46 100644
--- a/src/backend/partitioning/partbounds.c
+++ b/src/backend/partitioning/partbounds.c
@@ -3200,6 +3200,7 @@ check_default_partition_contents(Relation parent, Relation default_rel,
TableScanDesc scan;
MemoryContext oldCxt;
TupleTableSlot *tupslot;
+ Bitmapset *proj = NULL;
/* Lock already taken above. */
if (part_relid != RelationGetRelid(default_rel))
@@ -3264,7 +3265,15 @@ check_default_partition_contents(Relation parent, Relation default_rel,
econtext = GetPerTupleExprContext(estate);
snapshot = RegisterSnapshot(GetLatestSnapshot());
tupslot = table_slot_create(part_rel, &estate->es_tupleTable);
- scan = table_beginscan(part_rel, snapshot, 0, NULL);
+ if (table_scans_leverage_column_projection(part_rel))
+ {
+ PopulateNeededColumnsForNode((Node*)partqualstate->expr, tupslot->tts_tupleDescriptor->natts, &proj);
+ scan = table_beginscan_with_column_projection(part_rel, snapshot, 0, NULL, proj);
+ }
+ else
+ {
+ scan = table_beginscan(part_rel, snapshot, 0, NULL);
+ }
/*
* Switch to per-tuple memory context and reset it for each tuple
@@ -3295,6 +3304,9 @@ check_default_partition_contents(Relation parent, Relation default_rel,
if (RelationGetRelid(default_rel) != RelationGetRelid(part_rel))
table_close(part_rel, NoLock); /* keep the lock until commit */
+
+ if (proj)
+ pfree(proj);
}
}
diff --git a/src/backend/rewrite/rewriteHandler.c b/src/backend/rewrite/rewriteHandler.c
index 88a9e95e33..0e59f37e4b 100644
--- a/src/backend/rewrite/rewriteHandler.c
+++ b/src/backend/rewrite/rewriteHandler.c
@@ -1731,8 +1731,10 @@ ApplyRetrieveRule(Query *parsetree,
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
/*
* For the most part, Vars referencing the view should remain as
@@ -1832,15 +1834,19 @@ ApplyRetrieveRule(Query *parsetree,
subrte->checkAsUser = rte->checkAsUser;
subrte->selectedCols = rte->selectedCols;
subrte->insertedCols = rte->insertedCols;
+ subrte->returningCols = rte->returningCols;
subrte->updatedCols = rte->updatedCols;
subrte->extraUpdatedCols = rte->extraUpdatedCols;
+ subrte->scanCols = rte->scanCols;
rte->requiredPerms = 0; /* no permission check on subquery itself */
rte->checkAsUser = InvalidOid;
rte->selectedCols = NULL;
rte->insertedCols = NULL;
+ rte->returningCols = NULL;
rte->updatedCols = NULL;
rte->extraUpdatedCols = NULL;
+ rte->scanCols = NULL;
return parsetree;
}
@@ -3191,6 +3197,7 @@ rewriteTargetView(Query *parsetree, Relation view)
* base_rte instead of copying it.
*/
new_rte = base_rte;
+ new_rte->returningCols = bms_copy(view_rte->returningCols);
new_rte->rellockmode = RowExclusiveLock;
parsetree->rtable = lappend(parsetree->rtable, new_rte);
@@ -3543,6 +3550,7 @@ rewriteTargetView(Query *parsetree, Relation view)
}
}
+ new_rte->returningCols = bms_copy(view_rte->returningCols);
table_close(base_rel, NoLock);
return parsetree;
diff --git a/src/include/access/tableam.h b/src/include/access/tableam.h
index 9f1e4a1ac9..f8a86b5bd7 100644
--- a/src/include/access/tableam.h
+++ b/src/include/access/tableam.h
@@ -265,6 +265,7 @@ typedef struct TableAmRoutine
{
/* this must be set to T_TableAmRoutine */
NodeTag type;
+ bool scans_leverage_column_projection;
/* ------------------------------------------------------------------------
@@ -305,6 +306,30 @@ typedef struct TableAmRoutine
ParallelTableScanDesc pscan,
uint32 flags);
+ /*
+ * Variant of scan_begin() with a column projection bitmap that lists the
+ * ordinal attribute numbers to be fetched during the scan.
+ *
+ * If project_columns is an empty bitmap, none of the data columns are to be
+ * fetched.
+ *
+ * If project_columns is a singleton bitmap with a whole-row reference (0),
+ * all of the data columns are to be fetched.
+ *
+ * Please note: project_cols only deals with non system columns (attnum >= 0)
+ *
+ * Please note: Due to the limitations of the slot_get***() APIs, the
+ * scan_getnextslot() tableAM call must return a TupleTableSlot that is densely
+ * populated (missing cols indicated with isnull = true upto the largest
+ * attno in the projection list)
+ */
+ TableScanDesc (*scan_begin_with_column_projection)(Relation relation,
+ Snapshot snapshot,
+ int nkeys, struct ScanKeyData *key,
+ ParallelTableScanDesc parallel_scan,
+ uint32 flags,
+ Bitmapset *project_columns);
+
/*
* Release resources and deallocate scan. If TableScanDesc.temp_snap,
* TableScanDesc.rs_snapshot needs to be unregistered.
@@ -408,6 +433,26 @@ typedef struct TableAmRoutine
*/
void (*index_fetch_end) (struct IndexFetchTableData *data);
+ /*
+ * Set up a column projection list that can be used by index_fetch_tuple()
+ * to fetch a subset of columns for a tuple.
+ *
+ * If project_columns is an empty bitmap, none of the data columns are to be
+ * fetched.
+ *
+ * If project_columns is a singleton bitmap with a whole-row reference (0),
+ * all of the data columns are to be fetched.
+ *
+ * Please note: project_columns only deals with non system columns (attnum >= 0)
+ *
+ * Please note: Due to the limitations of the slot_get***() APIs,
+ * index_fetch_tuple() must return a TupleTableSlot that is densely
+ * populated (missing cols indicated with isnull = true upto the largest
+ * attno in the projection list)
+ */
+ void (*index_fetch_set_column_projection) (struct IndexFetchTableData *data,
+ Bitmapset *project_columns);
+
/*
* Fetch tuple at `tid` into `slot`, after doing a visibility test
* according to `snapshot`. If a tuple was found and passed the visibility
@@ -444,11 +489,27 @@ typedef struct TableAmRoutine
* Fetch tuple at `tid` into `slot`, after doing a visibility test
* according to `snapshot`. If a tuple was found and passed the visibility
* test, returns true, false otherwise.
+ *
+ * project_cols is a set of columns to be fetched for the given row.
+ *
+ * If project_cols is an empty bitmap, none of the data columns are to be
+ * fetched.
+ *
+ * If project_cols is a singleton bitmap with a whole-row reference (0),
+ * all of the data columns are to be fetched.
+ *
+ * Please note: project_cols only deals with non system columns (attnum >= 0)
+ *
+ * Please note: Due to the limitations of the slot_get***() APIs,
+ * tuple_fetch_row_version() must return a TupleTableSlot that is densely
+ * populated (missing cols indicated with isnull = true upto the largest
+ * attno in the projection list)
*/
bool (*tuple_fetch_row_version) (Relation rel,
ItemPointer tid,
Snapshot snapshot,
- TupleTableSlot *slot);
+ TupleTableSlot *slot,
+ Bitmapset *project_cols);
/*
* Is tid valid for a scan of this relation.
@@ -535,7 +596,8 @@ typedef struct TableAmRoutine
LockTupleMode mode,
LockWaitPolicy wait_policy,
uint8 flags,
- TM_FailureData *tmfd);
+ TM_FailureData *tmfd,
+ Bitmapset *project_cols);
/*
* Perform operations necessary to complete insertions made via
@@ -889,6 +951,12 @@ table_beginscan(Relation rel, Snapshot snapshot,
return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key, NULL, flags);
}
+static inline bool
+table_scans_leverage_column_projection(Relation relation)
+{
+ return relation->rd_tableam->scans_leverage_column_projection;
+}
+
/*
* Like table_beginscan(), but for scanning catalog. It'll automatically use a
* snapshot appropriate for scanning catalog relations.
@@ -918,6 +986,19 @@ table_beginscan_strat(Relation rel, Snapshot snapshot,
return rel->rd_tableam->scan_begin(rel, snapshot, nkeys, key, NULL, flags);
}
+static inline TableScanDesc
+table_beginscan_with_column_projection(Relation relation, Snapshot snapshot,
+ int nkeys, struct ScanKeyData *key,
+ Bitmapset *project_column)
+{
+ uint32 flags = SO_TYPE_SEQSCAN |
+ SO_ALLOW_STRAT | SO_ALLOW_SYNC | SO_ALLOW_PAGEMODE;
+
+ Assert(relation->rd_tableam->scans_leverage_column_projection);
+ return relation->rd_tableam->scan_begin_with_column_projection(
+ relation, snapshot, nkeys, key, NULL, flags, project_column);
+}
+
/*
* table_beginscan_bm is an alternative entry point for setting up a
* TableScanDesc for a bitmap heap scan. Although that scan technology is
@@ -1132,7 +1213,8 @@ extern void table_parallelscan_initialize(Relation rel,
* Caller must hold a suitable lock on the relation.
*/
extern TableScanDesc table_beginscan_parallel(Relation rel,
- ParallelTableScanDesc pscan);
+ ParallelTableScanDesc pscan,
+ Bitmapset *proj);
/*
* Restart a parallel scan. Call this in the leader process. Caller is
@@ -1182,6 +1264,13 @@ table_index_fetch_end(struct IndexFetchTableData *scan)
scan->rel->rd_tableam->index_fetch_end(scan);
}
+static inline void
+table_index_fetch_set_column_projection(struct IndexFetchTableData *scan,
+ Bitmapset *project_column)
+{
+ scan->rel->rd_tableam->index_fetch_set_column_projection(scan, project_column);
+}
+
/*
* Fetches, as part of an index scan, tuple at `tid` into `slot`, after doing
* a visibility test according to `snapshot`. If a tuple was found and passed
@@ -1257,7 +1346,8 @@ static inline bool
table_tuple_fetch_row_version(Relation rel,
ItemPointer tid,
Snapshot snapshot,
- TupleTableSlot *slot)
+ TupleTableSlot *slot,
+ Bitmapset *project_cols)
{
/*
* We don't expect direct calls to table_tuple_fetch_row_version with
@@ -1267,7 +1357,7 @@ table_tuple_fetch_row_version(Relation rel,
if (unlikely(TransactionIdIsValid(CheckXidAlive) && !bsysscan))
elog(ERROR, "unexpected table_tuple_fetch_row_version call during logical decoding");
- return rel->rd_tableam->tuple_fetch_row_version(rel, tid, snapshot, slot);
+ return rel->rd_tableam->tuple_fetch_row_version(rel, tid, snapshot, slot, project_cols);
}
/*
@@ -1528,6 +1618,20 @@ table_tuple_update(Relation rel, ItemPointer otid, TupleTableSlot *slot,
* also lock descendant tuples if lock modes don't conflict.
* If TUPLE_LOCK_FLAG_FIND_LAST_VERSION, follow the update chain and lock
* latest version.
+ * project_cols: It is a set of columns to be fetched for the tuple being locked.
+ *
+ * If project_cols is an empty bitmap, none of the data columns are to be
+ * fetched.
+ *
+ * If project_cols is a singleton bitmap with a whole-row reference (0),
+ * all of the data columns are to be fetched.
+ *
+ * Please note: project_cols only deals with non system columns (attnum >= 0)
+ *
+ * Please note: Due to the limitations of the slot_get***() APIs,
+ * tuple_lock() must return a TupleTableSlot that is densely
+ * populated (missing cols indicated with isnull = true upto the largest
+ * attno in the projection list)
*
* Output parameters:
* *slot: contains the target tuple
@@ -1549,11 +1653,11 @@ static inline TM_Result
table_tuple_lock(Relation rel, ItemPointer tid, Snapshot snapshot,
TupleTableSlot *slot, CommandId cid, LockTupleMode mode,
LockWaitPolicy wait_policy, uint8 flags,
- TM_FailureData *tmfd)
+ TM_FailureData *tmfd, Bitmapset *project_cols)
{
return rel->rd_tableam->tuple_lock(rel, tid, snapshot, slot,
cid, mode, wait_policy,
- flags, tmfd);
+ flags, tmfd, project_cols);
}
/*
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 3dc03c913e..6824ee79ce 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -646,6 +646,12 @@ extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
extern void CheckSubscriptionRelkind(char relkind, const char *nspname,
const char *relname);
+extern void
+PopulateNeededColumnsForNode(Node *expr, int ncol, Bitmapset **scanCols);
+extern Bitmapset *
+PopulateNeededColumnsForScan(ScanState *scanstate, int ncol);
+extern Bitmapset *PopulateNeededColumnsForEPQ(EPQState *epqstate, int ncol);
+extern void PopulateNeededColumnsForOnConflictUpdate(ResultRelInfo *resultRelInfo);
/*
* prototypes from functions in nodeModifyTable.c
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 7795a69490..94353b0060 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -381,6 +381,7 @@ typedef struct OnConflictSetState
TupleTableSlot *oc_ProjSlot; /* CONFLICT ... SET ... projection target */
ProjectionInfo *oc_ProjInfo; /* for ON CONFLICT DO UPDATE SET */
ExprState *oc_WhereClause; /* state for the WHERE clause */
+ Bitmapset *proj_cols; /* cols to be scanned during the operation */
} OnConflictSetState;
/*
diff --git a/src/include/nodes/parsenodes.h b/src/include/nodes/parsenodes.h
index ef73342019..5891e31c53 100644
--- a/src/include/nodes/parsenodes.h
+++ b/src/include/nodes/parsenodes.h
@@ -1149,6 +1149,20 @@ typedef struct RangeTblEntry
Bitmapset *updatedCols; /* columns needing UPDATE permission */
Bitmapset *extraUpdatedCols; /* generated columns being updated */
List *securityQuals; /* security barrier quals to apply, if any */
+
+ /*
+ * scanCols: Columns to be retrieved during a physical scan.
+ * returningCols: Columns to be retrieved to satisfy the RETURNING clause.
+ *
+ * Please note: These bitmaps only deal with non-system columns (attnum >= 0)
+ *
+ * These bitmaps have some special values:
+ * - A singleton bitmap with the element 0 indicates that all non-system
+ * columns must be fetched.
+ * - An empty bitmap indicates that no non-system column must be fetched.
+ */
+ Bitmapset *scanCols; /* columns to be fetched during a physical scan */
+ Bitmapset *returningCols; /* columns in the RETURNING clause */
} RangeTblEntry;
/*
diff --git a/src/include/utils/rel.h b/src/include/utils/rel.h
index 774ac5b2b1..66d1bea175 100644
--- a/src/include/utils/rel.h
+++ b/src/include/utils/rel.h
@@ -658,6 +658,20 @@ typedef struct ViewOptions
(relation)->rd_rel->relkind != RELKIND_FOREIGN_TABLE && \
!IsCatalogRelation(relation))
+static inline bool
+contains_whole_row_col(Bitmapset *cols)
+{
+ return bms_is_member(0, cols);
+}
+
+static inline Bitmapset *
+get_ordinal_attnos(Relation rel)
+{
+ Bitmapset *attnos = NULL;
+ attnos = bms_add_range(attnos, 1, RelationGetDescr(rel)->natts);
+ return attnos;
+}
+
/* routines in utils/cache/relcache.c */
extern void RelationIncrementReferenceCount(Relation rel);
extern void RelationDecrementReferenceCount(Relation rel);
--
2.25.1
