partition routing layering in nodeModifyTable.c

From 95de45e03e62abc284deb9ba1107e3b84ee7dfc9 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v1 1/2] Reduce es_result_relation_info usage

Change many places that access the currently active result relation
using es_result_relation_info to receive it directly via function
arguments.  Maintaining that state in es_result_relation_info has
become cumbersome, especially with partitioning where each partition
gets its own result relation info.  Having to set and reset it across
arbitrary operations has caused bugs in the past.

We still need to set it before calling extension code, such as FDWs,
because the existing interfaces leave them with no option but to
access the result relation via EState.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/nodeModifyTable.c   | 187 ++++++++++++++-----------------
 src/backend/replication/logical/worker.c |  39 ++++---
 src/include/executor/executor.h          |  19 +++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   8 +-
 9 files changed, 147 insertions(+), 162 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 4f04d122c3..2f682de785 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2445,9 +2445,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2480,7 +2477,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2845,7 +2843,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3117,11 +3114,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3225,7 +3217,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3296,7 +3288,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index fc1c4dfa4c..cedaecc6b9 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1745,7 +1745,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1875,7 +1874,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index d8b695d897..4a92ae17b4 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,50 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+		/* Result relation has changed, so update EState reference too. */
+		estate->es_result_relation_info = resultRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +411,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -428,7 +447,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -490,8 +509,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -551,7 +570,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -590,7 +610,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -676,6 +697,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -687,7 +709,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -697,10 +718,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1037,6 +1054,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1045,12 +1063,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1060,10 +1076,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1090,7 +1102,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1127,7 +1139,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1177,6 +1189,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1202,8 +1215,8 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate, false, false /* canSetTag */ ,
 					   true /* changingPart */ , &tuple_deleted, &epqslot);
 
 			/*
@@ -1245,16 +1258,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1271,18 +1274,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1448,7 +1451,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1687,7 +1691,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1844,41 +1848,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -1989,10 +1989,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2041,14 +2039,9 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
+	 * Save the result relation in EState, because that's the only place
+	 * where external modules such as FDWs may find it.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	estate->es_result_relation_info = resultRelInfo;
 
 	/*
@@ -2084,6 +2077,9 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
+				/*
+				 * Result relation has changed, so update EState reference too.
+				 */
 				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
@@ -2129,7 +2125,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2212,23 +2207,17 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
 								  true, node->canSetTag,
 								  false /* changingPart */ , NULL, NULL);
 				break;
@@ -2242,15 +2231,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2271,7 +2254,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2314,14 +2296,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2362,8 +2338,15 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			operation == CMD_UPDATE)
 			update_tuple_routing_needed = true;
 
-		/* Now init the plan for this result rel */
+		/*
+		 * Save the result relation in EState, because that's the only place
+		 * where external modules such as FDWs may find it.  (See
+		 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify()
+		 * as one example.)
+		 */
 		estate->es_result_relation_info = resultRelInfo;
+
+		/* Now init the plan for this result rel */
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2387,8 +2370,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 43edfef089..7df3e78b22 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -173,10 +173,10 @@ ensure_transaction(void)
  * This is based on similar code in copy.c
  */
 static EState *
-create_estate_for_relation(LogicalRepRelMapEntry *rel)
+create_estate_for_relation(LogicalRepRelMapEntry *rel,
+						   ResultRelInfo **resultRelInfo)
 {
 	EState	   *estate;
-	ResultRelInfo *resultRelInfo;
 	RangeTblEntry *rte;
 
 	estate = CreateExecutorState();
@@ -188,12 +188,11 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 	rte->rellockmode = AccessShareLock;
 	ExecInitRangeTable(estate, list_make1(rte));
 
-	resultRelInfo = makeNode(ResultRelInfo);
-	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
+	*resultRelInfo = makeNode(ResultRelInfo);
+	InitResultRelInfo(*resultRelInfo, rel->localrel, 1, NULL, 0);
 
-	estate->es_result_relations = resultRelInfo;
+	estate->es_result_relations = *resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -567,6 +566,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -589,7 +589,7 @@ apply_handle_insert(StringInfo s)
 	}
 
 	/* Initialize the executor state. */
-	estate = create_estate_for_relation(rel);
+	estate = create_estate_for_relation(rel, &resultRelInfo);
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -603,13 +603,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -664,6 +664,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -696,7 +697,7 @@ apply_handle_update(StringInfo s)
 	check_relation_updatable(rel);
 
 	/* Initialize the executor state. */
-	estate = create_estate_for_relation(rel);
+	estate = create_estate_for_relation(rel, &resultRelInfo);
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -705,7 +706,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -747,7 +748,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -763,7 +765,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -786,6 +788,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -815,7 +818,7 @@ apply_handle_delete(StringInfo s)
 	check_relation_updatable(rel);
 
 	/* Initialize the executor state. */
-	estate = create_estate_for_relation(rel);
+	estate = create_estate_for_relation(rel, &resultRelInfo);
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -824,7 +827,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -852,7 +855,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -864,7 +867,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 1fb28b4596..3ecdcc3a34 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 98bdcbcef5..9ce58049f5 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,7 +519,13 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
+
+	/*
+	 * Currently active result relation.  The core code no longer uses this
+	 * value, but it's still maintained for the convenience of third party
+	 * code.
+	 */
+	ResultRelInfo *es_result_relation_info;
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
-- 
2.11.0

From 8155bcd9dda6973963faa2149cb7a59d9f76194b Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v1 2/2] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecMove() which must be retried until
it finishes the movement.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 327 ++++++++++++++++++---------------
 1 file changed, 180 insertions(+), 147 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 4a92ae17b4..25a587723d 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -623,31 +622,29 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If this delete is a part of a partition key update, put this row into
+	 * the UPDATE trigger's NEW TABLE instead of that of an INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -713,7 +710,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -958,32 +954,29 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If this delete is a part of a partition key update, put this row into
+	 * the UPDATE trigger's OLD TABLE instead of that of a DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1030,6 +1023,139 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecMove
+ *		Move an updated tuple from the input result relation to the
+ *		new partition of its root parent table
+ *
+ *	This works by first deleting the tuple from the input result relation
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo.
+ *
+ *	Returns true if it's detected that the tuple we're trying to move has
+ *	been concurrently updated.
+ */
+static bool
+ExecMove(ModifyTableState *mtstate, ResultRelInfo *resultRelInfo,
+		 ItemPointer tupleid, HeapTuple oldtuple, TupleTableSlot *planSlot,
+		 EPQState *epqstate, bool canSetTag, TupleTableSlot **slot,
+		 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, *slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate, false, false /* canSetTag */ ,
+			   true /* changingPart */ , &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return false;
+		else
+		{
+			*slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return true;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		*slot = execute_attr_map_slot(tupconv_map->attrMap,
+									  *slot,
+									  mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, *slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return false;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1183,116 +1309,23 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecMove() will first DELETE the row from the current partition
+			 * and then insert it back into the root table, which will
+			 * re-route it to the correct partition.  The first part may have
+			 * to be repeated if ExecMove() detects that the tuple we're
+			 * trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+			retry = ExecMove(mtstate, resultRelInfo, tupleid, oldtuple,
+							 planSlot, epqstate, canSetTag,
+							 &slot, &inserted_tuple);
+			if (retry)
+				goto lreplace;
 
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
-			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
-			}
-
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.11.0

From 72c010158ea534e449fdbb998274951400415df0 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v2 3/3] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.
---
 src/backend/commands/copy.c            |  31 ++-----
 src/backend/commands/trigger.c         |  21 +++--
 src/backend/executor/execPartition.c   |  23 +++--
 src/backend/executor/nodeModifyTable.c | 156 +++++++++------------------------
 src/include/commands/trigger.h         |  10 +--
 src/include/nodes/execnodes.h          |   6 ++
 6 files changed, 84 insertions(+), 163 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 2f682de785..5d02c67389 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3114,32 +3114,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 2d9a8e9d54..43f796172c 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4644,9 +4645,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5750,14 +5749,26 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from the global variable and set
+		 * the latter to NULL because any given tuple must be read only once.
+		 * Note that the TransitionCaptureState is shared across many calls
+		 * to this function.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 729dc396a9..62b93f39d4 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -167,7 +167,8 @@ static void ExecInitRoutingInfo(ModifyTableState *mtstate,
 								PartitionTupleRouting *proute,
 								PartitionDispatch dispatch,
 								ResultRelInfo *partRelInfo,
-								int partidx);
+								int partidx,
+								bool is_update_result_rel);
 static PartitionDispatch ExecInitPartitionDispatchInfo(EState *estate,
 													   PartitionTupleRouting *proute,
 													   Oid partoid, PartitionDispatch parent_pd, int partidx);
@@ -389,7 +390,7 @@ ExecFindPartition(ModifyTableState *mtstate,
 
 						/* Set up the PartitionRoutingInfo for it */
 						ExecInitRoutingInfo(mtstate, estate, proute, dispatch,
-											rri, partidx);
+											rri, partidx, true);
 					}
 				}
 
@@ -676,7 +677,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 
 	/* Set up information needed for routing tuples to the partition. */
 	ExecInitRoutingInfo(mtstate, estate, proute, dispatch,
-						leaf_part_rri, partidx);
+						leaf_part_rri, partidx, false);
 
 	/*
 	 * If there is an ON CONFLICT clause, initialize state for it.
@@ -888,7 +889,8 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 					PartitionTupleRouting *proute,
 					PartitionDispatch dispatch,
 					ResultRelInfo *partRelInfo,
-					int partidx)
+					int partidx,
+					bool is_update_result_rel)
 {
 	MemoryContext oldcxt;
 	PartitionRoutingInfo *partrouteinfo;
@@ -935,10 +937,15 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot),
-								   gettext_noop("could not convert row type"));
+		/* If partition is an update target, then we already got the map. */
+		if (is_update_result_rel)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot),
+									   gettext_noop("could not convert row type"));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 253fe96e6c..8f4cf1e616 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -74,8 +74,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -339,10 +337,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1055,9 +1049,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1133,41 +1125,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1874,28 +1841,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1919,6 +1864,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1933,37 +1879,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -2018,20 +1944,6 @@ ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
 	}
 }
 
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2137,17 +2049,6 @@ ExecModifyTable(PlanState *pstate)
 				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2316,6 +2217,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2338,8 +2240,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2349,6 +2256,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2422,6 +2336,21 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc),
+									   gettext_noop("could not convert row type"));
 		resultRelInfo++;
 		i++;
 	}
@@ -2446,13 +2375,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
 	 * Construct mapping from each of the per-subplan partition attnos to the
 	 * root attno.  This is required when during update row movement the tuple
 	 * descriptor of a source partition does not match the root partitioned
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a46feeedb0..bb080980c0 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -65,14 +65,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 9ce58049f5..ffa14d3248 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -485,6 +485,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
-- 
2.11.0

From 2e62f7d72b4cc93a1bb5f3ae9ab75e09862734b3 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v2 2/3] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 27ad69aa1c..253fe96e6c 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -623,31 +622,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -713,7 +711,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -958,32 +955,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1030,6 +1025,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1183,119 +1325,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.11.0

From cf9609ec037ae91352e734e6e5aa2a9a6a9939f6 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v2 1/3] Reduce es_result_relation_info usage

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.

We still need to set it before calling third-party code, like FDWs,
because the existing interfaces leave them with no option but to
access the result relation via EState.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/nodeModifyTable.c   | 198 ++++++++++++++-----------------
 src/backend/replication/logical/worker.c |  26 ++--
 src/include/executor/executor.h          |  19 ++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   8 +-
 9 files changed, 150 insertions(+), 157 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 4f04d122c3..2f682de785 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2445,9 +2445,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2480,7 +2477,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2845,7 +2843,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3117,11 +3114,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3225,7 +3217,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3296,7 +3288,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index fb2be10794..e63b25bf25 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1746,7 +1746,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1876,7 +1875,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9e0c8794c4..27ad69aa1c 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,50 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+		/* Result relation has changed, so update EState reference too. */
+		estate->es_result_relation_info = resultRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +411,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -428,7 +447,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -490,8 +509,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -551,7 +570,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -590,7 +610,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -676,6 +697,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -687,7 +709,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -697,10 +718,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1037,6 +1054,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1045,12 +1063,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1060,10 +1076,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1090,7 +1102,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1127,7 +1139,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1177,6 +1189,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1202,9 +1215,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1245,16 +1261,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1271,18 +1277,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1448,7 +1454,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1687,7 +1694,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1844,41 +1851,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -1989,10 +1992,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2041,14 +2042,9 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
+	 * Save the result relation in EState, because that's the only place
+	 * where external modules such as FDWs may find it.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	estate->es_result_relation_info = resultRelInfo;
 
 	/*
@@ -2084,6 +2080,9 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
+				/*
+				 * Result relation has changed, so update EState reference too.
+				 */
 				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
@@ -2129,7 +2128,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2212,25 +2210,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2242,15 +2236,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2271,7 +2259,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2314,14 +2301,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2362,8 +2343,15 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			operation == CMD_UPDATE)
 			update_tuple_routing_needed = true;
 
-		/* Now init the plan for this result rel */
+		/*
+		 * Save the result relation in EState, because that's the only place
+		 * where external modules such as FDWs may find it.  (See
+		 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify()
+		 * as one example.)
+		 */
 		estate->es_result_relation_info = resultRelInfo;
+
+		/* Now init the plan for this result rel */
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2387,8 +2375,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 43edfef089..10ef6af3e7 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -193,7 +193,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -567,6 +566,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -590,6 +590,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -603,13 +604,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -664,6 +665,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -697,6 +699,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -705,7 +708,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -747,7 +750,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -763,7 +767,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -786,6 +790,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -816,6 +821,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -824,7 +830,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -852,7 +858,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -864,7 +870,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 1fb28b4596..3ecdcc3a34 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 98bdcbcef5..9ce58049f5 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,7 +519,13 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
+
+	/*
+	 * Currently active result relation.  The core code no longer uses this
+	 * value, but it's still maintained for the convenience of third party
+	 * code.
+	 */
+	ResultRelInfo *es_result_relation_info;
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
-- 
2.11.0

From fe4a05fcd0f16944ed27ee47a19393afa025f4be Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Wed, 31 Jul 2019 16:38:43 +0900
Subject: [PATCH v3 1/4] Revise BeginDirectModify API to pass ResultRelInfo
 directly

---
 contrib/postgres_fdw/postgres_fdw.c    | 22 ++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml           | 11 ++++++++++-
 src/backend/executor/nodeForeignscan.c |  7 +++----
 src/backend/executor/nodeModifyTable.c | 15 ++++++++++++---
 src/include/foreign/fdwapi.h           |  1 +
 5 files changed, 42 insertions(+), 14 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 033aeb2556..1b60ff88b1 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -205,6 +205,9 @@ typedef struct PgFdwDirectModifyState
 	List	   *retrieved_attrs;	/* attr numbers retrieved by RETURNING */
 	bool		set_processed;	/* do we set the command es_processed? */
 
+	/* Information about the relation being modified */
+	ResultRelInfo *resultRelInfo;
+
 	/* for remote query execution */
 	PGconn	   *conn;			/* connection for the update */
 	int			numParams;		/* number of parameters passed to query */
@@ -360,7 +363,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2340,7 +2345,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2368,7 +2375,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2414,6 +2421,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	dmstate->set_processed = intVal(list_nth(fsplan->fdw_private,
 											 FdwDirectModifyPrivateSetProcessed));
 
+	/* Save the ResultRelInfo of the relation being modified. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Create context for per-tuple temp workspace. */
 	dmstate->temp_cxt = AllocSetContextCreate(estate->es_query_cxt,
 											  "postgres_fdw temporary data",
@@ -2463,7 +2473,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4043,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4190,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 27b94fb611..04c2eccd1c 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -871,6 +871,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -883,7 +884,9 @@ BeginDirectModify(ForeignScanState *node,
      the table to modify is accessible through the
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
+     information provided by <function>PlanDirectModify</function>).  In
+     addition, <literal>rinfo</literal> also contains information describing
+     the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -895,6 +898,12 @@ BeginDirectModify(ForeignScanState *node,
      for <function>ExplainDirectModify</function> and <function>EndDirectModify</function>.
     </para>
 
+    <note>
+     Also note that it's a good idea to store the <literal>rinfo</literal>
+     in the <structfield>fdw_state</structfield> for
+     <function>IterateDirectModify</function> to use.
+    </node>
+
     <para>
      If the <function>BeginDirectModify</function> pointer is set to
      <literal>NULL</literal>, no attempts to execute a direct modification on the
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..88d9f5da10 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -221,11 +221,10 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan.  Direct modification scans
+	 * are initialized elsewhere; see ExecInitModifyTable().
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
 
 	return scanstate;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9e0c8794c4..55ce709adb 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -2370,9 +2370,18 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 								   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
 
 		/* Also let FDWs init themselves for foreign-table result rels */
-		if (!resultRelInfo->ri_usesFdwDirectModify &&
-			resultRelInfo->ri_FdwRoutine != NULL &&
-			resultRelInfo->ri_FdwRoutine->BeginForeignModify != NULL)
+		if (resultRelInfo->ri_usesFdwDirectModify)
+		{
+			ForeignScanState *scanstate;
+
+			Assert(IsA(mtstate->mt_plans[i], ForeignScanState));
+			scanstate = (ForeignScanState *) mtstate->mt_plans[i];
+			resultRelInfo->ri_FdwRoutine->BeginDirectModify(scanstate,
+															resultRelInfo,
+															eflags);
+		}
+		else if (resultRelInfo->ri_FdwRoutine != NULL &&
+				 resultRelInfo->ri_FdwRoutine->BeginForeignModify != NULL)
 		{
 			List	   *fdw_private = (List *) list_nth(node->fdwPrivLists, i);
 
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
-- 
2.11.0

From 88cb7d5a256f7d21268648f49595e513839030a2 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v3 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   4 -
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 188 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  26 +++--
 src/include/executor/executor.h          |  19 +++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   1 -
 11 files changed, 128 insertions(+), 168 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 4f04d122c3..2f682de785 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2445,9 +2445,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2480,7 +2477,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2845,7 +2843,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3117,11 +3114,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3225,7 +3217,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3296,7 +3288,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index fb2be10794..e63b25bf25 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1746,7 +1746,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1876,7 +1875,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index dbd7dd9bcd..2e8802df07 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -858,9 +858,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -904,7 +901,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index c1fc0d54e9..eaaf69bb93 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,8 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 55ce709adb..bf0679d753 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +409,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -428,7 +445,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -490,8 +507,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -551,7 +568,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -590,7 +608,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -676,6 +695,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -687,7 +707,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -697,10 +716,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1037,6 +1052,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1045,12 +1061,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1060,10 +1074,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1090,7 +1100,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1127,7 +1137,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1177,6 +1187,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1202,9 +1213,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1245,16 +1259,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1271,18 +1275,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1448,7 +1452,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1687,7 +1692,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1844,41 +1849,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -1989,10 +1990,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2041,17 +2040,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2084,7 +2072,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2129,7 +2116,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2212,25 +2198,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2242,15 +2224,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2271,7 +2247,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2314,14 +2289,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2363,7 +2332,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2396,8 +2364,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 43edfef089..10ef6af3e7 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -193,7 +193,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -567,6 +566,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -590,6 +590,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -603,13 +604,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -664,6 +665,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -697,6 +699,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -705,7 +708,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -747,7 +750,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -763,7 +767,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -786,6 +790,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -816,6 +821,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -824,7 +830,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -852,7 +858,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -864,7 +870,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 1fb28b4596..3ecdcc3a34 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 98bdcbcef5..b527cd93ed 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,7 +519,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
-- 
2.11.0

From 95eee81cff36bcbbfed75205d93f16dc1c53e1c5 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v3 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.
---
 src/backend/commands/copy.c            |  31 ++-----
 src/backend/commands/trigger.c         |  21 +++--
 src/backend/executor/execPartition.c   |  23 +++--
 src/backend/executor/nodeModifyTable.c | 156 +++++++++------------------------
 src/include/commands/trigger.h         |  10 +--
 src/include/nodes/execnodes.h          |   6 ++
 6 files changed, 84 insertions(+), 163 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 2f682de785..5d02c67389 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3114,32 +3114,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 2d9a8e9d54..43f796172c 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4644,9 +4645,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5750,14 +5749,26 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from the global variable and set
+		 * the latter to NULL because any given tuple must be read only once.
+		 * Note that the TransitionCaptureState is shared across many calls
+		 * to this function.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 729dc396a9..62b93f39d4 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -167,7 +167,8 @@ static void ExecInitRoutingInfo(ModifyTableState *mtstate,
 								PartitionTupleRouting *proute,
 								PartitionDispatch dispatch,
 								ResultRelInfo *partRelInfo,
-								int partidx);
+								int partidx,
+								bool is_update_result_rel);
 static PartitionDispatch ExecInitPartitionDispatchInfo(EState *estate,
 													   PartitionTupleRouting *proute,
 													   Oid partoid, PartitionDispatch parent_pd, int partidx);
@@ -389,7 +390,7 @@ ExecFindPartition(ModifyTableState *mtstate,
 
 						/* Set up the PartitionRoutingInfo for it */
 						ExecInitRoutingInfo(mtstate, estate, proute, dispatch,
-											rri, partidx);
+											rri, partidx, true);
 					}
 				}
 
@@ -676,7 +677,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 
 	/* Set up information needed for routing tuples to the partition. */
 	ExecInitRoutingInfo(mtstate, estate, proute, dispatch,
-						leaf_part_rri, partidx);
+						leaf_part_rri, partidx, false);
 
 	/*
 	 * If there is an ON CONFLICT clause, initialize state for it.
@@ -888,7 +889,8 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 					PartitionTupleRouting *proute,
 					PartitionDispatch dispatch,
 					ResultRelInfo *partRelInfo,
-					int partidx)
+					int partidx,
+					bool is_update_result_rel)
 {
 	MemoryContext oldcxt;
 	PartitionRoutingInfo *partrouteinfo;
@@ -935,10 +937,15 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot),
-								   gettext_noop("could not convert row type"));
+		/* If partition is an update target, then we already got the map. */
+		if (is_update_result_rel)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot),
+									   gettext_noop("could not convert row type"));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index fad8b928bb..4868e6f3a6 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -74,8 +74,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -339,10 +337,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1053,9 +1047,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1131,41 +1123,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1872,28 +1839,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1917,6 +1862,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1931,37 +1877,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -2016,20 +1942,6 @@ ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
 	}
 }
 
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2125,17 +2037,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2304,6 +2205,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2326,8 +2228,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2337,6 +2244,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2411,6 +2325,21 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc),
+									   gettext_noop("could not convert row type"));
 		resultRelInfo++;
 		i++;
 	}
@@ -2435,13 +2364,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
 	 * Construct mapping from each of the per-subplan partition attnos to the
 	 * root attno.  This is required when during update row movement the tuple
 	 * descriptor of a source partition does not match the root partitioned
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a46feeedb0..bb080980c0 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -65,14 +65,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index b527cd93ed..0e70a62b26 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -485,6 +485,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
-- 
2.11.0

From 9f66668cddb46f52db65024b3d05cb6654604cb8 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v3 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index bf0679d753..fad8b928bb 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -621,31 +620,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -711,7 +709,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -956,32 +953,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1028,6 +1023,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1181,119 +1323,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.11.0

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 033aeb2556..9982007082 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -197,6 +197,7 @@ typedef struct PgFdwModifyState
 typedef struct PgFdwDirectModifyState
 {
 	Relation	rel;			/* relcache entry for the foreign table */
+	ResultRelInfo *relInfo;		/* ResultRelInfo for the foreign table */
 	AttInMetadata *attinmeta;	/* attribute datatype conversion metadata */
 
 	/* extracted fdw_private data */
@@ -2303,6 +2304,7 @@ postgresPlanDirectModify(PlannerInfo *root,
 	 * Update the operation info.
 	 */
 	fscan->operation = operation;
+	fscan->resultRelation = resultRelation;
 
 	/*
 	 * Update the fdw_exprs list that will be available to the executor.
@@ -2368,7 +2370,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = fsplan->resultRelation;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2380,6 +2382,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	table = GetForeignTable(RelationGetRelid(dmstate->rel));
 	user = GetUserMapping(userid, table->serverid);
 
+	/* Get ResultRelInfo for foreign table. */
+	dmstate->relInfo = ExecFindResultRelInfo(estate, rtindex, false);
+
 	/*
 	 * Get connection to the foreign server.  Connection manager will
 	 * establish new connection if necessary.
@@ -2463,7 +2468,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->relInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4038,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->relInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4185,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->relInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index dbd7dd9bcd..4d0d0553be 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -1349,6 +1349,31 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
 	resultRelInfo->ri_CopyMultiInsertBuffer = NULL;
 }
 
+/*
+ * ExecFindResultRelInfo -- find the ResultRelInfo struct for given rangetable
+ * index
+ *
+ * This function only searches through the query result relations.  If no such
+ * struct, either return NULL or throw error depending on missing_ok
+ */
+ResultRelInfo *
+ExecFindResultRelInfo(EState *estate, Index rti, bool missing_ok)
+{
+	ResultRelInfo *rInfo = estate->es_result_relations;
+	int			nr  = estate->es_num_result_relations;
+
+	while (nr > 0)
+	{
+		if (rInfo->ri_RangeTableIndex == rti)
+			return rInfo;
+		rInfo++;
+		nr--;
+	}
+	if (!missing_ok)
+		elog(ERROR, "failed to find ResultRelInfo for rangetable index %u", rti);
+	return NULL;
+}
+
 /*
  * ExecGetTriggerResultRel
  *		Get a ResultRelInfo for a trigger target relation.
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 6414aded0e..9342b99aae 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -751,6 +751,7 @@ _copyForeignScan(const ForeignScan *from)
 	 * copy remainder of node
 	 */
 	COPY_SCALAR_FIELD(operation);
+	COPY_SCALAR_FIELD(resultRelation);
 	COPY_SCALAR_FIELD(fs_server);
 	COPY_NODE_FIELD(fdw_exprs);
 	COPY_NODE_FIELD(fdw_private);
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index 86c31a48c9..4dadb9ff7c 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -688,6 +688,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	_outScanInfo(str, (const Scan *) node);
 
 	WRITE_ENUM_FIELD(operation, CmdType);
+	WRITE_UINT_FIELD(resultRelation);
 	WRITE_OID_FIELD(fs_server);
 	WRITE_NODE_FIELD(fdw_exprs);
 	WRITE_NODE_FIELD(fdw_private);
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 6c2626ee62..04080c752a 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1976,6 +1976,7 @@ _readForeignScan(void)
 	ReadCommonScan(&local_node->scan);
 
 	READ_ENUM_FIELD(operation, CmdType);
+	READ_UINT_FIELD(resultRelation);
 	READ_OID_FIELD(fs_server);
 	READ_NODE_FIELD(fdw_exprs);
 	READ_NODE_FIELD(fdw_private);
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index c6b8553a08..cc5be409ec 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5415,6 +5415,7 @@ make_foreignscan(List *qptlist,
 	plan->righttree = NULL;
 	node->scan.scanrelid = scanrelid;
 	node->operation = CMD_SELECT;
+	node->resultRelation = 0;
 	/* fs_server will be filled in by create_foreignscan_plan */
 	node->fs_server = InvalidOid;
 	node->fdw_exprs = fdw_exprs;
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index dc11f098e0..25c1e7dfc7 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -1158,6 +1158,10 @@ set_foreignscan_references(PlannerInfo *root,
 	if (fscan->scan.scanrelid > 0)
 		fscan->scan.scanrelid += rtoffset;
 
+	/* Adjust resultRelation if it's valid */
+	if (fscan->resultRelation > 0)
+		fscan->resultRelation += rtoffset;
+
 	if (fscan->fdw_scan_tlist != NIL || fscan->scan.scanrelid == 0)
 	{
 		/*
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 1fb28b4596..a0533df779 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -184,6 +184,7 @@ extern void InitResultRelInfo(ResultRelInfo *resultRelInfo,
 							  Index resultRelationIndex,
 							  Relation partition_root,
 							  int instrument_options);
+extern ResultRelInfo *ExecFindResultRelInfo(EState *estate, Index rti, bool missing_ok);
 extern ResultRelInfo *ExecGetTriggerResultRel(EState *estate, Oid relid);
 extern void ExecCleanUpTriggerState(EState *estate);
 extern void ExecConstraints(ResultRelInfo *resultRelInfo,
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 70f8b8e22b..218626254f 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -609,6 +609,8 @@ typedef struct ForeignScan
 {
 	Scan		scan;
 	CmdType		operation;		/* SELECT/INSERT/UPDATE/DELETE */
+	Index		resultRelation;	/* rtable index of target relation for
+								 * INSERT/UPDATE/DELETE; 0 for SELECT */
 	Oid			fs_server;		/* OID of foreign server */
 	List	   *fdw_exprs;		/* expressions that FDW may evaluate */
 	List	   *fdw_private;	/* private data for FDW */

From f7231c6d04da106321f8b43b1a55dc10b8cb2b82 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Wed, 31 Jul 2019 16:38:43 +0900
Subject: [PATCH v4 1/4] Revise BeginDirectModify API to pass ResultRelInfo
 directly

For ExecInitForeignScan() to efficiently get the ResultRelInfo to
pass to BeginDirectModify(), add a field to ForeignScan that gives
the index of a given result relation in the query's list of result
relations.
---
 contrib/postgres_fdw/postgres_fdw.c     | 22 ++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            | 11 ++++++++++-
 src/backend/executor/nodeForeignscan.c  | 13 ++++++++++---
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 28 ++++++++++++++++++++++------
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  4 ++++
 9 files changed, 67 insertions(+), 16 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 033aeb2556..1b60ff88b1 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -205,6 +205,9 @@ typedef struct PgFdwDirectModifyState
 	List	   *retrieved_attrs;	/* attr numbers retrieved by RETURNING */
 	bool		set_processed;	/* do we set the command es_processed? */
 
+	/* Information about the relation being modified */
+	ResultRelInfo *resultRelInfo;
+
 	/* for remote query execution */
 	PGconn	   *conn;			/* connection for the update */
 	int			numParams;		/* number of parameters passed to query */
@@ -360,7 +363,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2340,7 +2345,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2368,7 +2375,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2414,6 +2421,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	dmstate->set_processed = intVal(list_nth(fsplan->fdw_private,
 											 FdwDirectModifyPrivateSetProcessed));
 
+	/* Save the ResultRelInfo of the relation being modified. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Create context for per-tuple temp workspace. */
 	dmstate->temp_cxt = AllocSetContextCreate(estate->es_query_cxt,
 											  "postgres_fdw temporary data",
@@ -2463,7 +2473,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4043,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4190,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 27b94fb611..04c2eccd1c 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -871,6 +871,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -883,7 +884,9 @@ BeginDirectModify(ForeignScanState *node,
      the table to modify is accessible through the
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
+     information provided by <function>PlanDirectModify</function>).  In
+     addition, <literal>rinfo</literal> also contains information describing
+     the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -895,6 +898,12 @@ BeginDirectModify(ForeignScanState *node,
      for <function>ExplainDirectModify</function> and <function>EndDirectModify</function>.
     </para>
 
+    <note>
+     Also note that it's a good idea to store the <literal>rinfo</literal>
+     in the <structfield>fdw_state</structfield> for
+     <function>IterateDirectModify</function> to use.
+    </node>
+
     <para>
      If the <function>BeginDirectModify</function> pointer is set to
      <literal>NULL</literal>, no attempts to execute a direct modification on the
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..9824c16e09 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -223,10 +223,17 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	/*
 	 * Tell the FDW to initialize the scan.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		/* Perform initializations for a direct modification. */
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index a2617c7cfd..e981298a75 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e6ce8e2110..80eedc4a24 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index f2325694c5..ff78167f79 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5455,6 +5455,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* this might be filled to a >= 0 value by set_plan_refs() */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 329ebd5f28..b233eb7dce 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -781,6 +781,7 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 		case T_ModifyTable:
 			{
 				ModifyTable *splan = (ModifyTable *) plan;
+				int		resultRelIndex;
 
 				Assert(splan->plan.targetlist == NIL);
 				Assert(splan->plan.qual == NIL);
@@ -877,12 +878,6 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
-				foreach(l, splan->plans)
-				{
-					lfirst(l) = set_plan_refs(root,
-											  (Plan *) lfirst(l),
-											  rtoffset);
-				}
 
 				/*
 				 * Append this ModifyTable node's final result relation RT
@@ -908,6 +903,27 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 						lappend_int(root->glob->rootResultRelations,
 									splan->rootRelation);
 				}
+
+				resultRelIndex = splan->resultRelIndex;
+				foreach(l, splan->plans)
+				{
+					lfirst(l) = set_plan_refs(root,
+											  (Plan *) lfirst(l),
+											  rtoffset);
+
+					/*
+					 * For foreign table result relations, save their index
+					 * in the global list of result relations into the
+					 * corresponding ForeignScan nodes.
+					 */
+					if (IsA(lfirst(l), ForeignScan))
+					{
+						ForeignScan *fscan = (ForeignScan *) lfirst(l);
+
+						fscan->resultRelIndex = resultRelIndex;
+					}
+					resultRelIndex++;
+				}
 			}
 			break;
 		case T_Append:
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 8e6594e355..047dd73dd1 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -616,6 +616,10 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* For non-SELECT operations, this contains
+								 * the offset of result relation in a
+								 * query-global list of result relations; -1
+								 * otherwise */
 } ForeignScan;
 
 /* ----------------
-- 
2.11.0

From bb48579528926d485a675bbedaac9dae8a4062ff Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v4 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   4 -
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 188 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  26 +++--
 src/include/executor/executor.h          |  19 +++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   1 -
 11 files changed, 128 insertions(+), 168 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 4f04d122c3..2f682de785 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2445,9 +2445,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2480,7 +2477,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2845,7 +2843,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3117,11 +3114,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3225,7 +3217,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3296,7 +3288,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index fb2be10794..e63b25bf25 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1746,7 +1746,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1876,7 +1875,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index dbd7dd9bcd..2e8802df07 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -858,9 +858,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -904,7 +901,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index c1fc0d54e9..eaaf69bb93 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,8 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9e0c8794c4..3316c089e9 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +409,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -428,7 +445,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -490,8 +507,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -551,7 +568,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -590,7 +608,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -676,6 +695,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -687,7 +707,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -697,10 +716,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1037,6 +1052,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1045,12 +1061,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1060,10 +1074,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1090,7 +1100,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1127,7 +1137,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1177,6 +1187,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1202,9 +1213,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1245,16 +1259,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1271,18 +1275,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1448,7 +1452,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1687,7 +1692,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1844,41 +1849,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -1989,10 +1990,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2041,17 +2040,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2084,7 +2072,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2129,7 +2116,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2212,25 +2198,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2242,15 +2224,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2271,7 +2247,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2314,14 +2289,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2363,7 +2332,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2387,8 +2355,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 43edfef089..10ef6af3e7 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -193,7 +193,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -567,6 +566,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -590,6 +590,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -603,13 +604,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -664,6 +665,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -697,6 +699,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -705,7 +708,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -747,7 +750,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -763,7 +767,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -786,6 +790,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -816,6 +821,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -824,7 +830,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -852,7 +858,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -864,7 +870,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 1fb28b4596..3ecdcc3a34 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 4ec78491f6..0b40d13bfa 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,7 +519,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
-- 
2.11.0

From 54778e45d8ed52171b30dcc1819a1934269690f1 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v4 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 ++---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  21 +++-
 src/backend/executor/nodeModifyTable.c | 209 +++++++--------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 87 insertions(+), 212 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 2f682de785..5d02c67389 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3114,32 +3114,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 2d9a8e9d54..a8faa5e1e4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4644,9 +4645,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5750,14 +5749,24 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 729dc396a9..743f54926a 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -935,10 +935,23 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot),
-								   gettext_noop("could not convert row type"));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot),
+									   gettext_noop("could not convert row type"));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index cbf3de6267..d327153a6a 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -339,10 +336,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1053,9 +1046,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1131,41 +1122,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1872,28 +1838,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1917,6 +1861,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1931,37 +1876,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -1977,59 +1902,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc,
-								   gettext_noop("could not convert row type"));
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2125,17 +1997,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2304,6 +2165,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2326,8 +2188,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2337,6 +2204,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2402,6 +2276,21 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc),
+									   gettext_noop("could not convert row type"));
 		resultRelInfo++;
 		i++;
 	}
@@ -2426,26 +2315,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a46feeedb0..bb080980c0 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -65,14 +65,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 0b40d13bfa..9571bbe328 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -485,6 +485,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1136,9 +1142,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
2.11.0

From 91a24b9eaef9949b14197fef3f44d938ea78772b Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v4 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 3316c089e9..cbf3de6267 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -621,31 +620,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -711,7 +709,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -956,32 +953,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1028,6 +1023,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1181,119 +1323,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.11.0

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 06a205877d..a61c2d8b5f 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -205,6 +205,9 @@ typedef struct PgFdwDirectModifyState
 	List	   *retrieved_attrs;	/* attr numbers retrieved by RETURNING */
 	bool		set_processed;	/* do we set the command es_processed? */
 
+	/* Information about the relation being modified */
+	ResultRelInfo *resultRelInfo;
+
 	/* for remote query execution */
 	PGconn	   *conn;			/* connection for the update */
 	int			numParams;		/* number of parameters passed to query */
@@ -360,7 +363,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2331,6 +2336,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2340,7 +2350,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2368,7 +2380,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2414,6 +2426,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	dmstate->set_processed = intVal(list_nth(fsplan->fdw_private,
 											 FdwDirectModifyPrivateSetProcessed));
 
+	/* Save the ResultRelInfo of the relation being modified. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Create context for per-tuple temp workspace. */
 	dmstate->temp_cxt = AllocSetContextCreate(estate->es_query_cxt,
 											  "postgres_fdw temporary data",
@@ -2463,7 +2478,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4048,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4195,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 27b94fb611..04c2eccd1c 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -871,6 +871,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -883,7 +884,9 @@ BeginDirectModify(ForeignScanState *node,
      the table to modify is accessible through the
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
+     information provided by <function>PlanDirectModify</function>).  In
+     addition, <literal>rinfo</literal> also contains information describing
+     the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -895,6 +898,12 @@ BeginDirectModify(ForeignScanState *node,
      for <function>ExplainDirectModify</function> and <function>EndDirectModify</function>.
     </para>
 
+    <note>
+     Also note that it's a good idea to store the <literal>rinfo</literal>
+     in the <structfield>fdw_state</structfield> for
+     <function>IterateDirectModify</function> to use.
+    </node>
+
     <para>
      If the <function>BeginDirectModify</function> pointer is set to
      <literal>NULL</literal>, no attempts to execute a direct modification on the
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..9824c16e09 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -223,10 +223,17 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	/*
 	 * Tell the FDW to initialize the scan.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		/* Perform initializations for a direct modification. */
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index a2617c7cfd..e981298a75 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e6ce8e2110..80eedc4a24 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 764e3bb90c..92cc90c0f0 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1983,6 +1983,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index f2325694c5..ff78167f79 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5455,6 +5455,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* this might be filled to a >= 0 value by set_plan_refs() */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 329ebd5f28..0bdfc50207 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -1225,6 +1225,14 @@ set_foreignscan_references(PlannerInfo *root,
 			tempset = bms_add_member(tempset, x + rtoffset);
 		fscan->fs_relids = tempset;
 	}
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 8e6594e355..047dd73dd1 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -616,6 +616,10 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* For non-SELECT operations, this contains
+								 * the offset of result relation in a
+								 * query-global list of result relations; -1
+								 * otherwise */
 } ForeignScan;
 
 /* ----------------

From b367546b5ad0a89bb0d11c95ba57bfab2ce3fdef Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Wed, 7 Aug 2019 09:46:27 +0900
Subject: [PATCH v5 1/4] Revise BeginDirectModify API to pass ResultRelInfo
 directly

For ExecInitForeignScan() to efficiently get the ResultRelInfo to
pass to BeginDirectModify(), add a field to ForeignScan that gives
the index of a given result relation in the query's list of result
relations.
---
 contrib/postgres_fdw/postgres_fdw.c     | 27 +++++++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            |  5 ++++-
 src/backend/executor/nodeForeignscan.c  | 13 ++++++++++---
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 15 +++++++++++++++
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  4 ++++
 10 files changed, 60 insertions(+), 10 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 06a205877d..a61c2d8b5f 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -205,6 +205,9 @@ typedef struct PgFdwDirectModifyState
 	List	   *retrieved_attrs;	/* attr numbers retrieved by RETURNING */
 	bool		set_processed;	/* do we set the command es_processed? */
 
+	/* Information about the relation being modified */
+	ResultRelInfo *resultRelInfo;
+
 	/* for remote query execution */
 	PGconn	   *conn;			/* connection for the update */
 	int			numParams;		/* number of parameters passed to query */
@@ -360,7 +363,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2331,6 +2336,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2340,7 +2350,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2368,7 +2380,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2414,6 +2426,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	dmstate->set_processed = intVal(list_nth(fsplan->fdw_private,
 											 FdwDirectModifyPrivateSetProcessed));
 
+	/* Save the ResultRelInfo of the relation being modified. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Create context for per-tuple temp workspace. */
 	dmstate->temp_cxt = AllocSetContextCreate(estate->es_query_cxt,
 											  "postgres_fdw temporary data",
@@ -2463,7 +2478,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4048,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4195,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 27b94fb611..123b02958a 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -871,6 +871,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -883,7 +884,9 @@ BeginDirectModify(ForeignScanState *node,
      the table to modify is accessible through the
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
+     information provided by <function>PlanDirectModify</function>).  In
+     addition, <literal>rinfo</literal> also contains information describing
+     the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..9824c16e09 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -223,10 +223,17 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	/*
 	 * Tell the FDW to initialize the scan.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		/* Perform initializations for a direct modification. */
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index a2617c7cfd..e981298a75 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e6ce8e2110..80eedc4a24 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 764e3bb90c..92cc90c0f0 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1983,6 +1983,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index f2325694c5..ff78167f79 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5455,6 +5455,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* this might be filled to a >= 0 value by set_plan_refs() */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 329ebd5f28..f18e94a879 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -877,6 +877,13 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
+				/*
+				 * Caution: Do not change the relative ordering of this loop
+				 * and the statement below that adds the result relations to
+				 * root->glob->resultRelations, because we need to use the
+				 * current value of list_length(root->glob->resultRelations)
+				 * in some plans.
+				 */
 				foreach(l, splan->plans)
 				{
 					lfirst(l) = set_plan_refs(root,
@@ -1225,6 +1232,14 @@ set_foreignscan_references(PlannerInfo *root,
 			tempset = bms_add_member(tempset, x + rtoffset);
 		fscan->fs_relids = tempset;
 	}
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 8e6594e355..dc2061cf89 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -616,6 +616,10 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* For "direct modification" operations, this
+								 * contains the offset of result relation in
+								 * query-global list of result relations; -1
+								 * otherwise */
 } ForeignScan;
 
 /* ----------------
-- 
2.11.0

From 7d51c519aa3b1232687acc79c00b89e5346181b2 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v5 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 3316c089e9..cbf3de6267 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -621,31 +620,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -711,7 +709,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -956,32 +953,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1028,6 +1023,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1181,119 +1323,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.11.0

From 648438dfe8223ade371ca6e81e6da5162ac673b1 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v5 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 ++---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  21 +++-
 src/backend/executor/nodeModifyTable.c | 209 +++++++--------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 87 insertions(+), 212 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 967dba6fcf..5e7153bb8d 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3113,32 +3113,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 2d9a8e9d54..a8faa5e1e4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4644,9 +4645,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5750,14 +5749,24 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 729dc396a9..743f54926a 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -935,10 +935,23 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot),
-								   gettext_noop("could not convert row type"));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot),
+									   gettext_noop("could not convert row type"));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index cbf3de6267..d327153a6a 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -339,10 +336,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1053,9 +1046,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1131,41 +1122,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1872,28 +1838,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1917,6 +1861,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1931,37 +1876,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -1977,59 +1902,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc,
-								   gettext_noop("could not convert row type"));
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2125,17 +1997,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2304,6 +2165,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2326,8 +2188,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2337,6 +2204,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2402,6 +2276,21 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc),
+									   gettext_noop("could not convert row type"));
 		resultRelInfo++;
 		i++;
 	}
@@ -2426,26 +2315,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a46feeedb0..bb080980c0 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -65,14 +65,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 0b40d13bfa..9571bbe328 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -485,6 +485,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1136,9 +1142,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
2.11.0

From 903b9d66e992de889a1247c20581b4ca511d17dc Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v5 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   4 -
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 188 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  26 +++--
 src/include/executor/executor.h          |  19 +++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   1 -
 11 files changed, 128 insertions(+), 168 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 3aeef30b28..967dba6fcf 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2444,9 +2444,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2479,7 +2476,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2844,7 +2842,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3116,11 +3113,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3224,7 +3216,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3295,7 +3287,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index fb2be10794..e63b25bf25 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1746,7 +1746,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1876,7 +1875,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index dbd7dd9bcd..2e8802df07 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -858,9 +858,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -904,7 +901,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index c1fc0d54e9..eaaf69bb93 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,8 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9e0c8794c4..3316c089e9 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +409,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -428,7 +445,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -490,8 +507,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -551,7 +568,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -590,7 +608,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -676,6 +695,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -687,7 +707,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -697,10 +716,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1037,6 +1052,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1045,12 +1061,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1060,10 +1074,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1090,7 +1100,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1127,7 +1137,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1177,6 +1187,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1202,9 +1213,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1245,16 +1259,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1271,18 +1275,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1448,7 +1452,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1687,7 +1692,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1844,41 +1849,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -1989,10 +1990,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2041,17 +2040,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2084,7 +2072,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2129,7 +2116,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2212,25 +2198,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2242,15 +2224,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2271,7 +2247,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2314,14 +2289,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2363,7 +2332,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2387,8 +2355,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 43edfef089..10ef6af3e7 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -193,7 +193,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -567,6 +566,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -590,6 +590,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -603,13 +604,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -664,6 +665,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -697,6 +699,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -705,7 +708,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -747,7 +750,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -763,7 +767,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -786,6 +790,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -816,6 +821,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -824,7 +830,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -852,7 +858,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -864,7 +870,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 1fb28b4596..3ecdcc3a34 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 4ec78491f6..0b40d13bfa 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,7 +519,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
-- 
2.11.0

From a85cc6820a7193c4398e6337ed66943f4ac63837 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Wed, 7 Aug 2019 09:46:27 +0900
Subject: [PATCH v6 1/4] Revise BeginDirectModify API to pass ResultRelInfo
 directly

For ExecInitForeignScan() to efficiently get the ResultRelInfo to
pass to BeginDirectModify(), add a field to ForeignScan that gives
the index of a given result relation in the query's list of result
relations.
---
 contrib/postgres_fdw/postgres_fdw.c     | 27 +++++++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            | 11 +++++++----
 src/backend/executor/nodeForeignscan.c  | 13 ++++++++++---
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 15 +++++++++++++++
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  4 ++++
 10 files changed, 63 insertions(+), 13 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 06a205877d..a61c2d8b5f 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -205,6 +205,9 @@ typedef struct PgFdwDirectModifyState
 	List	   *retrieved_attrs;	/* attr numbers retrieved by RETURNING */
 	bool		set_processed;	/* do we set the command es_processed? */
 
+	/* Information about the relation being modified */
+	ResultRelInfo *resultRelInfo;
+
 	/* for remote query execution */
 	PGconn	   *conn;			/* connection for the update */
 	int			numParams;		/* number of parameters passed to query */
@@ -360,7 +363,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2331,6 +2336,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2340,7 +2350,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2368,7 +2380,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2414,6 +2426,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	dmstate->set_processed = intVal(list_nth(fsplan->fdw_private,
 											 FdwDirectModifyPrivateSetProcessed));
 
+	/* Save the ResultRelInfo of the relation being modified. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Create context for per-tuple temp workspace. */
 	dmstate->temp_cxt = AllocSetContextCreate(estate->es_query_cxt,
 											  "postgres_fdw temporary data",
@@ -2463,7 +2478,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4048,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4195,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 27b94fb611..919b88a945 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -871,6 +871,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -883,7 +884,9 @@ BeginDirectModify(ForeignScanState *node,
      the table to modify is accessible through the
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
+     information provided by <function>PlanDirectModify</function>).  In
+     addition, <literal>rinfo</literal> also contains information describing
+     the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -915,9 +918,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
-     Return NULL if no more rows are available.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the result relation's <structname>ResultRelInfo</structname> that has
+     been made available via node.  Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
      <function>BeginDirectModify</function> if you need longer-lived storage, or use
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..9824c16e09 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -223,10 +223,17 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	/*
 	 * Tell the FDW to initialize the scan.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		/* Perform initializations for a direct modification. */
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index a2617c7cfd..e981298a75 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e6ce8e2110..80eedc4a24 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 764e3bb90c..92cc90c0f0 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1983,6 +1983,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index f2325694c5..ff78167f79 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5455,6 +5455,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* this might be filled to a >= 0 value by set_plan_refs() */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 329ebd5f28..f18e94a879 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -877,6 +877,13 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
+				/*
+				 * Caution: Do not change the relative ordering of this loop
+				 * and the statement below that adds the result relations to
+				 * root->glob->resultRelations, because we need to use the
+				 * current value of list_length(root->glob->resultRelations)
+				 * in some plans.
+				 */
 				foreach(l, splan->plans)
 				{
 					lfirst(l) = set_plan_refs(root,
@@ -1225,6 +1232,14 @@ set_foreignscan_references(PlannerInfo *root,
 			tempset = bms_add_member(tempset, x + rtoffset);
 		fscan->fs_relids = tempset;
 	}
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 8e6594e355..dc2061cf89 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -616,6 +616,10 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* For "direct modification" operations, this
+								 * contains the offset of result relation in
+								 * query-global list of result relations; -1
+								 * otherwise */
 } ForeignScan;
 
 /* ----------------
-- 
2.11.0

From 6cb135479a53bd88d3728193e35ec2cc8788a9ab Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v6 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   5 -
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 188 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  26 +++--
 src/include/executor/executor.h          |  19 +++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 130 insertions(+), 175 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 3aeef30b28..967dba6fcf 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2444,9 +2444,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2479,7 +2476,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2844,7 +2842,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3116,11 +3113,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3224,7 +3216,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3295,7 +3287,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index fb2be10794..e63b25bf25 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1746,7 +1746,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1876,7 +1875,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index dbd7dd9bcd..d6c3a94522 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -858,9 +858,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -904,7 +901,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
@@ -2822,7 +2818,6 @@ EvalPlanQualStart(EPQState *epqstate, EState *parentestate, Plan *planTree)
 			estate->es_num_root_result_relations = numRootResultRels;
 		}
 	}
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	estate->es_top_eflags = parentestate->es_top_eflags;
 	estate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index c1fc0d54e9..eaaf69bb93 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,8 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9e0c8794c4..3316c089e9 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +409,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -428,7 +445,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -490,8 +507,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -551,7 +568,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -590,7 +608,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -676,6 +695,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -687,7 +707,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -697,10 +716,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1037,6 +1052,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1045,12 +1061,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1060,10 +1074,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1090,7 +1100,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1127,7 +1137,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1177,6 +1187,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1202,9 +1213,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1245,16 +1259,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1271,18 +1275,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1448,7 +1452,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1687,7 +1692,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1844,41 +1849,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -1989,10 +1990,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2041,17 +2040,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2084,7 +2072,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2129,7 +2116,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2212,25 +2198,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2242,15 +2224,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2271,7 +2247,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2314,14 +2289,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2363,7 +2332,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2387,8 +2355,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 43edfef089..10ef6af3e7 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -193,7 +193,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -567,6 +566,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -590,6 +590,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -603,13 +604,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -664,6 +665,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -697,6 +699,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -705,7 +708,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -747,7 +750,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -763,7 +767,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -786,6 +790,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -816,6 +821,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -824,7 +830,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -852,7 +858,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -864,7 +870,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 1fb28b4596..3ecdcc3a34 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 4ec78491f6..0b40d13bfa 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,7 +519,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index 75e25cdf48..b73e26fc69 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index 23885f638c..e5a0a05d13 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
2.11.0

From 2afa538badeb6beb0c87b91c23f8de494b71e61f Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v6 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 ++---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  21 +++-
 src/backend/executor/nodeModifyTable.c | 209 +++++++--------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 87 insertions(+), 212 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 967dba6fcf..5e7153bb8d 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3113,32 +3113,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 2d9a8e9d54..a8faa5e1e4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4644,9 +4645,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5750,14 +5749,24 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 729dc396a9..743f54926a 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -935,10 +935,23 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot),
-								   gettext_noop("could not convert row type"));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot),
+									   gettext_noop("could not convert row type"));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index cbf3de6267..d327153a6a 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -339,10 +336,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1053,9 +1046,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1131,41 +1122,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1872,28 +1838,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1917,6 +1861,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1931,37 +1876,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -1977,59 +1902,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc,
-								   gettext_noop("could not convert row type"));
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2125,17 +1997,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2304,6 +2165,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2326,8 +2188,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2337,6 +2204,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2402,6 +2276,21 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc),
+									   gettext_noop("could not convert row type"));
 		resultRelInfo++;
 		i++;
 	}
@@ -2426,26 +2315,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a46feeedb0..bb080980c0 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -65,14 +65,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 0b40d13bfa..9571bbe328 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -485,6 +485,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1136,9 +1142,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
2.11.0

From a9e75e5c00d8cae7785f77c34b92461c8e029690 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v6 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 3316c089e9..cbf3de6267 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -621,31 +620,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -711,7 +709,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -956,32 +953,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1028,6 +1023,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1181,119 +1323,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.11.0

From 62ed7026523f19af95595e3cdbe51720b1bb266d Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Wed, 7 Aug 2019 09:46:27 +0900
Subject: [PATCH v7 1/4] Revise BeginDirectModify API to pass ResultRelInfo
 directly

For ExecInitForeignScan() to efficiently get the ResultRelInfo to
pass to BeginDirectModify(), add a field to ForeignScan that gives
the index of a given result relation in the query's list of result
relations.
---
 contrib/postgres_fdw/postgres_fdw.c     | 27 +++++++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            | 12 ++++++++----
 src/backend/executor/nodeForeignscan.c  | 13 ++++++++++---
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 15 +++++++++++++++
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  4 ++++
 10 files changed, 64 insertions(+), 13 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 06a205877d..a61c2d8b5f 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -205,6 +205,9 @@ typedef struct PgFdwDirectModifyState
 	List	   *retrieved_attrs;	/* attr numbers retrieved by RETURNING */
 	bool		set_processed;	/* do we set the command es_processed? */
 
+	/* Information about the relation being modified */
+	ResultRelInfo *resultRelInfo;
+
 	/* for remote query execution */
 	PGconn	   *conn;			/* connection for the update */
 	int			numParams;		/* number of parameters passed to query */
@@ -360,7 +363,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2331,6 +2336,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2340,7 +2350,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2368,7 +2380,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2414,6 +2426,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	dmstate->set_processed = intVal(list_nth(fsplan->fdw_private,
 											 FdwDirectModifyPrivateSetProcessed));
 
+	/* Save the ResultRelInfo of the relation being modified. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Create context for per-tuple temp workspace. */
 	dmstate->temp_cxt = AllocSetContextCreate(estate->es_query_cxt,
 											  "postgres_fdw temporary data",
@@ -2463,7 +2478,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4048,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4195,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 27b94fb611..907ce46ee6 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -871,6 +871,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -883,7 +884,9 @@ BeginDirectModify(ForeignScanState *node,
      the table to modify is accessible through the
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
+     information provided by <function>PlanDirectModify</function>).  In
+     addition, <literal>rinfo</literal> also contains information describing
+     the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -915,9 +918,10 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
-     Return NULL if no more rows are available.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     passed to <function>BeginDirectModify</function>.  Return NULL if no more
+     rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
      <function>BeginDirectModify</function> if you need longer-lived storage, or use
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..9824c16e09 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -223,10 +223,17 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	/*
 	 * Tell the FDW to initialize the scan.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		/* Perform initializations for a direct modification. */
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index a2617c7cfd..e981298a75 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e6ce8e2110..80eedc4a24 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 764e3bb90c..92cc90c0f0 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1983,6 +1983,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index f2325694c5..ff78167f79 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5455,6 +5455,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* this might be filled to a >= 0 value by set_plan_refs() */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 329ebd5f28..f18e94a879 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -877,6 +877,13 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
+				/*
+				 * Caution: Do not change the relative ordering of this loop
+				 * and the statement below that adds the result relations to
+				 * root->glob->resultRelations, because we need to use the
+				 * current value of list_length(root->glob->resultRelations)
+				 * in some plans.
+				 */
 				foreach(l, splan->plans)
 				{
 					lfirst(l) = set_plan_refs(root,
@@ -1225,6 +1232,14 @@ set_foreignscan_references(PlannerInfo *root,
 			tempset = bms_add_member(tempset, x + rtoffset);
 		fscan->fs_relids = tempset;
 	}
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 8e6594e355..dc2061cf89 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -616,6 +616,10 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* For "direct modification" operations, this
+								 * contains the offset of result relation in
+								 * query-global list of result relations; -1
+								 * otherwise */
 } ForeignScan;
 
 /* ----------------
-- 
2.11.0

From 0485b1889cb4c4f1ef4097223f8f26c5e8c349b0 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v7 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   5 -
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 188 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  26 +++--
 src/include/executor/executor.h          |  19 +++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 130 insertions(+), 175 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 3aeef30b28..967dba6fcf 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2444,9 +2444,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2479,7 +2476,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2844,7 +2842,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3116,11 +3113,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3224,7 +3216,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3295,7 +3287,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index fb2be10794..e63b25bf25 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1746,7 +1746,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1876,7 +1875,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index dbd7dd9bcd..d6c3a94522 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -858,9 +858,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -904,7 +901,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
@@ -2822,7 +2818,6 @@ EvalPlanQualStart(EPQState *epqstate, EState *parentestate, Plan *planTree)
 			estate->es_num_root_result_relations = numRootResultRels;
 		}
 	}
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	estate->es_top_eflags = parentestate->es_top_eflags;
 	estate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index c1fc0d54e9..eaaf69bb93 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,8 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9e0c8794c4..3316c089e9 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +409,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -428,7 +445,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -490,8 +507,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -551,7 +568,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -590,7 +608,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -676,6 +695,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -687,7 +707,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -697,10 +716,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1037,6 +1052,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1045,12 +1061,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1060,10 +1074,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1090,7 +1100,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1127,7 +1137,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1177,6 +1187,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1202,9 +1213,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1245,16 +1259,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1271,18 +1275,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1448,7 +1452,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1687,7 +1692,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1844,41 +1849,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -1989,10 +1990,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2041,17 +2040,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2084,7 +2072,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2129,7 +2116,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2212,25 +2198,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2242,15 +2224,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2271,7 +2247,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2314,14 +2289,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2363,7 +2332,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2387,8 +2355,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 43edfef089..10ef6af3e7 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -193,7 +193,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -567,6 +566,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -590,6 +590,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -603,13 +604,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -664,6 +665,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -697,6 +699,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -705,7 +708,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -747,7 +750,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -763,7 +767,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -786,6 +790,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -816,6 +821,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -824,7 +830,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -852,7 +858,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -864,7 +870,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 1fb28b4596..3ecdcc3a34 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 4ec78491f6..0b40d13bfa 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,7 +519,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index 75e25cdf48..b73e26fc69 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index 23885f638c..e5a0a05d13 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
2.11.0

From 1f05ac37b37f98e05a5cd468e2718ec4ab684c95 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v7 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 ++---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  21 +++-
 src/backend/executor/nodeModifyTable.c | 209 +++++++--------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 87 insertions(+), 212 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 967dba6fcf..5e7153bb8d 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3113,32 +3113,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 2d9a8e9d54..a8faa5e1e4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4644,9 +4645,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5750,14 +5749,24 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 729dc396a9..743f54926a 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -935,10 +935,23 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot),
-								   gettext_noop("could not convert row type"));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot),
+									   gettext_noop("could not convert row type"));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index cbf3de6267..d327153a6a 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -339,10 +336,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1053,9 +1046,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1131,41 +1122,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1872,28 +1838,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1917,6 +1861,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1931,37 +1876,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -1977,59 +1902,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc,
-								   gettext_noop("could not convert row type"));
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2125,17 +1997,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2304,6 +2165,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2326,8 +2188,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2337,6 +2204,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2402,6 +2276,21 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc),
+									   gettext_noop("could not convert row type"));
 		resultRelInfo++;
 		i++;
 	}
@@ -2426,26 +2315,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a46feeedb0..bb080980c0 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -65,14 +65,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 0b40d13bfa..9571bbe328 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -485,6 +485,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1136,9 +1142,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
2.11.0

From c60bad85be3ace4b84f6df185385de4c849b7c5d Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v7 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 3316c089e9..cbf3de6267 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -621,31 +620,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -711,7 +709,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -956,32 +953,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1028,6 +1023,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1181,119 +1323,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.11.0

From c11220d6376aafc8695155b209560a0ba1a5142c Mon Sep 17 00:00:00 2001
From: Etsuro Fujita <efujita@postgresql.org>
Date: Thu, 8 Aug 2019 21:41:12 +0900
Subject: [PATCH] Remove dependency on estate->es_result_relation_info from FDW
 APIs.

FDW APIs for executing a foreign table direct modification assumed that
the FDW would obtain the target foreign table's ResultRelInfo from
estate->es_result_relation_info of the passed-in ForeignScanState node,
but the upcoming patch(es) to refactor partitioning-related code in
nodeModifyTable.c will remove the es_result_relation_info variable.
Revise BeginDirectModify()'s API to pass the ResultRelInfo explicitly, to
remove the dependency on that variable from the FDW APIs.  For
ExecInitForeignScan() to efficiently get the ResultRelInfo to pass to
BeginDirectModify(), add a field to ForeignScan that gives the index of
the target foreign table in the list of the query result relations.

Patch by Amit Langote, following a proposal by Andres Freund, reviewed by
Andres Freund and me

Discussion: https://postgr.es/m/20190718010911.l6xcdv6birtxiei4@alap3.anarazel.de
---
 contrib/postgres_fdw/postgres_fdw.c     | 25 +++++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            |  8 ++++++--
 src/backend/executor/nodeForeignscan.c  | 14 ++++++++++----
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 15 +++++++++++++++
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  3 +++
 10 files changed, 59 insertions(+), 12 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 06a205877d..56b4b03cb0 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -217,6 +217,7 @@ typedef struct PgFdwDirectModifyState
 	int			num_tuples;		/* # of result tuples */
 	int			next_tuple;		/* index of next one to return */
 	Relation	resultRel;		/* relcache entry for the target relation */
+	ResultRelInfo *resultRelInfo;	/* ResultRelInfo for the target relation */
 	AttrNumber *attnoMap;		/* array of attnums of input user columns */
 	AttrNumber	ctidAttno;		/* attnum of input ctid column */
 	AttrNumber	oidAttno;		/* attnum of input oid column */
@@ -360,7 +361,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2331,6 +2334,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2340,7 +2348,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2368,7 +2378,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2401,6 +2411,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 		dmstate->rel = NULL;
 	}
 
+	/* Save the ResultRelInfo for the target relation. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Initialize state variable */
 	dmstate->num_tuples = -1;	/* -1 means not set yet */
 
@@ -2463,7 +2476,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4046,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4193,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 27b94fb611..cf57957ae3 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -871,6 +871,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -884,6 +885,8 @@ BeginDirectModify(ForeignScanState *node,
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
      information provided by <function>PlanDirectModify</function>).
+     In addition, the <structname>ResultRelInfo</structname> struct also
+     contains information about the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -915,8 +918,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     passed to <function>BeginDirectModify</function>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..84ef31ceef 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -221,12 +221,18 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan or the direct modification.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index a2617c7cfd..e981298a75 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e6ce8e2110..80eedc4a24 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 764e3bb90c..92cc90c0f0 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1983,6 +1983,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index f2325694c5..8f228f2b7e 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5455,6 +5455,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* resultRelIndex will be set by PlanDirectModify/setrefs.c, if needed */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 329ebd5f28..f18e94a879 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -877,6 +877,13 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
+				/*
+				 * Caution: Do not change the relative ordering of this loop
+				 * and the statement below that adds the result relations to
+				 * root->glob->resultRelations, because we need to use the
+				 * current value of list_length(root->glob->resultRelations)
+				 * in some plans.
+				 */
 				foreach(l, splan->plans)
 				{
 					lfirst(l) = set_plan_refs(root,
@@ -1225,6 +1232,14 @@ set_foreignscan_references(PlannerInfo *root,
 			tempset = bms_add_member(tempset, x + rtoffset);
 		fscan->fs_relids = tempset;
 	}
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 8e6594e355..fac1e6098a 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -616,6 +616,9 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* index of foreign table in the list of query
+								 * result relations for INSERT/UPDATE/DELETE;
+								 * -1 for SELECT */
 } ForeignScan;
 
 /* ----------------
-- 
2.19.2

From 50e876daa386484889e30e3f754d89f56c66f867 Mon Sep 17 00:00:00 2001
From: Etsuro Fujita <efujita@postgresql.org>
Date: Thu, 8 Aug 2019 21:41:12 +0900
Subject: [PATCH v8 1/4] Remove dependency on estate->es_result_relation_info
 from FDW APIs.

FDW APIs for executing a foreign table direct modification assumed that
the FDW would obtain the target foreign table's ResultRelInfo from
estate->es_result_relation_info of the passed-in ForeignScanState node,
but the upcoming patch(es) to refactor partitioning-related code in
nodeModifyTable.c will remove the es_result_relation_info variable.
Revise BeginDirectModify()'s API to pass the ResultRelInfo explicitly, to
remove the dependency on that variable from the FDW APIs.  For
ExecInitForeignScan() to efficiently get the ResultRelInfo to pass to
BeginDirectModify(), add a field to ForeignScan that gives the index of
the target foreign table in the list of the query result relations.

Patch by Amit Langote, following a proposal by Andres Freund, reviewed by
Andres Freund and me

Discussion: https://postgr.es/m/20190718010911.l6xcdv6birtxiei4@alap3.anarazel.de
---
 contrib/postgres_fdw/postgres_fdw.c     | 25 +++++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            |  8 ++++++--
 src/backend/executor/nodeForeignscan.c  | 14 ++++++++++----
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 15 +++++++++++++++
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  3 +++
 10 files changed, 59 insertions(+), 12 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 82d8140ba2..7ee783ae47 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -217,6 +217,7 @@ typedef struct PgFdwDirectModifyState
 	int			num_tuples;		/* # of result tuples */
 	int			next_tuple;		/* index of next one to return */
 	Relation	resultRel;		/* relcache entry for the target relation */
+	ResultRelInfo *resultRelInfo;	/* ResultRelInfo for the target relation */
 	AttrNumber *attnoMap;		/* array of attnums of input user columns */
 	AttrNumber	ctidAttno;		/* attnum of input ctid column */
 	AttrNumber	oidAttno;		/* attnum of input oid column */
@@ -360,7 +361,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2331,6 +2334,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2340,7 +2348,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2368,7 +2378,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2401,6 +2411,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 		dmstate->rel = NULL;
 	}
 
+	/* Save the ResultRelInfo for the target relation. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Initialize state variable */
 	dmstate->num_tuples = -1;	/* -1 means not set yet */
 
@@ -2463,7 +2476,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4033,7 +4046,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4180,7 +4193,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 6587678af2..0aff958415 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -873,6 +873,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -886,6 +887,8 @@ BeginDirectModify(ForeignScanState *node,
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
      information provided by <function>PlanDirectModify</function>).
+     In addition, the <structname>ResultRelInfo</structname> struct also
+     contains information about the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -917,8 +920,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     passed to <function>BeginDirectModify</function>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..84ef31ceef 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -221,12 +221,18 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan or the direct modification.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 3432bb921d..5f4cf0528a 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index b0dcd02ff6..c7d63522c2 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 764e3bb90c..92cc90c0f0 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1983,6 +1983,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 0c036209f0..756a04de79 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5455,6 +5455,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* resultRelIndex will be set by PlanDirectModify/setrefs.c, if needed */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 566ee96da8..8799c8f572 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -877,6 +877,13 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
+				/*
+				 * Caution: Do not change the relative ordering of this loop
+				 * and the statement below that adds the result relations to
+				 * root->glob->resultRelations, because we need to use the
+				 * current value of list_length(root->glob->resultRelations)
+				 * in some plans.
+				 */
 				foreach(l, splan->plans)
 				{
 					lfirst(l) = set_plan_refs(root,
@@ -1225,6 +1232,14 @@ set_foreignscan_references(PlannerInfo *root,
 			tempset = bms_add_member(tempset, x + rtoffset);
 		fscan->fs_relids = tempset;
 	}
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 8e6594e355..fac1e6098a 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -616,6 +616,9 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* index of foreign table in the list of query
+								 * result relations for INSERT/UPDATE/DELETE;
+								 * -1 for SELECT */
 } ForeignScan;
 
 /* ----------------
-- 
2.11.0

From f40fa0689657947a9d93e99c37041481fab11126 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v8 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 ++---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  19 ++-
 src/backend/executor/nodeModifyTable.c | 207 +++++++--------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 85 insertions(+), 210 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 967dba6fcf..5e7153bb8d 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3113,32 +3113,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index cdb1105b4a..17f60bfcd4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4643,9 +4644,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5749,14 +5748,24 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index d23f292cb0..d7ed3fcdbe 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -931,9 +931,22 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 956bab09af..1b89ed5c31 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -339,10 +336,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1052,9 +1045,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1130,41 +1121,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1869,28 +1835,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1914,6 +1858,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1928,37 +1873,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -1974,58 +1899,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc);
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2121,17 +1994,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2300,6 +2162,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2322,8 +2185,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2333,6 +2201,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2398,6 +2273,20 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc));
 		resultRelInfo++;
 		i++;
 	}
@@ -2422,26 +2311,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a46feeedb0..bb080980c0 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -65,14 +65,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 760e580281..175e4fb96f 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -486,6 +486,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1187,9 +1193,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
2.11.0

From 04eb853f2726e8bfbf1b97b784dff438e09bfbfe Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v8 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   5 -
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 188 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  26 +++--
 src/include/executor/executor.h          |  19 +++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 130 insertions(+), 175 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 3aeef30b28..967dba6fcf 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2444,9 +2444,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2479,7 +2476,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2844,7 +2842,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3116,11 +3113,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3224,7 +3216,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3295,7 +3287,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 05593f3316..706f3a258e 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1745,7 +1745,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1875,7 +1874,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index ea4b586984..d3a8302010 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -857,9 +857,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -903,7 +900,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
@@ -2823,7 +2819,6 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 			rcestate->es_num_root_result_relations = numRootResultRels;
 		}
 	}
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index ee0239b146..06bd715096 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -124,8 +124,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index c9d024ead5..57b774335b 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +409,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -428,7 +445,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -490,8 +507,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -551,7 +568,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -590,7 +608,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -676,6 +695,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -687,7 +707,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -697,10 +716,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1036,6 +1051,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1044,12 +1060,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1059,10 +1073,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1089,7 +1099,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1126,7 +1136,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1176,6 +1186,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1201,9 +1212,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1244,16 +1258,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1270,18 +1274,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1445,7 +1449,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1684,7 +1689,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1841,41 +1846,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -1985,10 +1986,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2037,17 +2036,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2080,7 +2068,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2125,7 +2112,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2208,25 +2194,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2238,15 +2220,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2267,7 +2243,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2310,14 +2285,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2359,7 +2328,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2383,8 +2351,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 11e6331f49..9a11cea7ce 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -193,7 +193,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -567,6 +566,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -590,6 +590,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -603,13 +604,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -664,6 +665,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -697,6 +699,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -705,7 +708,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -747,7 +750,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -763,7 +767,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -786,6 +790,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -816,6 +821,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -824,7 +830,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -852,7 +858,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -864,7 +870,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 6298c7c8ca..a12d02cf5e 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 44f76082e9..760e580281 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,7 +519,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index 75e25cdf48..b73e26fc69 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index 23885f638c..e5a0a05d13 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
2.11.0

From 2df9db3576f774b5653758d692d8cbc4e34efa31 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v8 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 57b774335b..956bab09af 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -621,31 +620,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -711,7 +709,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -955,32 +952,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1027,6 +1022,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1180,119 +1322,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.11.0

From a0f6939e5b2ee8f78c813545f45d59a346d22e5f Mon Sep 17 00:00:00 2001
From: Etsuro Fujita <efujita@postgresql.org>
Date: Thu, 8 Aug 2019 21:41:12 +0900
Subject: [PATCH v9 1/4] Remove dependency on estate->es_result_relation_info
 from FDW APIs.

FDW APIs for executing a foreign table direct modification assumed that
the FDW would obtain the target foreign table's ResultRelInfo from
estate->es_result_relation_info of the passed-in ForeignScanState node,
but the upcoming patch(es) to refactor partitioning-related code in
nodeModifyTable.c will remove the es_result_relation_info variable.
Revise BeginDirectModify()'s API to pass the ResultRelInfo explicitly, to
remove the dependency on that variable from the FDW APIs.  For
ExecInitForeignScan() to efficiently get the ResultRelInfo to pass to
BeginDirectModify(), add a field to ForeignScan that gives the index of
the target foreign table in the list of the query result relations.

Patch by Amit Langote, following a proposal by Andres Freund, reviewed by
Andres Freund and me

Discussion: https://postgr.es/m/20190718010911.l6xcdv6birtxiei4@alap3.anarazel.de
---
 contrib/postgres_fdw/postgres_fdw.c     | 25 +++++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            |  8 ++++++--
 src/backend/executor/nodeForeignscan.c  | 14 ++++++++++----
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 15 +++++++++++++++
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  3 +++
 10 files changed, 59 insertions(+), 12 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index bdc21b36d1..642deaf7cb 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -218,6 +218,7 @@ typedef struct PgFdwDirectModifyState
 	int			num_tuples;		/* # of result tuples */
 	int			next_tuple;		/* index of next one to return */
 	Relation	resultRel;		/* relcache entry for the target relation */
+	ResultRelInfo *resultRelInfo;	/* ResultRelInfo for the target relation */
 	AttrNumber *attnoMap;		/* array of attnums of input user columns */
 	AttrNumber	ctidAttno;		/* attnum of input ctid column */
 	AttrNumber	oidAttno;		/* attnum of input oid column */
@@ -361,7 +362,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2319,6 +2322,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2328,7 +2336,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2356,7 +2366,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2389,6 +2399,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 		dmstate->rel = NULL;
 	}
 
+	/* Save the ResultRelInfo for the target relation. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Initialize state variable */
 	dmstate->num_tuples = -1;	/* -1 means not set yet */
 
@@ -2451,7 +2464,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4087,7 +4100,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4234,7 +4247,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 6587678af2..0aff958415 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -873,6 +873,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -886,6 +887,8 @@ BeginDirectModify(ForeignScanState *node,
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
      information provided by <function>PlanDirectModify</function>).
+     In addition, the <structname>ResultRelInfo</structname> struct also
+     contains information about the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -917,8 +920,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     passed to <function>BeginDirectModify</function>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 52af1dac5c..84ef31ceef 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -221,12 +221,18 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan or the direct modification.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index a9b8b84b8f..a1094defb7 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -761,6 +761,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index ac02e5ec8d..91af796a00 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -698,6 +698,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 3f9ebc9044..d5514398f4 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -2013,6 +2013,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 8c8b4f8ed6..455740663b 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5460,6 +5460,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* resultRelIndex will be set by PlanDirectModify/setrefs.c, if needed */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index f581c5f532..c16282b082 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -901,6 +901,13 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
+				/*
+				 * Caution: Do not change the relative ordering of this loop
+				 * and the statement below that adds the result relations to
+				 * root->glob->resultRelations, because we need to use the
+				 * current value of list_length(root->glob->resultRelations)
+				 * in some plans.
+				 */
 				foreach(l, splan->plans)
 				{
 					lfirst(l) = set_plan_refs(root,
@@ -1240,6 +1247,14 @@ set_foreignscan_references(PlannerInfo *root,
 	}
 
 	fscan->fs_relids = offset_relid_set(fscan->fs_relids, rtoffset);
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 822686033e..adf39bc618 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 477b4da192..f3f2699e90 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -620,6 +620,9 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* index of foreign table in the list of query
+								 * result relations for INSERT/UPDATE/DELETE;
+								 * -1 for SELECT */
 } ForeignScan;
 
 /* ----------------
-- 
2.16.5

From 82a386601239c90da2a4bba69762df28351053cb Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v9 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 ++---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  19 ++-
 src/backend/executor/nodeModifyTable.c | 207 +++++++--------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 85 insertions(+), 210 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 38cffde583..61d9a1d3cf 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3110,32 +3110,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index faeea16d21..191f1418fe 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4652,9 +4653,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5748,13 +5747,23 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
+		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
 		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index d23f292cb0..d7ed3fcdbe 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -931,9 +931,22 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 1362b2f2d1..9de5a5371c 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -339,10 +336,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1051,9 +1044,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1129,41 +1120,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1868,28 +1834,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1913,6 +1857,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1927,37 +1872,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -1973,58 +1898,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc);
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2120,17 +1993,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2299,6 +2161,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2321,8 +2184,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2331,6 +2199,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	EvalPlanQualInit(&mtstate->mt_epqstate, estate, NULL, NIL, node->epqParam);
 	mtstate->fireBSTriggers = true;
 
+	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
 	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
@@ -2397,6 +2272,20 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc));
 		resultRelInfo++;
 		i++;
 	}
@@ -2421,26 +2310,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a46feeedb0..bb080980c0 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -64,14 +64,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_update_new_table;
 	bool		tcs_insert_new_table;
 
-	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
 	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 62e20fdfdc..62eed595ca 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -485,6 +485,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1185,9 +1191,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
2.16.5

From d3a17640835035e95c56859b4eed652444ec5b86 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v9 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  17 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   5 -
 src/backend/executor/execReplication.c   |  22 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 188 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  26 +++--
 src/include/executor/executor.h          |  19 +++-
 src/include/executor/nodeModifyTable.h   |   3 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 130 insertions(+), 175 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 42a147b67d..38cffde583 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2441,9 +2441,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2476,7 +2473,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2841,7 +2839,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3112,11 +3109,6 @@ CopyFrom(CopyState cstate)
 				prevResultRelInfo = resultRelInfo;
 			}
 
-			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
 			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
@@ -3221,7 +3213,7 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3292,7 +3284,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index daa80ec4aa..1495d2c60e 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1749,7 +1749,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1879,7 +1878,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 40bd8049f0..357bf17e31 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index c46eb8d646..879ac8fbcd 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -857,9 +857,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -903,7 +900,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
@@ -2814,7 +2810,6 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 			rcestate->es_num_root_result_relations = numRootResultRels;
 		}
 	}
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 95e027c970..14d11e75c3 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -390,10 +390,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -416,7 +416,7 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -428,7 +428,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -452,11 +453,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -482,7 +483,7 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -494,7 +495,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -513,11 +515,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index ee0239b146..06bd715096 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -124,8 +124,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9ba1d78344..b01601578a 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,9 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -334,32 +335,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot)
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
-
-	ExecMaterializeSlot(slot);
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 
 	/*
-	 * get information on the (current) result relation
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
+
+	ExecMaterializeSlot(slot);
+
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -392,7 +409,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -427,7 +444,7 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -489,8 +506,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -550,7 +567,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -589,7 +607,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -675,6 +694,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -686,7 +706,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -696,10 +715,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1035,6 +1050,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1043,12 +1059,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1058,10 +1072,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1088,7 +1098,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1125,7 +1135,7 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1175,6 +1185,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1200,9 +1211,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1242,16 +1256,6 @@ lreplace:;
 				}
 			}
 
-			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
@@ -1269,18 +1273,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
+			/* Tuple routing starts from the root table. */
+			Assert(mtstate->rootResultRelInfo != NULL);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1444,7 +1448,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1683,7 +1688,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1840,40 +1845,36 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
- *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
-	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
 	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
@@ -1984,10 +1985,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2035,17 +2034,6 @@ ExecModifyTable(PlanState *pstate)
 	subplanstate = node->mt_plans[node->mt_whichplan];
 	junkfilter = resultRelInfo->ri_junkFilter;
 
-	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
@@ -2079,7 +2067,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2124,7 +2111,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2207,25 +2193,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2237,15 +2219,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2266,7 +2242,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2309,14 +2284,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2358,7 +2327,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2382,8 +2350,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 4bf6f5e427..c3a6fc30d0 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -191,7 +191,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -581,6 +580,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -604,6 +604,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -617,13 +618,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -678,6 +679,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -711,6 +713,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -719,7 +722,7 @@ apply_handle_update(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -761,7 +764,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -777,7 +781,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -800,6 +804,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -830,6 +835,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -838,7 +844,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -866,7 +872,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -878,7 +884,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 6298c7c8ca..a12d02cf5e 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -567,10 +567,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -587,10 +591,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 891b119608..103d4cd6c3 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,8 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 0c2a77aaf8..62e20fdfdc 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -518,7 +518,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index 75e25cdf48..b73e26fc69 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index 23885f638c..e5a0a05d13 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
2.16.5

From 1bf84886550b8d31e8f1f8f4a46c43b14db3dd6c Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v9 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index b01601578a..1362b2f2d1 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -356,7 +356,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -620,31 +619,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -710,7 +708,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -954,32 +951,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1026,6 +1021,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1179,119 +1321,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
-
-			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
-			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (!tuple_deleted)
-			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
-			}
-
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.16.5

From 40e2aa8468cd539c7b6dbc6406efc3e8abc30f7b Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v10 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  18 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   5 -
 src/backend/executor/execReplication.c   |  24 +--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 193 ++++++++++-------------
 src/backend/replication/logical/worker.c |  26 +--
 src/include/executor/executor.h          |  19 ++-
 src/include/executor/nodeModifyTable.h   |   4 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 139 insertions(+), 175 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index e79ede4cb8..7ba2f5b522 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2442,9 +2442,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2477,7 +2474,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2842,7 +2840,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3113,11 +3110,6 @@ CopyFrom(CopyState cstate)
 				prevResultRelInfo = resultRelInfo;
 			}
 
-			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
 			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
@@ -3222,7 +3214,8 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot, CMD_INSERT);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot,
+											   CMD_INSERT);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3293,7 +3286,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 6a32812a1f..00ae661e4d 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1798,7 +1798,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1928,7 +1927,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 1862af621b..e1d34be2e9 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 7cb486b211..209b5d9c83 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -857,9 +857,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -903,7 +900,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
@@ -2816,7 +2812,6 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 			rcestate->es_num_root_result_relations = numRootResultRels;
 		}
 	}
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 7194becfd9..eb1cb621ca 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -393,10 +393,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -419,7 +419,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -431,7 +432,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -455,11 +457,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -485,7 +487,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -497,7 +500,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -516,11 +520,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index cc5177cc2b..26c7479dce 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -124,8 +124,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index d71c0a4322..dcbbf4c3bc 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,10 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -359,32 +361,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+
+	/*
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
+	 */
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -417,7 +435,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -452,7 +471,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -514,8 +534,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -575,7 +595,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -614,7 +635,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -700,6 +722,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -711,7 +734,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -721,10 +743,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1060,6 +1078,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1068,12 +1087,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1083,10 +1100,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1113,7 +1126,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1150,7 +1164,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1200,6 +1215,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1225,9 +1241,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1267,16 +1286,6 @@ lreplace:;
 				}
 			}
 
-			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
@@ -1294,18 +1303,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+			/* Tuple routing starts from the root table. */
 			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
-
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1469,7 +1478,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1708,7 +1718,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1865,40 +1875,36 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
- *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
-	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
 	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
@@ -2009,10 +2015,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2060,17 +2064,6 @@ ExecModifyTable(PlanState *pstate)
 	subplanstate = node->mt_plans[node->mt_whichplan];
 	junkfilter = resultRelInfo->ri_junkFilter;
 
-	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
@@ -2104,7 +2097,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2149,7 +2141,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2232,25 +2223,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2262,15 +2249,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2291,7 +2272,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2334,14 +2314,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2383,7 +2357,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2407,8 +2380,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index ad4a732fd2..643cf81e66 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -192,7 +192,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -585,6 +584,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -608,6 +608,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -621,13 +622,13 @@ apply_handle_insert(StringInfo s)
 	slot_fill_defaults(rel, estate, remoteslot);
 	MemoryContextSwitchTo(oldctx);
 
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(resultRelInfo, estate, remoteslot);
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -682,6 +683,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	Oid			idxoid;
@@ -716,6 +718,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -741,7 +744,7 @@ apply_handle_update(StringInfo s)
 	fill_extraUpdatedCols(target_rte, RelationGetDescr(rel->localrel));
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Build the search tuple. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -783,7 +786,8 @@ apply_handle_update(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(resultRelInfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -799,7 +803,7 @@ apply_handle_update(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
@@ -822,6 +826,7 @@ apply_handle_update(StringInfo s)
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -852,6 +857,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -860,7 +866,7 @@ apply_handle_delete(StringInfo s)
 	EvalPlanQualInit(&epqstate, estate, NULL, NIL, -1);
 
 	PushActiveSnapshot(GetTransactionSnapshot());
-	ExecOpenIndices(estate->es_result_relation_info, false);
+	ExecOpenIndices(resultRelInfo, false);
 
 	/* Find the tuple using the replica identity index. */
 	oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
@@ -888,7 +894,7 @@ apply_handle_delete(StringInfo s)
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(resultRelInfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -900,7 +906,7 @@ apply_handle_delete(StringInfo s)
 	}
 
 	/* Cleanup. */
-	ExecCloseIndices(estate->es_result_relation_info);
+	ExecCloseIndices(resultRelInfo);
 	PopActiveSnapshot();
 
 	/* Handle queued AFTER triggers. */
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 81fdfa4add..dda3410362 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -572,10 +572,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -592,10 +596,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 4ec4ebdabc..2518fe4f64 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,9 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index cd3ddf781f..f38ec22b97 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -522,7 +522,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index 45d77ba3a5..60ae509a4e 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index 23885f638c..e5a0a05d13 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
2.20.1 (Apple Git-117)

From 6246ba6121a5342d8976e5556e0d7238f024e02d Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v10 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 ++++++++++++++-----------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index dcbbf4c3bc..87fe1668c6 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -382,7 +382,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -648,31 +647,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -738,7 +736,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -982,32 +979,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1054,6 +1049,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1209,119 +1351,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
-
-			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
-			 */
-			if (!tuple_deleted)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
-			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-			}
-
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
2.20.1 (Apple Git-117)

From a04de2ad3697ed6815546cbb479b8796a4a65879 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v10 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 +---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  19 ++-
 src/backend/executor/nodeModifyTable.c | 207 +++++--------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 85 insertions(+), 210 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 7ba2f5b522..d5dc508274 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3111,32 +3111,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 6e8b7223fe..acbe6966c3 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4646,9 +4647,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5742,13 +5741,23 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
+		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
 		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index a5542b92c7..76d5094bf8 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -925,9 +925,22 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 87fe1668c6..29bebf9430 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -365,10 +362,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1079,9 +1072,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1157,41 +1148,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1898,28 +1864,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1943,6 +1887,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1957,37 +1902,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -2003,58 +1928,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc);
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2150,17 +2023,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2329,6 +2191,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2351,8 +2214,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2361,6 +2229,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	EvalPlanQualInit(&mtstate->mt_epqstate, estate, NULL, NIL, node->epqParam);
 	mtstate->fireBSTriggers = true;
 
+	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
 	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
@@ -2427,6 +2302,20 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc));
 		resultRelInfo++;
 		i++;
 	}
@@ -2451,26 +2340,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index 5d69192643..f1977cee49 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -45,7 +45,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -64,14 +64,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_update_new_table;
 	bool		tcs_insert_new_table;
 
-	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
 	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index f38ec22b97..550064a817 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -489,6 +489,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1189,9 +1195,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
2.20.1 (Apple Git-117)

From 72ae48a08cb0513e71555a6290eb230de7ba207e Mon Sep 17 00:00:00 2001
From: Etsuro Fujita <efujita@postgresql.org>
Date: Thu, 8 Aug 2019 21:41:12 +0900
Subject: [PATCH v10 1/4] Remove dependency on estate->es_result_relation_info
 from FDW APIs.

FDW APIs for executing a foreign table direct modification assumed that
the FDW would obtain the target foreign table's ResultRelInfo from
estate->es_result_relation_info of the passed-in ForeignScanState node,
but the upcoming patch(es) to refactor partitioning-related code in
nodeModifyTable.c will remove the es_result_relation_info variable.
Revise BeginDirectModify()'s API to pass the ResultRelInfo explicitly, to
remove the dependency on that variable from the FDW APIs.  For
ExecInitForeignScan() to efficiently get the ResultRelInfo to pass to
BeginDirectModify(), add a field to ForeignScan that gives the index of
the target foreign table in the list of the query result relations.

Patch by Amit Langote, following a proposal by Andres Freund, reviewed by
Andres Freund and me

Discussion: https://postgr.es/m/20190718010911.l6xcdv6birtxiei4@alap3.anarazel.de
---
 contrib/postgres_fdw/postgres_fdw.c     | 25 +++++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            |  8 ++++++--
 src/backend/executor/nodeForeignscan.c  | 14 ++++++++++----
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 15 +++++++++++++++
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  3 +++
 10 files changed, 59 insertions(+), 12 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 2175dff824..75761007af 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -218,6 +218,7 @@ typedef struct PgFdwDirectModifyState
 	int			num_tuples;		/* # of result tuples */
 	int			next_tuple;		/* index of next one to return */
 	Relation	resultRel;		/* relcache entry for the target relation */
+	ResultRelInfo *resultRelInfo;	/* ResultRelInfo for the target relation */
 	AttrNumber *attnoMap;		/* array of attnums of input user columns */
 	AttrNumber	ctidAttno;		/* attnum of input ctid column */
 	AttrNumber	oidAttno;		/* attnum of input oid column */
@@ -361,7 +362,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2319,6 +2322,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2328,7 +2336,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2356,7 +2366,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2389,6 +2399,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 		dmstate->rel = NULL;
 	}
 
+	/* Save the ResultRelInfo for the target relation. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Initialize state variable */
 	dmstate->num_tuples = -1;	/* -1 means not set yet */
 
@@ -2451,7 +2464,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4087,7 +4100,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4234,7 +4247,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 6587678af2..0aff958415 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -873,6 +873,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -886,6 +887,8 @@ BeginDirectModify(ForeignScanState *node,
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
      information provided by <function>PlanDirectModify</function>).
+     In addition, the <structname>ResultRelInfo</structname> struct also
+     contains information about the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -917,8 +920,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     passed to <function>BeginDirectModify</function>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 513471ab9b..302125284e 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -221,12 +221,18 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan or the direct modification.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index e04c33e4ad..1580e2d27c 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -761,6 +761,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e084c3f069..843ee5bdec 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -698,6 +698,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index d5b23a3479..e6f2358dab 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -2016,6 +2016,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index fc25908dc6..beba1607a1 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5462,6 +5462,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* resultRelIndex will be set by PlanDirectModify/setrefs.c, if needed */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 3dcded506b..fe6db3c7c9 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -903,6 +903,13 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
+				/*
+				 * Caution: Do not change the relative ordering of this loop
+				 * and the statement below that adds the result relations to
+				 * root->glob->resultRelations, because we need to use the
+				 * current value of list_length(root->glob->resultRelations)
+				 * in some plans.
+				 */
 				foreach(l, splan->plans)
 				{
 					lfirst(l) = set_plan_refs(root,
@@ -1242,6 +1249,14 @@ set_foreignscan_references(PlannerInfo *root,
 	}
 
 	fscan->fs_relids = offset_relid_set(fscan->fs_relids, rtoffset);
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 95556dfb15..5c605928b2 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 4869fe7b6d..81449ab92e 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -620,6 +620,9 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* index of foreign table in the list of query
+								 * result relations for INSERT/UPDATE/DELETE;
+								 * -1 for SELECT */
 } ForeignScan;
 
 /* ----------------
-- 
2.20.1 (Apple Git-117)

From db6849c6ef3c3329911bcba5d75a88d86164ca7b Mon Sep 17 00:00:00 2001
From: Etsuro Fujita <efujita@postgresql.org>
Date: Thu, 8 Aug 2019 21:41:12 +0900
Subject: [PATCH v11 1/4] Remove dependency on estate->es_result_relation_info
 from FDW APIs.

FDW APIs for executing a foreign table direct modification assumed that
the FDW would obtain the target foreign table's ResultRelInfo from
estate->es_result_relation_info of the passed-in ForeignScanState node,
but the upcoming patch(es) to refactor partitioning-related code in
nodeModifyTable.c will remove the es_result_relation_info variable.
Revise BeginDirectModify()'s API to pass the ResultRelInfo explicitly, to
remove the dependency on that variable from the FDW APIs.  For
ExecInitForeignScan() to efficiently get the ResultRelInfo to pass to
BeginDirectModify(), add a field to ForeignScan that gives the index of
the target foreign table in the list of the query result relations.

Patch by Amit Langote, following a proposal by Andres Freund, reviewed by
Andres Freund and me

Discussion: https://postgr.es/m/20190718010911.l6xcdv6birtxiei4@alap3.anarazel.de
---
 contrib/postgres_fdw/postgres_fdw.c     | 25 +++++++++++++++++++------
 doc/src/sgml/fdwhandler.sgml            |  8 ++++++--
 src/backend/executor/nodeForeignscan.c  | 14 ++++++++++----
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  2 ++
 src/backend/optimizer/plan/setrefs.c    | 15 +++++++++++++++
 src/include/foreign/fdwapi.h            |  1 +
 src/include/nodes/plannodes.h           |  3 +++
 10 files changed, 59 insertions(+), 12 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 9fc53ca..3af5b24 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -218,6 +218,7 @@ typedef struct PgFdwDirectModifyState
 	int			num_tuples;		/* # of result tuples */
 	int			next_tuple;		/* index of next one to return */
 	Relation	resultRel;		/* relcache entry for the target relation */
+	ResultRelInfo *resultRelInfo;	/* ResultRelInfo for the target relation */
 	AttrNumber *attnoMap;		/* array of attnums of input user columns */
 	AttrNumber	ctidAttno;		/* attnum of input ctid column */
 	AttrNumber	oidAttno;		/* attnum of input oid column */
@@ -361,7 +362,9 @@ static bool postgresPlanDirectModify(PlannerInfo *root,
 									 ModifyTable *plan,
 									 Index resultRelation,
 									 int subplan_index);
-static void postgresBeginDirectModify(ForeignScanState *node, int eflags);
+static void postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags);
 static TupleTableSlot *postgresIterateDirectModify(ForeignScanState *node);
 static void postgresEndDirectModify(ForeignScanState *node);
 static void postgresExplainForeignScan(ForeignScanState *node,
@@ -2319,6 +2322,11 @@ postgresPlanDirectModify(PlannerInfo *root,
 			rebuild_fdw_scan_tlist(fscan, returningList);
 	}
 
+	/*
+	 * Set the index of the subplan result rel.
+	 */
+	fscan->resultRelIndex = subplan_index;
+
 	table_close(rel, NoLock);
 	return true;
 }
@@ -2328,7 +2336,9 @@ postgresPlanDirectModify(PlannerInfo *root,
  *		Prepare a direct foreign table modification
  */
 static void
-postgresBeginDirectModify(ForeignScanState *node, int eflags)
+postgresBeginDirectModify(ForeignScanState *node,
+						  ResultRelInfo *rinfo,
+						  int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
@@ -2356,7 +2366,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = rinfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2389,6 +2399,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 		dmstate->rel = NULL;
 	}
 
+	/* Save the ResultRelInfo for the target relation. */
+	dmstate->resultRelInfo = rinfo;
+
 	/* Initialize state variable */
 	dmstate->num_tuples = -1;	/* -1 means not set yet */
 
@@ -2451,7 +2464,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4087,7 +4100,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4234,7 +4247,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *relInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 7479303..d2cfc6c 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -881,6 +881,7 @@ PlanDirectModify(PlannerInfo *root,
 <programlisting>
 void
 BeginDirectModify(ForeignScanState *node,
+                  ResultRelInfo *rinfo,
                   int eflags);
 </programlisting>
 
@@ -894,6 +895,8 @@ BeginDirectModify(ForeignScanState *node,
      <structname>ForeignScanState</structname> node (in particular, from the underlying
      <structname>ForeignScan</structname> plan node, which contains any FDW-private
      information provided by <function>PlanDirectModify</function>).
+     In addition, the <structname>ResultRelInfo</structname> struct also
+     contains information about the target foreign table.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -925,8 +928,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     passed to <function>BeginDirectModify</function>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 513471a..3021252 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -221,12 +221,18 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan or the direct modification.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
+	else
+	{
+		ResultRelInfo *resultRelInfo;
+
+		Assert(node->resultRelIndex >= 0);
+		resultRelInfo = &estate->es_result_relations[node->resultRelIndex];
+		fdwroutine->BeginDirectModify(scanstate, resultRelInfo, eflags);
+	}
 
 	return scanstate;
 }
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 89c409d..2afb195 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -761,6 +761,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index e2f1775..15fd85a 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -698,6 +698,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 42050ab..4024a80 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -2017,6 +2017,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index eb9543f..362bc44 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5559,6 +5559,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* resultRelIndex will be set by PlanDirectModify/setrefs.c, if needed */
+	node->resultRelIndex = -1;
 
 	return node;
 }
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index baefe0e..f3d1a12 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -904,6 +904,13 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 					rc->rti += rtoffset;
 					rc->prti += rtoffset;
 				}
+				/*
+				 * Caution: Do not change the relative ordering of this loop
+				 * and the statement below that adds the result relations to
+				 * root->glob->resultRelations, because we need to use the
+				 * current value of list_length(root->glob->resultRelations)
+				 * in some plans.
+				 */
 				foreach(l, splan->plans)
 				{
 					lfirst(l) = set_plan_refs(root,
@@ -1243,6 +1250,14 @@ set_foreignscan_references(PlannerInfo *root,
 	}
 
 	fscan->fs_relids = offset_relid_set(fscan->fs_relids, rtoffset);
+
+	/*
+	 * Adjust resultRelIndex if it's valid (note that we are called before
+	 * adding the RT indexes of ModifyTable result relations to the global
+	 * list)
+	 */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += list_length(root->glob->resultRelations);
 }
 
 /*
diff --git a/src/include/foreign/fdwapi.h b/src/include/foreign/fdwapi.h
index 95556df..5c60592 100644
--- a/src/include/foreign/fdwapi.h
+++ b/src/include/foreign/fdwapi.h
@@ -112,6 +112,7 @@ typedef bool (*PlanDirectModify_function) (PlannerInfo *root,
 										   int subplan_index);
 
 typedef void (*BeginDirectModify_function) (ForeignScanState *node,
+											ResultRelInfo *rinfo,
 											int eflags);
 
 typedef TupleTableSlot *(*IterateDirectModify_function) (ForeignScanState *node);
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 83e0107..7314d2f 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -620,6 +620,9 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* index of foreign table in the list of query
+								 * result relations for INSERT/UPDATE/DELETE;
+								 * -1 for SELECT */
 } ForeignScan;
 
 /* ----------------
-- 
1.8.3.1

From 8fc56638cbaece9ded770734e1369aefd1dd0e8d Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v11 3/4] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 0c481cb..dd97ef5 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -389,7 +389,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -655,31 +654,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -745,7 +743,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -989,32 +986,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1061,6 +1056,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1216,119 +1358,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
-
-			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
-			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (!tuple_deleted)
-			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
-			}
-
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
1.8.3.1

From b64d57f06a6074cbbd18c8587a0162b9a97dd8eb Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v11 2/4] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  18 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   5 -
 src/backend/executor/execReplication.c   |  24 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 193 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  44 +++----
 src/include/executor/executor.h          |  19 ++-
 src/include/executor/nodeModifyTable.h   |   4 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 146 insertions(+), 186 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 99d1457..735c621 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2437,9 +2437,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2472,7 +2469,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2789,7 +2787,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3061,11 +3058,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3169,7 +3161,8 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot, CMD_INSERT);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot,
+											   CMD_INSERT);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3240,7 +3233,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index ed553f7..f2258c4 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1801,7 +1801,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1931,7 +1930,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 1862af6..e1d34be 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 4fdffad..190f979 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -857,9 +857,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -903,7 +900,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
@@ -2816,7 +2812,6 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 			rcestate->es_num_root_result_relations = numRootResultRels;
 		}
 	}
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index 8f474fa..2e8d9c8 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -404,10 +404,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -430,7 +430,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -442,7 +443,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -466,11 +468,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -496,7 +498,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -508,7 +511,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -527,11 +531,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index d0e65b8..d8d7614 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,8 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 20a4c47..0c481cb 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,10 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -366,32 +368,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
-
-	ExecMaterializeSlot(slot);
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 
 	/*
-	 * get information on the (current) result relation
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
+
+	ExecMaterializeSlot(slot);
+
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -424,7 +442,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -459,7 +478,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -521,8 +541,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -582,7 +602,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -621,7 +642,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -707,6 +729,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -718,7 +741,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -728,10 +750,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1067,6 +1085,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1075,12 +1094,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1090,10 +1107,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1120,7 +1133,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1157,7 +1171,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1207,6 +1222,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1232,9 +1248,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1275,16 +1294,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1301,18 +1310,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
+			/* Tuple routing starts from the root table. */
+			Assert(mtstate->rootResultRelInfo != NULL);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1476,7 +1485,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1715,7 +1725,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1872,41 +1882,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
- *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -2016,10 +2022,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2068,17 +2072,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2111,7 +2104,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2156,7 +2148,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2239,25 +2230,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2269,15 +2256,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2298,7 +2279,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2341,14 +2321,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2390,7 +2364,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2414,8 +2387,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index f90a896..fae1598 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -215,7 +215,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -609,6 +608,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -632,6 +632,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -647,11 +648,10 @@ apply_handle_insert(StringInfo s)
 
 	/* For a partitioned table, insert the tuple into a partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, NULL, rel, CMD_INSERT);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, NULL,
+								   rel, CMD_INSERT);
 	else
-		apply_handle_insert_internal(estate->es_result_relation_info, estate,
-									 remoteslot);
+		apply_handle_insert_internal(resultRelInfo, estate, remoteslot);
 
 	PopActiveSnapshot();
 
@@ -674,7 +674,7 @@ apply_handle_insert_internal(ResultRelInfo *relinfo,
 	ExecOpenIndices(relinfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(relinfo, estate, remoteslot);
 
 	/* Cleanup. */
 	ExecCloseIndices(relinfo);
@@ -721,6 +721,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	EState	   *estate;
@@ -751,6 +752,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -782,11 +784,11 @@ apply_handle_update(StringInfo s)
 
 	/* For a partitioned table, apply update to correct partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, &newtup, rel, CMD_UPDATE);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, &newtup,
+								   rel, CMD_UPDATE);
 	else
-		apply_handle_update_internal(estate->es_result_relation_info, estate,
-									 remoteslot, &newtup, rel);
+		apply_handle_update_internal(resultRelInfo, estate, remoteslot,
+									 &newtup, rel);
 
 	PopActiveSnapshot();
 
@@ -838,7 +840,8 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(relinfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -866,6 +869,7 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -892,6 +896,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -905,11 +910,11 @@ apply_handle_delete(StringInfo s)
 
 	/* For a partitioned table, apply delete to correct partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, NULL, rel, CMD_DELETE);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, NULL,
+								   rel, CMD_DELETE);
 	else
-		apply_handle_delete_internal(estate->es_result_relation_info, estate,
-									 remoteslot, &rel->remoterel);
+		apply_handle_delete_internal(resultRelInfo, estate, remoteslot,
+									 &rel->remoterel);
 
 	PopActiveSnapshot();
 
@@ -947,7 +952,7 @@ apply_handle_delete_internal(ResultRelInfo *relinfo, EState *estate,
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(relinfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -1055,7 +1060,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 	}
 	MemoryContextSwitchTo(oldctx);
 
-	estate->es_result_relation_info = partrelinfo;
 	switch (operation)
 	{
 		case CMD_INSERT:
@@ -1136,8 +1140,8 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					ExecOpenIndices(partrelinfo, false);
 
 					EvalPlanQualSetSlot(&epqstate, remoteslot_part);
-					ExecSimpleRelationUpdate(estate, &epqstate, localslot,
-											 remoteslot_part);
+					ExecSimpleRelationUpdate(partrelinfo, estate, &epqstate,
+											 localslot, remoteslot_part);
 					ExecCloseIndices(partrelinfo);
 					EvalPlanQualEnd(&epqstate);
 				}
@@ -1178,7 +1182,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					Assert(partrelinfo_new != partrelinfo);
 
 					/* DELETE old tuple found in the old partition. */
-					estate->es_result_relation_info = partrelinfo;
 					apply_handle_delete_internal(partrelinfo, estate,
 												 localslot,
 												 &relmapentry->remoterel);
@@ -1210,7 +1213,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 						slot_getallattrs(remoteslot);
 					}
 					MemoryContextSwitchTo(oldctx);
-					estate->es_result_relation_info = partrelinfo_new;
 					apply_handle_insert_internal(partrelinfo_new, estate,
 												 remoteslot_part);
 				}
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index c7deeac..a4024b7 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -573,10 +573,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -593,10 +597,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 4ec4ebd..2518fe4 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,9 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 0187989..246a3f3 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -524,7 +524,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index eb9d45b..da50ee3 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index ffd4aac..963faa1 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
1.8.3.1

From d1be649eb56d7e83c064c502542256988eae6740 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v11 4/4] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 ++---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  19 ++-
 src/backend/executor/nodeModifyTable.c | 207 +++++++--------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 85 insertions(+), 210 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 735c621..891d935 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3058,32 +3058,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccf..86adbee 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4292,9 +4293,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5388,14 +5387,24 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index fb6ce49..d31f786 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -926,9 +926,22 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index dd97ef5..5b6c5d1 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -372,10 +369,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1086,9 +1079,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1164,41 +1155,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1905,28 +1871,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1950,6 +1894,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1964,37 +1909,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -2010,58 +1935,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc);
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2157,17 +2030,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2336,6 +2198,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2358,8 +2221,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2369,6 +2237,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2434,6 +2309,20 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc));
 		resultRelInfo++;
 		i++;
 	}
@@ -2458,26 +2347,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a40ddf5..e38d732 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -46,7 +46,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -66,14 +66,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 246a3f3..41bf27c 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -491,6 +491,12 @@ typedef struct ResultRelInfo
 
 	/* For use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1191,9 +1197,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
1.8.3.1

From ddccecf06b38ba950850c4528aa3d37fcf2b860f Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Wed, 7 Oct 2020 17:34:49 +0900
Subject: [PATCH v12 1/5] Refactor set_plan_refs()

Pass the information needed by set_plan_refs() in a "context" struct
instead of directly as arguments.  This will allow to expand the set
of information that can be passed without much code churn.
---
 src/backend/optimizer/plan/setrefs.c | 83 ++++++++++++++++++------------------
 1 file changed, 41 insertions(+), 42 deletions(-)

diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index dd8e2e9..e647f2d 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -71,6 +71,12 @@ typedef struct
 	double		num_exec;
 } fix_upper_expr_context;
 
+typedef struct
+{
+	PlannerInfo	   *root;
+	int				rtoffset;	/* offset in root->glob->finalrtable */
+} set_plan_refs_context;
+
 /*
  * Selecting the best alternative in an AlternativeSubPlan expression requires
  * estimating how many times that expression will be evaluated.  For an
@@ -108,7 +114,7 @@ static void add_rtes_to_flat_rtable(PlannerInfo *root, bool recursing);
 static void flatten_unplanned_rtes(PlannerGlobal *glob, RangeTblEntry *rte);
 static bool flatten_rtes_walker(Node *node, PlannerGlobal *glob);
 static void add_rte_to_flat_rtable(PlannerGlobal *glob, RangeTblEntry *rte);
-static Plan *set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset);
+static Plan *set_plan_refs(Plan *plan, set_plan_refs_context *context);
 static Plan *set_indexonlyscan_references(PlannerInfo *root,
 										  IndexOnlyScan *plan,
 										  int rtoffset);
@@ -120,15 +126,12 @@ static Plan *clean_up_removed_plan_level(Plan *parent, Plan *child);
 static void set_foreignscan_references(PlannerInfo *root,
 									   ForeignScan *fscan,
 									   int rtoffset);
-static void set_customscan_references(PlannerInfo *root,
-									  CustomScan *cscan,
-									  int rtoffset);
-static Plan *set_append_references(PlannerInfo *root,
-								   Append *aplan,
-								   int rtoffset);
-static Plan *set_mergeappend_references(PlannerInfo *root,
-										MergeAppend *mplan,
-										int rtoffset);
+static void set_customscan_references(CustomScan *cscan,
+									  set_plan_refs_context *context);
+static Plan *set_append_references(Append *aplan,
+								   set_plan_refs_context *context);
+static Plan *set_mergeappend_references(MergeAppend *mplan,
+										set_plan_refs_context *context);
 static void set_hash_references(PlannerInfo *root, Plan *plan, int rtoffset);
 static Relids offset_relid_set(Relids relids, int rtoffset);
 static Node *fix_scan_expr(PlannerInfo *root, Node *node,
@@ -252,6 +255,7 @@ set_plan_references(PlannerInfo *root, Plan *plan)
 	PlannerGlobal *glob = root->glob;
 	int			rtoffset = list_length(glob->finalrtable);
 	ListCell   *lc;
+	set_plan_refs_context	context;
 
 	/*
 	 * Add all the query's RTEs to the flattened rangetable.  The live ones
@@ -302,7 +306,9 @@ set_plan_references(PlannerInfo *root, Plan *plan)
 	}
 
 	/* Now fix the Plan tree */
-	return set_plan_refs(root, plan, rtoffset);
+	context.root = root;
+	context.rtoffset = rtoffset;
+	return set_plan_refs(plan, &context);
 }
 
 /*
@@ -496,8 +502,10 @@ add_rte_to_flat_rtable(PlannerGlobal *glob, RangeTblEntry *rte)
  * set_plan_refs: recurse through the Plan nodes of a single subquery level
  */
 static Plan *
-set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
+set_plan_refs(Plan *plan, set_plan_refs_context *context)
 {
+	PlannerInfo *root = context->root;
+	int			rtoffset = context->rtoffset;
 	ListCell   *l;
 
 	if (plan == NULL)
@@ -714,7 +722,7 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 			set_foreignscan_references(root, (ForeignScan *) plan, rtoffset);
 			break;
 		case T_CustomScan:
-			set_customscan_references(root, (CustomScan *) plan, rtoffset);
+			set_customscan_references((CustomScan *) plan, context);
 			break;
 
 		case T_NestLoop:
@@ -968,9 +976,7 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 				}
 				foreach(l, splan->plans)
 				{
-					lfirst(l) = set_plan_refs(root,
-											  (Plan *) lfirst(l),
-											  rtoffset);
+					lfirst(l) = set_plan_refs((Plan *) lfirst(l), context);
 				}
 
 				/*
@@ -1001,14 +1007,10 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 			break;
 		case T_Append:
 			/* Needs special treatment, see comments below */
-			return set_append_references(root,
-										 (Append *) plan,
-										 rtoffset);
+			return set_append_references((Append *) plan, context);
 		case T_MergeAppend:
 			/* Needs special treatment, see comments below */
-			return set_mergeappend_references(root,
-											  (MergeAppend *) plan,
-											  rtoffset);
+			return set_mergeappend_references((MergeAppend *) plan, context);
 		case T_RecursiveUnion:
 			/* This doesn't evaluate targetlist or check quals either */
 			set_dummy_tlist_references(plan, rtoffset);
@@ -1023,9 +1025,7 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 				Assert(splan->plan.qual == NIL);
 				foreach(l, splan->bitmapplans)
 				{
-					lfirst(l) = set_plan_refs(root,
-											  (Plan *) lfirst(l),
-											  rtoffset);
+					lfirst(l) = set_plan_refs((Plan *) lfirst(l), context);
 				}
 			}
 			break;
@@ -1038,9 +1038,7 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 				Assert(splan->plan.qual == NIL);
 				foreach(l, splan->bitmapplans)
 				{
-					lfirst(l) = set_plan_refs(root,
-											  (Plan *) lfirst(l),
-											  rtoffset);
+					lfirst(l) = set_plan_refs((Plan *) lfirst(l), context);
 				}
 			}
 			break;
@@ -1058,8 +1056,8 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 	 * reference-adjustments bottom-up, then we would fail to match this
 	 * plan's var nodes against the already-modified nodes of the children.
 	 */
-	plan->lefttree = set_plan_refs(root, plan->lefttree, rtoffset);
-	plan->righttree = set_plan_refs(root, plan->righttree, rtoffset);
+	plan->lefttree = set_plan_refs(plan->lefttree, context);
+	plan->righttree = set_plan_refs(plan->righttree, context);
 
 	return plan;
 }
@@ -1328,10 +1326,11 @@ set_foreignscan_references(PlannerInfo *root,
  *	   Do set_plan_references processing on a CustomScan
  */
 static void
-set_customscan_references(PlannerInfo *root,
-						  CustomScan *cscan,
-						  int rtoffset)
+set_customscan_references(CustomScan *cscan,
+						  set_plan_refs_context *context)
 {
+	PlannerInfo *root = context->root;
+	int			rtoffset = context->rtoffset;
 	ListCell   *lc;
 
 	/* Adjust scanrelid if it's valid */
@@ -1387,7 +1386,7 @@ set_customscan_references(PlannerInfo *root,
 	/* Adjust child plan-nodes recursively, if needed */
 	foreach(lc, cscan->custom_plans)
 	{
-		lfirst(lc) = set_plan_refs(root, (Plan *) lfirst(lc), rtoffset);
+		lfirst(lc) = set_plan_refs((Plan *) lfirst(lc), context);
 	}
 
 	cscan->custom_relids = offset_relid_set(cscan->custom_relids, rtoffset);
@@ -1401,10 +1400,10 @@ set_customscan_references(PlannerInfo *root,
  * to do the normal processing on it.
  */
 static Plan *
-set_append_references(PlannerInfo *root,
-					  Append *aplan,
-					  int rtoffset)
+set_append_references(Append *aplan,
+					  set_plan_refs_context *context)
 {
+	int			rtoffset = context->rtoffset;
 	ListCell   *l;
 
 	/*
@@ -1417,7 +1416,7 @@ set_append_references(PlannerInfo *root,
 	/* First, we gotta recurse on the children */
 	foreach(l, aplan->appendplans)
 	{
-		lfirst(l) = set_plan_refs(root, (Plan *) lfirst(l), rtoffset);
+		lfirst(l) = set_plan_refs((Plan *) lfirst(l), context);
 	}
 
 	/* Now, if there's just one, forget the Append and return that child */
@@ -1465,10 +1464,10 @@ set_append_references(PlannerInfo *root,
  * to do the normal processing on it.
  */
 static Plan *
-set_mergeappend_references(PlannerInfo *root,
-						   MergeAppend *mplan,
-						   int rtoffset)
+set_mergeappend_references(MergeAppend *mplan,
+						   set_plan_refs_context *context)
 {
+	int			rtoffset = context->rtoffset;
 	ListCell   *l;
 
 	/*
@@ -1481,7 +1480,7 @@ set_mergeappend_references(PlannerInfo *root,
 	/* First, we gotta recurse on the children */
 	foreach(l, mplan->mergeplans)
 	{
-		lfirst(l) = set_plan_refs(root, (Plan *) lfirst(l), rtoffset);
+		lfirst(l) = set_plan_refs((Plan *) lfirst(l), context);
 	}
 
 	/* Now, if there's just one, forget the MergeAppend and return that child */
-- 
1.8.3.1

From f92989ca32a771ac598381b657e6c3bb1bf2241e Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v12 5/5] Refactor transition tuple capture code a bit

In the case of inherited update and partitioned table inserts,
a child tuple needs to be converted back into the root table format.
The tuple conversion map needed to do that was previously stored in
ModifyTableState and adjusted every time the child relation changed,
an arrangement which is a bit cumbersome to maintain.  Instead save
the map in the child result relation's ResultRelInfo.

This allows to get rid of a bunch of code that was needed to
manipulate tcs_map.
---
 src/backend/commands/copy.c            |  31 ++---
 src/backend/commands/trigger.c         |  19 ++-
 src/backend/executor/execPartition.c   |  19 ++-
 src/backend/executor/nodeModifyTable.c | 207 +++++++--------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |   9 +-
 6 files changed, 85 insertions(+), 210 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 092c4e4..56efa58 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3113,32 +3113,15 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition, we can just use the original
+			 * unconverted tuple instead of converting the tuple in partition
+			 * format back to root format.  We must do the conversion if such
+			 * triggers exist because they may change the tuple.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccf..86adbee 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4292,9 +4293,7 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * If there are no triggers in 'trigdesc' that request relevant transition
  * tables, then return NULL.
  *
- * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * The resulting object can be passed to the ExecAR* functions.
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5388,14 +5387,24 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	 */
 	if (row_trigger && transition_capture != NULL)
 	{
-		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleTableSlot *original_insert_tuple;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ?
+								pinfo->pi_PartitionToRootMap :
+								relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
 		bool		insert_new_table = transition_capture->tcs_insert_new_table;
 
 		/*
+		 * Get the originally inserted tuple from TransitionCaptureState and
+		 * set the variable to NULL so that the same tuple is not read again.
+		 */
+		original_insert_tuple = transition_capture->tcs_original_insert_tuple;
+		transition_capture->tcs_original_insert_tuple = NULL;
+
+		/*
 		 * For INSERT events NEW should be non-NULL, for DELETE events OLD
 		 * should be non-NULL, whereas for UPDATE events normally both OLD and
 		 * NEW are non-NULL.  But for UPDATE events fired for capturing
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 33d2c6f..3069729 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -983,9 +983,22 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be an UPDATE result relation, the map
+		 * has already been initialized by ExecInitModifyTable(); use that one
+		 * instead of building one from scratch.  To distinguish UPDATE result
+		 * relations from tuple-routing result relations, we rely on the fact
+		 * that each of the former has a distinct RT index.
+		 */
+		if (node && node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap =
+				partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 6fcc9bd..a92d2c9 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -372,10 +369,6 @@ ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
  *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
- *
- *		This may change the currently active tuple conversion map in
- *		mtstate->mt_transition_capture, so the callers must take care to
- *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
@@ -1086,9 +1079,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
-	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
+	TupleConversionMap *tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1164,41 +1155,16 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 		}
 	}
 
-	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
-	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
-	if (mtstate->mt_transition_capture)
-	{
-		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
-
 	/* We're done moving. */
 	return true;
 }
@@ -1905,28 +1871,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1950,6 +1894,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1964,37 +1909,17 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE
+	 * triggers on the partition, we can just use the original
+	 * unconverted tuple instead of converting the tuple in partition
+	 * format back to root format.  We must do the conversion if such
+	 * triggers exist because they may change the tuple.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -2010,58 +1935,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc);
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2157,17 +2030,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2336,6 +2198,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRel;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2358,8 +2221,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	/* If modifying a partitioned table, initialize the root table info */
 	if (node->rootResultRelIndex >= 0)
+	{
 		mtstate->rootResultRelInfo = estate->es_root_result_relations +
 			node->rootResultRelIndex;
+		rootResultRel = mtstate->rootResultRelInfo;
+	}
+	else
+		rootResultRel = mtstate->resultRelInfo;
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2369,6 +2237,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2434,6 +2309,20 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRel->ri_RelationDesc));
 		resultRelInfo++;
 		i++;
 	}
@@ -2458,26 +2347,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a40ddf5..e38d732 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -46,7 +46,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -66,14 +66,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 8073a81..e5e0409 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -488,6 +488,12 @@ typedef struct ResultRelInfo
 
 	/* for use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child sublan tuples to root parent format, set only if
+	 * either update row movement or transition tuple capture is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1178,9 +1184,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
1.8.3.1

From 551c8b59925a09ea399c10d364fd52d978e2d456 Mon Sep 17 00:00:00 2001
From: Etsuro Fujita <efujita@postgresql.org>
Date: Thu, 8 Aug 2019 21:41:12 +0900
Subject: [PATCH v12 2/5] Include result relation index if any in ForeignScan

FDWs that can perform an UPDATE/DELETE remotely using the "direct
modify" set of APIs need in some cases to access the result relation
properties for which they can currently look at
EState.es_result_relation_info, which the core executor laboriously
makes sure is set correctly.  An upcoming patch will remove that
field from EState.  So this commit installs a new field resultRelIndex
in ForeignScan node which will be set by the core planner for an FDW
to peruse during a "direct modification" operation.

Amit Langote, Etsuro Fujita
---
 contrib/postgres_fdw/postgres_fdw.c     | 26 ++++++++++++++++++--------
 doc/src/sgml/fdwhandler.sgml            | 10 ++++++----
 src/backend/executor/nodeForeignscan.c  |  5 ++++-
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c | 13 +++++++++++++
 src/backend/optimizer/plan/setrefs.c    | 17 +++++++++++++----
 src/include/nodes/plannodes.h           |  8 ++++++++
 9 files changed, 65 insertions(+), 17 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index a31abce..13e256f 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -200,6 +200,9 @@ typedef struct PgFdwDirectModifyState
 	Relation	rel;			/* relcache entry for the foreign table */
 	AttInMetadata *attinmeta;	/* attribute datatype conversion metadata */
 
+	int			resultRelIndex;	/* index of ResultRelInfo for the foreign table
+								 * in EState.es_result_relations */
+
 	/* extracted fdw_private data */
 	char	   *query;			/* text of UPDATE/DELETE command */
 	bool		has_returning;	/* is there a RETURNING clause? */
@@ -446,11 +449,12 @@ static List *build_remote_returning(Index rtindex, Relation rel,
 									List *returningList);
 static void rebuild_fdw_scan_tlist(ForeignScan *fscan, List *tlist);
 static void execute_dml_stmt(ForeignScanState *node);
-static TupleTableSlot *get_returning_data(ForeignScanState *node);
+static TupleTableSlot *get_returning_data(ForeignScanState *node, ResultRelInfo *resultRelInfo);
 static void init_returning_filter(PgFdwDirectModifyState *dmstate,
 								  List *fdw_scan_tlist,
 								  Index rtindex);
 static TupleTableSlot *apply_returning_filter(PgFdwDirectModifyState *dmstate,
+											  ResultRelInfo *relInfo,
 											  TupleTableSlot *slot,
 											  EState *estate);
 static void prepare_query_params(PlanState *node,
@@ -2331,6 +2335,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 {
 	ForeignScan *fsplan = (ForeignScan *) node->ss.ps.plan;
 	EState	   *estate = node->ss.ps.state;
+	List	   *resultRelations = estate->es_plannedstmt->resultRelations;
 	PgFdwDirectModifyState *dmstate;
 	Index		rtindex;
 	RangeTblEntry *rte;
@@ -2355,7 +2360,9 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	Assert(fsplan->resultRelIndex >= 0);
+	dmstate->resultRelIndex = fsplan->resultRelIndex;
+	rtindex = list_nth_int(resultRelations, fsplan->resultRelIndex);
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2450,7 +2457,10 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = &estate->es_result_relations[dmstate->resultRelIndex];
+
+	/* The executor must have initialized the ResultRelInfo for us. */
+	Assert(resultRelInfo != NULL);
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -2482,7 +2492,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 	/*
 	 * Get the next RETURNING tuple.
 	 */
-	return get_returning_data(node);
+	return get_returning_data(node, resultRelInfo);
 }
 
 /*
@@ -4082,11 +4092,10 @@ execute_dml_stmt(ForeignScanState *node)
  * Get the result of a RETURNING clause.
  */
 static TupleTableSlot *
-get_returning_data(ForeignScanState *node)
+get_returning_data(ForeignScanState *node, ResultRelInfo *resultRelInfo)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4141,7 +4150,8 @@ get_returning_data(ForeignScanState *node)
 		if (dmstate->rel)
 			resultSlot = slot;
 		else
-			resultSlot = apply_returning_filter(dmstate, slot, estate);
+			resultSlot = apply_returning_filter(dmstate, resultRelInfo, slot,
+												estate);
 	}
 	dmstate->next_tuple++;
 
@@ -4230,10 +4240,10 @@ init_returning_filter(PgFdwDirectModifyState *dmstate,
  */
 static TupleTableSlot *
 apply_returning_filter(PgFdwDirectModifyState *dmstate,
+					   ResultRelInfo *relInfo,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 72fa127..1f6de9b 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -893,8 +893,9 @@ BeginDirectModify(ForeignScanState *node,
      its <structfield>fdw_state</structfield> field is still NULL.  Information about
      the table to modify is accessible through the
      <structname>ForeignScanState</structname> node (in particular, from the underlying
-     <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
+     <structname>ForeignScan</structname> plan node, which contains an integer field
+     giving the table's index in the query's list of result relations along with any
+     FDW-private information provided by <function>PlanDirectModify</function>.
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -926,8 +927,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     obtained using the information passed to <function>BeginDirectModify</function>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 513471a..19433b3 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -221,10 +221,13 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan or the direct modification.
 	 */
 	if (node->operation != CMD_SELECT)
+	{
+		Assert(node->resultRelIndex >= 0);
 		fdwroutine->BeginDirectModify(scanstate, eflags);
+	}
 	else
 		fdwroutine->BeginForeignScan(scanstate, eflags);
 
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 0409a40..de57744 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -761,6 +761,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelIndex);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index f038648..fe10e5d 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -698,6 +698,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelIndex);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 42050ab..4024a80 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -2017,6 +2017,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelIndex);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 3d7a4e3..b157848 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5541,6 +5541,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* resultRelIndex will be set by make_modifytable(), if needed */
+	node->resultRelIndex = -1;
 
 	return node;
 }
@@ -6900,7 +6902,18 @@ make_modifytable(PlannerInfo *root,
 			!has_stored_generated_columns(subroot, rti))
 			direct_modify = fdwroutine->PlanDirectModify(subroot, node, rti, i);
 		if (direct_modify)
+		{
+			ForeignScan   *fscan = (ForeignScan *) list_nth(node->plans, i);
+
+			/*
+			 * For result relations that will be modified directly, the FDW
+			 * needs to know where to find them.
+			 */
+			Assert(IsA(fscan, ForeignScan));
+			fscan->resultRelIndex = i;
+
 			direct_modify_plans = bms_add_member(direct_modify_plans, i);
+		}
 
 		if (!direct_modify &&
 			fdwroutine != NULL &&
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index e647f2d..577a911 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -75,6 +75,7 @@ typedef struct
 {
 	PlannerInfo	   *root;
 	int				rtoffset;	/* offset in root->glob->finalrtable */
+	int				rroffset;	/* offset in root->glob->resultRelations */
 } set_plan_refs_context;
 
 /*
@@ -125,7 +126,7 @@ static bool trivial_subqueryscan(SubqueryScan *plan);
 static Plan *clean_up_removed_plan_level(Plan *parent, Plan *child);
 static void set_foreignscan_references(PlannerInfo *root,
 									   ForeignScan *fscan,
-									   int rtoffset);
+									   int rtoffset, int rroffset);
 static void set_customscan_references(CustomScan *cscan,
 									  set_plan_refs_context *context);
 static Plan *set_append_references(Append *aplan,
@@ -254,6 +255,7 @@ set_plan_references(PlannerInfo *root, Plan *plan)
 {
 	PlannerGlobal *glob = root->glob;
 	int			rtoffset = list_length(glob->finalrtable);
+	int			rroffset = list_length(glob->resultRelations);
 	ListCell   *lc;
 	set_plan_refs_context	context;
 
@@ -308,6 +310,7 @@ set_plan_references(PlannerInfo *root, Plan *plan)
 	/* Now fix the Plan tree */
 	context.root = root;
 	context.rtoffset = rtoffset;
+	context.rroffset = rroffset;
 	return set_plan_refs(plan, &context);
 }
 
@@ -506,6 +509,7 @@ set_plan_refs(Plan *plan, set_plan_refs_context *context)
 {
 	PlannerInfo *root = context->root;
 	int			rtoffset = context->rtoffset;
+	int			rroffset = context->rroffset;
 	ListCell   *l;
 
 	if (plan == NULL)
@@ -719,7 +723,8 @@ set_plan_refs(Plan *plan, set_plan_refs_context *context)
 			}
 			break;
 		case T_ForeignScan:
-			set_foreignscan_references(root, (ForeignScan *) plan, rtoffset);
+			set_foreignscan_references(root, (ForeignScan *) plan, rtoffset,
+									   rroffset);
 			break;
 		case T_CustomScan:
 			set_customscan_references((CustomScan *) plan, context);
@@ -985,7 +990,7 @@ set_plan_refs(Plan *plan, set_plan_refs_context *context)
 				 * resultRelIndex to reflect their starting position in the
 				 * global list.
 				 */
-				splan->resultRelIndex = list_length(root->glob->resultRelations);
+				splan->resultRelIndex = rroffset;
 				root->glob->resultRelations =
 					list_concat(root->glob->resultRelations,
 								splan->resultRelations);
@@ -1250,7 +1255,7 @@ clean_up_removed_plan_level(Plan *parent, Plan *child)
 static void
 set_foreignscan_references(PlannerInfo *root,
 						   ForeignScan *fscan,
-						   int rtoffset)
+						   int rtoffset, int rroffset)
 {
 	/* Adjust scanrelid if it's valid */
 	if (fscan->scan.scanrelid > 0)
@@ -1319,6 +1324,10 @@ set_foreignscan_references(PlannerInfo *root,
 	}
 
 	fscan->fs_relids = offset_relid_set(fscan->fs_relids, rtoffset);
+
+	/* Adjust resultRelIndex if needed */
+	if (fscan->resultRelIndex >= 0)
+		fscan->resultRelIndex += rroffset;
 }
 
 /*
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 83e0107..b0a3924 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -607,6 +607,11 @@ typedef struct WorkTableScan
  * When the plan node represents a foreign join, scan.scanrelid is zero and
  * fs_relids must be consulted to identify the join relation.  (fs_relids
  * is valid for simple scans as well, but will always match scan.scanrelid.)
+ *
+ * If an FDW's PlanDirectModify() callback decides to repurpose a ForeignScan
+ * node to store the information about an UPDATE or DELETE operation to
+ * to perform on a given foreign table result relation, resultRelIndex is set
+ * to identify the result relation.
  * ----------------
  */
 typedef struct ForeignScan
@@ -620,6 +625,9 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	int			resultRelIndex;	/* index of foreign table in the list of query
+								 * result relations for UPDATE/DELETE; -1 for
+								 * other query types */
 } ForeignScan;
 
 /* ----------------
-- 
1.8.3.1

From 9f8d9a8a851c383fa6656407a3b9719ffa48df0d Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v12 4/5] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 221a82f..6fcc9bd 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -389,7 +389,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -655,31 +654,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -745,7 +743,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -989,32 +986,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1061,6 +1056,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1216,119 +1358,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
-
-			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
-			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (!tuple_deleted)
-			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
-			}
-
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
1.8.3.1

From 9573e3a38c43e2e2529e3b6300c8f654cae686ec Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v12 3/5] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  18 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   5 -
 src/backend/executor/execReplication.c   |  24 ++--
 src/backend/executor/execUtils.c         |   2 -
 src/backend/executor/nodeModifyTable.c   | 193 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  44 +++----
 src/include/executor/executor.h          |  19 ++-
 src/include/executor/nodeModifyTable.h   |   4 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 146 insertions(+), 186 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 3c7dbad..092c4e4 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2489,9 +2489,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2524,7 +2521,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2844,7 +2842,6 @@ CopyFrom(CopyState cstate)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	ExecInitRangeTable(estate, cstate->range_table);
 
@@ -3116,11 +3113,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3224,7 +3216,8 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot, CMD_INSERT);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot,
+											   CMD_INSERT);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3295,7 +3288,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index e0ac4e0..6277adf 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1815,7 +1815,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1945,7 +1944,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 1862af6..e1d34be 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 2e27e26..eec5d6b 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -857,9 +857,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		estate->es_result_relations = resultRelInfos;
 		estate->es_num_result_relations = numResultRelations;
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
 		/*
 		 * In the partitioned result relation case, also build ResultRelInfos
 		 * for all the partitioned table roots, because we will need them to
@@ -903,7 +900,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 		 */
 		estate->es_result_relations = NULL;
 		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
 		estate->es_root_result_relations = NULL;
 		estate->es_num_root_result_relations = 0;
 	}
@@ -2805,7 +2801,6 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 			rcestate->es_num_root_result_relations = numRootResultRels;
 		}
 	}
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index b29db7b..01d2688 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -404,10 +404,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -430,7 +430,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -442,7 +443,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -466,11 +468,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -496,7 +498,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -508,7 +511,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -527,11 +531,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index d0e65b8..d8d7614 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,8 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_num_result_relations = 0;
-	estate->es_result_relation_info = NULL;
-
 	estate->es_root_result_relations = NULL;
 	estate->es_num_root_result_relations = 0;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9812089..221a82f 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,10 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -366,32 +368,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
-
-	ExecMaterializeSlot(slot);
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 
 	/*
-	 * get information on the (current) result relation
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
+
+	ExecMaterializeSlot(slot);
+
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -424,7 +442,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -459,7 +478,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -521,8 +541,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -582,7 +602,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -621,7 +642,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -707,6 +729,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -718,7 +741,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -728,10 +750,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1067,6 +1085,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1075,12 +1094,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1090,10 +1107,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1120,7 +1133,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1157,7 +1171,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1207,6 +1222,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1232,9 +1248,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1275,16 +1294,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1301,18 +1310,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
+			/* Tuple routing starts from the root table. */
+			Assert(mtstate->rootResultRelInfo != NULL);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1476,7 +1485,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1715,7 +1725,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1872,41 +1882,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
- *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -2016,10 +2022,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2068,17 +2072,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2111,7 +2104,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2156,7 +2148,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2239,25 +2230,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2269,15 +2256,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2298,7 +2279,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l;
@@ -2341,14 +2321,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	foreach(l, node->plans)
@@ -2390,7 +2364,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2414,8 +2387,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 9c6fdee..ed3ba86 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -361,7 +361,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 
 	estate->es_result_relations = resultRelInfo;
 	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -1150,6 +1149,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -1176,6 +1176,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -1191,11 +1192,10 @@ apply_handle_insert(StringInfo s)
 
 	/* For a partitioned table, insert the tuple into a partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, NULL, rel, CMD_INSERT);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, NULL,
+								   rel, CMD_INSERT);
 	else
-		apply_handle_insert_internal(estate->es_result_relation_info, estate,
-									 remoteslot);
+		apply_handle_insert_internal(resultRelInfo, estate, remoteslot);
 
 	PopActiveSnapshot();
 
@@ -1218,7 +1218,7 @@ apply_handle_insert_internal(ResultRelInfo *relinfo,
 	ExecOpenIndices(relinfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(relinfo, estate, remoteslot);
 
 	/* Cleanup. */
 	ExecCloseIndices(relinfo);
@@ -1265,6 +1265,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	EState	   *estate;
@@ -1298,6 +1299,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -1337,11 +1339,11 @@ apply_handle_update(StringInfo s)
 
 	/* For a partitioned table, apply update to correct partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, &newtup, rel, CMD_UPDATE);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, &newtup,
+								   rel, CMD_UPDATE);
 	else
-		apply_handle_update_internal(estate->es_result_relation_info, estate,
-									 remoteslot, &newtup, rel);
+		apply_handle_update_internal(resultRelInfo, estate, remoteslot,
+									 &newtup, rel);
 
 	PopActiveSnapshot();
 
@@ -1392,7 +1394,8 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(relinfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -1420,6 +1423,7 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -1449,6 +1453,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = &estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -1462,11 +1467,11 @@ apply_handle_delete(StringInfo s)
 
 	/* For a partitioned table, apply delete to correct partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, NULL, rel, CMD_DELETE);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, NULL,
+								   rel, CMD_DELETE);
 	else
-		apply_handle_delete_internal(estate->es_result_relation_info, estate,
-									 remoteslot, &rel->remoterel);
+		apply_handle_delete_internal(resultRelInfo, estate, remoteslot,
+									 &rel->remoterel);
 
 	PopActiveSnapshot();
 
@@ -1504,7 +1509,7 @@ apply_handle_delete_internal(ResultRelInfo *relinfo, EState *estate,
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(relinfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -1612,7 +1617,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 	}
 	MemoryContextSwitchTo(oldctx);
 
-	estate->es_result_relation_info = partrelinfo;
 	switch (operation)
 	{
 		case CMD_INSERT:
@@ -1693,8 +1697,8 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					ExecOpenIndices(partrelinfo, false);
 
 					EvalPlanQualSetSlot(&epqstate, remoteslot_part);
-					ExecSimpleRelationUpdate(estate, &epqstate, localslot,
-											 remoteslot_part);
+					ExecSimpleRelationUpdate(partrelinfo, estate, &epqstate,
+											 localslot, remoteslot_part);
 					ExecCloseIndices(partrelinfo);
 					EvalPlanQualEnd(&epqstate);
 				}
@@ -1735,7 +1739,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					Assert(partrelinfo_new != partrelinfo);
 
 					/* DELETE old tuple found in the old partition. */
-					estate->es_result_relation_info = partrelinfo;
 					apply_handle_delete_internal(partrelinfo, estate,
 												 localslot,
 												 &relmapentry->remoterel);
@@ -1767,7 +1770,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 						slot_getallattrs(remoteslot);
 					}
 					MemoryContextSwitchTo(oldctx);
-					estate->es_result_relation_info = partrelinfo_new;
 					apply_handle_insert_internal(partrelinfo_new, estate,
 												 remoteslot_part);
 				}
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 415e117..719c66f 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -573,10 +573,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -593,10 +597,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 4ec4ebd..2518fe4 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,9 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index ef448d6..8073a81 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -521,7 +521,6 @@ typedef struct EState
 	/* Info about target table(s) for insert/update/delete queries: */
 	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
 	int			es_num_result_relations;	/* length of array */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	/*
 	 * Info about the partition root table(s) for insert/update/delete queries
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index eb9d45b..da50ee3 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index ffd4aac..963faa1 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
1.8.3.1

From 89db264f441062e6c938a44757d8faa4b977e0a6 Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Fri, 9 Oct 2020 13:52:29 +0900
Subject: [PATCH v13 1/5] Make es_result_relations array indexable by RT index

This allows us to get rid of the need to keep around a separate set
of indexes for each result relation.
---
 src/backend/commands/copy.c              |  20 ++--
 src/backend/commands/explain.c           |  17 ++-
 src/backend/commands/tablecmds.c         |  12 +-
 src/backend/executor/execMain.c          | 189 ++++++++-----------------------
 src/backend/executor/execUtils.c         |  73 +++++++++---
 src/backend/executor/nodeModifyTable.c   |  24 ++--
 src/backend/nodes/copyfuncs.c            |   2 -
 src/backend/nodes/outfuncs.c             |   2 -
 src/backend/nodes/readfuncs.c            |   2 -
 src/backend/optimizer/plan/createplan.c  |   2 -
 src/backend/optimizer/plan/setrefs.c     |  10 +-
 src/backend/replication/logical/worker.c |   4 +-
 src/include/executor/executor.h          |   5 +
 src/include/nodes/execnodes.h            |  21 ++--
 src/include/nodes/plannodes.h            |   2 -
 15 files changed, 162 insertions(+), 223 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 3c7dbad..6948214 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2829,25 +2829,18 @@ CopyFrom(CopyState cstate)
 	 * index-entry-making machinery.  (There used to be a huge amount of code
 	 * here that basically duplicated execUtils.c ...)
 	 */
-	resultRelInfo = makeNode(ResultRelInfo);
-	InitResultRelInfo(resultRelInfo,
-					  cstate->rel,
-					  1,		/* must match rel's position in range_table */
-					  NULL,
-					  0);
-	target_resultRelInfo = resultRelInfo;
+	ExecInitRangeTable(estate, cstate->range_table);
+	ExecInitResultRelationsArray(estate);
+	resultRelInfo = target_resultRelInfo = makeNode(ResultRelInfo);
+	ExecInitResultRelation(estate, resultRelInfo, 1);
 
 	/* Verify the named relation is a valid target for INSERT */
 	CheckValidResultRel(resultRelInfo, CMD_INSERT);
 
 	ExecOpenIndices(resultRelInfo, false);
 
-	estate->es_result_relations = resultRelInfo;
-	estate->es_num_result_relations = 1;
 	estate->es_result_relation_info = resultRelInfo;
 
-	ExecInitRangeTable(estate, cstate->range_table);
-
 	/*
 	 * Set up a ModifyTableState so we can let FDW(s) init themselves for
 	 * foreign-table result relation(s).
@@ -2856,7 +2849,7 @@ CopyFrom(CopyState cstate)
 	mtstate->ps.plan = NULL;
 	mtstate->ps.state = estate;
 	mtstate->operation = CMD_INSERT;
-	mtstate->resultRelInfo = estate->es_result_relations;
+	mtstate->resultRelInfo = resultRelInfo;
 
 	if (resultRelInfo->ri_FdwRoutine != NULL &&
 		resultRelInfo->ri_FdwRoutine->BeginForeignInsert != NULL)
@@ -3359,7 +3352,8 @@ CopyFrom(CopyState cstate)
 	if (insertMethod != CIM_SINGLE)
 		CopyMultiInsertInfoCleanup(&multiInsertInfo);
 
-	ExecCloseIndices(target_resultRelInfo);
+	ExecCloseResultRelations(estate);
+	ExecCloseRangeTableRelations(estate);
 
 	/* Close all the partitioned tables, leaf partitions, and their indices */
 	if (proute)
diff --git a/src/backend/commands/explain.c b/src/backend/commands/explain.c
index c98c9b5..3210f90 100644
--- a/src/backend/commands/explain.c
+++ b/src/backend/commands/explain.c
@@ -769,13 +769,14 @@ ExplainPrintTriggers(ExplainState *es, QueryDesc *queryDesc)
 {
 	ResultRelInfo *rInfo;
 	bool		show_relname;
-	int			numrels = queryDesc->estate->es_num_result_relations;
-	int			numrootrels = queryDesc->estate->es_num_root_result_relations;
+	int			numrels = list_length(queryDesc->plannedstmt->resultRelations);
+	int			numrootrels = list_length(queryDesc->plannedstmt->rootResultRelations);
+	List	   *resultrels;
 	List	   *routerels;
 	List	   *targrels;
-	int			nr;
 	ListCell   *l;
 
+	resultrels = queryDesc->estate->es_opened_result_relations;
 	routerels = queryDesc->estate->es_tuple_routing_result_relations;
 	targrels = queryDesc->estate->es_trig_target_relations;
 
@@ -783,13 +784,11 @@ ExplainPrintTriggers(ExplainState *es, QueryDesc *queryDesc)
 
 	show_relname = (numrels > 1 || numrootrels > 0 ||
 					routerels != NIL || targrels != NIL);
-	rInfo = queryDesc->estate->es_result_relations;
-	for (nr = 0; nr < numrels; rInfo++, nr++)
-		report_triggers(rInfo, show_relname, es);
-
-	rInfo = queryDesc->estate->es_root_result_relations;
-	for (nr = 0; nr < numrootrels; rInfo++, nr++)
+	foreach(l, resultrels)
+	{
+		rInfo = lfirst(l);
 		report_triggers(rInfo, show_relname, es);
+	}
 
 	foreach(l, routerels)
 	{
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index e0ac4e0..1cd8cfb 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1693,6 +1693,7 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	SubTransactionId mySubid;
 	ListCell   *cell;
 	Oid		   *logrelids;
+	int			dummy_rti;
 
 	/*
 	 * Check the explicitly-specified relations.
@@ -1792,19 +1793,24 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfos = (ResultRelInfo *)
 		palloc(list_length(rels) * sizeof(ResultRelInfo));
 	resultRelInfo = resultRelInfos;
+	estate->es_result_relations = (ResultRelInfo **)
+		palloc(list_length(rels) * sizeof(ResultRelInfo *));
+	dummy_rti = 1;
 	foreach(cell, rels)
 	{
 		Relation	rel = (Relation) lfirst(cell);
 
 		InitResultRelInfo(resultRelInfo,
 						  rel,
-						  0,	/* dummy rangetable index */
+						  dummy_rti,
 						  NULL,
 						  0);
+		estate->es_result_relations[dummy_rti - 1] = resultRelInfo;
+		estate->es_opened_result_relations =
+			lappend(estate->es_opened_result_relations, resultRelInfo);
 		resultRelInfo++;
+		dummy_rti++;
 	}
-	estate->es_result_relations = resultRelInfos;
-	estate->es_num_result_relations = list_length(rels);
 
 	/*
 	 * Process all BEFORE STATEMENT TRUNCATE triggers before we begin
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 2e27e26..f62fd8f 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -827,85 +827,12 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 
 	estate->es_plannedstmt = plannedstmt;
 
-	/*
-	 * Initialize ResultRelInfo data structures, and open the result rels.
-	 */
 	if (plannedstmt->resultRelations)
 	{
-		List	   *resultRelations = plannedstmt->resultRelations;
-		int			numResultRelations = list_length(resultRelations);
-		ResultRelInfo *resultRelInfos;
-		ResultRelInfo *resultRelInfo;
-
-		resultRelInfos = (ResultRelInfo *)
-			palloc(numResultRelations * sizeof(ResultRelInfo));
-		resultRelInfo = resultRelInfos;
-		foreach(l, resultRelations)
-		{
-			Index		resultRelationIndex = lfirst_int(l);
-			Relation	resultRelation;
-
-			resultRelation = ExecGetRangeTableRelation(estate,
-													   resultRelationIndex);
-			InitResultRelInfo(resultRelInfo,
-							  resultRelation,
-							  resultRelationIndex,
-							  NULL,
-							  estate->es_instrument);
-			resultRelInfo++;
-		}
-		estate->es_result_relations = resultRelInfos;
-		estate->es_num_result_relations = numResultRelations;
+		ExecInitResultRelationsArray(estate);
 
 		/* es_result_relation_info is NULL except when within ModifyTable */
 		estate->es_result_relation_info = NULL;
-
-		/*
-		 * In the partitioned result relation case, also build ResultRelInfos
-		 * for all the partitioned table roots, because we will need them to
-		 * fire statement-level triggers, if any.
-		 */
-		if (plannedstmt->rootResultRelations)
-		{
-			int			num_roots = list_length(plannedstmt->rootResultRelations);
-
-			resultRelInfos = (ResultRelInfo *)
-				palloc(num_roots * sizeof(ResultRelInfo));
-			resultRelInfo = resultRelInfos;
-			foreach(l, plannedstmt->rootResultRelations)
-			{
-				Index		resultRelIndex = lfirst_int(l);
-				Relation	resultRelDesc;
-
-				resultRelDesc = ExecGetRangeTableRelation(estate,
-														  resultRelIndex);
-				InitResultRelInfo(resultRelInfo,
-								  resultRelDesc,
-								  resultRelIndex,
-								  NULL,
-								  estate->es_instrument);
-				resultRelInfo++;
-			}
-
-			estate->es_root_result_relations = resultRelInfos;
-			estate->es_num_root_result_relations = num_roots;
-		}
-		else
-		{
-			estate->es_root_result_relations = NULL;
-			estate->es_num_root_result_relations = 0;
-		}
-	}
-	else
-	{
-		/*
-		 * if no result relation, then set state appropriately
-		 */
-		estate->es_result_relations = NULL;
-		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
-		estate->es_root_result_relations = NULL;
-		estate->es_num_root_result_relations = 0;
 	}
 
 	/*
@@ -1334,8 +1261,7 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
  *
  * Most of the time, triggers are fired on one of the result relations of the
  * query, and so we can just return a member of the es_result_relations array,
- * or the es_root_result_relations array (if any), or the
- * es_tuple_routing_result_relations list (if any).  (Note: in self-join
+ * or the es_tuple_routing_result_relations list (if any). (Note: in self-join
  * situations there might be multiple members with the same OID; if so it
  * doesn't matter which one we pick.)
  *
@@ -1352,30 +1278,16 @@ ResultRelInfo *
 ExecGetTriggerResultRel(EState *estate, Oid relid)
 {
 	ResultRelInfo *rInfo;
-	int			nr;
 	ListCell   *l;
 	Relation	rel;
 	MemoryContext oldcontext;
 
 	/* First, search through the query result relations */
-	rInfo = estate->es_result_relations;
-	nr = estate->es_num_result_relations;
-	while (nr > 0)
-	{
-		if (RelationGetRelid(rInfo->ri_RelationDesc) == relid)
-			return rInfo;
-		rInfo++;
-		nr--;
-	}
-	/* Second, search through the root result relations, if any */
-	rInfo = estate->es_root_result_relations;
-	nr = estate->es_num_root_result_relations;
-	while (nr > 0)
+	foreach(l, estate->es_opened_result_relations)
 	{
+		rInfo = lfirst(l);
 		if (RelationGetRelid(rInfo->ri_RelationDesc) == relid)
 			return rInfo;
-		rInfo++;
-		nr--;
 	}
 
 	/*
@@ -1512,9 +1424,6 @@ ExecPostprocessPlan(EState *estate)
 static void
 ExecEndPlan(PlanState *planstate, EState *estate)
 {
-	ResultRelInfo *resultRelInfo;
-	Index		num_relations;
-	Index		i;
 	ListCell   *l;
 
 	/*
@@ -1540,30 +1449,52 @@ ExecEndPlan(PlanState *planstate, EState *estate)
 	 */
 	ExecResetTupleTable(estate->es_tupleTable, false);
 
+	/* Close indexes of result relation(s), if any. */
+	ExecCloseResultRelations(estate);
+
+	/*
+	 * close whatever rangetable Relations have been opened.  We do not
+	 * release any locks we might hold on those rels.
+	 */
+	ExecCloseRangeTableRelations(estate);
+
+	/* likewise close any trigger target relations */
+	ExecCleanUpTriggerState(estate);
+}
+
+/*
+ * ExecCloseResultRelations
+ */
+void
+ExecCloseResultRelations(EState *estate)
+{
+	ListCell *l;
+
 	/*
 	 * close indexes of result relation(s) if any.  (Rels themselves get
 	 * closed next.)
 	 */
-	resultRelInfo = estate->es_result_relations;
-	for (i = estate->es_num_result_relations; i > 0; i--)
+	foreach(l, estate->es_opened_result_relations)
 	{
+		ResultRelInfo *resultRelInfo = lfirst(l);
+
 		ExecCloseIndices(resultRelInfo);
-		resultRelInfo++;
 	}
+}
 
-	/*
-	 * close whatever rangetable Relations have been opened.  We do not
-	 * release any locks we might hold on those rels.
-	 */
-	num_relations = estate->es_range_table_size;
-	for (i = 0; i < num_relations; i++)
+/*
+ * ExecCloseRangeTableRelations
+ */
+void
+ExecCloseRangeTableRelations(EState *estate)
+{
+	int		i;
+
+	for (i = 0; i < estate->es_range_table_size; i++)
 	{
 		if (estate->es_relations[i])
 			table_close(estate->es_relations[i], NoLock);
 	}
-
-	/* likewise close any trigger target relations */
-	ExecCleanUpTriggerState(estate);
 }
 
 /* ----------------------------------------------------------------
@@ -2758,17 +2689,9 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 
 	/*
 	 * Child EPQ EStates share the parent's copy of unchanging state such as
-	 * the snapshot, rangetable, result-rel info, and external Param info.
+	 * the snapshot, rangetable, and external Param info.
 	 * They need their own copies of local state, including a tuple table,
-	 * es_param_exec_vals, etc.
-	 *
-	 * The ResultRelInfo array management is trickier than it looks.  We
-	 * create fresh arrays for the child but copy all the content from the
-	 * parent.  This is because it's okay for the child to share any
-	 * per-relation state the parent has already created --- but if the child
-	 * sets up any ResultRelInfo fields, such as its own junkfilter, that
-	 * state must *not* propagate back to the parent.  (For one thing, the
-	 * pointed-to data is in a memory context that won't last long enough.)
+	 * es_param_exec_vals, result-rel info, etc.
 	 */
 	rcestate->es_direction = ForwardScanDirection;
 	rcestate->es_snapshot = parentestate->es_snapshot;
@@ -2781,30 +2704,12 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 	rcestate->es_plannedstmt = parentestate->es_plannedstmt;
 	rcestate->es_junkFilter = parentestate->es_junkFilter;
 	rcestate->es_output_cid = parentestate->es_output_cid;
-	if (parentestate->es_num_result_relations > 0)
-	{
-		int			numResultRelations = parentestate->es_num_result_relations;
-		int			numRootResultRels = parentestate->es_num_root_result_relations;
-		ResultRelInfo *resultRelInfos;
-
-		resultRelInfos = (ResultRelInfo *)
-			palloc(numResultRelations * sizeof(ResultRelInfo));
-		memcpy(resultRelInfos, parentestate->es_result_relations,
-			   numResultRelations * sizeof(ResultRelInfo));
-		rcestate->es_result_relations = resultRelInfos;
-		rcestate->es_num_result_relations = numResultRelations;
-
-		/* Also transfer partitioned root result relations. */
-		if (numRootResultRels > 0)
-		{
-			resultRelInfos = (ResultRelInfo *)
-				palloc(numRootResultRels * sizeof(ResultRelInfo));
-			memcpy(resultRelInfos, parentestate->es_root_result_relations,
-				   numRootResultRels * sizeof(ResultRelInfo));
-			rcestate->es_root_result_relations = resultRelInfos;
-			rcestate->es_num_root_result_relations = numRootResultRels;
-		}
-	}
+	/*
+	 * ResultRelInfos needed by subplans are initialized from scratch when
+	 * the subplans themselves are initialized.
+	 */
+	if (parentestate->es_result_relations)
+		ExecInitResultRelationsArray(rcestate);
 	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
@@ -2954,6 +2859,8 @@ EvalPlanQualEnd(EPQState *epqstate)
 		ExecEndNode(subplanstate);
 	}
 
+	ExecCloseResultRelations(estate);
+
 	/* throw away the per-estate tuple table, some node may have used it */
 	ExecResetTupleTable(estate->es_tupleTable, false);
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index d0e65b8..169dd92 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -124,14 +124,8 @@ CreateExecutorState(void)
 	estate->es_output_cid = (CommandId) 0;
 
 	estate->es_result_relations = NULL;
-	estate->es_num_result_relations = 0;
 	estate->es_result_relation_info = NULL;
-
-	estate->es_root_result_relations = NULL;
-	estate->es_num_root_result_relations = 0;
-
 	estate->es_tuple_routing_result_relations = NIL;
-
 	estate->es_trig_target_relations = NIL;
 
 	estate->es_param_list_info = NULL;
@@ -711,16 +705,10 @@ ExecCreateScanSlotFromOuterPlan(EState *estate,
 bool
 ExecRelationIsTargetRelation(EState *estate, Index scanrelid)
 {
-	ResultRelInfo *resultRelInfos;
-	int			i;
-
-	resultRelInfos = estate->es_result_relations;
-	for (i = 0; i < estate->es_num_result_relations; i++)
-	{
-		if (resultRelInfos[i].ri_RangeTableIndex == scanrelid)
-			return true;
-	}
-	return false;
+	return	list_member_int(estate->es_plannedstmt->resultRelations,
+							scanrelid) ||
+			list_member_int(estate->es_plannedstmt->rootResultRelations,
+							scanrelid);
 }
 
 /* ----------------------------------------------------------------
@@ -779,8 +767,8 @@ ExecInitRangeTable(EState *estate, List *rangeTable)
 		palloc0(estate->es_range_table_size * sizeof(Relation));
 
 	/*
-	 * es_rowmarks is also parallel to the es_range_table, but it's allocated
-	 * only if needed.
+	 * es_result_relations and es_rowmarks are also parallel to es_range_table,
+	 * but are only allocated if needed.
 	 */
 	estate->es_rowmarks = NULL;
 }
@@ -836,6 +824,55 @@ ExecGetRangeTableRelation(EState *estate, Index rti)
 }
 
 /*
+ * ExecInitResultRelationsArray
+ *		Allocate space to hold ResultRelInfo pointers of result relations
+ *
+ * Although not relations in the range table may be result relations, we
+ * allocate that many pointers, because that allows to access individual
+ * entries by RT index (minus 1 to be accurate), which is convenient.
+ */
+void
+ExecInitResultRelationsArray(EState *estate)
+{
+	/*
+	 * Individual pointers are assigned when ExecInitResultRelation() is
+	 * called one-by-one for each result relation.
+	 */
+	estate->es_result_relations = (ResultRelInfo **)
+		palloc0(estate->es_range_table_size * sizeof(ResultRelInfo *));
+}
+
+/*
+ * ExecInitResultRelation
+ *		Open result relation given by the passed-in RT index and fill its
+ *		ResultRelInfo node
+ *
+ * Here, we also save the ResultRelInfo in estate->es_result_relations array
+ * such that it can be accessed later using the RT index.
+ */
+void
+ExecInitResultRelation(EState *estate, ResultRelInfo *resultRelInfo,
+					   Index rti)
+{
+	Relation	resultRelationDesc;
+
+	resultRelationDesc = ExecGetRangeTableRelation(estate, rti);
+	InitResultRelInfo(resultRelInfo,
+					  resultRelationDesc,
+					  rti,
+					  NULL,
+					  estate->es_instrument);
+	estate->es_result_relations[rti - 1] = resultRelInfo;
+
+	/*
+	 * Saving in the list allows to avoid needlessly traversing the whole
+	 * array when only a few of its entries are possibly non-NULL.
+	 */
+	estate->es_opened_result_relations =
+		lappend(estate->es_opened_result_relations, resultRelInfo);
+}
+
+/*
  * UpdateChangedParamSet
  *		Add changed parameters to a plan node's chgParam set
  */
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9812089..9b27d34 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -2301,7 +2301,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
-	ListCell   *l;
+	ListCell   *l,
+			   *l1;
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
@@ -2322,13 +2323,17 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->mt_done = false;
 
 	mtstate->mt_plans = (PlanState **) palloc0(sizeof(PlanState *) * nplans);
-	mtstate->resultRelInfo = estate->es_result_relations + node->resultRelIndex;
+	mtstate->resultRelInfo = (ResultRelInfo *)
+		palloc(nplans * sizeof(ResultRelInfo));
 	mtstate->mt_scans = (TupleTableSlot **) palloc0(sizeof(TupleTableSlot *) * nplans);
 
 	/* If modifying a partitioned table, initialize the root table info */
-	if (node->rootResultRelIndex >= 0)
-		mtstate->rootResultRelInfo = estate->es_root_result_relations +
-			node->rootResultRelIndex;
+	if (node->rootRelation > 0)
+	{
+		mtstate->rootResultRelInfo = makeNode(ResultRelInfo);
+		ExecInitResultRelation(estate, mtstate->rootResultRelInfo,
+							   node->rootRelation);
+	}
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2351,9 +2356,14 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
-	foreach(l, node->plans)
+	forboth(l, node->resultRelations, l1, node->plans)
 	{
-		subplan = (Plan *) lfirst(l);
+		Index		resultRelation = lfirst_int(l);
+
+		subplan = (Plan *) lfirst(l1);
+
+		/* This opens result relation and fills ResultRelInfo. */
+		ExecInitResultRelation(estate, resultRelInfo, resultRelation);
 
 		/* Initialize the usesFdwDirectModify flag */
 		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(i,
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 0409a40..7974fa0 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -207,8 +207,6 @@ _copyModifyTable(const ModifyTable *from)
 	COPY_SCALAR_FIELD(rootRelation);
 	COPY_SCALAR_FIELD(partColsUpdated);
 	COPY_NODE_FIELD(resultRelations);
-	COPY_SCALAR_FIELD(resultRelIndex);
-	COPY_SCALAR_FIELD(rootResultRelIndex);
 	COPY_NODE_FIELD(plans);
 	COPY_NODE_FIELD(withCheckOptionLists);
 	COPY_NODE_FIELD(returningLists);
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index f038648..f35d2a0 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -408,8 +408,6 @@ _outModifyTable(StringInfo str, const ModifyTable *node)
 	WRITE_UINT_FIELD(rootRelation);
 	WRITE_BOOL_FIELD(partColsUpdated);
 	WRITE_NODE_FIELD(resultRelations);
-	WRITE_INT_FIELD(resultRelIndex);
-	WRITE_INT_FIELD(rootResultRelIndex);
 	WRITE_NODE_FIELD(plans);
 	WRITE_NODE_FIELD(withCheckOptionLists);
 	WRITE_NODE_FIELD(returningLists);
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 42050ab..73b8d9d 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1639,8 +1639,6 @@ _readModifyTable(void)
 	READ_UINT_FIELD(rootRelation);
 	READ_BOOL_FIELD(partColsUpdated);
 	READ_NODE_FIELD(resultRelations);
-	READ_INT_FIELD(resultRelIndex);
-	READ_INT_FIELD(rootResultRelIndex);
 	READ_NODE_FIELD(plans);
 	READ_NODE_FIELD(withCheckOptionLists);
 	READ_NODE_FIELD(returningLists);
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 3d7a4e3..881eaf4 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -6808,8 +6808,6 @@ make_modifytable(PlannerInfo *root,
 	node->rootRelation = rootRelation;
 	node->partColsUpdated = partColsUpdated;
 	node->resultRelations = resultRelations;
-	node->resultRelIndex = -1;	/* will be set correctly in setrefs.c */
-	node->rootResultRelIndex = -1;	/* will be set correctly in setrefs.c */
 	node->plans = subplans;
 	if (!onconflict)
 	{
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index dd8e2e9..66372d3 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -975,24 +975,18 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 
 				/*
 				 * Append this ModifyTable node's final result relation RT
-				 * index(es) to the global list for the plan, and set its
-				 * resultRelIndex to reflect their starting position in the
-				 * global list.
+				 * index(es) to the global list for the plan.
 				 */
-				splan->resultRelIndex = list_length(root->glob->resultRelations);
 				root->glob->resultRelations =
 					list_concat(root->glob->resultRelations,
 								splan->resultRelations);
 
 				/*
 				 * If the main target relation is a partitioned table, also
-				 * add the partition root's RT index to rootResultRelations,
-				 * and remember its index in that list in rootResultRelIndex.
+				 * add the partition root's RT index to rootResultRelations.
 				 */
 				if (splan->rootRelation)
 				{
-					splan->rootResultRelIndex =
-						list_length(root->glob->rootResultRelations);
 					root->glob->rootResultRelations =
 						lappend_int(root->glob->rootResultRelations,
 									splan->rootRelation);
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 9c6fdee..77f71e5 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -355,12 +355,12 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 	rte->relkind = rel->localrel->rd_rel->relkind;
 	rte->rellockmode = AccessShareLock;
 	ExecInitRangeTable(estate, list_make1(rte));
+	ExecInitResultRelationsArray(estate);
 
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
 
-	estate->es_result_relations = resultRelInfo;
-	estate->es_num_result_relations = 1;
+	estate->es_result_relations[0] = resultRelInfo;
 	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 415e117..54455e1 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -538,6 +538,9 @@ extern bool ExecRelationIsTargetRelation(EState *estate, Index scanrelid);
 extern Relation ExecOpenScanRelation(EState *estate, Index scanrelid, int eflags);
 
 extern void ExecInitRangeTable(EState *estate, List *rangeTable);
+extern void ExecInitResultRelationsArray(EState *estate);
+extern void ExecCloseRangeTableRelations(EState *estate);
+extern void ExecCloseResultRelations(EState *estate);
 
 static inline RangeTblEntry *
 exec_rt_fetch(Index rti, EState *estate)
@@ -546,6 +549,8 @@ exec_rt_fetch(Index rti, EState *estate)
 }
 
 extern Relation ExecGetRangeTableRelation(EState *estate, Index rti);
+extern void ExecInitResultRelation(EState *estate, ResultRelInfo *resultRelInfo,
+					   Index rti);
 
 extern int	executor_errposition(EState *estate, int location);
 
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index ef448d6..7eb5ca6 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -508,6 +508,14 @@ typedef struct EState
 	Index		es_range_table_size;	/* size of the range table arrays */
 	Relation   *es_relations;	/* Array of per-range-table-entry Relation
 								 * pointers, or NULL if not yet opened */
+	ResultRelInfo **es_result_relations;	/* Array of Per-range-table-entry
+											 * ResultRelInfo pointers, or
+											 * NULL if a given range table
+											 * relation not a target
+											 * table */
+	List		*es_opened_result_relations;	/* List of non-NULL entries
+												 * in es_result_relations added
+												 * in no specific order */
 	struct ExecRowMark **es_rowmarks;	/* Array of per-range-table-entry
 										 * ExecRowMarks, or NULL if none */
 	PlannedStmt *es_plannedstmt;	/* link to top of plan tree */
@@ -518,24 +526,13 @@ typedef struct EState
 	/* If query can insert/delete tuples, the command ID to mark them with */
 	CommandId	es_output_cid;
 
-	/* Info about target table(s) for insert/update/delete queries: */
-	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
-	int			es_num_result_relations;	/* length of array */
 	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
-	/*
-	 * Info about the partition root table(s) for insert/update/delete queries
-	 * targeting partitioned tables.  Only leaf partitions are mentioned in
-	 * es_result_relations, but we need access to the roots for firing
-	 * triggers and for runtime tuple routing.
-	 */
-	ResultRelInfo *es_root_result_relations;	/* array of ResultRelInfos */
-	int			es_num_root_result_relations;	/* length of the array */
 	PartitionDirectory es_partition_directory;	/* for PartitionDesc lookup */
 
 	/*
 	 * The following list contains ResultRelInfos created by the tuple routing
-	 * code for partitions that don't already have one.
+	 * code for partitions that aren't found in es_result_relations_array.
 	 */
 	List	   *es_tuple_routing_result_relations;
 
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 83e0107..0666546 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -224,8 +224,6 @@ typedef struct ModifyTable
 	Index		rootRelation;	/* Root RT index, if target is partitioned */
 	bool		partColsUpdated;	/* some part key in hierarchy updated */
 	List	   *resultRelations;	/* integer list of RT indexes */
-	int			resultRelIndex; /* index of first resultRel in plan's list */
-	int			rootResultRelIndex; /* index of the partitioned table root */
 	List	   *plans;			/* plan(s) producing source data */
 	List	   *withCheckOptionLists;	/* per-target-table WCO lists */
 	List	   *returningLists; /* per-target-table RETURNING tlists */
-- 
1.8.3.1

From b5a268bdb661f2fc4ecf327418d6cd7a03c98d69 Mon Sep 17 00:00:00 2001
From: Etsuro Fujita <efujita@postgresql.org>
Date: Thu, 8 Aug 2019 21:41:12 +0900
Subject: [PATCH v13 2/5] Include result relation index if any in ForeignScan

FDWs that can perform an UPDATE/DELETE remotely using the "direct
modify" set of APIs need in some cases to access the result relation
properties for which they can currently look at
EState.es_result_relation_info, which the core executor laboriously
makes sure is set correctly.  An upcoming patch will remove that
field from EState.  So this commit installs a new field resultRelation
in ForeignScan node which will be set by the core planner for an FDW
to peruse during a "direct modification" operation; it gives the
range table index of the target foreign table.

Amit Langote, Etsuro Fujita
---
 contrib/postgres_fdw/postgres_fdw.c     | 23 +++++++++++++++--------
 doc/src/sgml/fdwhandler.sgml            | 13 +++++++------
 src/backend/executor/nodeForeignscan.c  |  5 ++++-
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c | 13 +++++++++++++
 src/backend/optimizer/plan/setrefs.c    |  4 ++++
 src/include/nodes/plannodes.h           |  8 ++++++++
 9 files changed, 54 insertions(+), 15 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index a31abce..8eaec6c 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -200,6 +200,8 @@ typedef struct PgFdwDirectModifyState
 	Relation	rel;			/* relcache entry for the foreign table */
 	AttInMetadata *attinmeta;	/* attribute datatype conversion metadata */
 
+	Index			resultRelation;	/* Target foreign table's RT index */
+
 	/* extracted fdw_private data */
 	char	   *query;			/* text of UPDATE/DELETE command */
 	bool		has_returning;	/* is there a RETURNING clause? */
@@ -446,11 +448,12 @@ static List *build_remote_returning(Index rtindex, Relation rel,
 									List *returningList);
 static void rebuild_fdw_scan_tlist(ForeignScan *fscan, List *tlist);
 static void execute_dml_stmt(ForeignScanState *node);
-static TupleTableSlot *get_returning_data(ForeignScanState *node);
+static TupleTableSlot *get_returning_data(ForeignScanState *node, ResultRelInfo *resultRelInfo);
 static void init_returning_filter(PgFdwDirectModifyState *dmstate,
 								  List *fdw_scan_tlist,
 								  Index rtindex);
 static TupleTableSlot *apply_returning_filter(PgFdwDirectModifyState *dmstate,
+											  ResultRelInfo *relInfo,
 											  TupleTableSlot *slot,
 											  EState *estate);
 static void prepare_query_params(PlanState *node,
@@ -2355,7 +2358,8 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	Assert(fsplan->resultRelation > 0);
+	dmstate->resultRelation = rtindex = fsplan->resultRelation;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2450,7 +2454,10 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = estate->es_result_relations[dmstate->resultRelation - 1];
+
+	/* The executor must have initialized the ResultRelInfo for us. */
+	Assert(resultRelInfo != NULL);
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -2482,7 +2489,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 	/*
 	 * Get the next RETURNING tuple.
 	 */
-	return get_returning_data(node);
+	return get_returning_data(node, resultRelInfo);
 }
 
 /*
@@ -4082,11 +4089,10 @@ execute_dml_stmt(ForeignScanState *node)
  * Get the result of a RETURNING clause.
  */
 static TupleTableSlot *
-get_returning_data(ForeignScanState *node)
+get_returning_data(ForeignScanState *node, ResultRelInfo *resultRelInfo)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4141,7 +4147,8 @@ get_returning_data(ForeignScanState *node)
 		if (dmstate->rel)
 			resultSlot = slot;
 		else
-			resultSlot = apply_returning_filter(dmstate, slot, estate);
+			resultSlot = apply_returning_filter(dmstate, resultRelInfo, slot,
+												estate);
 	}
 	dmstate->next_tuple++;
 
@@ -4230,10 +4237,10 @@ init_returning_filter(PgFdwDirectModifyState *dmstate,
  */
 static TupleTableSlot *
 apply_returning_filter(PgFdwDirectModifyState *dmstate,
+					   ResultRelInfo *relInfo,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 72fa127..e995514 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -893,10 +893,10 @@ BeginDirectModify(ForeignScanState *node,
      its <structfield>fdw_state</structfield> field is still NULL.  Information about
      the table to modify is accessible through the
      <structname>ForeignScanState</structname> node (in particular, from the underlying
-     <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
-     <literal>eflags</literal> contains flag bits describing the executor's
-     operating mode for this plan node.
+     <structname>ForeignScan</structname> plan node, which contains the table's range
+     table index along with any FDW-private information provided by
+     <function>PlanDirectModify</function>.  <literal>eflags</literal> contains flag
+     bits describing the executor's operating mode for this plan node.
     </para>
 
     <para>
@@ -926,8 +926,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     obtained using the information passed to <function>BeginDirectModify</function>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 513471a..515860d 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -221,10 +221,13 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan or the direct modification.
 	 */
 	if (node->operation != CMD_SELECT)
+	{
+		Assert(node->resultRelation > 0);
 		fdwroutine->BeginDirectModify(scanstate, eflags);
+	}
 	else
 		fdwroutine->BeginForeignScan(scanstate, eflags);
 
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 7974fa0..70476d8 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -759,6 +759,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelation);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index f35d2a0..2d6efb5 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -696,6 +696,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelation);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 73b8d9d..1a9d01f 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -2015,6 +2015,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelation);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 881eaf4..fd13ac9 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5541,6 +5541,8 @@ make_foreignscan(List *qptlist,
 	node->fs_relids = NULL;
 	/* fsSystemCol will be filled in by create_foreignscan_plan */
 	node->fsSystemCol = false;
+	/* resultRelation will be set by make_modifytable(), if needed */
+	node->resultRelation = 0;
 
 	return node;
 }
@@ -6898,7 +6900,18 @@ make_modifytable(PlannerInfo *root,
 			!has_stored_generated_columns(subroot, rti))
 			direct_modify = fdwroutine->PlanDirectModify(subroot, node, rti, i);
 		if (direct_modify)
+		{
+			ForeignScan   *fscan = (ForeignScan *) list_nth(node->plans, i);
+
+			/*
+			 * For result relations that will be modified directly, the FDW
+			 * needs to know where to find them.
+			 */
+			Assert(IsA(fscan, ForeignScan));
+			fscan->resultRelation = rti;
+
 			direct_modify_plans = bms_add_member(direct_modify_plans, i);
+		}
 
 		if (!direct_modify &&
 			fdwroutine != NULL &&
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 66372d3..6db6e72 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -1315,6 +1315,10 @@ set_foreignscan_references(PlannerInfo *root,
 	}
 
 	fscan->fs_relids = offset_relid_set(fscan->fs_relids, rtoffset);
+
+	/* Adjust resultRelation if needed */
+	if (fscan->resultRelation > 0)
+		fscan->resultRelation += rtoffset;
 }
 
 /*
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 0666546..cadfdb9 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -605,6 +605,11 @@ typedef struct WorkTableScan
  * When the plan node represents a foreign join, scan.scanrelid is zero and
  * fs_relids must be consulted to identify the join relation.  (fs_relids
  * is valid for simple scans as well, but will always match scan.scanrelid.)
+ *
+ * If an FDW's PlanDirectModify() callback decides to repurpose a ForeignScan
+ * node to store the information about an UPDATE or DELETE operation to
+ * to perform on a given foreign table result relation, resultRelation is set
+ * to identify the relation.
  * ----------------
  */
 typedef struct ForeignScan
@@ -618,6 +623,9 @@ typedef struct ForeignScan
 	List	   *fdw_recheck_quals;	/* original quals not in scan.plan.qual */
 	Bitmapset  *fs_relids;		/* RTIs generated by this scan */
 	bool		fsSystemCol;	/* true if any "system column" is needed */
+	Index		resultRelation;	/* Target foreign table's RT index; valid for
+								 * result relations of UPDATE/DELETE; 0 for
+								 * other query types */
 } ForeignScan;
 
 /* ----------------
-- 
1.8.3.1

From 3975f8f3f17c2808f6b056b0558758261a59e514 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v13 5/5] Revise child-to-root tuple conversion map management

Transition tuple capture requires to convert child tuples to the
inheritance root table format because that's the format the
transition tuplestore stores tuple in.  For INSERTs into partitioned
tables, the conversion is handled by tuple routing code which
constructs the map for a given partition only if the partition is
targeted, but for UPDATE and DELETE, maps for all result relations
are made and stored in an array in ModifyTableState during
ExecInitModifyTable, which requires their ResultRelInfos to have been
already built. During execution, map for the currently active result
relation is set in TransitionCaptureState.tcs_map.

This commit removes TransitionCaptureMap.tcs_map in favor a new
map field in ResultRelInfo named ri_ChildToRootMap that is
initialized when the ResultRelInfo for a given result relation is.
This way is less confusing and less bug-prone than setting and
resetting tcs_map. Also, this will also allow us to delay creating
the map for a given result relation to when that relation is actually
processed during execution.
---
 src/backend/commands/copy.c            |  30 +----
 src/backend/commands/trigger.c         |   9 +-
 src/backend/executor/execPartition.c   |  20 ++-
 src/backend/executor/nodeModifyTable.c | 224 ++++++++++-----------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/nodes/execnodes.h          |  11 +-
 6 files changed, 102 insertions(+), 202 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 2d9e1f3..99b0f62 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3105,32 +3105,14 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition which may change the tuple, we can
+			 * just remember the original unconverted tuple to avoid a
+			 * needless round trip conversion.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccf..d1b5a03 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4293,8 +4294,8 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * tables, then return NULL.
  *
  * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * should set tcs_original_insert_tuple as appropriate when dealing with child
+ * tables
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5389,7 +5390,9 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	if (row_trigger && transition_capture != NULL)
 	{
 		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		PartitionRoutingInfo *pinfo = relinfo->ri_PartitionInfo;
+		TupleConversionMap *map = pinfo ? pinfo->pi_PartitionToRootMap :
+			relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 33d2c6f..1c20f6c 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -983,9 +983,23 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	if (mtstate &&
 		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
 	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot));
+		ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+
+		/*
+		 * If the partition appears to be a reused UPDATE result relation, the
+		 * necessary map would already have been set in ri_ChildToRootMap by
+		 * ExecInitModifyTable(), so use that one instead of building one from
+		 * scratch.  One can tell if it's actually a reused UPDATE result
+		 * relation by looking at its ri_RangeTableIndex which must be
+		 * different from the root RT index.
+		 */
+		if (node && node->operation == CMD_UPDATE &&
+			node->rootRelation != partRelInfo->ri_RangeTableIndex)
+			partrouteinfo->pi_PartitionToRootMap = partRelInfo->ri_ChildToRootMap;
+		else
+			partrouteinfo->pi_PartitionToRootMap =
+				convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
+									   RelationGetDescr(partRelInfo->ri_PartitionRoot));
 	}
 	else
 		partrouteinfo->pi_PartitionToRootMap = NULL;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 3405ef5..e5e2564 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -1086,9 +1083,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
 	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1165,39 +1160,29 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-	 * should convert the tuple into root's tuple descriptor, since
-	 * ExecInsert() starts the search from root.  The tuple conversion
-	 * map list is in the order of mtstate->resultRelInfo[], so to
-	 * retrieve the one for this resultRel, we need to know the
-	 * position of the resultRel in mtstate->resultRelInfo[].
+	 * resultRelInfo is one of the per-subplan resultRelInfos and the tuple
+	 * in the input slot is in its format.  Before we can call ExecInsert() to
+	 * route the tuple, which expects the root result relation to be passed,
+	 * we must convert the tuple into root relation's tuple descriptor, if at
+	 * all needed.
 	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-	 * so save the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
+	/*
+	 * Reset the transition state that may possibly have been written
+	 * by INSERT.
+	 */
 	if (mtstate->mt_transition_capture)
-	{
 		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
 
 	/* We're done moving. */
 	return true;
@@ -1905,28 +1890,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1950,6 +1913,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
@@ -1964,37 +1928,15 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE triggers
+	 * on the partition which may change the tuple, we can just remember the
+	 * original unconverted tuple to avoid a needless round trip conversion.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -2010,58 +1952,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc);
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2157,17 +2047,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2337,6 +2216,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRelInfo;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2358,13 +2238,24 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		palloc(nplans * sizeof(ResultRelInfo));
 	mtstate->mt_scans = (TupleTableSlot **) palloc0(sizeof(TupleTableSlot *) * nplans);
 
-	/* If modifying a partitioned table, initialize the root table info */
+	/*
+	 * Initialize the designated "root" result relation.  When modifying
+	 * partitioned tables, it's given by node->rootRelation, while in other
+	 * cases, it's the first relation in node->resultRelations.  We need to
+	 * initialize this one before any others, because
+	 * ExecSetupTransitionCaptureState() needs it.
+	 */
 	if (node->rootRelation > 0)
 	{
 		mtstate->rootResultRelInfo = makeNode(ResultRelInfo);
 		ExecInitResultRelation(estate, mtstate->rootResultRelInfo,
 							   node->rootRelation);
 	}
+	else
+		ExecInitResultRelation(estate, mtstate->resultRelInfo,
+							   linitial_int(node->resultRelations));
+
+	rootResultRelInfo = getTargetResultRelInfo(mtstate);
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2374,6 +2265,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2387,8 +2285,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 		subplan = (Plan *) lfirst(l1);
 
-		/* This opens result relation and fills ResultRelInfo. */
-		ExecInitResultRelation(estate, resultRelInfo, resultRelation);
+		/*
+		 * This opens result relation and fills ResultRelInfo.
+		 * ("root" relation already opened.)
+		 */
+		if (resultRelInfo != rootResultRelInfo)
+			ExecInitResultRelation(estate, resultRelInfo, resultRelation);
 
 		/* Initialize the usesFdwDirectModify flag */
 		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(i,
@@ -2444,12 +2346,28 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.  During INSERT, partition tuples to
+		 * store into the transition tuple store are converted using
+		 * PartitionToRoot map in the partition's PartitionRoutingInfo.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRelInfo->ri_RelationDesc));
 		resultRelInfo++;
 		i++;
 	}
 
 	/* Get the target relation */
-	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
+	rel = rootResultRelInfo->ri_RelationDesc;
 
 	/*
 	 * If it's not a partitioned table after all, UPDATE tuple routing should
@@ -2468,26 +2386,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a40ddf5..e38d732 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -46,7 +46,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -66,14 +66,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 36a3318..e657493 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -488,6 +488,14 @@ typedef struct ResultRelInfo
 
 	/* for use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child result relation tuples to the format of the
+	 * table actually mentioned in the query (called "root").  Set only
+	 * if either transition tuple capture or update partition row
+	 * movement is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1174,9 +1182,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
1.8.3.1

From 500e3459e84ed5291ec810580f28d65fa1eeb6eb Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 16:24:38 +0900
Subject: [PATCH v13 4/5] Rearrange partition update row movement code a bit

The block of code that does the actual moving (DELETE+INSERT) has
been moved to a function named ExecCrossPartitionUpdate() which must
be retried until it says the movement has been done or can't be done.

This also rearrange the code in ExecDelete() and ExecInsert() around
executing AFTER ROW DELETE and AFTER ROW INSERT triggers, resp.  In
the case of an update row movement, such triggers should not see the
affected tuple in their OLD/NEW transition table.
---
 src/backend/executor/nodeModifyTable.c | 347 +++++++++++++++++++--------------
 1 file changed, 199 insertions(+), 148 deletions(-)

diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9c51320..3405ef5 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -389,7 +389,6 @@ ExecInsert(ModifyTableState *mtstate,
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
-	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
@@ -655,31 +654,30 @@ ExecInsert(ModifyTableState *mtstate,
 	}
 
 	/*
-	 * If this insert is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition NEW TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the insert is a part of update row movement, put this row into the
+	 * UPDATE trigger's NEW TABLE (transition table) instead of that of an
+	 * INSERT trigger.
 	 */
-	ar_insert_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_new_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_new_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo, NULL,
-							 NULL,
-							 slot,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, NULL, NULL, slot,
+							 NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * INSERT trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW INSERT Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_insert_trig_tcs = NULL;
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW INSERT Triggers */
+		ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW INSERT Triggers */
-	ExecARInsertTriggers(estate, resultRelInfo, slot, recheckIndexes,
-						 ar_insert_trig_tcs);
 
 	list_free(recheckIndexes);
 
@@ -745,7 +743,6 @@ ExecDelete(ModifyTableState *mtstate,
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
-	TransitionCaptureState *ar_delete_trig_tcs;
 
 	if (tupleDeleted)
 		*tupleDeleted = false;
@@ -989,32 +986,30 @@ ldelete:;
 		*tupleDeleted = true;
 
 	/*
-	 * If this delete is the result of a partition key update that moved the
-	 * tuple to a new partition, put this row into the transition OLD TABLE,
-	 * if there is one. We need to do this separately for DELETE and INSERT
-	 * because they happen on different tables.
+	 * If the delete is a part of update row movement, put this row into the
+	 * UPDATE trigger's OLD TABLE (transition table) instead of that of an
+	 * DELETE trigger.
 	 */
-	ar_delete_trig_tcs = mtstate->mt_transition_capture;
-	if (mtstate->operation == CMD_UPDATE && mtstate->mt_transition_capture
-		&& mtstate->mt_transition_capture->tcs_update_old_table)
+	if (mtstate->operation == CMD_UPDATE &&
+		mtstate->mt_transition_capture &&
+		mtstate->mt_transition_capture->tcs_update_old_table)
 	{
-		ExecARUpdateTriggers(estate, resultRelInfo,
-							 tupleid,
-							 oldtuple,
-							 NULL,
-							 NULL,
-							 mtstate->mt_transition_capture);
+		ExecARUpdateTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL, NIL, mtstate->mt_transition_capture);
 
 		/*
-		 * We've already captured the NEW TABLE row, so make sure any AR
-		 * DELETE trigger fired below doesn't capture it again.
+		 * Execute AFTER ROW DELETE Triggers, but such that the row is not
+		 * captured again in the transition table if any.
 		 */
-		ar_delete_trig_tcs = NULL;
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 NULL);
+	}
+	else
+	{
+		/* AFTER ROW DELETE Triggers */
+		ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
+							 mtstate->mt_transition_capture);
 	}
-
-	/* AFTER ROW DELETE Triggers */
-	ExecARDeleteTriggers(estate, resultRelInfo, tupleid, oldtuple,
-						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
 	if (processReturning && resultRelInfo->ri_projectReturning)
@@ -1061,6 +1056,153 @@ ldelete:;
 	return NULL;
 }
 
+/*
+ *	ExecCrossPartitionUpdate
+ *		Move an updated tuple from a given partition to the correct partition
+ *		of its root parent table
+ *
+ *	This works by first deleting the tuple from the current partition,
+ *	followed by inserting it into the root parent table, that is,
+ *	mtstate->rootResultRelInfo, from where it's re-routed to the correct
+ *	partition.
+ *
+ *	Returns true if the tuple has been successfully moved or if it's found
+ *	that the tuple was concurrently deleted so there's nothing more to do
+ *	for the caller.
+ *
+ *	False is returned if the tuple we're trying to move is found to have been
+ *	concurrently updated.  Caller should check if the updated tuple that's
+ *	returned in *retry_slot still needs to be re-routed and call this function
+ *	again if needed.
+ */
+static bool
+ExecCrossPartitionUpdate(ModifyTableState *mtstate,
+						 ResultRelInfo *resultRelInfo,
+						 ItemPointer tupleid, HeapTuple oldtuple,
+						 TupleTableSlot *slot, TupleTableSlot *planSlot,
+						 EPQState *epqstate, bool canSetTag,
+						 TupleTableSlot **retry_slot,
+						 TupleTableSlot **inserted_tuple)
+{
+	EState	   *estate = mtstate->ps.state;
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	int			map_index;
+	TupleConversionMap *tupconv_map;
+	TupleConversionMap *saved_tcs_map = NULL;
+	bool		tuple_deleted;
+	TupleTableSlot *epqslot = NULL;
+
+	*inserted_tuple = NULL;
+	*retry_slot = NULL;
+
+	/*
+	 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
+	 * original row to migrate to a different partition.  Maybe this
+	 * can be implemented some day, but it seems a fringe feature with
+	 * little redeeming value.
+	 */
+	if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
+		ereport(ERROR,
+				(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
+				 errmsg("invalid ON UPDATE specification"),
+				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
+
+	/*
+	 * When an UPDATE is run on a leaf partition, we will not have
+	 * partition tuple routing set up. In that case, fail with
+	 * partition constraint violation error.
+	 */
+	if (proute == NULL)
+		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+	/*
+	 * Row movement, part 1.  Delete the tuple, but skip RETURNING
+	 * processing. We want to return rows from INSERT.
+	 */
+	ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+			   epqstate, estate,
+			   false,	/* processReturning */
+			   false,	/* canSetTag */
+			   true,	/* changingPart */
+			   &tuple_deleted, &epqslot);
+
+	/*
+	 * For some reason if DELETE didn't happen (e.g. trigger prevented
+	 * it, or it was already deleted by self, or it was concurrently
+	 * deleted by another transaction), then we should skip the insert
+	 * as well; otherwise, an UPDATE could cause an increase in the
+	 * total number of rows across all partitions, which is clearly
+	 * wrong.
+	 *
+	 * For a normal UPDATE, the case where the tuple has been the
+	 * subject of a concurrent UPDATE or DELETE would be handled by
+	 * the EvalPlanQual machinery, but for an UPDATE that we've
+	 * translated into a DELETE from this partition and an INSERT into
+	 * some other partition, that's not available, because CTID chains
+	 * can't span relation boundaries.  We mimic the semantics to a
+	 * limited extent by skipping the INSERT if the DELETE fails to
+	 * find a tuple. This ensures that two concurrent attempts to
+	 * UPDATE the same tuple at the same time can't turn one tuple
+	 * into two, and that an UPDATE of a just-deleted tuple can't
+	 * resurrect it.
+	 */
+	if (!tuple_deleted)
+	{
+		/*
+		 * epqslot will be typically NULL.  But when ExecDelete()
+		 * finds that another transaction has concurrently updated the
+		 * same row, it re-fetches the row, skips the delete, and
+		 * epqslot is set to the re-fetched tuple slot. In that case,
+		 * we need to do all the checks again.
+		 */
+		if (TupIsNull(epqslot))
+			return true;
+		else
+		{
+			*retry_slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
+			return false;
+		}
+	}
+
+	/*
+	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
+	 * should convert the tuple into root's tuple descriptor, since
+	 * ExecInsert() starts the search from root.  The tuple conversion
+	 * map list is in the order of mtstate->resultRelInfo[], so to
+	 * retrieve the one for this resultRel, we need to know the
+	 * position of the resultRel in mtstate->resultRelInfo[].
+	 */
+	map_index = resultRelInfo - mtstate->resultRelInfo;
+	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
+	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	if (tupconv_map != NULL)
+		slot = execute_attr_map_slot(tupconv_map->attrMap,
+									 slot,
+									 mtstate->mt_root_tuple_slot);
+
+	/*
+	 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+	 * so save the currently active map.
+	 */
+	if (mtstate->mt_transition_capture)
+		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
+
+	/* Tuple routing starts from the root table. */
+	Assert(mtstate->rootResultRelInfo != NULL);
+	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								 planSlot, estate, canSetTag);
+
+	/* Clear the INSERT's tuple and restore the saved map. */
+	if (mtstate->mt_transition_capture)
+	{
+		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
+		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+	}
+
+	/* We're done moving. */
+	return true;
+}
+
 /* ----------------------------------------------------------------
  *		ExecUpdate
  *
@@ -1216,119 +1358,28 @@ lreplace:;
 		 */
 		if (partition_constraint_failed)
 		{
-			bool		tuple_deleted;
-			TupleTableSlot *ret_slot;
-			TupleTableSlot *epqslot = NULL;
-			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-			int			map_index;
-			TupleConversionMap *tupconv_map;
-			TupleConversionMap *saved_tcs_map = NULL;
-
-			/*
-			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
-			 * original row to migrate to a different partition.  Maybe this
-			 * can be implemented some day, but it seems a fringe feature with
-			 * little redeeming value.
-			 */
-			if (((ModifyTable *) mtstate->ps.plan)->onConflictAction == ONCONFLICT_UPDATE)
-				ereport(ERROR,
-						(errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
-						 errmsg("invalid ON UPDATE specification"),
-						 errdetail("The result tuple would appear in a different partition than the original tuple.")));
-
-			/*
-			 * When an UPDATE is run on a leaf partition, we will not have
-			 * partition tuple routing set up. In that case, fail with
-			 * partition constraint violation error.
-			 */
-			if (proute == NULL)
-				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
-
-			/*
-			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
-			 * processing. We want to return rows from INSERT.
-			 */
-			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
-					   epqstate, estate,
-					   false,	/* processReturning */
-					   false,	/* canSetTag */
-					   true,	/* changingPart */
-					   &tuple_deleted, &epqslot);
+			TupleTableSlot *inserted_tuple,
+						   *retry_slot;
+			bool			retry;
 
 			/*
-			 * For some reason if DELETE didn't happen (e.g. trigger prevented
-			 * it, or it was already deleted by self, or it was concurrently
-			 * deleted by another transaction), then we should skip the insert
-			 * as well; otherwise, an UPDATE could cause an increase in the
-			 * total number of rows across all partitions, which is clearly
-			 * wrong.
-			 *
-			 * For a normal UPDATE, the case where the tuple has been the
-			 * subject of a concurrent UPDATE or DELETE would be handled by
-			 * the EvalPlanQual machinery, but for an UPDATE that we've
-			 * translated into a DELETE from this partition and an INSERT into
-			 * some other partition, that's not available, because CTID chains
-			 * can't span relation boundaries.  We mimic the semantics to a
-			 * limited extent by skipping the INSERT if the DELETE fails to
-			 * find a tuple. This ensures that two concurrent attempts to
-			 * UPDATE the same tuple at the same time can't turn one tuple
-			 * into two, and that an UPDATE of a just-deleted tuple can't
-			 * resurrect it.
+			 * ExecCrossPartitionUpdate will first DELETE the row from the
+			 * partition it's currently in and then insert it back into the
+			 * root table, which will re-route it to the correct partition.
+			 * The first part may have to be repeated if it is detected that
+			 * the tuple we're trying to move has been concurrently updated.
 			 */
-			if (!tuple_deleted)
-			{
-				/*
-				 * epqslot will be typically NULL.  But when ExecDelete()
-				 * finds that another transaction has concurrently updated the
-				 * same row, it re-fetches the row, skips the delete, and
-				 * epqslot is set to the re-fetched tuple slot. In that case,
-				 * we need to do all the checks again.
-				 */
-				if (TupIsNull(epqslot))
-					return NULL;
-				else
-				{
-					slot = ExecFilterJunk(resultRelInfo->ri_junkFilter, epqslot);
-					goto lreplace;
-				}
-			}
-
-			/*
-			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
-			 * should convert the tuple into root's tuple descriptor, since
-			 * ExecInsert() starts the search from root.  The tuple conversion
-			 * map list is in the order of mtstate->resultRelInfo[], so to
-			 * retrieve the one for this resultRel, we need to know the
-			 * position of the resultRel in mtstate->resultRelInfo[].
-			 */
-			map_index = resultRelInfo - mtstate->resultRelInfo;
-			Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-			tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
-			if (tupconv_map != NULL)
-				slot = execute_attr_map_slot(tupconv_map->attrMap,
-											 slot,
-											 mtstate->mt_root_tuple_slot);
-
-			/*
-			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
-			 * so save the currently active map.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/* Tuple routing starts from the root table. */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
-								  planSlot, estate, canSetTag);
-
-			/* Clear the INSERT's tuple and restore the saved map. */
-			if (mtstate->mt_transition_capture)
+			retry = !ExecCrossPartitionUpdate(mtstate, resultRelInfo, tupleid,
+											  oldtuple, slot, planSlot,
+											  epqstate, canSetTag,
+											  &retry_slot, &inserted_tuple);
+			if (retry)
 			{
-				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-				mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
+				slot = retry_slot;
+				goto lreplace;
 			}
 
-			return ret_slot;
+			return inserted_tuple;
 		}
 
 		/*
-- 
1.8.3.1

From e75bae4f6ad8f26faae0bef502af8a09ddbc4dfe Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Fri, 19 Jul 2019 14:53:20 +0900
Subject: [PATCH v13 3/5] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.
---
 src/backend/commands/copy.c              |  19 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |  12 +-
 src/backend/executor/execMain.c          |   6 -
 src/backend/executor/execReplication.c   |  24 ++--
 src/backend/executor/execUtils.c         |   1 -
 src/backend/executor/nodeModifyTable.c   | 193 +++++++++++++------------------
 src/backend/replication/logical/worker.c |  44 +++----
 src/include/executor/executor.h          |  19 ++-
 src/include/executor/nodeModifyTable.h   |   4 +-
 src/include/nodes/execnodes.h            |   2 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 146 insertions(+), 188 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 6948214..2d9e1f3 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2489,9 +2489,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2524,7 +2521,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2839,8 +2837,6 @@ CopyFrom(CopyState cstate)
 
 	ExecOpenIndices(resultRelInfo, false);
 
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Set up a ModifyTableState so we can let FDW(s) init themselves for
 	 * foreign-table result relation(s).
@@ -3109,11 +3105,6 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
-			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
 			 */
@@ -3217,7 +3208,8 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot, CMD_INSERT);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot,
+											   CMD_INSERT);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3288,7 +3280,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 1cd8cfb..feec994 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1821,7 +1821,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1951,7 +1950,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 1862af6..e1d34be 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -498,10 +496,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	ItemPointerSetInvalid(conflictTid);
 	ItemPointerSetInvalid(&invalidItemPtr);
 
-	/*
-	 * Get information from the result relation info structure.
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index f62fd8f..425d018 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -828,13 +828,8 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 	estate->es_plannedstmt = plannedstmt;
 
 	if (plannedstmt->resultRelations)
-	{
 		ExecInitResultRelationsArray(estate);
 
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-	}
-
 	/*
 	 * Next, build the ExecRowMark array from the PlanRowMark(s), if any.
 	 */
@@ -2710,7 +2705,6 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 	 */
 	if (parentestate->es_result_relations)
 		ExecInitResultRelationsArray(rcestate);
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index b29db7b..01d2688 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -404,10 +404,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -430,7 +430,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -442,7 +443,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -466,11 +468,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -496,7 +498,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -508,7 +511,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -527,11 +531,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index 169dd92..d96093f 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -124,7 +124,6 @@ CreateExecutorState(void)
 	estate->es_output_cid = (CommandId) 0;
 
 	estate->es_result_relations = NULL;
-	estate->es_result_relation_info = NULL;
 	estate->es_tuple_routing_result_relations = NIL;
 	estate->es_trig_target_relations = NIL;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9b27d34..9c51320 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,10 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -366,32 +368,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
-
-	ExecMaterializeSlot(slot);
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 
 	/*
-	 * get information on the (current) result relation
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
+
+	ExecMaterializeSlot(slot);
+
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -424,7 +442,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -459,7 +478,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -521,8 +541,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -582,7 +602,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -621,7 +642,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -707,6 +729,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -718,7 +741,6 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
@@ -728,10 +750,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW DELETE Triggers */
@@ -1067,6 +1085,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1075,12 +1094,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1090,10 +1107,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1120,7 +1133,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1157,7 +1171,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1207,6 +1222,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1232,9 +1248,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1275,16 +1294,6 @@ lreplace:;
 			}
 
 			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
-			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
 			 * ExecInsert() starts the search from root.  The tuple conversion
@@ -1301,18 +1310,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
+			/* Tuple routing starts from the root table. */
+			Assert(mtstate->rootResultRelInfo != NULL);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1476,7 +1485,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1715,7 +1725,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1872,41 +1882,37 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
- *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
-	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
 	 */
@@ -2016,10 +2022,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2068,17 +2072,6 @@ ExecModifyTable(PlanState *pstate)
 	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
-	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
 	 */
@@ -2111,7 +2104,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2156,7 +2148,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2239,25 +2230,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2269,15 +2256,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2298,7 +2279,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l,
@@ -2346,14 +2326,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	forboth(l, node->resultRelations, l1, node->plans)
@@ -2400,7 +2374,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2424,8 +2397,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 77f71e5..d92b90f 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -361,7 +361,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
 
 	estate->es_result_relations[0] = resultRelInfo;
-	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
 
@@ -1150,6 +1149,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -1176,6 +1176,7 @@ apply_handle_insert(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -1191,11 +1192,10 @@ apply_handle_insert(StringInfo s)
 
 	/* For a partitioned table, insert the tuple into a partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, NULL, rel, CMD_INSERT);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, NULL,
+								   rel, CMD_INSERT);
 	else
-		apply_handle_insert_internal(estate->es_result_relation_info, estate,
-									 remoteslot);
+		apply_handle_insert_internal(resultRelInfo, estate, remoteslot);
 
 	PopActiveSnapshot();
 
@@ -1218,7 +1218,7 @@ apply_handle_insert_internal(ResultRelInfo *relinfo,
 	ExecOpenIndices(relinfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(relinfo, estate, remoteslot);
 
 	/* Cleanup. */
 	ExecCloseIndices(relinfo);
@@ -1265,6 +1265,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	EState	   *estate;
@@ -1298,6 +1299,7 @@ apply_handle_update(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -1337,11 +1339,11 @@ apply_handle_update(StringInfo s)
 
 	/* For a partitioned table, apply update to correct partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, &newtup, rel, CMD_UPDATE);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, &newtup,
+								   rel, CMD_UPDATE);
 	else
-		apply_handle_update_internal(estate->es_result_relation_info, estate,
-									 remoteslot, &newtup, rel);
+		apply_handle_update_internal(resultRelInfo, estate, remoteslot,
+									 &newtup, rel);
 
 	PopActiveSnapshot();
 
@@ -1392,7 +1394,8 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(relinfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -1420,6 +1423,7 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -1449,6 +1453,7 @@ apply_handle_delete(StringInfo s)
 
 	/* Initialize the executor state. */
 	estate = create_estate_for_relation(rel);
+	resultRelInfo = estate->es_result_relations[0];
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
@@ -1462,11 +1467,11 @@ apply_handle_delete(StringInfo s)
 
 	/* For a partitioned table, apply delete to correct partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
-								   remoteslot, NULL, rel, CMD_DELETE);
+		apply_handle_tuple_routing(resultRelInfo, estate, remoteslot, NULL,
+								   rel, CMD_DELETE);
 	else
-		apply_handle_delete_internal(estate->es_result_relation_info, estate,
-									 remoteslot, &rel->remoterel);
+		apply_handle_delete_internal(resultRelInfo, estate, remoteslot,
+									 &rel->remoterel);
 
 	PopActiveSnapshot();
 
@@ -1504,7 +1509,7 @@ apply_handle_delete_internal(ResultRelInfo *relinfo, EState *estate,
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(relinfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -1612,7 +1617,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 	}
 	MemoryContextSwitchTo(oldctx);
 
-	estate->es_result_relation_info = partrelinfo;
 	switch (operation)
 	{
 		case CMD_INSERT:
@@ -1693,8 +1697,8 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					ExecOpenIndices(partrelinfo, false);
 
 					EvalPlanQualSetSlot(&epqstate, remoteslot_part);
-					ExecSimpleRelationUpdate(estate, &epqstate, localslot,
-											 remoteslot_part);
+					ExecSimpleRelationUpdate(partrelinfo, estate, &epqstate,
+											 localslot, remoteslot_part);
 					ExecCloseIndices(partrelinfo);
 					EvalPlanQualEnd(&epqstate);
 				}
@@ -1735,7 +1739,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					Assert(partrelinfo_new != partrelinfo);
 
 					/* DELETE old tuple found in the old partition. */
-					estate->es_result_relation_info = partrelinfo;
 					apply_handle_delete_internal(partrelinfo, estate,
 												 localslot,
 												 &relmapentry->remoterel);
@@ -1767,7 +1770,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 						slot_getallattrs(remoteslot);
 					}
 					MemoryContextSwitchTo(oldctx);
-					estate->es_result_relation_info = partrelinfo_new;
 					apply_handle_insert_internal(partrelinfo_new, estate,
 												 remoteslot_part);
 				}
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 54455e1..610169f 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -578,10 +578,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -598,10 +602,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 4ec4ebd..2518fe4 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,9 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 7eb5ca6..36a3318 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -526,8 +526,6 @@ typedef struct EState
 	/* If query can insert/delete tuples, the command ID to mark them with */
 	CommandId	es_output_cid;
 
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
-
 	PartitionDirectory es_partition_directory;	/* for PartitionDesc lookup */
 
 	/*
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index eb9d45b..da50ee3 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index ffd4aac..963faa1 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
1.8.3.1

From 17e22c13509941c561346f386d53fd4f4ff16ef8 Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Fri, 9 Oct 2020 13:52:29 +0900
Subject: [PATCH 1/2] Make es_result_relations array indexable by RT index

This allows us to get rid of the need to keep around a separate set
of indexes for each result relation.
---
 src/backend/commands/copy.c              |  20 +--
 src/backend/commands/explain.c           |  17 +-
 src/backend/commands/tablecmds.c         |  12 +-
 src/backend/executor/execMain.c          | 189 ++++++-----------------
 src/backend/executor/execUtils.c         |  73 ++++++---
 src/backend/executor/nodeModifyTable.c   |  24 ++-
 src/backend/nodes/copyfuncs.c            |   2 -
 src/backend/nodes/outfuncs.c             |   2 -
 src/backend/nodes/readfuncs.c            |   2 -
 src/backend/optimizer/plan/createplan.c  |   2 -
 src/backend/optimizer/plan/setrefs.c     |  10 +-
 src/backend/replication/logical/worker.c |   4 +-
 src/include/executor/executor.h          |   5 +
 src/include/nodes/execnodes.h            |  21 ++-
 src/include/nodes/plannodes.h            |   2 -
 15 files changed, 162 insertions(+), 223 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 3c7dbad27a..6948214334 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2829,25 +2829,18 @@ CopyFrom(CopyState cstate)
 	 * index-entry-making machinery.  (There used to be a huge amount of code
 	 * here that basically duplicated execUtils.c ...)
 	 */
-	resultRelInfo = makeNode(ResultRelInfo);
-	InitResultRelInfo(resultRelInfo,
-					  cstate->rel,
-					  1,		/* must match rel's position in range_table */
-					  NULL,
-					  0);
-	target_resultRelInfo = resultRelInfo;
+	ExecInitRangeTable(estate, cstate->range_table);
+	ExecInitResultRelationsArray(estate);
+	resultRelInfo = target_resultRelInfo = makeNode(ResultRelInfo);
+	ExecInitResultRelation(estate, resultRelInfo, 1);
 
 	/* Verify the named relation is a valid target for INSERT */
 	CheckValidResultRel(resultRelInfo, CMD_INSERT);
 
 	ExecOpenIndices(resultRelInfo, false);
 
-	estate->es_result_relations = resultRelInfo;
-	estate->es_num_result_relations = 1;
 	estate->es_result_relation_info = resultRelInfo;
 
-	ExecInitRangeTable(estate, cstate->range_table);
-
 	/*
 	 * Set up a ModifyTableState so we can let FDW(s) init themselves for
 	 * foreign-table result relation(s).
@@ -2856,7 +2849,7 @@ CopyFrom(CopyState cstate)
 	mtstate->ps.plan = NULL;
 	mtstate->ps.state = estate;
 	mtstate->operation = CMD_INSERT;
-	mtstate->resultRelInfo = estate->es_result_relations;
+	mtstate->resultRelInfo = resultRelInfo;
 
 	if (resultRelInfo->ri_FdwRoutine != NULL &&
 		resultRelInfo->ri_FdwRoutine->BeginForeignInsert != NULL)
@@ -3359,7 +3352,8 @@ CopyFrom(CopyState cstate)
 	if (insertMethod != CIM_SINGLE)
 		CopyMultiInsertInfoCleanup(&multiInsertInfo);
 
-	ExecCloseIndices(target_resultRelInfo);
+	ExecCloseResultRelations(estate);
+	ExecCloseRangeTableRelations(estate);
 
 	/* Close all the partitioned tables, leaf partitions, and their indices */
 	if (proute)
diff --git a/src/backend/commands/explain.c b/src/backend/commands/explain.c
index c98c9b5547..3210f90bd1 100644
--- a/src/backend/commands/explain.c
+++ b/src/backend/commands/explain.c
@@ -769,13 +769,14 @@ ExplainPrintTriggers(ExplainState *es, QueryDesc *queryDesc)
 {
 	ResultRelInfo *rInfo;
 	bool		show_relname;
-	int			numrels = queryDesc->estate->es_num_result_relations;
-	int			numrootrels = queryDesc->estate->es_num_root_result_relations;
+	int			numrels = list_length(queryDesc->plannedstmt->resultRelations);
+	int			numrootrels = list_length(queryDesc->plannedstmt->rootResultRelations);
+	List	   *resultrels;
 	List	   *routerels;
 	List	   *targrels;
-	int			nr;
 	ListCell   *l;
 
+	resultrels = queryDesc->estate->es_opened_result_relations;
 	routerels = queryDesc->estate->es_tuple_routing_result_relations;
 	targrels = queryDesc->estate->es_trig_target_relations;
 
@@ -783,13 +784,11 @@ ExplainPrintTriggers(ExplainState *es, QueryDesc *queryDesc)
 
 	show_relname = (numrels > 1 || numrootrels > 0 ||
 					routerels != NIL || targrels != NIL);
-	rInfo = queryDesc->estate->es_result_relations;
-	for (nr = 0; nr < numrels; rInfo++, nr++)
-		report_triggers(rInfo, show_relname, es);
-
-	rInfo = queryDesc->estate->es_root_result_relations;
-	for (nr = 0; nr < numrootrels; rInfo++, nr++)
+	foreach(l, resultrels)
+	{
+		rInfo = lfirst(l);
 		report_triggers(rInfo, show_relname, es);
+	}
 
 	foreach(l, routerels)
 	{
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index e0ac4e05e5..1cd8cfb41d 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1693,6 +1693,7 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	SubTransactionId mySubid;
 	ListCell   *cell;
 	Oid		   *logrelids;
+	int			dummy_rti;
 
 	/*
 	 * Check the explicitly-specified relations.
@@ -1792,19 +1793,24 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfos = (ResultRelInfo *)
 		palloc(list_length(rels) * sizeof(ResultRelInfo));
 	resultRelInfo = resultRelInfos;
+	estate->es_result_relations = (ResultRelInfo **)
+		palloc(list_length(rels) * sizeof(ResultRelInfo *));
+	dummy_rti = 1;
 	foreach(cell, rels)
 	{
 		Relation	rel = (Relation) lfirst(cell);
 
 		InitResultRelInfo(resultRelInfo,
 						  rel,
-						  0,	/* dummy rangetable index */
+						  dummy_rti,
 						  NULL,
 						  0);
+		estate->es_result_relations[dummy_rti - 1] = resultRelInfo;
+		estate->es_opened_result_relations =
+			lappend(estate->es_opened_result_relations, resultRelInfo);
 		resultRelInfo++;
+		dummy_rti++;
 	}
-	estate->es_result_relations = resultRelInfos;
-	estate->es_num_result_relations = list_length(rels);
 
 	/*
 	 * Process all BEFORE STATEMENT TRUNCATE triggers before we begin
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 2e27e26ba4..f62fd8f7aa 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -827,85 +827,12 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 
 	estate->es_plannedstmt = plannedstmt;
 
-	/*
-	 * Initialize ResultRelInfo data structures, and open the result rels.
-	 */
 	if (plannedstmt->resultRelations)
 	{
-		List	   *resultRelations = plannedstmt->resultRelations;
-		int			numResultRelations = list_length(resultRelations);
-		ResultRelInfo *resultRelInfos;
-		ResultRelInfo *resultRelInfo;
-
-		resultRelInfos = (ResultRelInfo *)
-			palloc(numResultRelations * sizeof(ResultRelInfo));
-		resultRelInfo = resultRelInfos;
-		foreach(l, resultRelations)
-		{
-			Index		resultRelationIndex = lfirst_int(l);
-			Relation	resultRelation;
-
-			resultRelation = ExecGetRangeTableRelation(estate,
-													   resultRelationIndex);
-			InitResultRelInfo(resultRelInfo,
-							  resultRelation,
-							  resultRelationIndex,
-							  NULL,
-							  estate->es_instrument);
-			resultRelInfo++;
-		}
-		estate->es_result_relations = resultRelInfos;
-		estate->es_num_result_relations = numResultRelations;
+		ExecInitResultRelationsArray(estate);
 
 		/* es_result_relation_info is NULL except when within ModifyTable */
 		estate->es_result_relation_info = NULL;
-
-		/*
-		 * In the partitioned result relation case, also build ResultRelInfos
-		 * for all the partitioned table roots, because we will need them to
-		 * fire statement-level triggers, if any.
-		 */
-		if (plannedstmt->rootResultRelations)
-		{
-			int			num_roots = list_length(plannedstmt->rootResultRelations);
-
-			resultRelInfos = (ResultRelInfo *)
-				palloc(num_roots * sizeof(ResultRelInfo));
-			resultRelInfo = resultRelInfos;
-			foreach(l, plannedstmt->rootResultRelations)
-			{
-				Index		resultRelIndex = lfirst_int(l);
-				Relation	resultRelDesc;
-
-				resultRelDesc = ExecGetRangeTableRelation(estate,
-														  resultRelIndex);
-				InitResultRelInfo(resultRelInfo,
-								  resultRelDesc,
-								  resultRelIndex,
-								  NULL,
-								  estate->es_instrument);
-				resultRelInfo++;
-			}
-
-			estate->es_root_result_relations = resultRelInfos;
-			estate->es_num_root_result_relations = num_roots;
-		}
-		else
-		{
-			estate->es_root_result_relations = NULL;
-			estate->es_num_root_result_relations = 0;
-		}
-	}
-	else
-	{
-		/*
-		 * if no result relation, then set state appropriately
-		 */
-		estate->es_result_relations = NULL;
-		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
-		estate->es_root_result_relations = NULL;
-		estate->es_num_root_result_relations = 0;
 	}
 
 	/*
@@ -1334,8 +1261,7 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
  *
  * Most of the time, triggers are fired on one of the result relations of the
  * query, and so we can just return a member of the es_result_relations array,
- * or the es_root_result_relations array (if any), or the
- * es_tuple_routing_result_relations list (if any).  (Note: in self-join
+ * or the es_tuple_routing_result_relations list (if any). (Note: in self-join
  * situations there might be multiple members with the same OID; if so it
  * doesn't matter which one we pick.)
  *
@@ -1352,30 +1278,16 @@ ResultRelInfo *
 ExecGetTriggerResultRel(EState *estate, Oid relid)
 {
 	ResultRelInfo *rInfo;
-	int			nr;
 	ListCell   *l;
 	Relation	rel;
 	MemoryContext oldcontext;
 
 	/* First, search through the query result relations */
-	rInfo = estate->es_result_relations;
-	nr = estate->es_num_result_relations;
-	while (nr > 0)
-	{
-		if (RelationGetRelid(rInfo->ri_RelationDesc) == relid)
-			return rInfo;
-		rInfo++;
-		nr--;
-	}
-	/* Second, search through the root result relations, if any */
-	rInfo = estate->es_root_result_relations;
-	nr = estate->es_num_root_result_relations;
-	while (nr > 0)
+	foreach(l, estate->es_opened_result_relations)
 	{
+		rInfo = lfirst(l);
 		if (RelationGetRelid(rInfo->ri_RelationDesc) == relid)
 			return rInfo;
-		rInfo++;
-		nr--;
 	}
 
 	/*
@@ -1512,9 +1424,6 @@ ExecPostprocessPlan(EState *estate)
 static void
 ExecEndPlan(PlanState *planstate, EState *estate)
 {
-	ResultRelInfo *resultRelInfo;
-	Index		num_relations;
-	Index		i;
 	ListCell   *l;
 
 	/*
@@ -1540,30 +1449,52 @@ ExecEndPlan(PlanState *planstate, EState *estate)
 	 */
 	ExecResetTupleTable(estate->es_tupleTable, false);
 
+	/* Close indexes of result relation(s), if any. */
+	ExecCloseResultRelations(estate);
+
+	/*
+	 * close whatever rangetable Relations have been opened.  We do not
+	 * release any locks we might hold on those rels.
+	 */
+	ExecCloseRangeTableRelations(estate);
+
+	/* likewise close any trigger target relations */
+	ExecCleanUpTriggerState(estate);
+}
+
+/*
+ * ExecCloseResultRelations
+ */
+void
+ExecCloseResultRelations(EState *estate)
+{
+	ListCell *l;
+
 	/*
 	 * close indexes of result relation(s) if any.  (Rels themselves get
 	 * closed next.)
 	 */
-	resultRelInfo = estate->es_result_relations;
-	for (i = estate->es_num_result_relations; i > 0; i--)
+	foreach(l, estate->es_opened_result_relations)
 	{
+		ResultRelInfo *resultRelInfo = lfirst(l);
+
 		ExecCloseIndices(resultRelInfo);
-		resultRelInfo++;
 	}
+}
 
-	/*
-	 * close whatever rangetable Relations have been opened.  We do not
-	 * release any locks we might hold on those rels.
-	 */
-	num_relations = estate->es_range_table_size;
-	for (i = 0; i < num_relations; i++)
+/*
+ * ExecCloseRangeTableRelations
+ */
+void
+ExecCloseRangeTableRelations(EState *estate)
+{
+	int		i;
+
+	for (i = 0; i < estate->es_range_table_size; i++)
 	{
 		if (estate->es_relations[i])
 			table_close(estate->es_relations[i], NoLock);
 	}
-
-	/* likewise close any trigger target relations */
-	ExecCleanUpTriggerState(estate);
 }
 
 /* ----------------------------------------------------------------
@@ -2758,17 +2689,9 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 
 	/*
 	 * Child EPQ EStates share the parent's copy of unchanging state such as
-	 * the snapshot, rangetable, result-rel info, and external Param info.
+	 * the snapshot, rangetable, and external Param info.
 	 * They need their own copies of local state, including a tuple table,
-	 * es_param_exec_vals, etc.
-	 *
-	 * The ResultRelInfo array management is trickier than it looks.  We
-	 * create fresh arrays for the child but copy all the content from the
-	 * parent.  This is because it's okay for the child to share any
-	 * per-relation state the parent has already created --- but if the child
-	 * sets up any ResultRelInfo fields, such as its own junkfilter, that
-	 * state must *not* propagate back to the parent.  (For one thing, the
-	 * pointed-to data is in a memory context that won't last long enough.)
+	 * es_param_exec_vals, result-rel info, etc.
 	 */
 	rcestate->es_direction = ForwardScanDirection;
 	rcestate->es_snapshot = parentestate->es_snapshot;
@@ -2781,30 +2704,12 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 	rcestate->es_plannedstmt = parentestate->es_plannedstmt;
 	rcestate->es_junkFilter = parentestate->es_junkFilter;
 	rcestate->es_output_cid = parentestate->es_output_cid;
-	if (parentestate->es_num_result_relations > 0)
-	{
-		int			numResultRelations = parentestate->es_num_result_relations;
-		int			numRootResultRels = parentestate->es_num_root_result_relations;
-		ResultRelInfo *resultRelInfos;
-
-		resultRelInfos = (ResultRelInfo *)
-			palloc(numResultRelations * sizeof(ResultRelInfo));
-		memcpy(resultRelInfos, parentestate->es_result_relations,
-			   numResultRelations * sizeof(ResultRelInfo));
-		rcestate->es_result_relations = resultRelInfos;
-		rcestate->es_num_result_relations = numResultRelations;
-
-		/* Also transfer partitioned root result relations. */
-		if (numRootResultRels > 0)
-		{
-			resultRelInfos = (ResultRelInfo *)
-				palloc(numRootResultRels * sizeof(ResultRelInfo));
-			memcpy(resultRelInfos, parentestate->es_root_result_relations,
-				   numRootResultRels * sizeof(ResultRelInfo));
-			rcestate->es_root_result_relations = resultRelInfos;
-			rcestate->es_num_root_result_relations = numRootResultRels;
-		}
-	}
+	/*
+	 * ResultRelInfos needed by subplans are initialized from scratch when
+	 * the subplans themselves are initialized.
+	 */
+	if (parentestate->es_result_relations)
+		ExecInitResultRelationsArray(rcestate);
 	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
@@ -2954,6 +2859,8 @@ EvalPlanQualEnd(EPQState *epqstate)
 		ExecEndNode(subplanstate);
 	}
 
+	ExecCloseResultRelations(estate);
+
 	/* throw away the per-estate tuple table, some node may have used it */
 	ExecResetTupleTable(estate->es_tupleTable, false);
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index d0e65b8647..169dd925ad 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -124,14 +124,8 @@ CreateExecutorState(void)
 	estate->es_output_cid = (CommandId) 0;
 
 	estate->es_result_relations = NULL;
-	estate->es_num_result_relations = 0;
 	estate->es_result_relation_info = NULL;
-
-	estate->es_root_result_relations = NULL;
-	estate->es_num_root_result_relations = 0;
-
 	estate->es_tuple_routing_result_relations = NIL;
-
 	estate->es_trig_target_relations = NIL;
 
 	estate->es_param_list_info = NULL;
@@ -711,16 +705,10 @@ ExecCreateScanSlotFromOuterPlan(EState *estate,
 bool
 ExecRelationIsTargetRelation(EState *estate, Index scanrelid)
 {
-	ResultRelInfo *resultRelInfos;
-	int			i;
-
-	resultRelInfos = estate->es_result_relations;
-	for (i = 0; i < estate->es_num_result_relations; i++)
-	{
-		if (resultRelInfos[i].ri_RangeTableIndex == scanrelid)
-			return true;
-	}
-	return false;
+	return	list_member_int(estate->es_plannedstmt->resultRelations,
+							scanrelid) ||
+			list_member_int(estate->es_plannedstmt->rootResultRelations,
+							scanrelid);
 }
 
 /* ----------------------------------------------------------------
@@ -779,8 +767,8 @@ ExecInitRangeTable(EState *estate, List *rangeTable)
 		palloc0(estate->es_range_table_size * sizeof(Relation));
 
 	/*
-	 * es_rowmarks is also parallel to the es_range_table, but it's allocated
-	 * only if needed.
+	 * es_result_relations and es_rowmarks are also parallel to es_range_table,
+	 * but are only allocated if needed.
 	 */
 	estate->es_rowmarks = NULL;
 }
@@ -835,6 +823,55 @@ ExecGetRangeTableRelation(EState *estate, Index rti)
 	return rel;
 }
 
+/*
+ * ExecInitResultRelationsArray
+ *		Allocate space to hold ResultRelInfo pointers of result relations
+ *
+ * Although not relations in the range table may be result relations, we
+ * allocate that many pointers, because that allows to access individual
+ * entries by RT index (minus 1 to be accurate), which is convenient.
+ */
+void
+ExecInitResultRelationsArray(EState *estate)
+{
+	/*
+	 * Individual pointers are assigned when ExecInitResultRelation() is
+	 * called one-by-one for each result relation.
+	 */
+	estate->es_result_relations = (ResultRelInfo **)
+		palloc0(estate->es_range_table_size * sizeof(ResultRelInfo *));
+}
+
+/*
+ * ExecInitResultRelation
+ *		Open result relation given by the passed-in RT index and fill its
+ *		ResultRelInfo node
+ *
+ * Here, we also save the ResultRelInfo in estate->es_result_relations array
+ * such that it can be accessed later using the RT index.
+ */
+void
+ExecInitResultRelation(EState *estate, ResultRelInfo *resultRelInfo,
+					   Index rti)
+{
+	Relation	resultRelationDesc;
+
+	resultRelationDesc = ExecGetRangeTableRelation(estate, rti);
+	InitResultRelInfo(resultRelInfo,
+					  resultRelationDesc,
+					  rti,
+					  NULL,
+					  estate->es_instrument);
+	estate->es_result_relations[rti - 1] = resultRelInfo;
+
+	/*
+	 * Saving in the list allows to avoid needlessly traversing the whole
+	 * array when only a few of its entries are possibly non-NULL.
+	 */
+	estate->es_opened_result_relations =
+		lappend(estate->es_opened_result_relations, resultRelInfo);
+}
+
 /*
  * UpdateChangedParamSet
  *		Add changed parameters to a plan node's chgParam set
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9812089161..9b27d34ba5 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -2301,7 +2301,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
-	ListCell   *l;
+	ListCell   *l,
+			   *l1;
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
@@ -2322,13 +2323,17 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->mt_done = false;
 
 	mtstate->mt_plans = (PlanState **) palloc0(sizeof(PlanState *) * nplans);
-	mtstate->resultRelInfo = estate->es_result_relations + node->resultRelIndex;
+	mtstate->resultRelInfo = (ResultRelInfo *)
+		palloc(nplans * sizeof(ResultRelInfo));
 	mtstate->mt_scans = (TupleTableSlot **) palloc0(sizeof(TupleTableSlot *) * nplans);
 
 	/* If modifying a partitioned table, initialize the root table info */
-	if (node->rootResultRelIndex >= 0)
-		mtstate->rootResultRelInfo = estate->es_root_result_relations +
-			node->rootResultRelIndex;
+	if (node->rootRelation > 0)
+	{
+		mtstate->rootResultRelInfo = makeNode(ResultRelInfo);
+		ExecInitResultRelation(estate, mtstate->rootResultRelInfo,
+							   node->rootRelation);
+	}
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2351,9 +2356,14 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
-	foreach(l, node->plans)
+	forboth(l, node->resultRelations, l1, node->plans)
 	{
-		subplan = (Plan *) lfirst(l);
+		Index		resultRelation = lfirst_int(l);
+
+		subplan = (Plan *) lfirst(l1);
+
+		/* This opens result relation and fills ResultRelInfo. */
+		ExecInitResultRelation(estate, resultRelInfo, resultRelation);
 
 		/* Initialize the usesFdwDirectModify flag */
 		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(i,
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 0409a40b82..7974fa01e6 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -207,8 +207,6 @@ _copyModifyTable(const ModifyTable *from)
 	COPY_SCALAR_FIELD(rootRelation);
 	COPY_SCALAR_FIELD(partColsUpdated);
 	COPY_NODE_FIELD(resultRelations);
-	COPY_SCALAR_FIELD(resultRelIndex);
-	COPY_SCALAR_FIELD(rootResultRelIndex);
 	COPY_NODE_FIELD(plans);
 	COPY_NODE_FIELD(withCheckOptionLists);
 	COPY_NODE_FIELD(returningLists);
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index f0386480ab..f35d2a04a7 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -408,8 +408,6 @@ _outModifyTable(StringInfo str, const ModifyTable *node)
 	WRITE_UINT_FIELD(rootRelation);
 	WRITE_BOOL_FIELD(partColsUpdated);
 	WRITE_NODE_FIELD(resultRelations);
-	WRITE_INT_FIELD(resultRelIndex);
-	WRITE_INT_FIELD(rootResultRelIndex);
 	WRITE_NODE_FIELD(plans);
 	WRITE_NODE_FIELD(withCheckOptionLists);
 	WRITE_NODE_FIELD(returningLists);
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 42050ab719..73b8d9d7f6 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1639,8 +1639,6 @@ _readModifyTable(void)
 	READ_UINT_FIELD(rootRelation);
 	READ_BOOL_FIELD(partColsUpdated);
 	READ_NODE_FIELD(resultRelations);
-	READ_INT_FIELD(resultRelIndex);
-	READ_INT_FIELD(rootResultRelIndex);
 	READ_NODE_FIELD(plans);
 	READ_NODE_FIELD(withCheckOptionLists);
 	READ_NODE_FIELD(returningLists);
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 3d7a4e373f..881eaf4813 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -6808,8 +6808,6 @@ make_modifytable(PlannerInfo *root,
 	node->rootRelation = rootRelation;
 	node->partColsUpdated = partColsUpdated;
 	node->resultRelations = resultRelations;
-	node->resultRelIndex = -1;	/* will be set correctly in setrefs.c */
-	node->rootResultRelIndex = -1;	/* will be set correctly in setrefs.c */
 	node->plans = subplans;
 	if (!onconflict)
 	{
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index dd8e2e966d..66372d30fe 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -975,24 +975,18 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 
 				/*
 				 * Append this ModifyTable node's final result relation RT
-				 * index(es) to the global list for the plan, and set its
-				 * resultRelIndex to reflect their starting position in the
-				 * global list.
+				 * index(es) to the global list for the plan.
 				 */
-				splan->resultRelIndex = list_length(root->glob->resultRelations);
 				root->glob->resultRelations =
 					list_concat(root->glob->resultRelations,
 								splan->resultRelations);
 
 				/*
 				 * If the main target relation is a partitioned table, also
-				 * add the partition root's RT index to rootResultRelations,
-				 * and remember its index in that list in rootResultRelIndex.
+				 * add the partition root's RT index to rootResultRelations.
 				 */
 				if (splan->rootRelation)
 				{
-					splan->rootResultRelIndex =
-						list_length(root->glob->rootResultRelations);
 					root->glob->rootResultRelations =
 						lappend_int(root->glob->rootResultRelations,
 									splan->rootRelation);
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 9c6fdeeb56..77f71e52d4 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -355,12 +355,12 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 	rte->relkind = rel->localrel->rd_rel->relkind;
 	rte->rellockmode = AccessShareLock;
 	ExecInitRangeTable(estate, list_make1(rte));
+	ExecInitResultRelationsArray(estate);
 
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
 
-	estate->es_result_relations = resultRelInfo;
-	estate->es_num_result_relations = 1;
+	estate->es_result_relations[0] = resultRelInfo;
 	estate->es_result_relation_info = resultRelInfo;
 
 	estate->es_output_cid = GetCurrentCommandId(true);
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 415e117407..54455e1446 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -538,6 +538,9 @@ extern bool ExecRelationIsTargetRelation(EState *estate, Index scanrelid);
 extern Relation ExecOpenScanRelation(EState *estate, Index scanrelid, int eflags);
 
 extern void ExecInitRangeTable(EState *estate, List *rangeTable);
+extern void ExecInitResultRelationsArray(EState *estate);
+extern void ExecCloseRangeTableRelations(EState *estate);
+extern void ExecCloseResultRelations(EState *estate);
 
 static inline RangeTblEntry *
 exec_rt_fetch(Index rti, EState *estate)
@@ -546,6 +549,8 @@ exec_rt_fetch(Index rti, EState *estate)
 }
 
 extern Relation ExecGetRangeTableRelation(EState *estate, Index rti);
+extern void ExecInitResultRelation(EState *estate, ResultRelInfo *resultRelInfo,
+					   Index rti);
 
 extern int	executor_errposition(EState *estate, int location);
 
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index ef448d67c7..7eb5ca6a04 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -508,6 +508,14 @@ typedef struct EState
 	Index		es_range_table_size;	/* size of the range table arrays */
 	Relation   *es_relations;	/* Array of per-range-table-entry Relation
 								 * pointers, or NULL if not yet opened */
+	ResultRelInfo **es_result_relations;	/* Array of Per-range-table-entry
+											 * ResultRelInfo pointers, or
+											 * NULL if a given range table
+											 * relation not a target
+											 * table */
+	List		*es_opened_result_relations;	/* List of non-NULL entries
+												 * in es_result_relations added
+												 * in no specific order */
 	struct ExecRowMark **es_rowmarks;	/* Array of per-range-table-entry
 										 * ExecRowMarks, or NULL if none */
 	PlannedStmt *es_plannedstmt;	/* link to top of plan tree */
@@ -518,24 +526,13 @@ typedef struct EState
 	/* If query can insert/delete tuples, the command ID to mark them with */
 	CommandId	es_output_cid;
 
-	/* Info about target table(s) for insert/update/delete queries: */
-	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
-	int			es_num_result_relations;	/* length of array */
 	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
-	/*
-	 * Info about the partition root table(s) for insert/update/delete queries
-	 * targeting partitioned tables.  Only leaf partitions are mentioned in
-	 * es_result_relations, but we need access to the roots for firing
-	 * triggers and for runtime tuple routing.
-	 */
-	ResultRelInfo *es_root_result_relations;	/* array of ResultRelInfos */
-	int			es_num_root_result_relations;	/* length of the array */
 	PartitionDirectory es_partition_directory;	/* for PartitionDesc lookup */
 
 	/*
 	 * The following list contains ResultRelInfos created by the tuple routing
-	 * code for partitions that don't already have one.
+	 * code for partitions that aren't found in es_result_relations_array.
 	 */
 	List	   *es_tuple_routing_result_relations;
 
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 83e01074ed..06665468a5 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -224,8 +224,6 @@ typedef struct ModifyTable
 	Index		rootRelation;	/* Root RT index, if target is partitioned */
 	bool		partColsUpdated;	/* some part key in hierarchy updated */
 	List	   *resultRelations;	/* integer list of RT indexes */
-	int			resultRelIndex; /* index of first resultRel in plan's list */
-	int			rootResultRelIndex; /* index of the partitioned table root */
 	List	   *plans;			/* plan(s) producing source data */
 	List	   *withCheckOptionLists;	/* per-target-table WCO lists */
 	List	   *returningLists; /* per-target-table RETURNING tlists */
-- 
2.20.1

From 807dde3d299a2d4ae89d96ea7422d0feb9c58e4b Mon Sep 17 00:00:00 2001
From: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Date: Fri, 9 Oct 2020 22:09:08 +0300
Subject: [PATCH 2/2] Merge ExecCleanUpTriggerState() and
 ExecCloseResultRelations().

And some other kibitzing.
---
 src/backend/commands/copy.c              |  9 ++-
 src/backend/commands/explain.c           |  2 +-
 src/backend/commands/trigger.c           |  2 +-
 src/backend/executor/execMain.c          | 83 ++++++++++--------------
 src/backend/executor/execUtils.c         |  7 +-
 src/backend/replication/logical/worker.c |  6 +-
 src/include/executor/executor.h          |  1 -
 src/include/nodes/execnodes.h            | 17 ++---
 8 files changed, 59 insertions(+), 68 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 6948214334..bfdd366139 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2727,6 +2727,7 @@ CopyFrom(CopyState cstate)
 	bool		leafpart_use_multi_insert = false;
 
 	Assert(cstate->rel);
+	Assert(list_length(cstate->range_table) == 1);
 
 	/*
 	 * The target must be a plain, foreign, or partitioned relation, or have
@@ -3352,15 +3353,13 @@ CopyFrom(CopyState cstate)
 	if (insertMethod != CIM_SINGLE)
 		CopyMultiInsertInfoCleanup(&multiInsertInfo);
 
-	ExecCloseResultRelations(estate);
-	ExecCloseRangeTableRelations(estate);
-
 	/* Close all the partitioned tables, leaf partitions, and their indices */
 	if (proute)
 		ExecCleanupTupleRouting(mtstate, proute);
 
-	/* Close any trigger target relations */
-	ExecCleanUpTriggerState(estate);
+	/* Close the result relations */
+	ExecCloseResultRelations(estate);
+	ExecCloseRangeTableRelations(estate);
 
 	FreeExecutorState(estate);
 
diff --git a/src/backend/commands/explain.c b/src/backend/commands/explain.c
index 3210f90bd1..1cc47b0e77 100644
--- a/src/backend/commands/explain.c
+++ b/src/backend/commands/explain.c
@@ -786,7 +786,7 @@ ExplainPrintTriggers(ExplainState *es, QueryDesc *queryDesc)
 					routerels != NIL || targrels != NIL);
 	foreach(l, resultrels)
 	{
-		rInfo = lfirst(l);
+		rInfo = (ResultRelInfo *) lfirst(l);
 		report_triggers(rInfo, show_relname, es);
 	}
 
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..3b4fbdadf4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -4227,7 +4227,7 @@ afterTriggerInvokeEvents(AfterTriggerEventList *events,
 
 	if (local_estate)
 	{
-		ExecCleanUpTriggerState(estate);
+		ExecCloseResultRelations(estate);
 		ExecResetTupleTable(estate->es_tupleTable, false);
 		FreeExecutorState(estate);
 	}
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index f62fd8f7aa..403d62f2b6 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -1340,35 +1340,6 @@ ExecGetTriggerResultRel(EState *estate, Oid relid)
 	return rInfo;
 }
 
-/*
- * Close any relations that have been opened by ExecGetTriggerResultRel().
- */
-void
-ExecCleanUpTriggerState(EState *estate)
-{
-	ListCell   *l;
-
-	foreach(l, estate->es_trig_target_relations)
-	{
-		ResultRelInfo *resultRelInfo = (ResultRelInfo *) lfirst(l);
-
-		/*
-		 * Assert this is a "dummy" ResultRelInfo, see above.  Otherwise we
-		 * might be issuing a duplicate close against a Relation opened by
-		 * ExecGetRangeTableRelation.
-		 */
-		Assert(resultRelInfo->ri_RangeTableIndex == 0);
-
-		/*
-		 * Since ExecGetTriggerResultRel doesn't call ExecOpenIndices for
-		 * these rels, we needn't call ExecCloseIndices either.
-		 */
-		Assert(resultRelInfo->ri_NumIndices == 0);
-
-		table_close(resultRelInfo->ri_RelationDesc, NoLock);
-	}
-}
-
 /* ----------------------------------------------------------------
  *		ExecPostprocessPlan
  *
@@ -1449,7 +1420,7 @@ ExecEndPlan(PlanState *planstate, EState *estate)
 	 */
 	ExecResetTupleTable(estate->es_tupleTable, false);
 
-	/* Close indexes of result relation(s), if any. */
+	/* Close result relation(s), if any. */
 	ExecCloseResultRelations(estate);
 
 	/*
@@ -1457,22 +1428,19 @@ ExecEndPlan(PlanState *planstate, EState *estate)
 	 * release any locks we might hold on those rels.
 	 */
 	ExecCloseRangeTableRelations(estate);
-
-	/* likewise close any trigger target relations */
-	ExecCleanUpTriggerState(estate);
 }
 
 /*
- * ExecCloseResultRelations
+ * Close any relations that have been opened for ResultRelInfos.
  */
 void
 ExecCloseResultRelations(EState *estate)
 {
-	ListCell *l;
+	ListCell   *l;
 
 	/*
-	 * close indexes of result relation(s) if any.  (Rels themselves get
-	 * closed next.)
+	 * close indexes of result relation(s) if any.  (Rels themselves are
+	 * closed in ExecCloseRangeTableRelations())
 	 */
 	foreach(l, estate->es_opened_result_relations)
 	{
@@ -1480,15 +1448,36 @@ ExecCloseResultRelations(EState *estate)
 
 		ExecCloseIndices(resultRelInfo);
 	}
+
+	/* Close any relations that have been opened by ExecGetTriggerResultRel(). */
+	foreach(l, estate->es_trig_target_relations)
+	{
+		ResultRelInfo *resultRelInfo = (ResultRelInfo *) lfirst(l);
+
+		/*
+		 * Assert this is a "dummy" ResultRelInfo, see above.  Otherwise we
+		 * might be issuing a duplicate close against a Relation opened by
+		 * ExecGetRangeTableRelation.
+		 */
+		Assert(resultRelInfo->ri_RangeTableIndex == 0);
+
+		/*
+		 * Since ExecGetTriggerResultRel doesn't call ExecOpenIndices for
+		 * these rels, we needn't call ExecCloseIndices either.
+		 */
+		Assert(resultRelInfo->ri_NumIndices == 0);
+
+		table_close(resultRelInfo->ri_RelationDesc, NoLock);
+	}
 }
 
 /*
- * ExecCloseRangeTableRelations
+ * Close all relations opened by ExecGetRangeTableRelation()
  */
 void
 ExecCloseRangeTableRelations(EState *estate)
 {
-	int		i;
+	int			i;
 
 	for (i = 0; i < estate->es_range_table_size; i++)
 	{
@@ -2689,9 +2678,9 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 
 	/*
 	 * Child EPQ EStates share the parent's copy of unchanging state such as
-	 * the snapshot, rangetable, and external Param info.
-	 * They need their own copies of local state, including a tuple table,
-	 * es_param_exec_vals, result-rel info, etc.
+	 * the snapshot, rangetable, and external Param info.  They need their own
+	 * copies of local state, including a tuple table, es_param_exec_vals,
+	 * result-rel info, etc.
 	 */
 	rcestate->es_direction = ForwardScanDirection;
 	rcestate->es_snapshot = parentestate->es_snapshot;
@@ -2704,9 +2693,10 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 	rcestate->es_plannedstmt = parentestate->es_plannedstmt;
 	rcestate->es_junkFilter = parentestate->es_junkFilter;
 	rcestate->es_output_cid = parentestate->es_output_cid;
+
 	/*
-	 * ResultRelInfos needed by subplans are initialized from scratch when
-	 * the subplans themselves are initialized.
+	 * ResultRelInfos needed by subplans are initialized from scratch when the
+	 * subplans themselves are initialized.
 	 */
 	if (parentestate->es_result_relations)
 		ExecInitResultRelationsArray(rcestate);
@@ -2859,13 +2849,10 @@ EvalPlanQualEnd(EPQState *epqstate)
 		ExecEndNode(subplanstate);
 	}
 
-	ExecCloseResultRelations(estate);
-
 	/* throw away the per-estate tuple table, some node may have used it */
 	ExecResetTupleTable(estate->es_tupleTable, false);
 
-	/* close any trigger target relations attached to this EState */
-	ExecCleanUpTriggerState(estate);
+	ExecCloseResultRelations(estate);
 
 	MemoryContextSwitchTo(oldcontext);
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index 169dd925ad..eda84bca45 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -770,6 +770,7 @@ ExecInitRangeTable(EState *estate, List *rangeTable)
 	 * es_result_relations and es_rowmarks are also parallel to es_range_table,
 	 * but are only allocated if needed.
 	 */
+	estate->es_result_relations = NULL;
 	estate->es_rowmarks = NULL;
 }
 
@@ -827,9 +828,9 @@ ExecGetRangeTableRelation(EState *estate, Index rti)
  * ExecInitResultRelationsArray
  *		Allocate space to hold ResultRelInfo pointers of result relations
  *
- * Although not relations in the range table may be result relations, we
- * allocate that many pointers, because that allows to access individual
- * entries by RT index (minus 1 to be accurate), which is convenient.
+ * Usually, only some relations in the range table are result relations, but
+ * we allocate an array with the same size as the range table, so that we
+ * can index it by the RT index (minus 1 to be accurate).
  */
 void
 ExecInitResultRelationsArray(EState *estate)
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 77f71e52d4..07889f7a7b 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -357,9 +357,13 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 	ExecInitRangeTable(estate, list_make1(rte));
 	ExecInitResultRelationsArray(estate);
 
+	/*
+	 * Initialize a ResultRelInfo for the target relation.  Note that we
+	 * intentionally don't add it to the es_opened_result_relations list,
+	 * because we do our own cleanup and don't use ExecCloseResultRelations().
+	 */
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
-
 	estate->es_result_relations[0] = resultRelInfo;
 	estate->es_result_relation_info = resultRelInfo;
 
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 54455e1446..76ef5cd91c 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -191,7 +191,6 @@ extern void InitResultRelInfo(ResultRelInfo *resultRelInfo,
 							  Relation partition_root,
 							  int instrument_options);
 extern ResultRelInfo *ExecGetTriggerResultRel(EState *estate, Oid relid);
-extern void ExecCleanUpTriggerState(EState *estate);
 extern void ExecConstraints(ResultRelInfo *resultRelInfo,
 							TupleTableSlot *slot, EState *estate);
 extern bool ExecPartitionCheck(ResultRelInfo *resultRelInfo,
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 7eb5ca6a04..b6500abc24 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -508,14 +508,6 @@ typedef struct EState
 	Index		es_range_table_size;	/* size of the range table arrays */
 	Relation   *es_relations;	/* Array of per-range-table-entry Relation
 								 * pointers, or NULL if not yet opened */
-	ResultRelInfo **es_result_relations;	/* Array of Per-range-table-entry
-											 * ResultRelInfo pointers, or
-											 * NULL if a given range table
-											 * relation not a target
-											 * table */
-	List		*es_opened_result_relations;	/* List of non-NULL entries
-												 * in es_result_relations added
-												 * in no specific order */
 	struct ExecRowMark **es_rowmarks;	/* Array of per-range-table-entry
 										 * ExecRowMarks, or NULL if none */
 	PlannedStmt *es_plannedstmt;	/* link to top of plan tree */
@@ -526,6 +518,15 @@ typedef struct EState
 	/* If query can insert/delete tuples, the command ID to mark them with */
 	CommandId	es_output_cid;
 
+	/* Info about target table(s) for insert/update/delete queries: */
+	ResultRelInfo **es_result_relations;	/* Array of Per-range-table-entry
+											 * ResultRelInfo pointers, or
+											 * NULL if a given range table
+											 * relation not a target
+											 * table */
+	List		*es_opened_result_relations;	/* List of non-NULL entries
+												 * in es_result_relations added
+												 * in no specific order */
 	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	PartitionDirectory es_partition_directory;	/* for PartitionDesc lookup */
-- 
2.20.1

From c3f24ca219aa3d535df48c7add376ee3f8f4bc1e Mon Sep 17 00:00:00 2001
From: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Date: Mon, 12 Oct 2020 19:43:55 +0300
Subject: [PATCH v15 1/1] Make es_result_relations array indexable by RT index.

Create each ResultRelInfo on demand in ExecInitModifyTable, instead of
initializing the whole array at once in InitPlan. And instead of having
a separate concept of a "result rel index", access the array using the
range table indexes. This allows several simplifications. The arrays of
regular result rels and root result are merged into one array. And we
don't need the global resultRelations list in PlannedStmt anymore.

Author: Amit Langote
Discussion: https://www.postgresql.org/message-id/CA%2BHiwqGEmiib8FLiHMhKB%2BCH5dRgHSLc5N5wnvc4kym%2BZYpQEQ%40mail.gmail.com
---
 src/backend/commands/copy.c              |  24 +--
 src/backend/commands/explain.c           |  17 +-
 src/backend/commands/tablecmds.c         |  11 +-
 src/backend/commands/trigger.c           |   2 +-
 src/backend/executor/execMain.c          | 256 +++++++----------------
 src/backend/executor/execParallel.c      |   2 -
 src/backend/executor/execUtils.c         |  58 +++--
 src/backend/executor/nodeModifyTable.c   |  24 ++-
 src/backend/nodes/copyfuncs.c            |   4 -
 src/backend/nodes/outfuncs.c             |   6 -
 src/backend/nodes/readfuncs.c            |   4 -
 src/backend/optimizer/plan/createplan.c  |   2 -
 src/backend/optimizer/plan/planner.c     |   6 -
 src/backend/optimizer/plan/setrefs.c     |  30 +--
 src/backend/replication/logical/worker.c |  32 +--
 src/include/executor/executor.h          |   5 +-
 src/include/nodes/execnodes.h            |  19 +-
 src/include/nodes/pathnodes.h            |   4 -
 src/include/nodes/plannodes.h            |  11 -
 19 files changed, 188 insertions(+), 329 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 3c7dbad27a..ca05702139 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2727,6 +2727,7 @@ CopyFrom(CopyState cstate)
 	bool		leafpart_use_multi_insert = false;
 
 	Assert(cstate->rel);
+	Assert(list_length(cstate->range_table) == 1);
 
 	/*
 	 * The target must be a plain, foreign, or partitioned relation, or have
@@ -2829,25 +2830,17 @@ CopyFrom(CopyState cstate)
 	 * index-entry-making machinery.  (There used to be a huge amount of code
 	 * here that basically duplicated execUtils.c ...)
 	 */
-	resultRelInfo = makeNode(ResultRelInfo);
-	InitResultRelInfo(resultRelInfo,
-					  cstate->rel,
-					  1,		/* must match rel's position in range_table */
-					  NULL,
-					  0);
-	target_resultRelInfo = resultRelInfo;
+	ExecInitRangeTable(estate, cstate->range_table);
+	resultRelInfo = target_resultRelInfo = makeNode(ResultRelInfo);
+	ExecInitResultRelation(estate, resultRelInfo, 1);
 
 	/* Verify the named relation is a valid target for INSERT */
 	CheckValidResultRel(resultRelInfo, CMD_INSERT);
 
 	ExecOpenIndices(resultRelInfo, false);
 
-	estate->es_result_relations = resultRelInfo;
-	estate->es_num_result_relations = 1;
 	estate->es_result_relation_info = resultRelInfo;
 
-	ExecInitRangeTable(estate, cstate->range_table);
-
 	/*
 	 * Set up a ModifyTableState so we can let FDW(s) init themselves for
 	 * foreign-table result relation(s).
@@ -2856,7 +2849,7 @@ CopyFrom(CopyState cstate)
 	mtstate->ps.plan = NULL;
 	mtstate->ps.state = estate;
 	mtstate->operation = CMD_INSERT;
-	mtstate->resultRelInfo = estate->es_result_relations;
+	mtstate->resultRelInfo = resultRelInfo;
 
 	if (resultRelInfo->ri_FdwRoutine != NULL &&
 		resultRelInfo->ri_FdwRoutine->BeginForeignInsert != NULL)
@@ -3359,14 +3352,13 @@ CopyFrom(CopyState cstate)
 	if (insertMethod != CIM_SINGLE)
 		CopyMultiInsertInfoCleanup(&multiInsertInfo);
 
-	ExecCloseIndices(target_resultRelInfo);
-
 	/* Close all the partitioned tables, leaf partitions, and their indices */
 	if (proute)
 		ExecCleanupTupleRouting(mtstate, proute);
 
-	/* Close any trigger target relations */
-	ExecCleanUpTriggerState(estate);
+	/* Close the result relations */
+	ExecCloseResultRelations(estate);
+	ExecCloseRangeTableRelations(estate);
 
 	FreeExecutorState(estate);
 
diff --git a/src/backend/commands/explain.c b/src/backend/commands/explain.c
index c98c9b5547..c8e292adfa 100644
--- a/src/backend/commands/explain.c
+++ b/src/backend/commands/explain.c
@@ -769,27 +769,24 @@ ExplainPrintTriggers(ExplainState *es, QueryDesc *queryDesc)
 {
 	ResultRelInfo *rInfo;
 	bool		show_relname;
-	int			numrels = queryDesc->estate->es_num_result_relations;
-	int			numrootrels = queryDesc->estate->es_num_root_result_relations;
+	List	   *resultrels;
 	List	   *routerels;
 	List	   *targrels;
-	int			nr;
 	ListCell   *l;
 
+	resultrels = queryDesc->estate->es_opened_result_relations;
 	routerels = queryDesc->estate->es_tuple_routing_result_relations;
 	targrels = queryDesc->estate->es_trig_target_relations;
 
 	ExplainOpenGroup("Triggers", "Triggers", false, es);
 
-	show_relname = (numrels > 1 || numrootrels > 0 ||
+	show_relname = (list_length(resultrels) > 1 ||
 					routerels != NIL || targrels != NIL);
-	rInfo = queryDesc->estate->es_result_relations;
-	for (nr = 0; nr < numrels; rInfo++, nr++)
-		report_triggers(rInfo, show_relname, es);
-
-	rInfo = queryDesc->estate->es_root_result_relations;
-	for (nr = 0; nr < numrootrels; rInfo++, nr++)
+	foreach(l, resultrels)
+	{
+		rInfo = (ResultRelInfo *) lfirst(l);
 		report_triggers(rInfo, show_relname, es);
+	}
 
 	foreach(l, routerels)
 	{
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index e0ac4e05e5..1f0b0f3ef9 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1787,11 +1787,18 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	/*
 	 * To fire triggers, we'll need an EState as well as a ResultRelInfo for
 	 * each relation.  We don't need to call ExecOpenIndices, though.
+	 *
+	 * We put the ResultRelInfos in the es_opened_result_relations list, even
+	 * though we don't have a range table and don't populate the
+	 * es_result_relations array.  That's a big bogus, but it's enough to make
+	 * ExecGetTriggerResultRel() find them.
 	 */
 	estate = CreateExecutorState();
 	resultRelInfos = (ResultRelInfo *)
 		palloc(list_length(rels) * sizeof(ResultRelInfo));
 	resultRelInfo = resultRelInfos;
+	estate->es_result_relations = (ResultRelInfo **)
+		palloc(list_length(rels) * sizeof(ResultRelInfo *));
 	foreach(cell, rels)
 	{
 		Relation	rel = (Relation) lfirst(cell);
@@ -1801,10 +1808,10 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 						  0,	/* dummy rangetable index */
 						  NULL,
 						  0);
+		estate->es_opened_result_relations =
+			lappend(estate->es_opened_result_relations, resultRelInfo);
 		resultRelInfo++;
 	}
-	estate->es_result_relations = resultRelInfos;
-	estate->es_num_result_relations = list_length(rels);
 
 	/*
 	 * Process all BEFORE STATEMENT TRUNCATE triggers before we begin
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 672fccff5b..3b4fbdadf4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -4227,7 +4227,7 @@ afterTriggerInvokeEvents(AfterTriggerEventList *events,
 
 	if (local_estate)
 	{
-		ExecCleanUpTriggerState(estate);
+		ExecCloseResultRelations(estate);
 		ExecResetTupleTable(estate->es_tupleTable, false);
 		FreeExecutorState(estate);
 	}
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 2e27e26ba4..e32f5984e5 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -827,86 +827,8 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 
 	estate->es_plannedstmt = plannedstmt;
 
-	/*
-	 * Initialize ResultRelInfo data structures, and open the result rels.
-	 */
-	if (plannedstmt->resultRelations)
-	{
-		List	   *resultRelations = plannedstmt->resultRelations;
-		int			numResultRelations = list_length(resultRelations);
-		ResultRelInfo *resultRelInfos;
-		ResultRelInfo *resultRelInfo;
-
-		resultRelInfos = (ResultRelInfo *)
-			palloc(numResultRelations * sizeof(ResultRelInfo));
-		resultRelInfo = resultRelInfos;
-		foreach(l, resultRelations)
-		{
-			Index		resultRelationIndex = lfirst_int(l);
-			Relation	resultRelation;
-
-			resultRelation = ExecGetRangeTableRelation(estate,
-													   resultRelationIndex);
-			InitResultRelInfo(resultRelInfo,
-							  resultRelation,
-							  resultRelationIndex,
-							  NULL,
-							  estate->es_instrument);
-			resultRelInfo++;
-		}
-		estate->es_result_relations = resultRelInfos;
-		estate->es_num_result_relations = numResultRelations;
-
-		/* es_result_relation_info is NULL except when within ModifyTable */
-		estate->es_result_relation_info = NULL;
-
-		/*
-		 * In the partitioned result relation case, also build ResultRelInfos
-		 * for all the partitioned table roots, because we will need them to
-		 * fire statement-level triggers, if any.
-		 */
-		if (plannedstmt->rootResultRelations)
-		{
-			int			num_roots = list_length(plannedstmt->rootResultRelations);
-
-			resultRelInfos = (ResultRelInfo *)
-				palloc(num_roots * sizeof(ResultRelInfo));
-			resultRelInfo = resultRelInfos;
-			foreach(l, plannedstmt->rootResultRelations)
-			{
-				Index		resultRelIndex = lfirst_int(l);
-				Relation	resultRelDesc;
-
-				resultRelDesc = ExecGetRangeTableRelation(estate,
-														  resultRelIndex);
-				InitResultRelInfo(resultRelInfo,
-								  resultRelDesc,
-								  resultRelIndex,
-								  NULL,
-								  estate->es_instrument);
-				resultRelInfo++;
-			}
-
-			estate->es_root_result_relations = resultRelInfos;
-			estate->es_num_root_result_relations = num_roots;
-		}
-		else
-		{
-			estate->es_root_result_relations = NULL;
-			estate->es_num_root_result_relations = 0;
-		}
-	}
-	else
-	{
-		/*
-		 * if no result relation, then set state appropriately
-		 */
-		estate->es_result_relations = NULL;
-		estate->es_num_result_relations = 0;
-		estate->es_result_relation_info = NULL;
-		estate->es_root_result_relations = NULL;
-		estate->es_num_root_result_relations = 0;
-	}
+	/* es_result_relation_info is NULL except when within ModifyTable */
+	estate->es_result_relation_info = NULL;
 
 	/*
 	 * Next, build the ExecRowMark array from the PlanRowMark(s), if any.
@@ -1334,8 +1256,7 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
  *
  * Most of the time, triggers are fired on one of the result relations of the
  * query, and so we can just return a member of the es_result_relations array,
- * or the es_root_result_relations array (if any), or the
- * es_tuple_routing_result_relations list (if any).  (Note: in self-join
+ * or the es_tuple_routing_result_relations list (if any). (Note: in self-join
  * situations there might be multiple members with the same OID; if so it
  * doesn't matter which one we pick.)
  *
@@ -1352,30 +1273,16 @@ ResultRelInfo *
 ExecGetTriggerResultRel(EState *estate, Oid relid)
 {
 	ResultRelInfo *rInfo;
-	int			nr;
 	ListCell   *l;
 	Relation	rel;
 	MemoryContext oldcontext;
 
 	/* First, search through the query result relations */
-	rInfo = estate->es_result_relations;
-	nr = estate->es_num_result_relations;
-	while (nr > 0)
-	{
-		if (RelationGetRelid(rInfo->ri_RelationDesc) == relid)
-			return rInfo;
-		rInfo++;
-		nr--;
-	}
-	/* Second, search through the root result relations, if any */
-	rInfo = estate->es_root_result_relations;
-	nr = estate->es_num_root_result_relations;
-	while (nr > 0)
+	foreach(l, estate->es_opened_result_relations)
 	{
+		rInfo = lfirst(l);
 		if (RelationGetRelid(rInfo->ri_RelationDesc) == relid)
 			return rInfo;
-		rInfo++;
-		nr--;
 	}
 
 	/*
@@ -1428,35 +1335,6 @@ ExecGetTriggerResultRel(EState *estate, Oid relid)
 	return rInfo;
 }
 
-/*
- * Close any relations that have been opened by ExecGetTriggerResultRel().
- */
-void
-ExecCleanUpTriggerState(EState *estate)
-{
-	ListCell   *l;
-
-	foreach(l, estate->es_trig_target_relations)
-	{
-		ResultRelInfo *resultRelInfo = (ResultRelInfo *) lfirst(l);
-
-		/*
-		 * Assert this is a "dummy" ResultRelInfo, see above.  Otherwise we
-		 * might be issuing a duplicate close against a Relation opened by
-		 * ExecGetRangeTableRelation.
-		 */
-		Assert(resultRelInfo->ri_RangeTableIndex == 0);
-
-		/*
-		 * Since ExecGetTriggerResultRel doesn't call ExecOpenIndices for
-		 * these rels, we needn't call ExecCloseIndices either.
-		 */
-		Assert(resultRelInfo->ri_NumIndices == 0);
-
-		table_close(resultRelInfo->ri_RelationDesc, NoLock);
-	}
-}
-
 /* ----------------------------------------------------------------
  *		ExecPostprocessPlan
  *
@@ -1512,9 +1390,6 @@ ExecPostprocessPlan(EState *estate)
 static void
 ExecEndPlan(PlanState *planstate, EState *estate)
 {
-	ResultRelInfo *resultRelInfo;
-	Index		num_relations;
-	Index		i;
 	ListCell   *l;
 
 	/*
@@ -1540,30 +1415,70 @@ ExecEndPlan(PlanState *planstate, EState *estate)
 	 */
 	ExecResetTupleTable(estate->es_tupleTable, false);
 
+	/* Close result relation(s), if any. */
+	ExecCloseResultRelations(estate);
+
+	/*
+	 * close whatever rangetable Relations have been opened.  We do not
+	 * release any locks we might hold on those rels.
+	 */
+	ExecCloseRangeTableRelations(estate);
+}
+
+/*
+ * Close any relations that have been opened for ResultRelInfos.
+ */
+void
+ExecCloseResultRelations(EState *estate)
+{
+	ListCell   *l;
+
 	/*
-	 * close indexes of result relation(s) if any.  (Rels themselves get
-	 * closed next.)
+	 * close indexes of result relation(s) if any.  (Rels themselves are
+	 * closed in ExecCloseRangeTableRelations())
 	 */
-	resultRelInfo = estate->es_result_relations;
-	for (i = estate->es_num_result_relations; i > 0; i--)
+	foreach(l, estate->es_opened_result_relations)
 	{
+		ResultRelInfo *resultRelInfo = lfirst(l);
+
 		ExecCloseIndices(resultRelInfo);
-		resultRelInfo++;
 	}
 
-	/*
-	 * close whatever rangetable Relations have been opened.  We do not
-	 * release any locks we might hold on those rels.
-	 */
-	num_relations = estate->es_range_table_size;
-	for (i = 0; i < num_relations; i++)
+	/* Close any relations that have been opened by ExecGetTriggerResultRel(). */
+	foreach(l, estate->es_trig_target_relations)
+	{
+		ResultRelInfo *resultRelInfo = (ResultRelInfo *) lfirst(l);
+
+		/*
+		 * Assert this is a "dummy" ResultRelInfo, see above.  Otherwise we
+		 * might be issuing a duplicate close against a Relation opened by
+		 * ExecGetRangeTableRelation.
+		 */
+		Assert(resultRelInfo->ri_RangeTableIndex == 0);
+
+		/*
+		 * Since ExecGetTriggerResultRel doesn't call ExecOpenIndices for
+		 * these rels, we needn't call ExecCloseIndices either.
+		 */
+		Assert(resultRelInfo->ri_NumIndices == 0);
+
+		table_close(resultRelInfo->ri_RelationDesc, NoLock);
+	}
+}
+
+/*
+ * Close all relations opened by ExecGetRangeTableRelation()
+ */
+void
+ExecCloseRangeTableRelations(EState *estate)
+{
+	int			i;
+
+	for (i = 0; i < estate->es_range_table_size; i++)
 	{
 		if (estate->es_relations[i])
 			table_close(estate->es_relations[i], NoLock);
 	}
-
-	/* likewise close any trigger target relations */
-	ExecCleanUpTriggerState(estate);
 }
 
 /* ----------------------------------------------------------------
@@ -2758,17 +2673,9 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 
 	/*
 	 * Child EPQ EStates share the parent's copy of unchanging state such as
-	 * the snapshot, rangetable, result-rel info, and external Param info.
-	 * They need their own copies of local state, including a tuple table,
-	 * es_param_exec_vals, etc.
-	 *
-	 * The ResultRelInfo array management is trickier than it looks.  We
-	 * create fresh arrays for the child but copy all the content from the
-	 * parent.  This is because it's okay for the child to share any
-	 * per-relation state the parent has already created --- but if the child
-	 * sets up any ResultRelInfo fields, such as its own junkfilter, that
-	 * state must *not* propagate back to the parent.  (For one thing, the
-	 * pointed-to data is in a memory context that won't last long enough.)
+	 * the snapshot, rangetable, and external Param info.  They need their own
+	 * copies of local state, including a tuple table, es_param_exec_vals,
+	 * result-rel info, etc.
 	 */
 	rcestate->es_direction = ForwardScanDirection;
 	rcestate->es_snapshot = parentestate->es_snapshot;
@@ -2781,30 +2688,12 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 	rcestate->es_plannedstmt = parentestate->es_plannedstmt;
 	rcestate->es_junkFilter = parentestate->es_junkFilter;
 	rcestate->es_output_cid = parentestate->es_output_cid;
-	if (parentestate->es_num_result_relations > 0)
-	{
-		int			numResultRelations = parentestate->es_num_result_relations;
-		int			numRootResultRels = parentestate->es_num_root_result_relations;
-		ResultRelInfo *resultRelInfos;
-
-		resultRelInfos = (ResultRelInfo *)
-			palloc(numResultRelations * sizeof(ResultRelInfo));
-		memcpy(resultRelInfos, parentestate->es_result_relations,
-			   numResultRelations * sizeof(ResultRelInfo));
-		rcestate->es_result_relations = resultRelInfos;
-		rcestate->es_num_result_relations = numResultRelations;
-
-		/* Also transfer partitioned root result relations. */
-		if (numRootResultRels > 0)
-		{
-			resultRelInfos = (ResultRelInfo *)
-				palloc(numRootResultRels * sizeof(ResultRelInfo));
-			memcpy(resultRelInfos, parentestate->es_root_result_relations,
-				   numRootResultRels * sizeof(ResultRelInfo));
-			rcestate->es_root_result_relations = resultRelInfos;
-			rcestate->es_num_root_result_relations = numRootResultRels;
-		}
-	}
+
+	/*
+	 * ResultRelInfos needed by subplans are initialized from scratch when the
+	 * subplans themselves are initialized.
+	 */
+	parentestate->es_result_relations = NULL;
 	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
@@ -2914,8 +2803,9 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
  * This is a cut-down version of ExecutorEnd(); basically we want to do most
  * of the normal cleanup, but *not* close result relations (which we are
  * just sharing from the outer query).  We do, however, have to close any
- * trigger target relations that got opened, since those are not shared.
- * (There probably shouldn't be any of the latter, but just in case...)
+ * result and trigger target relations that got opened, since those are not
+ * shared.  (There probably shouldn't be any of the latter, but just in
+ * case...)
  */
 void
 EvalPlanQualEnd(EPQState *epqstate)
@@ -2957,8 +2847,8 @@ EvalPlanQualEnd(EPQState *epqstate)
 	/* throw away the per-estate tuple table, some node may have used it */
 	ExecResetTupleTable(estate->es_tupleTable, false);
 
-	/* close any trigger target relations attached to this EState */
-	ExecCleanUpTriggerState(estate);
+	/* Close any result and trigger target relations attached to this EState */
+	ExecCloseResultRelations(estate);
 
 	MemoryContextSwitchTo(oldcontext);
 
diff --git a/src/backend/executor/execParallel.c b/src/backend/executor/execParallel.c
index 382e78fb7f..d85d0fdb64 100644
--- a/src/backend/executor/execParallel.c
+++ b/src/backend/executor/execParallel.c
@@ -183,8 +183,6 @@ ExecSerializePlan(Plan *plan, EState *estate)
 	pstmt->parallelModeNeeded = false;
 	pstmt->planTree = plan;
 	pstmt->rtable = estate->es_range_table;
-	pstmt->resultRelations = NIL;
-	pstmt->rootResultRelations = NIL;
 	pstmt->appendRelations = NIL;
 
 	/*
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index d0e65b8647..c41cdc79e4 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -124,14 +124,9 @@ CreateExecutorState(void)
 	estate->es_output_cid = (CommandId) 0;
 
 	estate->es_result_relations = NULL;
-	estate->es_num_result_relations = 0;
+	estate->es_opened_result_relations = NIL;
 	estate->es_result_relation_info = NULL;
-
-	estate->es_root_result_relations = NULL;
-	estate->es_num_root_result_relations = 0;
-
 	estate->es_tuple_routing_result_relations = NIL;
-
 	estate->es_trig_target_relations = NIL;
 
 	estate->es_param_list_info = NULL;
@@ -711,16 +706,10 @@ ExecCreateScanSlotFromOuterPlan(EState *estate,
 bool
 ExecRelationIsTargetRelation(EState *estate, Index scanrelid)
 {
-	ResultRelInfo *resultRelInfos;
-	int			i;
+	if (!estate->es_result_relations)
+		return false;
 
-	resultRelInfos = estate->es_result_relations;
-	for (i = 0; i < estate->es_num_result_relations; i++)
-	{
-		if (resultRelInfos[i].ri_RangeTableIndex == scanrelid)
-			return true;
-	}
-	return false;
+	return estate->es_result_relations[scanrelid - 1] != NULL;
 }
 
 /* ----------------------------------------------------------------
@@ -779,9 +768,10 @@ ExecInitRangeTable(EState *estate, List *rangeTable)
 		palloc0(estate->es_range_table_size * sizeof(Relation));
 
 	/*
-	 * es_rowmarks is also parallel to the es_range_table, but it's allocated
-	 * only if needed.
+	 * es_result_relations and es_rowmarks are also parallel to es_range_table,
+	 * but are allocated only if needed.
 	 */
+	estate->es_result_relations = NULL;
 	estate->es_rowmarks = NULL;
 }
 
@@ -835,6 +825,40 @@ ExecGetRangeTableRelation(EState *estate, Index rti)
 	return rel;
 }
 
+/*
+ * ExecInitResultRelation
+ *		Open result relation given by the passed-in RT index and fill its
+ *		ResultRelInfo node
+ *
+ * Here, we also save the ResultRelInfo in estate->es_result_relations array
+ * such that it can be accessed later using the RT index.
+ */
+void
+ExecInitResultRelation(EState *estate, ResultRelInfo *resultRelInfo,
+					   Index rti)
+{
+	Relation	resultRelationDesc;
+
+	resultRelationDesc = ExecGetRangeTableRelation(estate, rti);
+	InitResultRelInfo(resultRelInfo,
+					  resultRelationDesc,
+					  rti,
+					  NULL,
+					  estate->es_instrument);
+
+	if (estate->es_result_relations == NULL)
+		estate->es_result_relations = (ResultRelInfo **)
+			palloc0(estate->es_range_table_size * sizeof(ResultRelInfo *));
+	estate->es_result_relations[rti - 1] = resultRelInfo;
+
+	/*
+	 * Saving in the list allows to avoid needlessly traversing the whole
+	 * array when only a few of its entries are possibly non-NULL.
+	 */
+	estate->es_opened_result_relations =
+		lappend(estate->es_opened_result_relations, resultRelInfo);
+}
+
 /*
  * UpdateChangedParamSet
  *		Add changed parameters to a plan node's chgParam set
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 9812089161..9b27d34ba5 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -2301,7 +2301,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
-	ListCell   *l;
+	ListCell   *l,
+			   *l1;
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
@@ -2322,13 +2323,17 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->mt_done = false;
 
 	mtstate->mt_plans = (PlanState **) palloc0(sizeof(PlanState *) * nplans);
-	mtstate->resultRelInfo = estate->es_result_relations + node->resultRelIndex;
+	mtstate->resultRelInfo = (ResultRelInfo *)
+		palloc(nplans * sizeof(ResultRelInfo));
 	mtstate->mt_scans = (TupleTableSlot **) palloc0(sizeof(TupleTableSlot *) * nplans);
 
 	/* If modifying a partitioned table, initialize the root table info */
-	if (node->rootResultRelIndex >= 0)
-		mtstate->rootResultRelInfo = estate->es_root_result_relations +
-			node->rootResultRelIndex;
+	if (node->rootRelation > 0)
+	{
+		mtstate->rootResultRelInfo = makeNode(ResultRelInfo);
+		ExecInitResultRelation(estate, mtstate->rootResultRelInfo,
+							   node->rootRelation);
+	}
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2351,9 +2356,14 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
-	foreach(l, node->plans)
+	forboth(l, node->resultRelations, l1, node->plans)
 	{
-		subplan = (Plan *) lfirst(l);
+		Index		resultRelation = lfirst_int(l);
+
+		subplan = (Plan *) lfirst(l1);
+
+		/* This opens result relation and fills ResultRelInfo. */
+		ExecInitResultRelation(estate, resultRelInfo, resultRelation);
 
 		/* Initialize the usesFdwDirectModify flag */
 		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(i,
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 0409a40b82..e6b48c73ff 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -90,8 +90,6 @@ _copyPlannedStmt(const PlannedStmt *from)
 	COPY_SCALAR_FIELD(jitFlags);
 	COPY_NODE_FIELD(planTree);
 	COPY_NODE_FIELD(rtable);
-	COPY_NODE_FIELD(resultRelations);
-	COPY_NODE_FIELD(rootResultRelations);
 	COPY_NODE_FIELD(appendRelations);
 	COPY_NODE_FIELD(subplans);
 	COPY_BITMAPSET_FIELD(rewindPlanIDs);
@@ -207,8 +205,6 @@ _copyModifyTable(const ModifyTable *from)
 	COPY_SCALAR_FIELD(rootRelation);
 	COPY_SCALAR_FIELD(partColsUpdated);
 	COPY_NODE_FIELD(resultRelations);
-	COPY_SCALAR_FIELD(resultRelIndex);
-	COPY_SCALAR_FIELD(rootResultRelIndex);
 	COPY_NODE_FIELD(plans);
 	COPY_NODE_FIELD(withCheckOptionLists);
 	COPY_NODE_FIELD(returningLists);
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index f0386480ab..fae01357d8 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -308,8 +308,6 @@ _outPlannedStmt(StringInfo str, const PlannedStmt *node)
 	WRITE_INT_FIELD(jitFlags);
 	WRITE_NODE_FIELD(planTree);
 	WRITE_NODE_FIELD(rtable);
-	WRITE_NODE_FIELD(resultRelations);
-	WRITE_NODE_FIELD(rootResultRelations);
 	WRITE_NODE_FIELD(appendRelations);
 	WRITE_NODE_FIELD(subplans);
 	WRITE_BITMAPSET_FIELD(rewindPlanIDs);
@@ -408,8 +406,6 @@ _outModifyTable(StringInfo str, const ModifyTable *node)
 	WRITE_UINT_FIELD(rootRelation);
 	WRITE_BOOL_FIELD(partColsUpdated);
 	WRITE_NODE_FIELD(resultRelations);
-	WRITE_INT_FIELD(resultRelIndex);
-	WRITE_INT_FIELD(rootResultRelIndex);
 	WRITE_NODE_FIELD(plans);
 	WRITE_NODE_FIELD(withCheckOptionLists);
 	WRITE_NODE_FIELD(returningLists);
@@ -2193,8 +2189,6 @@ _outPlannerGlobal(StringInfo str, const PlannerGlobal *node)
 	WRITE_BITMAPSET_FIELD(rewindPlanIDs);
 	WRITE_NODE_FIELD(finalrtable);
 	WRITE_NODE_FIELD(finalrowmarks);
-	WRITE_NODE_FIELD(resultRelations);
-	WRITE_NODE_FIELD(rootResultRelations);
 	WRITE_NODE_FIELD(appendRelations);
 	WRITE_NODE_FIELD(relationOids);
 	WRITE_NODE_FIELD(invalItems);
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 42050ab719..867d1360b8 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -1541,8 +1541,6 @@ _readPlannedStmt(void)
 	READ_INT_FIELD(jitFlags);
 	READ_NODE_FIELD(planTree);
 	READ_NODE_FIELD(rtable);
-	READ_NODE_FIELD(resultRelations);
-	READ_NODE_FIELD(rootResultRelations);
 	READ_NODE_FIELD(appendRelations);
 	READ_NODE_FIELD(subplans);
 	READ_BITMAPSET_FIELD(rewindPlanIDs);
@@ -1639,8 +1637,6 @@ _readModifyTable(void)
 	READ_UINT_FIELD(rootRelation);
 	READ_BOOL_FIELD(partColsUpdated);
 	READ_NODE_FIELD(resultRelations);
-	READ_INT_FIELD(resultRelIndex);
-	READ_INT_FIELD(rootResultRelIndex);
 	READ_NODE_FIELD(plans);
 	READ_NODE_FIELD(withCheckOptionLists);
 	READ_NODE_FIELD(returningLists);
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 3d7a4e373f..881eaf4813 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -6808,8 +6808,6 @@ make_modifytable(PlannerInfo *root,
 	node->rootRelation = rootRelation;
 	node->partColsUpdated = partColsUpdated;
 	node->resultRelations = resultRelations;
-	node->resultRelIndex = -1;	/* will be set correctly in setrefs.c */
-	node->rootResultRelIndex = -1;	/* will be set correctly in setrefs.c */
 	node->plans = subplans;
 	if (!onconflict)
 	{
diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c
index f331f82a6c..3e4de96b8e 100644
--- a/src/backend/optimizer/plan/planner.c
+++ b/src/backend/optimizer/plan/planner.c
@@ -304,8 +304,6 @@ standard_planner(Query *parse, const char *query_string, int cursorOptions,
 	glob->rewindPlanIDs = NULL;
 	glob->finalrtable = NIL;
 	glob->finalrowmarks = NIL;
-	glob->resultRelations = NIL;
-	glob->rootResultRelations = NIL;
 	glob->appendRelations = NIL;
 	glob->relationOids = NIL;
 	glob->invalItems = NIL;
@@ -492,8 +490,6 @@ standard_planner(Query *parse, const char *query_string, int cursorOptions,
 	/* final cleanup of the plan */
 	Assert(glob->finalrtable == NIL);
 	Assert(glob->finalrowmarks == NIL);
-	Assert(glob->resultRelations == NIL);
-	Assert(glob->rootResultRelations == NIL);
 	Assert(glob->appendRelations == NIL);
 	top_plan = set_plan_references(root, top_plan);
 	/* ... and the subplans (both regular subplans and initplans) */
@@ -519,8 +515,6 @@ standard_planner(Query *parse, const char *query_string, int cursorOptions,
 	result->parallelModeNeeded = glob->parallelModeNeeded;
 	result->planTree = top_plan;
 	result->rtable = glob->finalrtable;
-	result->resultRelations = glob->resultRelations;
-	result->rootResultRelations = glob->rootResultRelations;
 	result->appendRelations = glob->appendRelations;
 	result->subplans = glob->subplans;
 	result->rewindPlanIDs = glob->rewindPlanIDs;
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index dd8e2e966d..be151416af 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -235,9 +235,8 @@ static List *set_returning_clause_references(PlannerInfo *root,
  * different when the passed-in Plan is a node we decide isn't needed.
  *
  * The flattened rangetable entries are appended to root->glob->finalrtable.
- * Also, rowmarks entries are appended to root->glob->finalrowmarks, and the
- * RT indexes of ModifyTable result relations to root->glob->resultRelations,
- * and flattened AppendRelInfos are appended to root->glob->appendRelations.
+ * Also, rowmarks entries are appended to root->glob->finalrowmarks, and
+ * flattened AppendRelInfos are appended to root->glob->appendRelations.
  * Plan dependencies are appended to root->glob->relationOids (for relations)
  * and root->glob->invalItems (for everything else).
  *
@@ -972,31 +971,6 @@ set_plan_refs(PlannerInfo *root, Plan *plan, int rtoffset)
 											  (Plan *) lfirst(l),
 											  rtoffset);
 				}
-
-				/*
-				 * Append this ModifyTable node's final result relation RT
-				 * index(es) to the global list for the plan, and set its
-				 * resultRelIndex to reflect their starting position in the
-				 * global list.
-				 */
-				splan->resultRelIndex = list_length(root->glob->resultRelations);
-				root->glob->resultRelations =
-					list_concat(root->glob->resultRelations,
-								splan->resultRelations);
-
-				/*
-				 * If the main target relation is a partitioned table, also
-				 * add the partition root's RT index to rootResultRelations,
-				 * and remember its index in that list in rootResultRelIndex.
-				 */
-				if (splan->rootRelation)
-				{
-					splan->rootResultRelIndex =
-						list_length(root->glob->rootResultRelations);
-					root->glob->rootResultRelations =
-						lappend_int(root->glob->rootResultRelations,
-									splan->rootRelation);
-				}
 			}
 			break;
 		case T_Append:
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 9c6fdeeb56..8d5d9e05b3 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -344,7 +344,6 @@ static EState *
 create_estate_for_relation(LogicalRepRelMapEntry *rel)
 {
 	EState	   *estate;
-	ResultRelInfo *resultRelInfo;
 	RangeTblEntry *rte;
 
 	estate = CreateExecutorState();
@@ -356,13 +355,6 @@ create_estate_for_relation(LogicalRepRelMapEntry *rel)
 	rte->rellockmode = AccessShareLock;
 	ExecInitRangeTable(estate, list_make1(rte));
 
-	resultRelInfo = makeNode(ResultRelInfo);
-	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
-
-	estate->es_result_relations = resultRelInfo;
-	estate->es_num_result_relations = 1;
-	estate->es_result_relation_info = resultRelInfo;
-
 	estate->es_output_cid = GetCurrentCommandId(true);
 
 	/* Prepare to catch AFTER triggers. */
@@ -1150,6 +1142,7 @@ GetRelationIdentityOrPK(Relation rel)
 static void
 apply_handle_insert(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData newtup;
 	LogicalRepRelId relid;
@@ -1179,6 +1172,9 @@ apply_handle_insert(StringInfo s)
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
+	resultRelInfo = makeNode(ResultRelInfo);
+	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
+	estate->es_result_relation_info = resultRelInfo;
 
 	/* Input functions may need an active snapshot, so get one */
 	PushActiveSnapshot(GetTransactionSnapshot());
@@ -1191,10 +1187,10 @@ apply_handle_insert(StringInfo s)
 
 	/* For a partitioned table, insert the tuple into a partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
+		apply_handle_tuple_routing(resultRelInfo, estate,
 								   remoteslot, NULL, rel, CMD_INSERT);
 	else
-		apply_handle_insert_internal(estate->es_result_relation_info, estate,
+		apply_handle_insert_internal(resultRelInfo, estate,
 									 remoteslot);
 
 	PopActiveSnapshot();
@@ -1265,6 +1261,7 @@ check_relation_updatable(LogicalRepRelMapEntry *rel)
 static void
 apply_handle_update(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepRelId relid;
 	EState	   *estate;
@@ -1301,6 +1298,9 @@ apply_handle_update(StringInfo s)
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
+	resultRelInfo = makeNode(ResultRelInfo);
+	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
+	estate->es_result_relation_info = resultRelInfo;
 
 	/*
 	 * Populate updatedCols so that per-column triggers can fire.  This could
@@ -1337,10 +1337,10 @@ apply_handle_update(StringInfo s)
 
 	/* For a partitioned table, apply update to correct partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
+		apply_handle_tuple_routing(resultRelInfo, estate,
 								   remoteslot, &newtup, rel, CMD_UPDATE);
 	else
-		apply_handle_update_internal(estate->es_result_relation_info, estate,
+		apply_handle_update_internal(resultRelInfo, estate,
 									 remoteslot, &newtup, rel);
 
 	PopActiveSnapshot();
@@ -1420,6 +1420,7 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 static void
 apply_handle_delete(StringInfo s)
 {
+	ResultRelInfo *resultRelInfo;
 	LogicalRepRelMapEntry *rel;
 	LogicalRepTupleData oldtup;
 	LogicalRepRelId relid;
@@ -1452,6 +1453,9 @@ apply_handle_delete(StringInfo s)
 	remoteslot = ExecInitExtraTupleSlot(estate,
 										RelationGetDescr(rel->localrel),
 										&TTSOpsVirtual);
+	resultRelInfo = makeNode(ResultRelInfo);
+	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
+	estate->es_result_relation_info = resultRelInfo;
 
 	PushActiveSnapshot(GetTransactionSnapshot());
 
@@ -1462,10 +1466,10 @@ apply_handle_delete(StringInfo s)
 
 	/* For a partitioned table, apply delete to correct partition. */
 	if (rel->localrel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE)
-		apply_handle_tuple_routing(estate->es_result_relation_info, estate,
+		apply_handle_tuple_routing(resultRelInfo, estate,
 								   remoteslot, NULL, rel, CMD_DELETE);
 	else
-		apply_handle_delete_internal(estate->es_result_relation_info, estate,
+		apply_handle_delete_internal(resultRelInfo, estate,
 									 remoteslot, &rel->remoterel);
 
 	PopActiveSnapshot();
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index 415e117407..c283bf1454 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -191,7 +191,6 @@ extern void InitResultRelInfo(ResultRelInfo *resultRelInfo,
 							  Relation partition_root,
 							  int instrument_options);
 extern ResultRelInfo *ExecGetTriggerResultRel(EState *estate, Oid relid);
-extern void ExecCleanUpTriggerState(EState *estate);
 extern void ExecConstraints(ResultRelInfo *resultRelInfo,
 							TupleTableSlot *slot, EState *estate);
 extern bool ExecPartitionCheck(ResultRelInfo *resultRelInfo,
@@ -538,6 +537,8 @@ extern bool ExecRelationIsTargetRelation(EState *estate, Index scanrelid);
 extern Relation ExecOpenScanRelation(EState *estate, Index scanrelid, int eflags);
 
 extern void ExecInitRangeTable(EState *estate, List *rangeTable);
+extern void ExecCloseRangeTableRelations(EState *estate);
+extern void ExecCloseResultRelations(EState *estate);
 
 static inline RangeTblEntry *
 exec_rt_fetch(Index rti, EState *estate)
@@ -546,6 +547,8 @@ exec_rt_fetch(Index rti, EState *estate)
 }
 
 extern Relation ExecGetRangeTableRelation(EState *estate, Index rti);
+extern void ExecInitResultRelation(EState *estate, ResultRelInfo *resultRelInfo,
+								   Index rti);
 
 extern int	executor_errposition(EState *estate, int location);
 
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index ef448d67c7..8683484050 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -519,23 +519,20 @@ typedef struct EState
 	CommandId	es_output_cid;
 
 	/* Info about target table(s) for insert/update/delete queries: */
-	ResultRelInfo *es_result_relations; /* array of ResultRelInfos */
-	int			es_num_result_relations;	/* length of array */
+	ResultRelInfo **es_result_relations;	/* Array of per-range-table-entry
+											 * ResultRelInfo pointers, or NULL
+											 * if a given range table relation
+											 * not a target table */
+	List	   *es_opened_result_relations; /* List of non-NULL entries in
+											 * es_result_relations added in no
+											 * specific order */
 	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
-	/*
-	 * Info about the partition root table(s) for insert/update/delete queries
-	 * targeting partitioned tables.  Only leaf partitions are mentioned in
-	 * es_result_relations, but we need access to the roots for firing
-	 * triggers and for runtime tuple routing.
-	 */
-	ResultRelInfo *es_root_result_relations;	/* array of ResultRelInfos */
-	int			es_num_root_result_relations;	/* length of the array */
 	PartitionDirectory es_partition_directory;	/* for PartitionDesc lookup */
 
 	/*
 	 * The following list contains ResultRelInfos created by the tuple routing
-	 * code for partitions that don't already have one.
+	 * code for partitions that aren't found in the es_result_relations array.
 	 */
 	List	   *es_tuple_routing_result_relations;
 
diff --git a/src/include/nodes/pathnodes.h b/src/include/nodes/pathnodes.h
index dbe86e7af6..c3ed0ee8d4 100644
--- a/src/include/nodes/pathnodes.h
+++ b/src/include/nodes/pathnodes.h
@@ -118,10 +118,6 @@ typedef struct PlannerGlobal
 
 	List	   *finalrowmarks;	/* "flat" list of PlanRowMarks */
 
-	List	   *resultRelations;	/* "flat" list of integer RT indexes */
-
-	List	   *rootResultRelations;	/* "flat" list of integer RT indexes */
-
 	List	   *appendRelations;	/* "flat" list of AppendRelInfos */
 
 	List	   *relationOids;	/* OIDs of relations the plan depends on */
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index 83e01074ed..1943ec4f48 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -65,15 +65,6 @@ typedef struct PlannedStmt
 
 	List	   *rtable;			/* list of RangeTblEntry nodes */
 
-	/* rtable indexes of target relations for INSERT/UPDATE/DELETE */
-	List	   *resultRelations;	/* integer list of RT indexes, or NIL */
-
-	/*
-	 * rtable indexes of partitioned table roots that are UPDATE/DELETE
-	 * targets; needed for trigger firing.
-	 */
-	List	   *rootResultRelations;
-
 	List	   *appendRelations;	/* list of AppendRelInfo nodes */
 
 	List	   *subplans;		/* Plan trees for SubPlan expressions; note
@@ -224,8 +215,6 @@ typedef struct ModifyTable
 	Index		rootRelation;	/* Root RT index, if target is partitioned */
 	bool		partColsUpdated;	/* some part key in hierarchy updated */
 	List	   *resultRelations;	/* integer list of RT indexes */
-	int			resultRelIndex; /* index of first resultRel in plan's list */
-	int			rootResultRelIndex; /* index of the partitioned table root */
 	List	   *plans;			/* plan(s) producing source data */
 	List	   *withCheckOptionLists;	/* per-target-table WCO lists */
 	List	   *returningLists; /* per-target-table RETURNING tlists */
-- 
2.20.1

From fbf1940d054bdafdac3795b62abe1ad17eaa3dce Mon Sep 17 00:00:00 2001
From: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Date: Tue, 13 Oct 2020 18:28:43 +0300
Subject: [PATCH v16 1/2] Include result relation index in ForeignScan for
 direct modify plans.

FDWs that can perform an UPDATE/DELETE remotely using the "direct
modify" set of APIs need in some cases to access the result relation
properties for which they can currently look at
EState.es_result_relation_info, which the core executor laboriously
makes sure is set correctly.  An upcoming patch will remove that field
from EState.

This commit adds a new resultRelation field in ForeignScan, to store
the target relation's RT index. The FDW's PlanDirectModify callback is
expected to set it along with 'operation'. The core code doesn't need
it for anything, but the FDW's Begin- and IterateDirectModify
callbacks can use it to get the target relation's ResultRelInfo.

Amit Langote, Etsuro Fujita
Discussion: https://www.postgresql.org/message-id/CA%2BHiwqGEmiib8FLiHMhKB%2BCH5dRgHSLc5N5wnvc4kym%2BZYpQEQ%40mail.gmail.com
---
 contrib/postgres_fdw/postgres_fdw.c     | 41 +++++++++++++++++--------
 doc/src/sgml/fdwhandler.sgml            | 22 ++++++++-----
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  4 +++
 src/backend/optimizer/plan/setrefs.c    |  4 +++
 src/include/nodes/plannodes.h           |  8 +++++
 8 files changed, 61 insertions(+), 21 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index a31abce7c9..bfd73b40f2 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -218,6 +218,7 @@ typedef struct PgFdwDirectModifyState
 	int			num_tuples;		/* # of result tuples */
 	int			next_tuple;		/* index of next one to return */
 	Relation	resultRel;		/* relcache entry for the target relation */
+	ResultRelInfo *resultRelInfo;	/* ResultRelInfo for the target relation */
 	AttrNumber *attnoMap;		/* array of attnums of input user columns */
 	AttrNumber	ctidAttno;		/* attnum of input ctid column */
 	AttrNumber	oidAttno;		/* attnum of input oid column */
@@ -2287,9 +2288,10 @@ postgresPlanDirectModify(PlannerInfo *root,
 	}
 
 	/*
-	 * Update the operation info.
+	 * Update the operation and target relation info.
 	 */
 	fscan->operation = operation;
+	fscan->resultRelation = resultRelation;
 
 	/*
 	 * Update the fdw_exprs list that will be available to the executor.
@@ -2333,6 +2335,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	EState	   *estate = node->ss.ps.state;
 	PgFdwDirectModifyState *dmstate;
 	Index		rtindex;
+	Relation	rel;
 	RangeTblEntry *rte;
 	Oid			userid;
 	ForeignTable *table;
@@ -2355,18 +2358,31 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	Assert(fsplan->resultRelation > 0);
+	rtindex = fsplan->resultRelation;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
-	/* Get info about foreign table. */
+	/*
+	 * Get info about target table.  For a simple scan on a single foreign
+	 * table, the target table is the table being scanned.  For a join, it's
+	 * one of the tables being joined.
+	 */
 	if (fsplan->scan.scanrelid == 0)
-		dmstate->rel = ExecOpenScanRelation(estate, rtindex, eflags);
+		rel = ExecOpenScanRelation(estate, rtindex, eflags);
 	else
-		dmstate->rel = node->ss.ss_currentRelation;
-	table = GetForeignTable(RelationGetRelid(dmstate->rel));
+	{
+		Assert(rtindex == fsplan->scan.scanrelid);
+		rel = node->ss.ss_currentRelation;
+	}
+	table = GetForeignTable(RelationGetRelid(rel));
 	user = GetUserMapping(userid, table->serverid);
 
+	dmstate->resultRelInfo = estate->es_result_relations[rtindex - 1];
+	/* the executor must have initialized the ResultRelInfo for us. */
+	Assert(dmstate->resultRelInfo != NULL);
+	dmstate->resultRel = rel;
+
 	/*
 	 * Get connection to the foreign server.  Connection manager will
 	 * establish new connection if necessary.
@@ -2376,9 +2392,6 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	/* Update the foreign-join-related fields. */
 	if (fsplan->scan.scanrelid == 0)
 	{
-		/* Save info about foreign table. */
-		dmstate->resultRel = dmstate->rel;
-
 		/*
 		 * Set dmstate->rel to NULL to teach get_returning_data() and
 		 * make_tuple_from_result_row() that columns fetched from the remote
@@ -2387,6 +2400,8 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 		 */
 		dmstate->rel = NULL;
 	}
+	else
+		dmstate->rel = rel;
 
 	/* Initialize state variable */
 	dmstate->num_tuples = -1;	/* -1 means not set yet */
@@ -2450,7 +2465,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4085,8 +4100,8 @@ static TupleTableSlot *
 get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4233,7 +4248,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = dmstate->resultRelInfo;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
@@ -4245,7 +4260,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 	/*
 	 * Use the return tuple slot as a place to store the result tuple.
 	 */
-	resultSlot = ExecGetReturningSlot(estate, relInfo);
+	resultSlot = ExecGetReturningSlot(estate, resultRelInfo);
 
 	/*
 	 * Extract all the values of the scan tuple.
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 72fa127212..43f287a29c 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -861,11 +861,15 @@ PlanDirectModify(PlannerInfo *root,
      To execute the direct modification on the remote server, this function
      must rewrite the target subplan with a <structname>ForeignScan</structname> plan
      node that executes the direct modification on the remote server.  The
-     <structfield>operation</structfield> field of the <structname>ForeignScan</structname> must
-     be set to the <literal>CmdType</literal> enumeration appropriately; that is,
+     <structfield>operation</structfield> and <structfield>resultRelation</structfield> fields
+     of the <structname>ForeignScan</structname> must be set appropriately.
+     <structfield>operation</structfield> must be set to the <literal>CmdType</literal>
+     enumeration corresponding to the statement kind (that is,
      <literal>CMD_UPDATE</literal> for <command>UPDATE</command>,
      <literal>CMD_INSERT</literal> for <command>INSERT</command>, and
-     <literal>CMD_DELETE</literal> for <command>DELETE</command>.
+     <literal>CMD_DELETE</literal> for <command>DELETE</command>), and the
+     <literal>resultRelation</literal> argument must be copied to the
+     <structfield>resultRelation</structfield> field.
     </para>
 
     <para>
@@ -892,9 +896,10 @@ BeginDirectModify(ForeignScanState *node,
      The <structname>ForeignScanState</structname> node has already been created, but
      its <structfield>fdw_state</structfield> field is still NULL.  Information about
      the table to modify is accessible through the
-     <structname>ForeignScanState</structname> node (in particular, from the underlying
-     <structname>ForeignScan</structname> plan node, which contains any FDW-private
-     information provided by <function>PlanDirectModify</function>).
+     <structname>ForeignScanState</structname> node (in particular, from the
+     underlying <structname>ForeignScan</structname> plan node, which contains
+     the target table's range table index and any FDW-private information
+     provided by <function>PlanDirectModify</function>).
      <literal>eflags</literal> contains flag bits describing the executor's
      operating mode for this plan node.
     </para>
@@ -926,8 +931,9 @@ IterateDirectModify(ForeignScanState *node);
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
      or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     <literal>ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
+     of the target foreign table's <structname>ResultRelInfo</structname>
+     obtained using the information passed to <function>BeginDirectModify</function>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 4d79f70950..2b4d7654cc 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelation);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index f441ae3c51..08a049232e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelation);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 3a54765f5c..ab7b535caa 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -2014,6 +2014,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelation);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 881eaf4813..94280a730c 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5530,7 +5530,11 @@ make_foreignscan(List *qptlist,
 	plan->lefttree = outer_plan;
 	plan->righttree = NULL;
 	node->scan.scanrelid = scanrelid;
+
+	/* these may be overridden by the FDW's PlanDirectModify callback. */
 	node->operation = CMD_SELECT;
+	node->resultRelation = 0;
+
 	/* fs_server will be filled in by create_foreignscan_plan */
 	node->fs_server = InvalidOid;
 	node->fdw_exprs = fdw_exprs;
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 6847ff6f44..8b43371425 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -1310,6 +1310,10 @@ set_foreignscan_references(PlannerInfo *root,
 	}
 
 	fscan->fs_relids = offset_relid_set(fscan->fs_relids, rtoffset);
+
+	/* Adjust resultRelation if it's valid */
+	if (fscan->resultRelation > 0)
+		fscan->resultRelation += rtoffset;
 }
 
 /*
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index a7bdf3497e..7e6b10f86b 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -599,12 +599,20 @@ typedef struct WorkTableScan
  * When the plan node represents a foreign join, scan.scanrelid is zero and
  * fs_relids must be consulted to identify the join relation.  (fs_relids
  * is valid for simple scans as well, but will always match scan.scanrelid.)
+ *
+ * If the FDW's PlanDirectModify() callback decides to repurpose a ForeignScan
+ * node to perform the UPDATE or DELETE operation directly in the remote
+ * server, it sets 'operation' and 'resultRelation' to identify the operation
+ * type and target relation.  Note that these fields are only set if the
+ * modification is performed *fully* remotely; otherwise, the modification is
+ * driven by a local ModifyTable node and 'operation' is left to CMD_SELECT.
  * ----------------
  */
 typedef struct ForeignScan
 {
 	Scan		scan;
 	CmdType		operation;		/* SELECT/INSERT/UPDATE/DELETE */
+	Index		resultRelation; /* direct modification target's RT index */
 	Oid			fs_server;		/* OID of foreign server */
 	List	   *fdw_exprs;		/* expressions that FDW may evaluate */
 	List	   *fdw_private;	/* private data for FDW */
-- 
2.20.1

From f02ec7543e4d17075b62734f3c1938f307c5b07c Mon Sep 17 00:00:00 2001
From: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Date: Tue, 13 Oct 2020 18:37:57 +0300
Subject: [PATCH v16 2/2] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.

Author: Amit Langote
Discussion: https://www.postgresql.org/message-id/CA%2BHiwqGEmiib8FLiHMhKB%2BCH5dRgHSLc5N5wnvc4kym%2BZYpQEQ%40mail.gmail.com
---
 src/backend/commands/copy.c              |  19 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |   9 +-
 src/backend/executor/execMain.c          |   4 -
 src/backend/executor/execReplication.c   |  24 +--
 src/backend/executor/execUtils.c         |   1 -
 src/backend/executor/nodeModifyTable.c   | 201 ++++++++++-------------
 src/backend/replication/logical/worker.c |  17 +-
 src/include/executor/executor.h          |  19 ++-
 src/include/executor/nodeModifyTable.h   |   4 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 131 insertions(+), 178 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 71d48d4574..531bd7c73a 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2489,9 +2489,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2524,7 +2521,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2839,8 +2837,6 @@ CopyFrom(CopyState cstate)
 
 	ExecOpenIndices(resultRelInfo, false);
 
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Set up a ModifyTableState so we can let FDW(s) init themselves for
 	 * foreign-table result relation(s).
@@ -3108,11 +3104,6 @@ CopyFrom(CopyState cstate)
 				prevResultRelInfo = resultRelInfo;
 			}
 
-			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
 			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
@@ -3217,7 +3208,8 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot, CMD_INSERT);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot,
+											   CMD_INSERT);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3288,7 +3280,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 80fedad5e0..511f015a86 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1820,7 +1820,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1950,7 +1949,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 1862af621b..c6b5bcba7b 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -501,7 +499,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 783eecbc13..293f53d07c 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -827,9 +827,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 
 	estate->es_plannedstmt = plannedstmt;
 
-	/* es_result_relation_info is NULL except when within ModifyTable */
-	estate->es_result_relation_info = NULL;
-
 	/*
 	 * Next, build the ExecRowMark array from the PlanRowMark(s), if any.
 	 */
@@ -2694,7 +2691,6 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 	 * subplans themselves are initialized.
 	 */
 	parentestate->es_result_relations = NULL;
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index b29db7bf4f..01d26881e7 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -404,10 +404,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -430,7 +430,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -442,7 +443,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -466,11 +468,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -496,7 +498,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -508,7 +511,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -527,11 +531,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index 6d8c112e2f..071a0007eb 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,7 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_opened_result_relations = NIL;
-	estate->es_result_relation_info = NULL;
 	estate->es_tuple_routing_result_relations = NIL;
 	estate->es_trig_target_relations = NIL;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index b3f7012e38..ad9920883b 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,10 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -366,32 +368,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
-
-	ExecMaterializeSlot(slot);
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 
 	/*
-	 * get information on the (current) result relation
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
+
+	ExecMaterializeSlot(slot);
+
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -424,7 +442,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -459,7 +478,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -521,8 +541,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -582,7 +602,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -621,7 +642,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -707,6 +729,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -718,8 +741,7 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
-	Relation	resultRelationDesc;
+	Relation	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
@@ -728,12 +750,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
-	resultRelationDesc = resultRelInfo->ri_RelationDesc;
-
 	/* BEFORE ROW DELETE Triggers */
 	if (resultRelInfo->ri_TrigDesc &&
 		resultRelInfo->ri_TrigDesc->trig_delete_before_row)
@@ -1067,6 +1083,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1075,12 +1092,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
-	Relation	resultRelationDesc;
+	Relation	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1090,12 +1105,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
-	resultRelationDesc = resultRelInfo->ri_RelationDesc;
-
 	/* BEFORE ROW UPDATE Triggers */
 	if (resultRelInfo->ri_TrigDesc &&
 		resultRelInfo->ri_TrigDesc->trig_update_before_row)
@@ -1120,7 +1129,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1157,7 +1167,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1207,6 +1218,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1232,9 +1244,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1274,16 +1289,6 @@ lreplace:;
 				}
 			}
 
-			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
@@ -1301,18 +1306,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
+			/* Tuple routing starts from the root table. */
+			Assert(mtstate->rootResultRelInfo != NULL);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1476,7 +1481,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1715,7 +1721,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1872,40 +1878,36 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
- *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
-	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
 	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
@@ -2016,10 +2018,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2067,17 +2067,6 @@ ExecModifyTable(PlanState *pstate)
 	subplanstate = node->mt_plans[node->mt_whichplan];
 	junkfilter = resultRelInfo->ri_junkFilter;
 
-	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
@@ -2111,7 +2100,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2156,7 +2144,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2239,25 +2226,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2269,15 +2252,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2298,7 +2275,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l,
@@ -2346,14 +2322,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	forboth(l, node->resultRelations, l1, node->plans)
@@ -2400,7 +2370,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2424,8 +2393,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 8d5d9e05b3..4f32dc74c8 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -1174,7 +1174,6 @@ apply_handle_insert(StringInfo s)
 										&TTSOpsVirtual);
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
-	estate->es_result_relation_info = resultRelInfo;
 
 	/* Input functions may need an active snapshot, so get one */
 	PushActiveSnapshot(GetTransactionSnapshot());
@@ -1214,7 +1213,7 @@ apply_handle_insert_internal(ResultRelInfo *relinfo,
 	ExecOpenIndices(relinfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(relinfo, estate, remoteslot);
 
 	/* Cleanup. */
 	ExecCloseIndices(relinfo);
@@ -1300,7 +1299,6 @@ apply_handle_update(StringInfo s)
 										&TTSOpsVirtual);
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
-	estate->es_result_relation_info = resultRelInfo;
 
 	/*
 	 * Populate updatedCols so that per-column triggers can fire.  This could
@@ -1392,7 +1390,8 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(relinfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -1455,7 +1454,6 @@ apply_handle_delete(StringInfo s)
 										&TTSOpsVirtual);
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
-	estate->es_result_relation_info = resultRelInfo;
 
 	PushActiveSnapshot(GetTransactionSnapshot());
 
@@ -1508,7 +1506,7 @@ apply_handle_delete_internal(ResultRelInfo *relinfo, EState *estate,
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(relinfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -1616,7 +1614,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 	}
 	MemoryContextSwitchTo(oldctx);
 
-	estate->es_result_relation_info = partrelinfo;
 	switch (operation)
 	{
 		case CMD_INSERT:
@@ -1697,8 +1694,8 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					ExecOpenIndices(partrelinfo, false);
 
 					EvalPlanQualSetSlot(&epqstate, remoteslot_part);
-					ExecSimpleRelationUpdate(estate, &epqstate, localslot,
-											 remoteslot_part);
+					ExecSimpleRelationUpdate(partrelinfo, estate, &epqstate,
+											 localslot, remoteslot_part);
 					ExecCloseIndices(partrelinfo);
 					EvalPlanQualEnd(&epqstate);
 				}
@@ -1739,7 +1736,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					Assert(partrelinfo_new != partrelinfo);
 
 					/* DELETE old tuple found in the old partition. */
-					estate->es_result_relation_info = partrelinfo;
 					apply_handle_delete_internal(partrelinfo, estate,
 												 localslot,
 												 &relmapentry->remoterel);
@@ -1771,7 +1767,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 						slot_getallattrs(remoteslot);
 					}
 					MemoryContextSwitchTo(oldctx);
-					estate->es_result_relation_info = partrelinfo_new;
 					apply_handle_insert_internal(partrelinfo_new, estate,
 												 remoteslot_part);
 				}
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index c283bf1454..2f221d7114 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -576,10 +576,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -596,10 +600,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 4ec4ebdabc..2518fe4f64 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,9 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index a926ff1711..0310e640fe 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -525,7 +525,6 @@ typedef struct EState
 	List	   *es_opened_result_relations; /* List of non-NULL entries in
 											 * es_result_relations in no
 											 * specific order */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	PartitionDirectory es_partition_directory;	/* for PartitionDesc lookup */
 
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index eb9d45be5e..da50ee3b67 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index ffd4aacbc4..963faa1614 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
2.20.1

From 0206a1429171949de1141c0e9a99a244f7bbc115 Mon Sep 17 00:00:00 2001
From: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Date: Tue, 13 Oct 2020 19:28:00 +0300
Subject: [PATCH v17 1/2] Include result relation index in ForeignScan for
 direct modify plans.

FDWs that can perform an UPDATE/DELETE remotely using the "direct
modify" set of APIs need in some cases to access the result relation
properties for which they can currently look at
EState.es_result_relation_info, which the core executor laboriously
makes sure is set correctly.  An upcoming patch will remove that field
from EState.

This commit adds a new resultRelation field in ForeignScan, to store
the target relation's RT index. The FDW's PlanDirectModify callback is
expected to set it along with 'operation'. The core code doesn't need
it for anything, but the FDW's Begin- and IterateDirectModify
callbacks can use it to get the target relation's ResultRelInfo.

Amit Langote, Etsuro Fujita
Discussion: https://www.postgresql.org/message-id/CA%2BHiwqGEmiib8FLiHMhKB%2BCH5dRgHSLc5N5wnvc4kym%2BZYpQEQ%40mail.gmail.com
---
 contrib/postgres_fdw/postgres_fdw.c     | 16 +++++++++-------
 doc/src/sgml/fdwhandler.sgml            | 15 +++++++++------
 src/backend/executor/nodeForeignscan.c  |  7 +++++++
 src/backend/nodes/copyfuncs.c           |  1 +
 src/backend/nodes/outfuncs.c            |  1 +
 src/backend/nodes/readfuncs.c           |  1 +
 src/backend/optimizer/plan/createplan.c |  4 ++++
 src/backend/optimizer/plan/setrefs.c    |  4 ++++
 src/include/nodes/execnodes.h           |  2 ++
 src/include/nodes/plannodes.h           |  8 ++++++++
 10 files changed, 46 insertions(+), 13 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index a31abce7c9..78facb8ebf 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -451,6 +451,7 @@ static void init_returning_filter(PgFdwDirectModifyState *dmstate,
 								  List *fdw_scan_tlist,
 								  Index rtindex);
 static TupleTableSlot *apply_returning_filter(PgFdwDirectModifyState *dmstate,
+											  ResultRelInfo *resultRelInfo,
 											  TupleTableSlot *slot,
 											  EState *estate);
 static void prepare_query_params(PlanState *node,
@@ -2287,9 +2288,10 @@ postgresPlanDirectModify(PlannerInfo *root,
 	}
 
 	/*
-	 * Update the operation info.
+	 * Update the operation and target relation info.
 	 */
 	fscan->operation = operation;
+	fscan->resultRelation = resultRelation;
 
 	/*
 	 * Update the fdw_exprs list that will be available to the executor.
@@ -2355,7 +2357,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = estate->es_result_relation_info->ri_RangeTableIndex;
+	rtindex = node->resultRelInfo->ri_RangeTableIndex;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
@@ -2450,7 +2452,7 @@ postgresIterateDirectModify(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = node->resultRelInfo;
 
 	/*
 	 * If this is the first call after Begin, execute the statement.
@@ -4086,7 +4088,7 @@ get_returning_data(ForeignScanState *node)
 {
 	PgFdwDirectModifyState *dmstate = (PgFdwDirectModifyState *) node->fdw_state;
 	EState	   *estate = node->ss.ps.state;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
+	ResultRelInfo *resultRelInfo = node->resultRelInfo;
 	TupleTableSlot *slot = node->ss.ss_ScanTupleSlot;
 	TupleTableSlot *resultSlot;
 
@@ -4141,7 +4143,7 @@ get_returning_data(ForeignScanState *node)
 		if (dmstate->rel)
 			resultSlot = slot;
 		else
-			resultSlot = apply_returning_filter(dmstate, slot, estate);
+			resultSlot = apply_returning_filter(dmstate, resultRelInfo, slot, estate);
 	}
 	dmstate->next_tuple++;
 
@@ -4230,10 +4232,10 @@ init_returning_filter(PgFdwDirectModifyState *dmstate,
  */
 static TupleTableSlot *
 apply_returning_filter(PgFdwDirectModifyState *dmstate,
+					   ResultRelInfo *resultRelInfo,
 					   TupleTableSlot *slot,
 					   EState *estate)
 {
-	ResultRelInfo *relInfo = estate->es_result_relation_info;
 	TupleDesc	resultTupType = RelationGetDescr(dmstate->resultRel);
 	TupleTableSlot *resultSlot;
 	Datum	   *values;
@@ -4245,7 +4247,7 @@ apply_returning_filter(PgFdwDirectModifyState *dmstate,
 	/*
 	 * Use the return tuple slot as a place to store the result tuple.
 	 */
-	resultSlot = ExecGetReturningSlot(estate, relInfo);
+	resultSlot = ExecGetReturningSlot(estate, resultRelInfo);
 
 	/*
 	 * Extract all the values of the scan tuple.
diff --git a/doc/src/sgml/fdwhandler.sgml b/doc/src/sgml/fdwhandler.sgml
index 72fa127212..9c9293414c 100644
--- a/doc/src/sgml/fdwhandler.sgml
+++ b/doc/src/sgml/fdwhandler.sgml
@@ -861,11 +861,15 @@ PlanDirectModify(PlannerInfo *root,
      To execute the direct modification on the remote server, this function
      must rewrite the target subplan with a <structname>ForeignScan</structname> plan
      node that executes the direct modification on the remote server.  The
-     <structfield>operation</structfield> field of the <structname>ForeignScan</structname> must
-     be set to the <literal>CmdType</literal> enumeration appropriately; that is,
+     <structfield>operation</structfield> and <structfield>resultRelation</structfield> fields
+     of the <structname>ForeignScan</structname> must be set appropriately.
+     <structfield>operation</structfield> must be set to the <literal>CmdType</literal>
+     enumeration corresponding to the statement kind (that is,
      <literal>CMD_UPDATE</literal> for <command>UPDATE</command>,
      <literal>CMD_INSERT</literal> for <command>INSERT</command>, and
-     <literal>CMD_DELETE</literal> for <command>DELETE</command>.
+     <literal>CMD_DELETE</literal> for <command>DELETE</command>), and the
+     <literal>resultRelation</literal> argument must be copied to the
+     <structfield>resultRelation</structfield> field.
     </para>
 
     <para>
@@ -925,9 +929,8 @@ IterateDirectModify(ForeignScanState *node);
      needed for the <literal>RETURNING</literal> calculation, returning it in a
      tuple table slot (the node's <structfield>ScanTupleSlot</structfield> should be
      used for this purpose).  The data that was actually inserted, updated
-     or deleted must be stored in the
-     <literal>es_result_relation_info-&gt;ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>
-     of the node's <structname>EState</structname>.
+     or deleted must be stored in
+     <literal>node->resultRelInfo->ri_projectReturning-&gt;pi_exprContext-&gt;ecxt_scantuple</literal>.
      Return NULL if no more rows are available.
      Note that this is called in a short-lived memory context that will be
      reset between invocations.  Create a memory context in
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 513471ab9b..0b20f94035 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -215,6 +215,13 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	scanstate->fdwroutine = fdwroutine;
 	scanstate->fdw_state = NULL;
 
+	/*
+	 * For the FDW's convenience, look up the modification target relation's.
+	 * ResultRelInfo.
+	 */
+	if (node->resultRelation > 0)
+		scanstate->resultRelInfo = estate->es_result_relations[node->resultRelation - 1];
+
 	/* Initialize any outer plan. */
 	if (outerPlan(node))
 		outerPlanState(scanstate) =
diff --git a/src/backend/nodes/copyfuncs.c b/src/backend/nodes/copyfuncs.c
index 4d79f70950..2b4d7654cc 100644
--- a/src/backend/nodes/copyfuncs.c
+++ b/src/backend/nodes/copyfuncs.c
@@ -758,6 +758,7 @@ _copyForeignScan(const ForeignScan *from)
 	COPY_NODE_FIELD(fdw_recheck_quals);
 	COPY_BITMAPSET_FIELD(fs_relids);
 	COPY_SCALAR_FIELD(fsSystemCol);
+	COPY_SCALAR_FIELD(resultRelation);
 
 	return newnode;
 }
diff --git a/src/backend/nodes/outfuncs.c b/src/backend/nodes/outfuncs.c
index f441ae3c51..08a049232e 100644
--- a/src/backend/nodes/outfuncs.c
+++ b/src/backend/nodes/outfuncs.c
@@ -695,6 +695,7 @@ _outForeignScan(StringInfo str, const ForeignScan *node)
 	WRITE_NODE_FIELD(fdw_recheck_quals);
 	WRITE_BITMAPSET_FIELD(fs_relids);
 	WRITE_BOOL_FIELD(fsSystemCol);
+	WRITE_INT_FIELD(resultRelation);
 }
 
 static void
diff --git a/src/backend/nodes/readfuncs.c b/src/backend/nodes/readfuncs.c
index 3a54765f5c..ab7b535caa 100644
--- a/src/backend/nodes/readfuncs.c
+++ b/src/backend/nodes/readfuncs.c
@@ -2014,6 +2014,7 @@ _readForeignScan(void)
 	READ_NODE_FIELD(fdw_recheck_quals);
 	READ_BITMAPSET_FIELD(fs_relids);
 	READ_BOOL_FIELD(fsSystemCol);
+	READ_INT_FIELD(resultRelation);
 
 	READ_DONE();
 }
diff --git a/src/backend/optimizer/plan/createplan.c b/src/backend/optimizer/plan/createplan.c
index 881eaf4813..94280a730c 100644
--- a/src/backend/optimizer/plan/createplan.c
+++ b/src/backend/optimizer/plan/createplan.c
@@ -5530,7 +5530,11 @@ make_foreignscan(List *qptlist,
 	plan->lefttree = outer_plan;
 	plan->righttree = NULL;
 	node->scan.scanrelid = scanrelid;
+
+	/* these may be overridden by the FDW's PlanDirectModify callback. */
 	node->operation = CMD_SELECT;
+	node->resultRelation = 0;
+
 	/* fs_server will be filled in by create_foreignscan_plan */
 	node->fs_server = InvalidOid;
 	node->fdw_exprs = fdw_exprs;
diff --git a/src/backend/optimizer/plan/setrefs.c b/src/backend/optimizer/plan/setrefs.c
index 6847ff6f44..8b43371425 100644
--- a/src/backend/optimizer/plan/setrefs.c
+++ b/src/backend/optimizer/plan/setrefs.c
@@ -1310,6 +1310,10 @@ set_foreignscan_references(PlannerInfo *root,
 	}
 
 	fscan->fs_relids = offset_relid_set(fscan->fs_relids, rtoffset);
+
+	/* Adjust resultRelation if it's valid */
+	if (fscan->resultRelation > 0)
+		fscan->resultRelation += rtoffset;
 }
 
 /*
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index a926ff1711..da1f3f269a 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -1780,6 +1780,8 @@ typedef struct ForeignScanState
 	/* use struct pointer to avoid including fdwapi.h here */
 	struct FdwRoutine *fdwroutine;
 	void	   *fdw_state;		/* foreign-data wrapper can keep state here */
+
+	ResultRelInfo *resultRelInfo;	/* result rel info, if UPDATE or DELETE */
 } ForeignScanState;
 
 /* ----------------
diff --git a/src/include/nodes/plannodes.h b/src/include/nodes/plannodes.h
index a7bdf3497e..7e6b10f86b 100644
--- a/src/include/nodes/plannodes.h
+++ b/src/include/nodes/plannodes.h
@@ -599,12 +599,20 @@ typedef struct WorkTableScan
  * When the plan node represents a foreign join, scan.scanrelid is zero and
  * fs_relids must be consulted to identify the join relation.  (fs_relids
  * is valid for simple scans as well, but will always match scan.scanrelid.)
+ *
+ * If the FDW's PlanDirectModify() callback decides to repurpose a ForeignScan
+ * node to perform the UPDATE or DELETE operation directly in the remote
+ * server, it sets 'operation' and 'resultRelation' to identify the operation
+ * type and target relation.  Note that these fields are only set if the
+ * modification is performed *fully* remotely; otherwise, the modification is
+ * driven by a local ModifyTable node and 'operation' is left to CMD_SELECT.
  * ----------------
  */
 typedef struct ForeignScan
 {
 	Scan		scan;
 	CmdType		operation;		/* SELECT/INSERT/UPDATE/DELETE */
+	Index		resultRelation; /* direct modification target's RT index */
 	Oid			fs_server;		/* OID of foreign server */
 	List	   *fdw_exprs;		/* expressions that FDW may evaluate */
 	List	   *fdw_private;	/* private data for FDW */
-- 
2.20.1

From 4e7aa355561303ca9998993964d5cf393eff50f3 Mon Sep 17 00:00:00 2001
From: Heikki Linnakangas <heikki.linnakangas@iki.fi>
Date: Tue, 13 Oct 2020 18:37:57 +0300
Subject: [PATCH v17 2/2] Remove es_result_relation_info

This changes many places that access the currently active result
relation via es_result_relation_info to instead receive it directly
via function parameters.  Maintaining that state in
es_result_relation_info has become cumbersome, especially with
partitioning where each partition gets its own result relation info.
Having to set and reset it across arbitrary operations has caused
bugs in the past.

Author: Amit Langote
Discussion: https://www.postgresql.org/message-id/CA%2BHiwqGEmiib8FLiHMhKB%2BCH5dRgHSLc5N5wnvc4kym%2BZYpQEQ%40mail.gmail.com
---
 src/backend/commands/copy.c              |  19 +--
 src/backend/commands/tablecmds.c         |   2 -
 src/backend/executor/execIndexing.c      |   9 +-
 src/backend/executor/execMain.c          |   4 -
 src/backend/executor/execReplication.c   |  24 +--
 src/backend/executor/execUtils.c         |   1 -
 src/backend/executor/nodeModifyTable.c   | 201 ++++++++++-------------
 src/backend/replication/logical/worker.c |  17 +-
 src/include/executor/executor.h          |  19 ++-
 src/include/executor/nodeModifyTable.h   |   4 +-
 src/include/nodes/execnodes.h            |   1 -
 src/test/regress/expected/insert.out     |   4 +-
 src/test/regress/sql/insert.sql          |   4 +-
 13 files changed, 131 insertions(+), 178 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 71d48d4574..531bd7c73a 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -2489,9 +2489,6 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 	ResultRelInfo *resultRelInfo = buffer->resultRelInfo;
 	TupleTableSlot **slots = buffer->slots;
 
-	/* Set es_result_relation_info to the ResultRelInfo we're flushing. */
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Print error context information correctly, if one of the operations
 	 * below fail.
@@ -2524,7 +2521,8 @@ CopyMultiInsertBufferFlush(CopyMultiInsertInfo *miinfo,
 
 			cstate->cur_lineno = buffer->linenos[i];
 			recheckIndexes =
-				ExecInsertIndexTuples(buffer->slots[i], estate, false, NULL,
+				ExecInsertIndexTuples(resultRelInfo,
+									  buffer->slots[i], estate, false, NULL,
 									  NIL);
 			ExecARInsertTriggers(estate, resultRelInfo,
 								 slots[i], recheckIndexes,
@@ -2839,8 +2837,6 @@ CopyFrom(CopyState cstate)
 
 	ExecOpenIndices(resultRelInfo, false);
 
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Set up a ModifyTableState so we can let FDW(s) init themselves for
 	 * foreign-table result relation(s).
@@ -3108,11 +3104,6 @@ CopyFrom(CopyState cstate)
 				prevResultRelInfo = resultRelInfo;
 			}
 
-			/*
-			 * For ExecInsertIndexTuples() to work on the partition's indexes
-			 */
-			estate->es_result_relation_info = resultRelInfo;
-
 			/*
 			 * If we're capturing transition tuples, we might need to convert
 			 * from the partition rowtype to root rowtype.
@@ -3217,7 +3208,8 @@ CopyFrom(CopyState cstate)
 				/* Compute stored generated columns */
 				if (resultRelInfo->ri_RelationDesc->rd_att->constr &&
 					resultRelInfo->ri_RelationDesc->rd_att->constr->has_generated_stored)
-					ExecComputeStoredGenerated(estate, myslot, CMD_INSERT);
+					ExecComputeStoredGenerated(resultRelInfo, estate, myslot,
+											   CMD_INSERT);
 
 				/*
 				 * If the target is a plain table, check the constraints of
@@ -3288,7 +3280,8 @@ CopyFrom(CopyState cstate)
 										   myslot, mycid, ti_options, bistate);
 
 						if (resultRelInfo->ri_NumIndices > 0)
-							recheckIndexes = ExecInsertIndexTuples(myslot,
+							recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+																   myslot,
 																   estate,
 																   false,
 																   NULL,
diff --git a/src/backend/commands/tablecmds.c b/src/backend/commands/tablecmds.c
index 80fedad5e0..511f015a86 100644
--- a/src/backend/commands/tablecmds.c
+++ b/src/backend/commands/tablecmds.c
@@ -1820,7 +1820,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecBSTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
@@ -1950,7 +1949,6 @@ ExecuteTruncateGuts(List *explicit_rels, List *relids, List *relids_logged,
 	resultRelInfo = resultRelInfos;
 	foreach(cell, rels)
 	{
-		estate->es_result_relation_info = resultRelInfo;
 		ExecASTruncateTriggers(estate, resultRelInfo);
 		resultRelInfo++;
 	}
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 1862af621b..c6b5bcba7b 100644
--- a/src/backend/executor/execIndexing.c
+++ b/src/backend/executor/execIndexing.c
@@ -270,7 +270,8 @@ ExecCloseIndices(ResultRelInfo *resultRelInfo)
  * ----------------------------------------------------------------
  */
 List *
-ExecInsertIndexTuples(TupleTableSlot *slot,
+ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+					  TupleTableSlot *slot,
 					  EState *estate,
 					  bool noDupErr,
 					  bool *specConflict,
@@ -278,7 +279,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 {
 	ItemPointer tupleid = &slot->tts_tid;
 	List	   *result = NIL;
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -293,7 +293,6 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
@@ -479,11 +478,10 @@ ExecInsertIndexTuples(TupleTableSlot *slot,
  * ----------------------------------------------------------------
  */
 bool
-ExecCheckIndexConstraints(TupleTableSlot *slot,
+ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo, TupleTableSlot *slot,
 						  EState *estate, ItemPointer conflictTid,
 						  List *arbiterIndexes)
 {
-	ResultRelInfo *resultRelInfo;
 	int			i;
 	int			numIndices;
 	RelationPtr relationDescs;
@@ -501,7 +499,6 @@ ExecCheckIndexConstraints(TupleTableSlot *slot,
 	/*
 	 * Get information from the result relation info structure.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
 	numIndices = resultRelInfo->ri_NumIndices;
 	relationDescs = resultRelInfo->ri_IndexRelationDescs;
 	indexInfoArray = resultRelInfo->ri_IndexRelationInfo;
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 783eecbc13..293f53d07c 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -827,9 +827,6 @@ InitPlan(QueryDesc *queryDesc, int eflags)
 
 	estate->es_plannedstmt = plannedstmt;
 
-	/* es_result_relation_info is NULL except when within ModifyTable */
-	estate->es_result_relation_info = NULL;
-
 	/*
 	 * Next, build the ExecRowMark array from the PlanRowMark(s), if any.
 	 */
@@ -2694,7 +2691,6 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 	 * subplans themselves are initialized.
 	 */
 	parentestate->es_result_relations = NULL;
-	/* es_result_relation_info must NOT be copied */
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
diff --git a/src/backend/executor/execReplication.c b/src/backend/executor/execReplication.c
index b29db7bf4f..01d26881e7 100644
--- a/src/backend/executor/execReplication.c
+++ b/src/backend/executor/execReplication.c
@@ -404,10 +404,10 @@ retry:
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
+ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+						 EState *estate, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 
 	/* For now we support only tables. */
@@ -430,7 +430,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -442,7 +443,8 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
 		simple_table_tuple_insert(resultRelInfo->ri_RelationDesc, slot);
 
 		if (resultRelInfo->ri_NumIndices > 0)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW INSERT Triggers */
@@ -466,11 +468,11 @@ ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot)
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot, TupleTableSlot *slot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &(searchslot->tts_tid);
 
@@ -496,7 +498,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 		/* Compute stored generated columns */
 		if (rel->rd_att->constr &&
 			rel->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/* Check the constraints of the tuple */
 		if (rel->rd_att->constr)
@@ -508,7 +511,8 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
 								  &update_indexes);
 
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL,
 												   NIL);
 
 		/* AFTER ROW UPDATE Triggers */
@@ -527,11 +531,11 @@ ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
  * Caller is responsible for opening the indexes.
  */
 void
-ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+						 EState *estate, EPQState *epqstate,
 						 TupleTableSlot *searchslot)
 {
 	bool		skip_tuple = false;
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	ItemPointer tid = &searchslot->tts_tid;
 
diff --git a/src/backend/executor/execUtils.c b/src/backend/executor/execUtils.c
index 6d8c112e2f..071a0007eb 100644
--- a/src/backend/executor/execUtils.c
+++ b/src/backend/executor/execUtils.c
@@ -125,7 +125,6 @@ CreateExecutorState(void)
 
 	estate->es_result_relations = NULL;
 	estate->es_opened_result_relations = NIL;
-	estate->es_result_relation_info = NULL;
 	estate->es_tuple_routing_result_relations = NIL;
 	estate->es_trig_target_relations = NIL;
 
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index b3f7012e38..ad9920883b 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -70,7 +70,8 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   EState *estate,
 											   PartitionTupleRouting *proute,
 											   ResultRelInfo *targetRelInfo,
-											   TupleTableSlot *slot);
+											   TupleTableSlot *slot,
+											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
 static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
 static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
@@ -246,9 +247,10 @@ ExecCheckTIDVisible(EState *estate,
  * Compute stored generated columns for a tuple
  */
 void
-ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype)
+ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype)
 {
-	ResultRelInfo *resultRelInfo = estate->es_result_relation_info;
 	Relation	rel = resultRelInfo->ri_RelationDesc;
 	TupleDesc	tupdesc = RelationGetDescr(rel);
 	int			natts = tupdesc->natts;
@@ -366,32 +368,48 @@ ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype
  *		ExecInsert
  *
  *		For INSERT, we have to insert the tuple into the target relation
- *		and insert appropriate tuples into the index relations.
+ *		(or partition thereof) and insert appropriate tuples into the index
+ *		relations.
  *
  *		Returns RETURNING result if any, otherwise NULL.
+ *
+ *		This may change the currently active tuple conversion map in
+ *		mtstate->mt_transition_capture, so the callers must take care to
+ *		save the previous value to avoid losing track of it.
  * ----------------------------------------------------------------
  */
 static TupleTableSlot *
 ExecInsert(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   TupleTableSlot *slot,
 		   TupleTableSlot *planSlot,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
 	Relation	resultRelationDesc;
 	List	   *recheckIndexes = NIL;
 	TupleTableSlot *result = NULL;
 	TransitionCaptureState *ar_insert_trig_tcs;
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	OnConflictAction onconflict = node->onConflictAction;
-
-	ExecMaterializeSlot(slot);
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 
 	/*
-	 * get information on the (current) result relation
+	 * If the input result relation is a partitioned table, find the leaf
+	 * partition to insert the tuple into.
 	 */
-	resultRelInfo = estate->es_result_relation_info;
+	if (proute)
+	{
+		ResultRelInfo *partRelInfo;
+
+		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
+									   resultRelInfo, slot,
+									   &partRelInfo);
+		resultRelInfo = partRelInfo;
+	}
+
+	ExecMaterializeSlot(slot);
+
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 
 	/*
@@ -424,7 +442,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * insert into foreign table: let the FDW do it
@@ -459,7 +478,8 @@ ExecInsert(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_INSERT);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_INSERT);
 
 		/*
 		 * Check any RLS WITH CHECK policies.
@@ -521,8 +541,8 @@ ExecInsert(ModifyTableState *mtstate,
 			 */
 	vlock:
 			specConflict = false;
-			if (!ExecCheckIndexConstraints(slot, estate, &conflictTid,
-										   arbiterIndexes))
+			if (!ExecCheckIndexConstraints(resultRelInfo, slot, estate,
+										   &conflictTid, arbiterIndexes))
 			{
 				/* committed conflict tuple found */
 				if (onconflict == ONCONFLICT_UPDATE)
@@ -582,7 +602,8 @@ ExecInsert(ModifyTableState *mtstate,
 										   specToken);
 
 			/* insert index entries for tuple */
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, true,
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, true,
 												   &specConflict,
 												   arbiterIndexes);
 
@@ -621,7 +642,8 @@ ExecInsert(ModifyTableState *mtstate,
 
 			/* insert index entries for tuple */
 			if (resultRelInfo->ri_NumIndices > 0)
-				recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL,
+				recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+													   slot, estate, false, NULL,
 													   NIL);
 		}
 	}
@@ -707,6 +729,7 @@ ExecInsert(ModifyTableState *mtstate,
  */
 static TupleTableSlot *
 ExecDelete(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *planSlot,
@@ -718,8 +741,7 @@ ExecDelete(ModifyTableState *mtstate,
 		   bool *tupleDeleted,
 		   TupleTableSlot **epqreturnslot)
 {
-	ResultRelInfo *resultRelInfo;
-	Relation	resultRelationDesc;
+	Relation	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	TupleTableSlot *slot = NULL;
@@ -728,12 +750,6 @@ ExecDelete(ModifyTableState *mtstate,
 	if (tupleDeleted)
 		*tupleDeleted = false;
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
-	resultRelationDesc = resultRelInfo->ri_RelationDesc;
-
 	/* BEFORE ROW DELETE Triggers */
 	if (resultRelInfo->ri_TrigDesc &&
 		resultRelInfo->ri_TrigDesc->trig_delete_before_row)
@@ -1067,6 +1083,7 @@ ldelete:;
  */
 static TupleTableSlot *
 ExecUpdate(ModifyTableState *mtstate,
+		   ResultRelInfo *resultRelInfo,
 		   ItemPointer tupleid,
 		   HeapTuple oldtuple,
 		   TupleTableSlot *slot,
@@ -1075,12 +1092,10 @@ ExecUpdate(ModifyTableState *mtstate,
 		   EState *estate,
 		   bool canSetTag)
 {
-	ResultRelInfo *resultRelInfo;
-	Relation	resultRelationDesc;
+	Relation	resultRelationDesc = resultRelInfo->ri_RelationDesc;
 	TM_Result	result;
 	TM_FailureData tmfd;
 	List	   *recheckIndexes = NIL;
-	TupleConversionMap *saved_tcs_map = NULL;
 
 	/*
 	 * abort the operation if not running transactions
@@ -1090,12 +1105,6 @@ ExecUpdate(ModifyTableState *mtstate,
 
 	ExecMaterializeSlot(slot);
 
-	/*
-	 * get information on the (current) result relation
-	 */
-	resultRelInfo = estate->es_result_relation_info;
-	resultRelationDesc = resultRelInfo->ri_RelationDesc;
-
 	/* BEFORE ROW UPDATE Triggers */
 	if (resultRelInfo->ri_TrigDesc &&
 		resultRelInfo->ri_TrigDesc->trig_update_before_row)
@@ -1120,7 +1129,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * update in foreign table: let the FDW do it
@@ -1157,7 +1167,8 @@ ExecUpdate(ModifyTableState *mtstate,
 		 */
 		if (resultRelationDesc->rd_att->constr &&
 			resultRelationDesc->rd_att->constr->has_generated_stored)
-			ExecComputeStoredGenerated(estate, slot, CMD_UPDATE);
+			ExecComputeStoredGenerated(resultRelInfo, estate, slot,
+									   CMD_UPDATE);
 
 		/*
 		 * Check any RLS UPDATE WITH CHECK policies
@@ -1207,6 +1218,7 @@ lreplace:;
 			PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 			int			map_index;
 			TupleConversionMap *tupconv_map;
+			TupleConversionMap *saved_tcs_map = NULL;
 
 			/*
 			 * Disallow an INSERT ON CONFLICT DO UPDATE that causes the
@@ -1232,9 +1244,12 @@ lreplace:;
 			 * Row movement, part 1.  Delete the tuple, but skip RETURNING
 			 * processing. We want to return rows from INSERT.
 			 */
-			ExecDelete(mtstate, tupleid, oldtuple, planSlot, epqstate,
-					   estate, false, false /* canSetTag */ ,
-					   true /* changingPart */ , &tuple_deleted, &epqslot);
+			ExecDelete(mtstate, resultRelInfo, tupleid, oldtuple, planSlot,
+					   epqstate, estate,
+					   false,	/* processReturning */
+					   false,	/* canSetTag */
+					   true,	/* changingPart */
+					   &tuple_deleted, &epqslot);
 
 			/*
 			 * For some reason if DELETE didn't happen (e.g. trigger prevented
@@ -1274,16 +1289,6 @@ lreplace:;
 				}
 			}
 
-			/*
-			 * Updates set the transition capture map only when a new subplan
-			 * is chosen.  But for inserts, it is set for each row. So after
-			 * INSERT, we need to revert back to the map created for UPDATE;
-			 * otherwise the next UPDATE will incorrectly use the one created
-			 * for INSERT.  So first save the one created for UPDATE.
-			 */
-			if (mtstate->mt_transition_capture)
-				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 			/*
 			 * resultRelInfo is one of the per-subplan resultRelInfos.  So we
 			 * should convert the tuple into root's tuple descriptor, since
@@ -1301,18 +1306,18 @@ lreplace:;
 											 mtstate->mt_root_tuple_slot);
 
 			/*
-			 * Prepare for tuple routing, making it look like we're inserting
-			 * into the root.
+			 * ExecInsert() may scribble on mtstate->mt_transition_capture,
+			 * so save the currently active map.
 			 */
-			Assert(mtstate->rootResultRelInfo != NULL);
-			slot = ExecPrepareTupleRouting(mtstate, estate, proute,
-										   mtstate->rootResultRelInfo, slot);
+			if (mtstate->mt_transition_capture)
+				saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
 
-			ret_slot = ExecInsert(mtstate, slot, planSlot,
-								  estate, canSetTag);
+			/* Tuple routing starts from the root table. */
+			Assert(mtstate->rootResultRelInfo != NULL);
+			ret_slot = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
+								  planSlot, estate, canSetTag);
 
-			/* Revert ExecPrepareTupleRouting's node change. */
-			estate->es_result_relation_info = resultRelInfo;
+			/* Clear the INSERT's tuple and restore the saved map. */
 			if (mtstate->mt_transition_capture)
 			{
 				mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
@@ -1476,7 +1481,8 @@ lreplace:;
 
 		/* insert index entries for tuple if necessary */
 		if (resultRelInfo->ri_NumIndices > 0 && update_indexes)
-			recheckIndexes = ExecInsertIndexTuples(slot, estate, false, NULL, NIL);
+			recheckIndexes = ExecInsertIndexTuples(resultRelInfo,
+												   slot, estate, false, NULL, NIL);
 	}
 
 	if (canSetTag)
@@ -1715,7 +1721,7 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	 */
 
 	/* Execute UPDATE with projection */
-	*returning = ExecUpdate(mtstate, conflictTid, NULL,
+	*returning = ExecUpdate(mtstate, resultRelInfo, conflictTid, NULL,
 							resultRelInfo->ri_onConflict->oc_ProjSlot,
 							planSlot,
 							&mtstate->mt_epqstate, mtstate->ps.state,
@@ -1872,40 +1878,36 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
  * ExecPrepareTupleRouting --- prepare for routing one tuple
  *
  * Determine the partition in which the tuple in slot is to be inserted,
- * and modify mtstate and estate to prepare for it.
- *
- * Caller must revert the estate changes after executing the insertion!
- * In mtstate, transition capture changes may also need to be reverted.
+ * and return its ResultRelInfo in *partRelInfo.  The returned value is
+ * a slot holding the tuple of the partition rowtype.
  *
- * Returns a slot holding the tuple of the partition rowtype.
+ * This also sets the transition table information in mtstate based on the
+ * selected partition.
  */
 static TupleTableSlot *
 ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						EState *estate,
 						PartitionTupleRouting *proute,
 						ResultRelInfo *targetRelInfo,
-						TupleTableSlot *slot)
+						TupleTableSlot *slot,
+						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 
 	/*
-	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
-	 * not find a valid partition for the tuple in 'slot' then an error is
+	 * Look up the target partition's ResultRelInfo.  If ExecFindPartition
+	 * doesn't find a valid partition for the tuple in 'slot' then an error is
 	 * raised.  An error may also be raised if the found partition is not a
 	 * valid target for INSERTs.  This is required since a partitioned table
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
+	*partRelInfo = partrel;
 	partrouteinfo = partrel->ri_PartitionInfo;
 	Assert(partrouteinfo != NULL);
 
-	/*
-	 * Make it look like we are inserting into the partition.
-	 */
-	estate->es_result_relation_info = partrel;
-
 	/*
 	 * If we're capturing transition tuples, we might need to convert from the
 	 * partition rowtype to root partitioned table's rowtype.
@@ -2016,10 +2018,8 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
-	PartitionTupleRouting *proute = node->mt_partition_tuple_routing;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	PlanState  *subplanstate;
 	JunkFilter *junkfilter;
@@ -2067,17 +2067,6 @@ ExecModifyTable(PlanState *pstate)
 	subplanstate = node->mt_plans[node->mt_whichplan];
 	junkfilter = resultRelInfo->ri_junkFilter;
 
-	/*
-	 * es_result_relation_info must point to the currently active result
-	 * relation while we are within this ModifyTable node.  Even though
-	 * ModifyTable nodes can't be nested statically, they can be nested
-	 * dynamically (since our subplan could include a reference to a modifying
-	 * CTE).  So we have to save and restore the caller's value.
-	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
-	estate->es_result_relation_info = resultRelInfo;
-
 	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
 	 * for each row.
@@ -2111,7 +2100,6 @@ ExecModifyTable(PlanState *pstate)
 				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
 				junkfilter = resultRelInfo->ri_junkFilter;
-				estate->es_result_relation_info = resultRelInfo;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				/* Prepare to convert transition tuples from this child. */
@@ -2156,7 +2144,6 @@ ExecModifyTable(PlanState *pstate)
 			 */
 			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
 
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
 		}
 
@@ -2239,25 +2226,21 @@ ExecModifyTable(PlanState *pstate)
 		switch (operation)
 		{
 			case CMD_INSERT:
-				/* Prepare for tuple routing if needed. */
-				if (proute)
-					slot = ExecPrepareTupleRouting(node, estate, proute,
-												   resultRelInfo, slot);
-				slot = ExecInsert(node, slot, planSlot,
+				slot = ExecInsert(node, resultRelInfo, slot, planSlot,
 								  estate, node->canSetTag);
-				/* Revert ExecPrepareTupleRouting's state change. */
-				if (proute)
-					estate->es_result_relation_info = resultRelInfo;
 				break;
 			case CMD_UPDATE:
-				slot = ExecUpdate(node, tupleid, oldtuple, slot, planSlot,
-								  &node->mt_epqstate, estate, node->canSetTag);
+				slot = ExecUpdate(node, resultRelInfo, tupleid, oldtuple, slot,
+								  planSlot, &node->mt_epqstate, estate,
+								  node->canSetTag);
 				break;
 			case CMD_DELETE:
-				slot = ExecDelete(node, tupleid, oldtuple, planSlot,
-								  &node->mt_epqstate, estate,
-								  true, node->canSetTag,
-								  false /* changingPart */ , NULL, NULL);
+				slot = ExecDelete(node, resultRelInfo, tupleid, oldtuple,
+								  planSlot, &node->mt_epqstate, estate,
+								  true,		/* processReturning */
+								  node->canSetTag,
+								  false,	/* changingPart */
+								  NULL, NULL);
 				break;
 			default:
 				elog(ERROR, "unknown operation");
@@ -2269,15 +2252,9 @@ ExecModifyTable(PlanState *pstate)
 		 * the work on next call.
 		 */
 		if (slot)
-		{
-			estate->es_result_relation_info = saved_resultRelInfo;
 			return slot;
-		}
 	}
 
-	/* Restore es_result_relation_info before exiting */
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/*
 	 * We're done, but fire AFTER STATEMENT triggers before exiting.
 	 */
@@ -2298,7 +2275,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *saved_resultRelInfo;
 	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
 	ListCell   *l,
@@ -2346,14 +2322,8 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.  Note we *must* set
-	 * estate->es_result_relation_info correctly while we initialize each
-	 * sub-plan; external modules such as FDWs may depend on that (see
-	 * contrib/postgres_fdw/postgres_fdw.c: postgresBeginDirectModify() as one
-	 * example).
+	 * indexes for insertion of new index entries.
 	 */
-	saved_resultRelInfo = estate->es_result_relation_info;
-
 	resultRelInfo = mtstate->resultRelInfo;
 	i = 0;
 	forboth(l, node->resultRelations, l1, node->plans)
@@ -2400,7 +2370,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			update_tuple_routing_needed = true;
 
 		/* Now init the plan for this result rel */
-		estate->es_result_relation_info = resultRelInfo;
 		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
 		mtstate->mt_scans[i] =
 			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
@@ -2424,8 +2393,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		i++;
 	}
 
-	estate->es_result_relation_info = saved_resultRelInfo;
-
 	/* Get the target relation */
 	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
 
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index 8d5d9e05b3..4f32dc74c8 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -1174,7 +1174,6 @@ apply_handle_insert(StringInfo s)
 										&TTSOpsVirtual);
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
-	estate->es_result_relation_info = resultRelInfo;
 
 	/* Input functions may need an active snapshot, so get one */
 	PushActiveSnapshot(GetTransactionSnapshot());
@@ -1214,7 +1213,7 @@ apply_handle_insert_internal(ResultRelInfo *relinfo,
 	ExecOpenIndices(relinfo, false);
 
 	/* Do the insert. */
-	ExecSimpleRelationInsert(estate, remoteslot);
+	ExecSimpleRelationInsert(relinfo, estate, remoteslot);
 
 	/* Cleanup. */
 	ExecCloseIndices(relinfo);
@@ -1300,7 +1299,6 @@ apply_handle_update(StringInfo s)
 										&TTSOpsVirtual);
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
-	estate->es_result_relation_info = resultRelInfo;
 
 	/*
 	 * Populate updatedCols so that per-column triggers can fire.  This could
@@ -1392,7 +1390,8 @@ apply_handle_update_internal(ResultRelInfo *relinfo,
 		EvalPlanQualSetSlot(&epqstate, remoteslot);
 
 		/* Do the actual update. */
-		ExecSimpleRelationUpdate(estate, &epqstate, localslot, remoteslot);
+		ExecSimpleRelationUpdate(relinfo, estate, &epqstate, localslot,
+								 remoteslot);
 	}
 	else
 	{
@@ -1455,7 +1454,6 @@ apply_handle_delete(StringInfo s)
 										&TTSOpsVirtual);
 	resultRelInfo = makeNode(ResultRelInfo);
 	InitResultRelInfo(resultRelInfo, rel->localrel, 1, NULL, 0);
-	estate->es_result_relation_info = resultRelInfo;
 
 	PushActiveSnapshot(GetTransactionSnapshot());
 
@@ -1508,7 +1506,7 @@ apply_handle_delete_internal(ResultRelInfo *relinfo, EState *estate,
 		EvalPlanQualSetSlot(&epqstate, localslot);
 
 		/* Do the actual delete. */
-		ExecSimpleRelationDelete(estate, &epqstate, localslot);
+		ExecSimpleRelationDelete(relinfo, estate, &epqstate, localslot);
 	}
 	else
 	{
@@ -1616,7 +1614,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 	}
 	MemoryContextSwitchTo(oldctx);
 
-	estate->es_result_relation_info = partrelinfo;
 	switch (operation)
 	{
 		case CMD_INSERT:
@@ -1697,8 +1694,8 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					ExecOpenIndices(partrelinfo, false);
 
 					EvalPlanQualSetSlot(&epqstate, remoteslot_part);
-					ExecSimpleRelationUpdate(estate, &epqstate, localslot,
-											 remoteslot_part);
+					ExecSimpleRelationUpdate(partrelinfo, estate, &epqstate,
+											 localslot, remoteslot_part);
 					ExecCloseIndices(partrelinfo);
 					EvalPlanQualEnd(&epqstate);
 				}
@@ -1739,7 +1736,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					Assert(partrelinfo_new != partrelinfo);
 
 					/* DELETE old tuple found in the old partition. */
-					estate->es_result_relation_info = partrelinfo;
 					apply_handle_delete_internal(partrelinfo, estate,
 												 localslot,
 												 &relmapentry->remoterel);
@@ -1771,7 +1767,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 						slot_getallattrs(remoteslot);
 					}
 					MemoryContextSwitchTo(oldctx);
-					estate->es_result_relation_info = partrelinfo_new;
 					apply_handle_insert_internal(partrelinfo_new, estate,
 												 remoteslot_part);
 				}
diff --git a/src/include/executor/executor.h b/src/include/executor/executor.h
index c283bf1454..2f221d7114 100644
--- a/src/include/executor/executor.h
+++ b/src/include/executor/executor.h
@@ -576,10 +576,14 @@ extern TupleTableSlot *ExecGetReturningSlot(EState *estate, ResultRelInfo *relIn
  */
 extern void ExecOpenIndices(ResultRelInfo *resultRelInfo, bool speculative);
 extern void ExecCloseIndices(ResultRelInfo *resultRelInfo);
-extern List *ExecInsertIndexTuples(TupleTableSlot *slot, EState *estate, bool noDupErr,
+extern List *ExecInsertIndexTuples(ResultRelInfo *resultRelInfo,
+								   TupleTableSlot *slot, EState *estate,
+								   bool noDupErr,
 								   bool *specConflict, List *arbiterIndexes);
-extern bool ExecCheckIndexConstraints(TupleTableSlot *slot, EState *estate,
-									  ItemPointer conflictTid, List *arbiterIndexes);
+extern bool ExecCheckIndexConstraints(ResultRelInfo *resultRelInfo,
+						  TupleTableSlot *slot,
+						  EState *estate, ItemPointer conflictTid,
+						  List *arbiterIndexes);
 extern void check_exclusion_constraint(Relation heap, Relation index,
 									   IndexInfo *indexInfo,
 									   ItemPointer tupleid,
@@ -596,10 +600,13 @@ extern bool RelationFindReplTupleByIndex(Relation rel, Oid idxoid,
 extern bool RelationFindReplTupleSeq(Relation rel, LockTupleMode lockmode,
 									 TupleTableSlot *searchslot, TupleTableSlot *outslot);
 
-extern void ExecSimpleRelationInsert(EState *estate, TupleTableSlot *slot);
-extern void ExecSimpleRelationUpdate(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationInsert(ResultRelInfo *resultRelInfo,
+									 EState *estate, TupleTableSlot *slot);
+extern void ExecSimpleRelationUpdate(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot, TupleTableSlot *slot);
-extern void ExecSimpleRelationDelete(EState *estate, EPQState *epqstate,
+extern void ExecSimpleRelationDelete(ResultRelInfo *resultRelInfo,
+									 EState *estate, EPQState *epqstate,
 									 TupleTableSlot *searchslot);
 extern void CheckCmdReplicaIdentity(Relation rel, CmdType cmd);
 
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 4ec4ebdabc..2518fe4f64 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -15,7 +15,9 @@
 
 #include "nodes/execnodes.h"
 
-extern void ExecComputeStoredGenerated(EState *estate, TupleTableSlot *slot, CmdType cmdtype);
+extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
+						   EState *estate, TupleTableSlot *slot,
+						   CmdType cmdtype);
 
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index da1f3f269a..c1e41c77c7 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -525,7 +525,6 @@ typedef struct EState
 	List	   *es_opened_result_relations; /* List of non-NULL entries in
 											 * es_result_relations in no
 											 * specific order */
-	ResultRelInfo *es_result_relation_info; /* currently active array elt */
 
 	PartitionDirectory es_partition_directory;	/* for PartitionDesc lookup */
 
diff --git a/src/test/regress/expected/insert.out b/src/test/regress/expected/insert.out
index eb9d45be5e..da50ee3b67 100644
--- a/src/test/regress/expected/insert.out
+++ b/src/test/regress/expected/insert.out
@@ -818,9 +818,7 @@ drop role regress_coldesc_role;
 drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
diff --git a/src/test/regress/sql/insert.sql b/src/test/regress/sql/insert.sql
index ffd4aacbc4..963faa1614 100644
--- a/src/test/regress/sql/insert.sql
+++ b/src/test/regress/sql/insert.sql
@@ -542,9 +542,7 @@ drop table inserttest3;
 drop table brtrigpartcon;
 drop function brtrigpartcon1trigf();
 
--- check that "do nothing" BR triggers work with tuple-routing (this checks
--- that estate->es_result_relation_info is appropriately set/reset for each
--- routed tuple)
+-- check that "do nothing" BR triggers work with tuple-routing
 create table donothingbrtrig_test (a int, b text) partition by list (a);
 create table donothingbrtrig_test1 (b text, a int);
 create table donothingbrtrig_test2 (c text, b text, a int);
-- 
2.20.1

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 78facb8..e2b4b56 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -2357,7 +2357,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = node->resultRelInfo->ri_RangeTableIndex;
+	rtindex = fsplan->resultRelation;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 0b20f94..ce1ff17 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -49,7 +49,18 @@ ForeignNext(ForeignScanState *node)
 	/* Call the Iterate function in short-lived context */
 	oldcontext = MemoryContextSwitchTo(econtext->ecxt_per_tuple_memory);
 	if (plan->operation != CMD_SELECT)
+	{
+		EState   *estate = node->ss.ps.state;
+
+		/*
+		 * For the FDW's convenience, look up the modification target
+		 * relation's ResultRelInfo.
+		 */
+		Assert(plan->resultRelation > 0);
+		if (node->resultRelInfo == NULL)
+			node->resultRelInfo = estate->es_result_relations[plan->resultRelation - 1];
 		slot = node->fdwroutine->IterateDirectModify(node);
+	}
 	else
 		slot = node->fdwroutine->IterateForeignScan(node);
 	MemoryContextSwitchTo(oldcontext);
@@ -215,13 +226,6 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	scanstate->fdwroutine = fdwroutine;
 	scanstate->fdw_state = NULL;
 
-	/*
-	 * For the FDW's convenience, look up the modification target relation's.
-	 * ResultRelInfo.
-	 */
-	if (node->resultRelation > 0)
-		scanstate->resultRelInfo = estate->es_result_relations[node->resultRelation - 1];
-
 	/* Initialize any outer plan. */
 	if (outerPlan(node))
 		outerPlanState(scanstate) =

From 15b84595a7fd9e081bba8443cbc6116544440a2e Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v18] Revise child-to-root tuple conversion map management

Transition tuple capture requires to convert child tuples to the
inheritance root table format because that's the format the
transition tuplestore stores tuple in.  For INSERTs into partitioned
tables, the conversion is handled by tuple routing code which
constructs the map for a given partition only if the partition is
targeted, but for UPDATE and DELETE, maps for all result relations
are made and stored in an array in ModifyTableState during
ExecInitModifyTable, which requires their ResultRelInfos to have been
already built. During execution, map for the currently active result
relation is set in TransitionCaptureState.tcs_map.

This commit removes TransitionCaptureMap.tcs_map in favor a new
map field in ResultRelInfo named ri_ChildToRootMap that is
initialized when the ResultRelInfo for a given result relation is.
This way is less confusing and less bug-prone than setting and
resetting tcs_map. Also, this will also allow us to delay creating
the map for a given result relation to when that relation is actually
processed during execution.
---
 src/backend/commands/copy.c            |  30 +----
 src/backend/commands/trigger.c         |   7 +-
 src/backend/executor/execPartition.c   |  23 ++--
 src/backend/executor/nodeModifyTable.c | 220 +++++++++------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/executor/execPartition.h   |   6 -
 src/include/nodes/execnodes.h          |  11 +-
 7 files changed, 89 insertions(+), 218 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 531bd7c..eb326de 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3105,32 +3105,14 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition which may change the tuple, we can
+			 * just remember the original unconverted tuple to avoid a
+			 * needless round trip conversion.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 3b4fbda..e76f5d4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4293,8 +4294,8 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * tables, then return NULL.
  *
  * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * should set tcs_original_insert_tuple as appropriate when dealing with child
+ * tables
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5389,7 +5390,7 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	if (row_trigger && transition_capture != NULL)
 	{
 		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleConversionMap *map = relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 33d2c6f..08f91e5 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -908,6 +908,15 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 	}
 
 	/*
+	 * Also, if transition capture is required, store a map to convert tuples
+	 * from partition's rowtype to the root partition table's.
+	 */
+	if (mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture)
+		leaf_part_rri->ri_ChildToRootMap =
+			convert_tuples_by_name(RelationGetDescr(leaf_part_rri->ri_RelationDesc),
+								   RelationGetDescr(leaf_part_rri->ri_PartitionRoot));
+
+	/*
 	 * Since we've just initialized this ResultRelInfo, it's not in any list
 	 * attached to the estate as yet.  Add it, so that it can be found later.
 	 *
@@ -977,20 +986,6 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 		partrouteinfo->pi_PartitionTupleSlot = NULL;
 
 	/*
-	 * Also, if transition capture is required, store a map to convert tuples
-	 * from partition's rowtype to the root partition table's.
-	 */
-	if (mtstate &&
-		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
-	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot));
-	}
-	else
-		partrouteinfo->pi_PartitionToRootMap = NULL;
-
-	/*
 	 * If the partition is a foreign table, let the FDW init itself for
 	 * routing tuples to the partition.
 	 */
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 0c055ed..01ce3ec 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -1087,9 +1084,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
 	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1164,38 +1159,26 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 
 	/*
 	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we should
-	 * convert the tuple into root's tuple descriptor, since ExecInsert()
-	 * starts the search from root.  The tuple conversion map list is in the
-	 * order of mtstate->resultRelInfo[], so to retrieve the one for this
-	 * resultRel, we need to know the position of the resultRel in
-	 * mtstate->resultRelInfo[].
+	 * convert the tuple into root's tuple descriptor if needed, since
+	 * ExecInsert() starts the search from root.
 	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture, so save
-	 * the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
+	/*
+	 * Reset the transition state that may possibly have been written
+	 * by INSERT.
+	 */
 	if (mtstate->mt_transition_capture)
-	{
 		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
 
 	/* We're done moving. */
 	return true;
@@ -1902,28 +1885,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1947,6 +1908,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
@@ -1960,37 +1922,15 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE triggers
+	 * on the partition which may change the tuple, we can just remember the
+	 * original unconverted tuple to avoid a needless round trip conversion.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -2007,58 +1947,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc);
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2154,17 +2042,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2334,6 +2211,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRelInfo;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2355,13 +2233,24 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		palloc(nplans * sizeof(ResultRelInfo));
 	mtstate->mt_scans = (TupleTableSlot **) palloc0(sizeof(TupleTableSlot *) * nplans);
 
-	/* If modifying a partitioned table, initialize the root table info */
+	/*
+	 * Initialize the designated "root" result relation.  When modifying
+	 * partitioned tables, it's given by node->rootRelation, while in other
+	 * cases, it's the first relation in node->resultRelations.  We need to
+	 * initialize this one before any others, because
+	 * ExecSetupTransitionCaptureState() needs it.
+	 */
 	if (node->rootRelation > 0)
 	{
 		mtstate->rootResultRelInfo = makeNode(ResultRelInfo);
 		ExecInitResultRelation(estate, mtstate->rootResultRelInfo,
 							   node->rootRelation);
 	}
+	else
+		ExecInitResultRelation(estate, mtstate->resultRelInfo,
+							   linitial_int(node->resultRelations));
+
+	rootResultRelInfo = getTargetResultRelInfo(mtstate);
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2371,6 +2260,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2384,8 +2280,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 		subplan = (Plan *) lfirst(l1);
 
-		/* This opens the relation and fills ResultRelInfo. */
-		ExecInitResultRelation(estate, resultRelInfo, resultRelation);
+		/*
+		 * This opens result relation and fills ResultRelInfo.
+		 * ("root" relation already opened.)
+		 */
+		if (resultRelInfo != rootResultRelInfo)
+			ExecInitResultRelation(estate, resultRelInfo, resultRelation);
 
 		/* Initialize the usesFdwDirectModify flag */
 		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(i,
@@ -2441,12 +2341,28 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.  During INSERT, partition tuples to
+		 * store into the transition tuple store are converted using
+		 * PartitionToRoot map in the partition's PartitionRoutingInfo.
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRelInfo->ri_RelationDesc));
 		resultRelInfo++;
 		i++;
 	}
 
 	/* Get the target relation */
-	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
+	rel = rootResultRelInfo->ri_RelationDesc;
 
 	/*
 	 * If it's not a partitioned table after all, UPDATE tuple routing should
@@ -2465,26 +2381,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a40ddf5..e38d732 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -46,7 +46,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -66,14 +66,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/executor/execPartition.h b/src/include/executor/execPartition.h
index 6d1b722..74c3991 100644
--- a/src/include/executor/execPartition.h
+++ b/src/include/executor/execPartition.h
@@ -37,12 +37,6 @@ typedef struct PartitionRoutingInfo
 	TupleConversionMap *pi_RootToPartitionMap;
 
 	/*
-	 * Map for converting tuples in partition format into the root partitioned
-	 * table format, or NULL if no conversion is required.
-	 */
-	TupleConversionMap *pi_PartitionToRootMap;
-
-	/*
 	 * Slot to store tuples in partition format, or NULL when no translation
 	 * is required between root and partition.
 	 */
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index b7e9e5d..ce83b81 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -488,6 +488,14 @@ typedef struct ResultRelInfo
 
 	/* for use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child result relation tuples to the format of the
+	 * table actually mentioned in the query (called "root").  Set only
+	 * if either transition tuple capture or update partition row
+	 * movement is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1174,9 +1182,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
1.8.3.1

From 14de4960934c101e860df1249eeb3c8996617ee5 Mon Sep 17 00:00:00 2001
From: amit <amitlangote09@gmail.com>
Date: Tue, 30 Jul 2019 10:51:35 +0900
Subject: [PATCH v19 1/2] Revise child-to-root tuple conversion map management

Transition tuple capture requires to convert child tuples to the
inheritance root table format because that's the format the
transition tuplestore stores tuple in.  For INSERTs into partitioned
tables, the conversion is handled by tuple routing code which
constructs the map for a given partition only if the partition is
targeted, but for UPDATE and DELETE, maps for all result relations
are made and stored in an array in ModifyTableState during
ExecInitModifyTable, which requires their ResultRelInfos to have been
already built. During execution, map for the currently active result
relation is set in TransitionCaptureState.tcs_map.

This commit removes TransitionCaptureMap.tcs_map in favor a new
map field in ResultRelInfo named ri_ChildToRootMap that is
initialized when the ResultRelInfo for a given result relation is.
This way is less confusing and less bug-prone than setting and
resetting tcs_map. Also, this will also allow us to delay creating
the map for a given result relation to when that relation is actually
processed during execution.
---
 src/backend/commands/copy.c            |  30 +----
 src/backend/commands/trigger.c         |   7 +-
 src/backend/executor/execPartition.c   |  23 ++--
 src/backend/executor/nodeModifyTable.c | 221 +++++++++------------------------
 src/include/commands/trigger.h         |  10 +-
 src/include/executor/execPartition.h   |   6 -
 src/include/nodes/execnodes.h          |  11 +-
 7 files changed, 90 insertions(+), 218 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index 531bd7c..eb326de 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3105,32 +3105,14 @@ CopyFrom(CopyState cstate)
 			}
 
 			/*
-			 * If we're capturing transition tuples, we might need to convert
-			 * from the partition rowtype to root rowtype.
+			 * If we're capturing transition tuples and there are no BEFORE
+			 * triggers on the partition which may change the tuple, we can
+			 * just remember the original unconverted tuple to avoid a
+			 * needless round trip conversion.
 			 */
 			if (cstate->transition_capture != NULL)
-			{
-				if (has_before_insert_row_trig)
-				{
-					/*
-					 * If there are any BEFORE triggers on the partition,
-					 * we'll have to be ready to convert their result back to
-					 * tuplestore format.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = NULL;
-					cstate->transition_capture->tcs_map =
-						resultRelInfo->ri_PartitionInfo->pi_PartitionToRootMap;
-				}
-				else
-				{
-					/*
-					 * Otherwise, just remember the original unconverted
-					 * tuple, to avoid a needless round trip conversion.
-					 */
-					cstate->transition_capture->tcs_original_insert_tuple = myslot;
-					cstate->transition_capture->tcs_map = NULL;
-				}
-			}
+				cstate->transition_capture->tcs_original_insert_tuple =
+					!has_before_insert_row_trig ? myslot : NULL;
 
 			/*
 			 * We might need to convert from the root rowtype to the partition
diff --git a/src/backend/commands/trigger.c b/src/backend/commands/trigger.c
index 3b4fbda..e76f5d4 100644
--- a/src/backend/commands/trigger.c
+++ b/src/backend/commands/trigger.c
@@ -35,6 +35,7 @@
 #include "commands/defrem.h"
 #include "commands/trigger.h"
 #include "executor/executor.h"
+#include "executor/execPartition.h"
 #include "miscadmin.h"
 #include "nodes/bitmapset.h"
 #include "nodes/makefuncs.h"
@@ -4293,8 +4294,8 @@ GetAfterTriggersTableData(Oid relid, CmdType cmdType)
  * tables, then return NULL.
  *
  * The resulting object can be passed to the ExecAR* functions.  The caller
- * should set tcs_map or tcs_original_insert_tuple as appropriate when dealing
- * with child tables.
+ * should set tcs_original_insert_tuple as appropriate when dealing with child
+ * tables
  *
  * Note that we copy the flags from a parent table into this struct (rather
  * than subsequently using the relation's TriggerDesc directly) so that we can
@@ -5389,7 +5390,7 @@ AfterTriggerSaveEvent(EState *estate, ResultRelInfo *relinfo,
 	if (row_trigger && transition_capture != NULL)
 	{
 		TupleTableSlot *original_insert_tuple = transition_capture->tcs_original_insert_tuple;
-		TupleConversionMap *map = transition_capture->tcs_map;
+		TupleConversionMap *map = relinfo->ri_ChildToRootMap;
 		bool		delete_old_table = transition_capture->tcs_delete_old_table;
 		bool		update_old_table = transition_capture->tcs_update_old_table;
 		bool		update_new_table = transition_capture->tcs_update_new_table;
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 33d2c6f..08f91e5 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -908,6 +908,15 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 	}
 
 	/*
+	 * Also, if transition capture is required, store a map to convert tuples
+	 * from partition's rowtype to the root partition table's.
+	 */
+	if (mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture)
+		leaf_part_rri->ri_ChildToRootMap =
+			convert_tuples_by_name(RelationGetDescr(leaf_part_rri->ri_RelationDesc),
+								   RelationGetDescr(leaf_part_rri->ri_PartitionRoot));
+
+	/*
 	 * Since we've just initialized this ResultRelInfo, it's not in any list
 	 * attached to the estate as yet.  Add it, so that it can be found later.
 	 *
@@ -977,20 +986,6 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 		partrouteinfo->pi_PartitionTupleSlot = NULL;
 
 	/*
-	 * Also, if transition capture is required, store a map to convert tuples
-	 * from partition's rowtype to the root partition table's.
-	 */
-	if (mtstate &&
-		(mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture))
-	{
-		partrouteinfo->pi_PartitionToRootMap =
-			convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_RelationDesc),
-								   RelationGetDescr(partRelInfo->ri_PartitionRoot));
-	}
-	else
-		partrouteinfo->pi_PartitionToRootMap = NULL;
-
-	/*
 	 * If the partition is a foreign table, let the FDW init itself for
 	 * routing tuples to the partition.
 	 */
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 0c055ed..5efbfb6 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -73,9 +73,6 @@ static TupleTableSlot *ExecPrepareTupleRouting(ModifyTableState *mtstate,
 											   TupleTableSlot *slot,
 											   ResultRelInfo **partRelInfo);
 static ResultRelInfo *getTargetResultRelInfo(ModifyTableState *node);
-static void ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate);
-static TupleConversionMap *tupconv_map_for_subplan(ModifyTableState *node,
-												   int whichplan);
 
 /*
  * Verify that the tuples to be produced by INSERT or UPDATE match the
@@ -1087,9 +1084,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 {
 	EState	   *estate = mtstate->ps.state;
 	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
-	int			map_index;
 	TupleConversionMap *tupconv_map;
-	TupleConversionMap *saved_tcs_map = NULL;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
 
@@ -1164,38 +1159,26 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 
 	/*
 	 * resultRelInfo is one of the per-subplan resultRelInfos.  So we should
-	 * convert the tuple into root's tuple descriptor, since ExecInsert()
-	 * starts the search from root.  The tuple conversion map list is in the
-	 * order of mtstate->resultRelInfo[], so to retrieve the one for this
-	 * resultRel, we need to know the position of the resultRel in
-	 * mtstate->resultRelInfo[].
+	 * convert the tuple into root's tuple descriptor if needed, since
+	 * ExecInsert() starts the search from root.
 	 */
-	map_index = resultRelInfo - mtstate->resultRelInfo;
-	Assert(map_index >= 0 && map_index < mtstate->mt_nplans);
-	tupconv_map = tupconv_map_for_subplan(mtstate, map_index);
+	tupconv_map = resultRelInfo->ri_ChildToRootMap;
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
 									 mtstate->mt_root_tuple_slot);
 
-	/*
-	 * ExecInsert() may scribble on mtstate->mt_transition_capture, so save
-	 * the currently active map.
-	 */
-	if (mtstate->mt_transition_capture)
-		saved_tcs_map = mtstate->mt_transition_capture->tcs_map;
-
 	/* Tuple routing starts from the root table. */
 	Assert(mtstate->rootResultRelInfo != NULL);
 	*inserted_tuple = ExecInsert(mtstate, mtstate->rootResultRelInfo, slot,
 								 planSlot, estate, canSetTag);
 
-	/* Clear the INSERT's tuple and restore the saved map. */
+	/*
+	 * Reset the transition state that may possibly have been written
+	 * by INSERT.
+	 */
 	if (mtstate->mt_transition_capture)
-	{
 		mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-		mtstate->mt_transition_capture->tcs_map = saved_tcs_map;
-	}
 
 	/* We're done moving. */
 	return true;
@@ -1902,28 +1885,6 @@ ExecSetupTransitionCaptureState(ModifyTableState *mtstate, EState *estate)
 			MakeTransitionCaptureState(targetRelInfo->ri_TrigDesc,
 									   RelationGetRelid(targetRelInfo->ri_RelationDesc),
 									   CMD_UPDATE);
-
-	/*
-	 * If we found that we need to collect transition tuples then we may also
-	 * need tuple conversion maps for any children that have TupleDescs that
-	 * aren't compatible with the tuplestores.  (We can share these maps
-	 * between the regular and ON CONFLICT cases.)
-	 */
-	if (mtstate->mt_transition_capture != NULL ||
-		mtstate->mt_oc_transition_capture != NULL)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-		/*
-		 * Install the conversion map for the first plan for UPDATE and DELETE
-		 * operations.  It will be advanced each time we switch to the next
-		 * plan.  (INSERT operations set it every time, so we need not update
-		 * mtstate->mt_oc_transition_capture here.)
-		 */
-		if (mtstate->mt_transition_capture && mtstate->operation != CMD_INSERT)
-			mtstate->mt_transition_capture->tcs_map =
-				tupconv_map_for_subplan(mtstate, 0);
-	}
 }
 
 /*
@@ -1947,6 +1908,7 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	ResultRelInfo *partrel;
 	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
+	bool		has_before_insert_row_trig;
 
 	/*
 	 * Lookup the target partition's ResultRelInfo.  If ExecFindPartition does
@@ -1960,37 +1922,15 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	Assert(partrouteinfo != NULL);
 
 	/*
-	 * If we're capturing transition tuples, we might need to convert from the
-	 * partition rowtype to root partitioned table's rowtype.
+	 * If we're capturing transition tuples and there are no BEFORE triggers
+	 * on the partition which may change the tuple, we can just remember the
+	 * original unconverted tuple to avoid a needless round trip conversion.
 	 */
+	has_before_insert_row_trig = (partrel->ri_TrigDesc &&
+								  partrel->ri_TrigDesc->trig_insert_before_row);
 	if (mtstate->mt_transition_capture != NULL)
-	{
-		if (partrel->ri_TrigDesc &&
-			partrel->ri_TrigDesc->trig_insert_before_row)
-		{
-			/*
-			 * If there are any BEFORE triggers on the partition, we'll have
-			 * to be ready to convert their result back to tuplestore format.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = NULL;
-			mtstate->mt_transition_capture->tcs_map =
-				partrouteinfo->pi_PartitionToRootMap;
-		}
-		else
-		{
-			/*
-			 * Otherwise, just remember the original unconverted tuple, to
-			 * avoid a needless round trip conversion.
-			 */
-			mtstate->mt_transition_capture->tcs_original_insert_tuple = slot;
-			mtstate->mt_transition_capture->tcs_map = NULL;
-		}
-	}
-	if (mtstate->mt_oc_transition_capture != NULL)
-	{
-		mtstate->mt_oc_transition_capture->tcs_map =
-			partrouteinfo->pi_PartitionToRootMap;
-	}
+		mtstate->mt_transition_capture->tcs_original_insert_tuple =
+			!has_before_insert_row_trig ? slot : NULL;
 
 	/*
 	 * Convert the tuple, if necessary.
@@ -2007,58 +1947,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	return slot;
 }
 
-/*
- * Initialize the child-to-root tuple conversion map array for UPDATE subplans.
- *
- * This map array is required to convert the tuple from the subplan result rel
- * to the target table descriptor. This requirement arises for two independent
- * scenarios:
- * 1. For update-tuple-routing.
- * 2. For capturing tuples in transition tables.
- */
-static void
-ExecSetupChildParentMapForSubplan(ModifyTableState *mtstate)
-{
-	ResultRelInfo *targetRelInfo = getTargetResultRelInfo(mtstate);
-	ResultRelInfo *resultRelInfos = mtstate->resultRelInfo;
-	TupleDesc	outdesc;
-	int			numResultRelInfos = mtstate->mt_nplans;
-	int			i;
-
-	/*
-	 * Build array of conversion maps from each child's TupleDesc to the one
-	 * used in the target relation.  The map pointers may be NULL when no
-	 * conversion is necessary, which is hopefully a common case.
-	 */
-
-	/* Get tuple descriptor of the target rel. */
-	outdesc = RelationGetDescr(targetRelInfo->ri_RelationDesc);
-
-	mtstate->mt_per_subplan_tupconv_maps = (TupleConversionMap **)
-		palloc(sizeof(TupleConversionMap *) * numResultRelInfos);
-
-	for (i = 0; i < numResultRelInfos; ++i)
-	{
-		mtstate->mt_per_subplan_tupconv_maps[i] =
-			convert_tuples_by_name(RelationGetDescr(resultRelInfos[i].ri_RelationDesc),
-								   outdesc);
-	}
-}
-
-/*
- * For a given subplan index, get the tuple conversion map.
- */
-static TupleConversionMap *
-tupconv_map_for_subplan(ModifyTableState *mtstate, int whichplan)
-{
-	/* If nobody else set the per-subplan array of maps, do so ourselves. */
-	if (mtstate->mt_per_subplan_tupconv_maps == NULL)
-		ExecSetupChildParentMapForSubplan(mtstate);
-
-	Assert(whichplan >= 0 && whichplan < mtstate->mt_nplans);
-	return mtstate->mt_per_subplan_tupconv_maps[whichplan];
-}
-
 /* ----------------------------------------------------------------
  *	   ExecModifyTable
  *
@@ -2154,17 +2042,6 @@ ExecModifyTable(PlanState *pstate)
 				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
-				/* Prepare to convert transition tuples from this child. */
-				if (node->mt_transition_capture != NULL)
-				{
-					node->mt_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
-				if (node->mt_oc_transition_capture != NULL)
-				{
-					node->mt_oc_transition_capture->tcs_map =
-						tupconv_map_for_subplan(node, node->mt_whichplan);
-				}
 				continue;
 			}
 			else
@@ -2334,6 +2211,7 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	int			i;
 	Relation	rel;
 	bool		update_tuple_routing_needed = node->partColsUpdated;
+	ResultRelInfo *rootResultRelInfo;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2355,13 +2233,24 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		palloc(nplans * sizeof(ResultRelInfo));
 	mtstate->mt_scans = (TupleTableSlot **) palloc0(sizeof(TupleTableSlot *) * nplans);
 
-	/* If modifying a partitioned table, initialize the root table info */
+	/*
+	 * Initialize the designated "root" result relation.  When modifying
+	 * partitioned tables, it's given by node->rootRelation, while in other
+	 * cases, it's the first relation in node->resultRelations.  We need to
+	 * initialize this one before any others, because
+	 * ExecSetupTransitionCaptureState() needs it.
+	 */
 	if (node->rootRelation > 0)
 	{
 		mtstate->rootResultRelInfo = makeNode(ResultRelInfo);
 		ExecInitResultRelation(estate, mtstate->rootResultRelInfo,
 							   node->rootRelation);
 	}
+	else
+		ExecInitResultRelation(estate, mtstate->resultRelInfo,
+							   linitial_int(node->resultRelations));
+
+	rootResultRelInfo = getTargetResultRelInfo(mtstate);
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
 	mtstate->mt_nplans = nplans;
@@ -2371,6 +2260,13 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->fireBSTriggers = true;
 
 	/*
+	 * Build state for collecting transition tuples.  This requires having a
+	 * valid trigger query context, so skip it in explain-only mode.
+	 */
+	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
+		ExecSetupTransitionCaptureState(mtstate, estate);
+
+	/*
 	 * call ExecInitNode on each of the plans to be executed and save the
 	 * results into the array "mt_plans".  This is also a convenient place to
 	 * verify that the proposed target relations are valid and open their
@@ -2384,8 +2280,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 		subplan = (Plan *) lfirst(l1);
 
-		/* This opens the relation and fills ResultRelInfo. */
-		ExecInitResultRelation(estate, resultRelInfo, resultRelation);
+		/*
+		 * This opens result relation and fills ResultRelInfo.
+		 * ("root" relation already opened.)
+		 */
+		if (resultRelInfo != rootResultRelInfo)
+			ExecInitResultRelation(estate, resultRelInfo, resultRelation);
 
 		/* Initialize the usesFdwDirectModify flag */
 		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(i,
@@ -2441,12 +2341,29 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 															 eflags);
 		}
 
+		/*
+		 * If needed, initialize a map to convert tuples in the child format
+		 * to the format of the table mentioned in the query (root relation).
+		 * It's needed for update tuple routing, because the routing starts
+		 * from the root relation.  It's also needed for capturing transition
+		 * tuples, because the transition tuple store can only store tuples
+		 * in the root table format.
+		 *
+		 * For INSERT, the map is only initialized for a given partition when
+		 * the partition itself is first initialized by ExecFindPartition().
+		 */
+		if (update_tuple_routing_needed ||
+			(mtstate->mt_transition_capture &&
+			 mtstate->operation != CMD_INSERT))
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
+									   RelationGetDescr(rootResultRelInfo->ri_RelationDesc));
 		resultRelInfo++;
 		i++;
 	}
 
 	/* Get the target relation */
-	rel = (getTargetResultRelInfo(mtstate))->ri_RelationDesc;
+	rel = rootResultRelInfo->ri_RelationDesc;
 
 	/*
 	 * If it's not a partitioned table after all, UPDATE tuple routing should
@@ -2465,26 +2382,12 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
-	 * Build state for collecting transition tuples.  This requires having a
-	 * valid trigger query context, so skip it in explain-only mode.
-	 */
-	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
-		ExecSetupTransitionCaptureState(mtstate, estate);
-
-	/*
-	 * Construct mapping from each of the per-subplan partition attnos to the
-	 * root attno.  This is required when during update row movement the tuple
-	 * descriptor of a source partition does not match the root partitioned
-	 * table descriptor.  In such a case we need to convert tuples to the root
-	 * tuple descriptor, because the search for destination partition starts
-	 * from the root.  We'll also need a slot to store these converted tuples.
-	 * We can skip this setup if it's not a partition key update.
+	 * For update row movement we'll need a dedicated slot to store the
+	 * tuples that have been converted from partition format to the root
+	 * table format.
 	 */
 	if (update_tuple_routing_needed)
-	{
-		ExecSetupChildParentMapForSubplan(mtstate);
 		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-	}
 
 	/*
 	 * Initialize any WITH CHECK OPTION constraints if needed.
diff --git a/src/include/commands/trigger.h b/src/include/commands/trigger.h
index a40ddf5..e38d732 100644
--- a/src/include/commands/trigger.h
+++ b/src/include/commands/trigger.h
@@ -46,7 +46,7 @@ typedef struct TriggerData
  * The state for capturing old and new tuples into transition tables for a
  * single ModifyTable node (or other operation source, e.g. copy.c).
  *
- * This is per-caller to avoid conflicts in setting tcs_map or
+ * This is per-caller to avoid conflicts in setting
  * tcs_original_insert_tuple.  Note, however, that the pointed-to
  * private data may be shared across multiple callers.
  */
@@ -66,14 +66,6 @@ typedef struct TransitionCaptureState
 	bool		tcs_insert_new_table;
 
 	/*
-	 * For UPDATE and DELETE, AfterTriggerSaveEvent may need to convert the
-	 * new and old tuples from a child table's format to the format of the
-	 * relation named in a query so that it is compatible with the transition
-	 * tuplestores.  The caller must store the conversion map here if so.
-	 */
-	TupleConversionMap *tcs_map;
-
-	/*
 	 * For INSERT and COPY, it would be wasteful to convert tuples from child
 	 * format to parent format after they have already been converted in the
 	 * opposite direction during routing.  In that case we bypass conversion
diff --git a/src/include/executor/execPartition.h b/src/include/executor/execPartition.h
index 6d1b722..74c3991 100644
--- a/src/include/executor/execPartition.h
+++ b/src/include/executor/execPartition.h
@@ -37,12 +37,6 @@ typedef struct PartitionRoutingInfo
 	TupleConversionMap *pi_RootToPartitionMap;
 
 	/*
-	 * Map for converting tuples in partition format into the root partitioned
-	 * table format, or NULL if no conversion is required.
-	 */
-	TupleConversionMap *pi_PartitionToRootMap;
-
-	/*
 	 * Slot to store tuples in partition format, or NULL when no translation
 	 * is required between root and partition.
 	 */
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index b7e9e5d..ce83b81 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -488,6 +488,14 @@ typedef struct ResultRelInfo
 
 	/* for use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
+
+	/*
+	 * Map to convert child result relation tuples to the format of the
+	 * table actually mentioned in the query (called "root").  Set only
+	 * if either transition tuple capture or update partition row
+	 * movement is active.
+	 */
+	TupleConversionMap *ri_ChildToRootMap;
 } ResultRelInfo;
 
 /* ----------------
@@ -1174,9 +1182,6 @@ typedef struct ModifyTableState
 
 	/* controls transition table population for INSERT...ON CONFLICT UPDATE */
 	struct TransitionCaptureState *mt_oc_transition_capture;
-
-	/* Per plan map for tuple conversion from child to root */
-	TupleConversionMap **mt_per_subplan_tupconv_maps;
 } ModifyTableState;
 
 /* ----------------
-- 
1.8.3.1

From 7d321c1627b9ce16357a35704fee030b3ecbdedd Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Mon, 19 Oct 2020 12:32:09 +0900
Subject: [PATCH v19 2/2] Remove PartitionRoutingInfo struct

The extra indirection neeeded to access its members via its enclosing
ResultRelInfo seems pointless.
---
 src/backend/commands/copy.c              |  4 ++--
 src/backend/executor/execMain.c          |  3 ++-
 src/backend/executor/execPartition.c     | 32 ++++++++++++++++----------------
 src/backend/executor/nodeModifyTable.c   |  7 ++-----
 src/backend/replication/logical/worker.c | 11 ++++-------
 src/include/executor/execPartition.h     | 21 ---------------------
 src/include/nodes/execnodes.h            | 15 ++++++++++-----
 7 files changed, 36 insertions(+), 57 deletions(-)

diff --git a/src/backend/commands/copy.c b/src/backend/commands/copy.c
index eb326de..0c3672e 100644
--- a/src/backend/commands/copy.c
+++ b/src/backend/commands/copy.c
@@ -3118,7 +3118,7 @@ CopyFrom(CopyState cstate)
 			 * We might need to convert from the root rowtype to the partition
 			 * rowtype.
 			 */
-			map = resultRelInfo->ri_PartitionInfo->pi_RootToPartitionMap;
+			map = resultRelInfo->ri_RootToPartitionMap;
 			if (insertMethod == CIM_SINGLE || !leafpart_use_multi_insert)
 			{
 				/* non batch insert */
@@ -3126,7 +3126,7 @@ CopyFrom(CopyState cstate)
 				{
 					TupleTableSlot *new_slot;
 
-					new_slot = resultRelInfo->ri_PartitionInfo->pi_PartitionTupleSlot;
+					new_slot = resultRelInfo->ri_PartitionTupleSlot;
 					myslot = execute_attr_map_slot(map->attrMap, myslot, new_slot);
 				}
 			}
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 293f53d..eef3516 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -1243,7 +1243,8 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
 	resultRelInfo->ri_TrigOldSlot = NULL;
 	resultRelInfo->ri_TrigNewSlot = NULL;
 	resultRelInfo->ri_PartitionRoot = partition_root;
-	resultRelInfo->ri_PartitionInfo = NULL; /* may be set later */
+	resultRelInfo->ri_RootToPartitionMap = NULL; /* set by ExecInitPartitionInfo */
+	resultRelInfo->ri_PartitionTupleSlot = NULL; /* ditto */
 	resultRelInfo->ri_CopyMultiInsertBuffer = NULL;
 }
 
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 08f91e5..e72487f 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -261,7 +261,7 @@ ExecSetupPartitionTupleRouting(EState *estate, ModifyTableState *mtstate,
  * If the partition's ResultRelInfo does not yet exist in 'proute' then we set
  * one up or reuse one from mtstate's resultRelInfo array.  When reusing a
  * ResultRelInfo from the mtstate we verify that the relation is a valid
- * target for INSERTs and then set up a PartitionRoutingInfo for it.
+ * target for INSERTs and initialize tuple routing information.
  *
  * rootResultRelInfo is the relation named in the query.
  *
@@ -307,6 +307,7 @@ ExecFindPartition(ModifyTableState *mtstate,
 	while (dispatch != NULL)
 	{
 		int			partidx = -1;
+		bool		is_leaf = false;
 
 		CHECK_FOR_INTERRUPTS();
 
@@ -348,6 +349,8 @@ ExecFindPartition(ModifyTableState *mtstate,
 
 		if (partdesc->is_leaf[partidx])
 		{
+			is_leaf = true;
+
 			/*
 			 * We've reached the leaf -- hurray, we're done.  Look to see if
 			 * we've already got a ResultRelInfo for this partition.
@@ -382,7 +385,10 @@ ExecFindPartition(ModifyTableState *mtstate,
 						/* Verify this ResultRelInfo allows INSERTs */
 						CheckValidResultRel(rri, CMD_INSERT);
 
-						/* Set up the PartitionRoutingInfo for it */
+						/*
+						 * Initialize information needed to insert this and
+						 * subsequent tuples routed to this partition.
+						 */
 						ExecInitRoutingInfo(mtstate, estate, proute, dispatch,
 											rri, partidx);
 					}
@@ -464,8 +470,6 @@ ExecFindPartition(ModifyTableState *mtstate,
 		 */
 		if (partidx == partdesc->boundinfo->default_index)
 		{
-			PartitionRoutingInfo *partrouteinfo = rri->ri_PartitionInfo;
-
 			/*
 			 * The tuple must match the partition's layout for the constraint
 			 * expression to be evaluated successfully.  If the partition is
@@ -478,13 +482,13 @@ ExecFindPartition(ModifyTableState *mtstate,
 			 * So if we have to convert, do it from the root slot; if not, use
 			 * the root slot as-is.
 			 */
-			if (partrouteinfo)
+			if (is_leaf)
 			{
-				TupleConversionMap *map = partrouteinfo->pi_RootToPartitionMap;
+				TupleConversionMap *map = rri->ri_RootToPartitionMap;
 
 				if (map)
 					slot = execute_attr_map_slot(map->attrMap, rootslot,
-												 partrouteinfo->pi_PartitionTupleSlot);
+												 rri->ri_PartitionTupleSlot);
 				else
 					slot = rootslot;
 			}
@@ -788,7 +792,7 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 		{
 			TupleConversionMap *map;
 
-			map = leaf_part_rri->ri_PartitionInfo->pi_RootToPartitionMap;
+			map = leaf_part_rri->ri_RootToPartitionMap;
 
 			Assert(node->onConflictSet != NIL);
 			Assert(rootResultRelInfo->ri_onConflict != NULL);
@@ -949,18 +953,15 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 					int partidx)
 {
 	MemoryContext oldcxt;
-	PartitionRoutingInfo *partrouteinfo;
 	int			rri_index;
 
 	oldcxt = MemoryContextSwitchTo(proute->memcxt);
 
-	partrouteinfo = palloc(sizeof(PartitionRoutingInfo));
-
 	/*
 	 * Set up a tuple conversion map to convert a tuple routed to the
 	 * partition from the parent's type to the partition's.
 	 */
-	partrouteinfo->pi_RootToPartitionMap =
+	partRelInfo->ri_RootToPartitionMap =
 		convert_tuples_by_name(RelationGetDescr(partRelInfo->ri_PartitionRoot),
 							   RelationGetDescr(partRelInfo->ri_RelationDesc));
 
@@ -970,7 +971,7 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 	 * for various operations that are applied to tuples after routing, such
 	 * as checking constraints.
 	 */
-	if (partrouteinfo->pi_RootToPartitionMap != NULL)
+	if (partRelInfo->ri_RootToPartitionMap != NULL)
 	{
 		Relation	partrel = partRelInfo->ri_RelationDesc;
 
@@ -979,11 +980,11 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 		 * partition's TupleDesc; TupleDesc reference will be released at the
 		 * end of the command.
 		 */
-		partrouteinfo->pi_PartitionTupleSlot =
+		partRelInfo->ri_PartitionTupleSlot =
 			table_slot_create(partrel, &estate->es_tupleTable);
 	}
 	else
-		partrouteinfo->pi_PartitionTupleSlot = NULL;
+		partRelInfo->ri_PartitionTupleSlot = NULL;
 
 	/*
 	 * If the partition is a foreign table, let the FDW init itself for
@@ -993,7 +994,6 @@ ExecInitRoutingInfo(ModifyTableState *mtstate,
 		partRelInfo->ri_FdwRoutine->BeginForeignInsert != NULL)
 		partRelInfo->ri_FdwRoutine->BeginForeignInsert(mtstate, partRelInfo);
 
-	partRelInfo->ri_PartitionInfo = partrouteinfo;
 	partRelInfo->ri_CopyMultiInsertBuffer = NULL;
 
 	/*
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 5efbfb6..4a26a07 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -1906,7 +1906,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 						ResultRelInfo **partRelInfo)
 {
 	ResultRelInfo *partrel;
-	PartitionRoutingInfo *partrouteinfo;
 	TupleConversionMap *map;
 	bool		has_before_insert_row_trig;
 
@@ -1918,8 +1917,6 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	 * UPDATE to another partition becomes a DELETE+INSERT.
 	 */
 	partrel = ExecFindPartition(mtstate, targetRelInfo, proute, slot, estate);
-	partrouteinfo = partrel->ri_PartitionInfo;
-	Assert(partrouteinfo != NULL);
 
 	/*
 	 * If we're capturing transition tuples and there are no BEFORE triggers
@@ -1935,10 +1932,10 @@ ExecPrepareTupleRouting(ModifyTableState *mtstate,
 	/*
 	 * Convert the tuple, if necessary.
 	 */
-	map = partrouteinfo->pi_RootToPartitionMap;
+	map = partrel->ri_RootToPartitionMap;
 	if (map != NULL)
 	{
-		TupleTableSlot *new_slot = partrouteinfo->pi_PartitionTupleSlot;
+		TupleTableSlot *new_slot = partrel->ri_PartitionTupleSlot;
 
 		slot = execute_attr_map_slot(map->attrMap, slot, new_slot);
 	}
diff --git a/src/backend/replication/logical/worker.c b/src/backend/replication/logical/worker.c
index b8e297c..3a5b733 100644
--- a/src/backend/replication/logical/worker.c
+++ b/src/backend/replication/logical/worker.c
@@ -1572,7 +1572,6 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 	ResultRelInfo *partrelinfo;
 	Relation	partrel;
 	TupleTableSlot *remoteslot_part;
-	PartitionRoutingInfo *partinfo;
 	TupleConversionMap *map;
 	MemoryContext oldctx;
 
@@ -1599,11 +1598,10 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 	 * partition's rowtype. Convert if needed or just copy, using a dedicated
 	 * slot to store the tuple in any case.
 	 */
-	partinfo = partrelinfo->ri_PartitionInfo;
-	remoteslot_part = partinfo->pi_PartitionTupleSlot;
+	remoteslot_part = partrelinfo->ri_PartitionTupleSlot;
 	if (remoteslot_part == NULL)
 		remoteslot_part = table_slot_create(partrel, &estate->es_tupleTable);
-	map = partinfo->pi_RootToPartitionMap;
+	map = partrelinfo->ri_RootToPartitionMap;
 	if (map != NULL)
 		remoteslot_part = execute_attr_map_slot(map->attrMap, remoteslot,
 												remoteslot_part);
@@ -1748,12 +1746,11 @@ apply_handle_tuple_routing(ResultRelInfo *relinfo,
 					 */
 					oldctx = MemoryContextSwitchTo(GetPerTupleMemoryContext(estate));
 					partrel = partrelinfo_new->ri_RelationDesc;
-					partinfo = partrelinfo_new->ri_PartitionInfo;
-					remoteslot_part = partinfo->pi_PartitionTupleSlot;
+					remoteslot_part = partrelinfo_new->ri_PartitionTupleSlot;
 					if (remoteslot_part == NULL)
 						remoteslot_part = table_slot_create(partrel,
 															&estate->es_tupleTable);
-					map = partinfo->pi_RootToPartitionMap;
+					map = partrelinfo_new->ri_RootToPartitionMap;
 					if (map != NULL)
 					{
 						remoteslot_part = execute_attr_map_slot(map->attrMap,
diff --git a/src/include/executor/execPartition.h b/src/include/executor/execPartition.h
index 74c3991..473c4cd 100644
--- a/src/include/executor/execPartition.h
+++ b/src/include/executor/execPartition.h
@@ -23,27 +23,6 @@ typedef struct PartitionDispatchData *PartitionDispatch;
 typedef struct PartitionTupleRouting PartitionTupleRouting;
 
 /*
- * PartitionRoutingInfo
- *
- * Additional result relation information specific to routing tuples to a
- * table partition.
- */
-typedef struct PartitionRoutingInfo
-{
-	/*
-	 * Map for converting tuples in root partitioned table format into
-	 * partition format, or NULL if no conversion is required.
-	 */
-	TupleConversionMap *pi_RootToPartitionMap;
-
-	/*
-	 * Slot to store tuples in partition format, or NULL when no translation
-	 * is required between root and partition.
-	 */
-	TupleTableSlot *pi_PartitionTupleSlot;
-} PartitionRoutingInfo;
-
-/*
  * PartitionedRelPruningData - Per-partitioned-table data for run-time pruning
  * of partitions.  For a multilevel partitioned table, we have one of these
  * for the topmost partition plus one for each non-leaf child partition.
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index ce83b81..1d9265c 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -33,7 +33,6 @@
 #include "utils/tuplestore.h"
 
 struct PlanState;				/* forward references in this file */
-struct PartitionRoutingInfo;
 struct ParallelHashJoinState;
 struct ExecRowMark;
 struct ExprState;
@@ -480,11 +479,17 @@ typedef struct ResultRelInfo
 	/* partition check expression state (NULL if not set up yet) */
 	ExprState  *ri_PartitionCheckExpr;
 
-	/* relation descriptor for partitioned table's root, if any */
+	/*
+	 * information needed by tuple routing target relations
+	 *
+	 * PartitionRoot gives the target relation mentioned in the query.
+	 * RootToPartitionMap and PartitionTupleSlot, initialized by
+	 * ExecInitRoutingInfo, are non-NULL if partition has a different tuple
+	 * format than the root table.
+	 */
 	Relation	ri_PartitionRoot;
-
-	/* info for partition tuple routing (NULL if not set up yet) */
-	struct PartitionRoutingInfo *ri_PartitionInfo;
+	TupleConversionMap *ri_RootToPartitionMap;
+	TupleTableSlot *ri_PartitionTupleSlot;
 
 	/* for use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
-- 
1.8.3.1

From b4933ee484d71bd90c96cc0cafabbc2529be4ea3 Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Mon, 19 Oct 2020 17:17:33 +0900
Subject: [PATCH v1 2/3] Initialize ForeignScanState.resultRelInfo in
 ExecForeignScan()

An upcoming patch will make the initialization of the individual
elements of es_result_relations[] to be delayed until the ModifyTable
node, from where the result relations come from, actually begins
executing.  To allow that, initialize ForeignScanState.resultRelInfo
in ExecForeignScan instead of in ExecInitForeignScan(), because
child result relations' ResultRelInfos are now not available during
the latter.
---
 contrib/postgres_fdw/postgres_fdw.c    |  2 +-
 src/backend/executor/nodeForeignscan.c | 19 ++++++++++++-------
 2 files changed, 13 insertions(+), 8 deletions(-)

diff --git a/contrib/postgres_fdw/postgres_fdw.c b/contrib/postgres_fdw/postgres_fdw.c
index 9c5aaac..523ffac 100644
--- a/contrib/postgres_fdw/postgres_fdw.c
+++ b/contrib/postgres_fdw/postgres_fdw.c
@@ -2357,7 +2357,7 @@ postgresBeginDirectModify(ForeignScanState *node, int eflags)
 	 * Identify which user to do the remote access as.  This should match what
 	 * ExecCheckRTEPerms() does.
 	 */
-	rtindex = node->resultRelInfo->ri_RangeTableIndex;
+	rtindex = fsplan->resultRelation;
 	rte = exec_rt_fetch(rtindex, estate);
 	userid = rte->checkAsUser ? rte->checkAsUser : GetUserId();
 
diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 0b20f94..503ef8b 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -111,6 +111,18 @@ static TupleTableSlot *
 ExecForeignScan(PlanState *pstate)
 {
 	ForeignScanState *node = castNode(ForeignScanState, pstate);
+	ForeignScan *plan = (ForeignScan *) node->ss.ps.plan;
+
+	/*
+	 * For the FDW's convenience, look up the modification target relation's
+	 * ResultRelInfo.
+	 */
+	if (plan->resultRelation > 0 && node->resultRelInfo == NULL)
+	{
+		EState   *estate = node->ss.ps.state;
+
+		node->resultRelInfo = estate->es_result_relations[plan->resultRelation - 1];
+	}
 
 	return ExecScan(&node->ss,
 					(ExecScanAccessMtd) ForeignNext,
@@ -215,13 +227,6 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	scanstate->fdwroutine = fdwroutine;
 	scanstate->fdw_state = NULL;
 
-	/*
-	 * For the FDW's convenience, look up the modification target relation's.
-	 * ResultRelInfo.
-	 */
-	if (node->resultRelation > 0)
-		scanstate->resultRelInfo = estate->es_result_relations[node->resultRelation - 1];
-
 	/* Initialize any outer plan. */
 	if (outerPlan(node))
 		outerPlanState(scanstate) =
-- 
1.8.3.1

From f4b779e7d0e4a329073d6588decd5c8ea53e5f7d Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Thu, 22 Oct 2020 14:45:47 +0900
Subject: [PATCH v1 1/3] Fix a thinko of 1375422c782

EvalPlanQualStart() was mistakenly modified to reset parent EState's
es_result_relations.
---
 src/backend/executor/execMain.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index aea0479..7179f58 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -2693,7 +2693,7 @@ EvalPlanQualStart(EPQState *epqstate, Plan *planTree)
 	 * ResultRelInfos needed by subplans are initialized from scratch when the
 	 * subplans themselves are initialized.
 	 */
-	parentestate->es_result_relations = NULL;
+	rcestate->es_result_relations = NULL;
 	/* es_trig_target_relations must NOT be copied */
 	rcestate->es_top_eflags = parentestate->es_top_eflags;
 	rcestate->es_instrument = parentestate->es_instrument;
-- 
1.8.3.1

From a38c0349471c9b338528a6296a6d73b2242dc0b9 Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Thu, 2 Jul 2020 10:51:45 +0900
Subject: [PATCH v1 3/3] Initialize result relation information lazily

Currently, all elements of the ModifyTableState.resultRelInfo array
are initialized in ExecInitModifyTable(), possibly wastefully,
because only one or a handful of potentially many result relations
appearing in that array may actually have any rows to update or
delete.

This commit refactors all places that directly access the individual
elements of the array to instead go through a lazy-initialization-on-
access function, such that only the elements corresponding to result
relations that are actually operated on are initialized.

Also, extend this lazy initialization approach to some of the
individual fields of ResultRelInfo such that even for the result
relations that are initialized, those fields are only initialized on
first access.  While no performance improvement is to be expected
there, it can lead to a simpler initialization logic of the
ResultRelInfo itself, because the conditions for whether a given
field is needed or not tends to look confusing.  One side-effect
of this is that any "SubPlans" referenced in the expressions of
those fields are also lazily initialized and hence changes the
output of EXPLAIN (without ANALYZE) in some regression tests.

Another unrelated regression test output change is in update.out,
which is caused by deferred initialization of PartitionTupleRouting
for update tuple routing.  Whereas previously a partition constraint
violation error would be reported as occurring on a leaf partition,
due to the aforementioned change, it is now shown as occurring on
the query's target relation, which is valid because it is really
that table's (which is a sub-partitioned table) partition constraint
that is actually violated in the affected test cases.
---
 src/backend/commands/explain.c                |    6 +-
 src/backend/executor/execMain.c               |    7 +
 src/backend/executor/execPartition.c          |  116 ++-
 src/backend/executor/nodeModifyTable.c        | 1067 ++++++++++++++-----------
 src/include/executor/nodeModifyTable.h        |    1 +
 src/include/nodes/execnodes.h                 |    2 +
 src/test/regress/expected/insert_conflict.out |    5 +-
 src/test/regress/expected/updatable_views.out |   18 +-
 src/test/regress/expected/update.out          |   12 +-
 9 files changed, 707 insertions(+), 527 deletions(-)

diff --git a/src/backend/commands/explain.c b/src/backend/commands/explain.c
index 41317f1..e51d0f2 100644
--- a/src/backend/commands/explain.c
+++ b/src/backend/commands/explain.c
@@ -18,7 +18,9 @@
 #include "commands/createas.h"
 #include "commands/defrem.h"
 #include "commands/prepare.h"
+#include "executor/executor.h"
 #include "executor/nodeHash.h"
+#include "executor/nodeModifyTable.h"
 #include "foreign/fdwapi.h"
 #include "jit/jit.h"
 #include "nodes/extensible.h"
@@ -3678,14 +3680,14 @@ show_modifytable_info(ModifyTableState *mtstate, List *ancestors,
 	/* Should we explicitly label target relations? */
 	labeltargets = (mtstate->mt_nplans > 1 ||
 					(mtstate->mt_nplans == 1 &&
-					 mtstate->resultRelInfo->ri_RangeTableIndex != node->nominalRelation));
+					 ExecGetResultRelation(mtstate, 0)->ri_RangeTableIndex != node->nominalRelation));
 
 	if (labeltargets)
 		ExplainOpenGroup("Target Tables", "Target Tables", false, es);
 
 	for (j = 0; j < mtstate->mt_nplans; j++)
 	{
-		ResultRelInfo *resultRelInfo = mtstate->resultRelInfo + j;
+		ResultRelInfo *resultRelInfo = ExecGetResultRelation(mtstate, j);
 		FdwRoutine *fdwroutine = resultRelInfo->ri_FdwRoutine;
 
 		if (labeltargets)
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 7179f58..f484e6a 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -1236,6 +1236,7 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
 	resultRelInfo->ri_ConstraintExprs = NULL;
 	resultRelInfo->ri_GeneratedExprs = NULL;
 	resultRelInfo->ri_junkFilter = NULL;
+	resultRelInfo->ri_junkFilterValid = false;;
 	resultRelInfo->ri_projectReturning = NULL;
 	resultRelInfo->ri_onConflictArbiterIndexes = NIL;
 	resultRelInfo->ri_onConflict = NULL;
@@ -1247,6 +1248,7 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
 													 * ExecInitRoutingInfo */
 	resultRelInfo->ri_PartitionTupleSlot = NULL;	/* ditto */
 	resultRelInfo->ri_ChildToRootMap = NULL;
+	resultRelInfo->ri_ChildToRootMapValid = false;
 	resultRelInfo->ri_CopyMultiInsertBuffer = NULL;
 }
 
@@ -1440,6 +1442,11 @@ ExecCloseResultRelations(EState *estate)
 		ResultRelInfo *resultRelInfo = lfirst(l);
 
 		ExecCloseIndices(resultRelInfo);
+		if (!resultRelInfo->ri_usesFdwDirectModify &&
+			resultRelInfo->ri_FdwRoutine != NULL &&
+			resultRelInfo->ri_FdwRoutine->EndForeignModify != NULL)
+			resultRelInfo->ri_FdwRoutine->EndForeignModify(estate,
+														   resultRelInfo);
 	}
 
 	/* Close any relations that have been opened by ExecGetTriggerResultRel(). */
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 86594bd..8265db2 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -20,6 +20,7 @@
 #include "catalog/pg_type.h"
 #include "executor/execPartition.h"
 #include "executor/executor.h"
+#include "executor/nodeModifyTable.h"
 #include "foreign/fdwapi.h"
 #include "mb/pg_wchar.h"
 #include "miscadmin.h"
@@ -157,10 +158,11 @@ typedef struct PartitionDispatchData
 typedef struct SubplanResultRelHashElem
 {
 	Oid			relid;			/* hash key -- must be first */
-	ResultRelInfo *rri;
+	int			index;
 } SubplanResultRelHashElem;
 
 
+static ResultRelInfo *ExecLookupUpdateResultRelByOid(ModifyTableState *mtstate, Oid reloid);
 static void ExecHashSubPlanResultRelsByOid(ModifyTableState *mtstate,
 										   PartitionTupleRouting *proute);
 static ResultRelInfo *ExecInitPartitionInfo(ModifyTableState *mtstate,
@@ -218,7 +220,6 @@ ExecSetupPartitionTupleRouting(EState *estate, ModifyTableState *mtstate,
 							   Relation rel)
 {
 	PartitionTupleRouting *proute;
-	ModifyTable *node = mtstate ? (ModifyTable *) mtstate->ps.plan : NULL;
 
 	/*
 	 * Here we attempt to expend as little effort as possible in setting up
@@ -240,17 +241,6 @@ ExecSetupPartitionTupleRouting(EState *estate, ModifyTableState *mtstate,
 	ExecInitPartitionDispatchInfo(estate, proute, RelationGetRelid(rel),
 								  NULL, 0);
 
-	/*
-	 * If performing an UPDATE with tuple routing, we can reuse partition
-	 * sub-plan result rels.  We build a hash table to map the OIDs of
-	 * partitions present in mtstate->resultRelInfo to their ResultRelInfos.
-	 * Every time a tuple is routed to a partition that we've yet to set the
-	 * ResultRelInfo for, before we go to the trouble of making one, we check
-	 * for a pre-made one in the hash table.
-	 */
-	if (node && node->operation == CMD_UPDATE)
-		ExecHashSubPlanResultRelsByOid(mtstate, proute);
-
 	return proute;
 }
 
@@ -350,7 +340,6 @@ ExecFindPartition(ModifyTableState *mtstate,
 		is_leaf = partdesc->is_leaf[partidx];
 		if (is_leaf)
 		{
-
 			/*
 			 * We've reached the leaf -- hurray, we're done.  Look to see if
 			 * we've already got a ResultRelInfo for this partition.
@@ -367,20 +356,19 @@ ExecFindPartition(ModifyTableState *mtstate,
 
 				/*
 				 * We have not yet set up a ResultRelInfo for this partition,
-				 * but if we have a subplan hash table, we might have one
-				 * there.  If not, we'll have to create one.
+				 * but if the partition is also an UPDATE result relation, use
+				 * the one in mtstate->resultRelInfo instead of creating a new
+				 * one with ExecInitPartitionInfo().
 				 */
-				if (proute->subplan_resultrel_htab)
+				if (mtstate->operation == CMD_UPDATE && mtstate->ps.plan)
 				{
 					Oid			partoid = partdesc->oids[partidx];
-					SubplanResultRelHashElem *elem;
 
-					elem = hash_search(proute->subplan_resultrel_htab,
-									   &partoid, HASH_FIND, NULL);
-					if (elem)
+					rri = ExecLookupUpdateResultRelByOid(mtstate, partoid);
+
+					if (rri)
 					{
 						found = true;
-						rri = elem->rri;
 
 						/* Verify this ResultRelInfo allows INSERTs */
 						CheckValidResultRel(rri, CMD_INSERT);
@@ -508,6 +496,41 @@ ExecFindPartition(ModifyTableState *mtstate,
 }
 
 /*
+ * ExecLookupUpdateResultRelByOid
+ * 		If the table with given OID appears in the list of result relations
+ * 		to be updated by the given ModifyTable node, return its
+ * 		ResultRelInfo, NULL otherwise.
+ */
+static ResultRelInfo *
+ExecLookupUpdateResultRelByOid(ModifyTableState *mtstate, Oid reloid)
+{
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	SubplanResultRelHashElem *elem;
+	ResultRelInfo *result = NULL;
+
+	Assert(proute != NULL);
+	if (proute->subplan_resultrel_htab == NULL)
+		ExecHashSubPlanResultRelsByOid(mtstate, proute);
+
+	elem = hash_search(proute->subplan_resultrel_htab, &reloid,
+					   HASH_FIND, NULL);
+
+	if (elem)
+	{
+		result = ExecGetResultRelation(mtstate, elem->index);
+
+		/*
+		 * This is required in order to convert the partition's tuple to be
+		 * compatible with the root partitioned table's tuple descriptor. When
+		 * generating the per-subplan result rels, this was not set.
+		 */
+		result->ri_PartitionRoot = proute->partition_root;
+	}
+
+	return result;
+}
+
+/*
  * ExecHashSubPlanResultRelsByOid
  *		Build a hash table to allow fast lookups of subplan ResultRelInfos by
  *		partition Oid.  We also populate the subplan ResultRelInfo with an
@@ -517,9 +540,13 @@ static void
 ExecHashSubPlanResultRelsByOid(ModifyTableState *mtstate,
 							   PartitionTupleRouting *proute)
 {
+	EState	   *estate = mtstate->ps.state;
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	ListCell   *l;
 	HASHCTL		ctl;
 	HTAB	   *htab;
 	int			i;
+	MemoryContext oldcxt = MemoryContextSwitchTo(estate->es_query_cxt);
 
 	memset(&ctl, 0, sizeof(ctl));
 	ctl.keysize = sizeof(Oid);
@@ -530,26 +557,26 @@ ExecHashSubPlanResultRelsByOid(ModifyTableState *mtstate,
 					   &ctl, HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
 	proute->subplan_resultrel_htab = htab;
 
-	/* Hash all subplans by their Oid */
-	for (i = 0; i < mtstate->mt_nplans; i++)
+	/*
+	 * Map each result relation's OID to its ordinal position in
+	 * plan->resultRelations.
+	 */
+	i = 0;
+	foreach(l, plan->resultRelations)
 	{
-		ResultRelInfo *rri = &mtstate->resultRelInfo[i];
+		Index		rti = lfirst_int(l);
+		RangeTblEntry *rte = exec_rt_fetch(rti, estate);
+		Oid			partoid = rte->relid;
 		bool		found;
-		Oid			partoid = RelationGetRelid(rri->ri_RelationDesc);
 		SubplanResultRelHashElem *elem;
 
 		elem = (SubplanResultRelHashElem *)
 			hash_search(htab, &partoid, HASH_ENTER, &found);
 		Assert(!found);
-		elem->rri = rri;
-
-		/*
-		 * This is required in order to convert the partition's tuple to be
-		 * compatible with the root partitioned table's tuple descriptor. When
-		 * generating the per-subplan result rels, this was not set.
-		 */
-		rri->ri_PartitionRoot = proute->partition_root;
+		elem->index = i++;
 	}
+
+	MemoryContextSwitchTo(oldcxt);
 }
 
 /*
@@ -570,7 +597,8 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	Relation	rootrel = rootResultRelInfo->ri_RelationDesc,
 				partrel;
-	Relation	firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
+	Relation	firstResultRel = NULL;
+	Index		firstVarno = 0;
 	ResultRelInfo *leaf_part_rri;
 	MemoryContext oldcxt;
 	AttrMap    *part_attmap = NULL;
@@ -606,19 +634,26 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 						(node != NULL &&
 						 node->onConflictAction != ONCONFLICT_NONE));
 
+	if (node)
+	{
+		ResultRelInfo *firstResultRelInfo = ExecGetResultRelation(mtstate, 0);
+
+		firstResultRel = firstResultRelInfo->ri_RelationDesc;
+		firstVarno = firstResultRelInfo->ri_RangeTableIndex;
+	}
+
 	/*
 	 * Build WITH CHECK OPTION constraints for the partition.  Note that we
 	 * didn't build the withCheckOptionList for partitions within the planner,
 	 * but simple translation of varattnos will suffice.  This only occurs for
 	 * the INSERT case or in the case of UPDATE tuple routing where we didn't
-	 * find a result rel to reuse in ExecSetupPartitionTupleRouting().
+	 * find a result rel to reuse.
 	 */
 	if (node && node->withCheckOptionLists != NIL)
 	{
 		List	   *wcoList;
 		List	   *wcoExprs = NIL;
 		ListCell   *ll;
-		int			firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
 
 		/*
 		 * In the case of INSERT on a partitioned table, there is only one
@@ -682,7 +717,6 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 		TupleTableSlot *slot;
 		ExprContext *econtext;
 		List	   *returningList;
-		int			firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
 
 		/* See the comment above for WCO lists. */
 		Assert((node->operation == CMD_INSERT &&
@@ -741,7 +775,6 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 	 */
 	if (node && node->onConflictAction != ONCONFLICT_NONE)
 	{
-		int			firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
 		TupleDesc	partrelDesc = RelationGetDescr(partrel);
 		ExprContext *econtext = mtstate->ps.ps_ExprContext;
 		ListCell   *lc;
@@ -916,9 +949,14 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 	 * from partition's rowtype to the root partition table's.
 	 */
 	if (mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture)
+	{
 		leaf_part_rri->ri_ChildToRootMap =
 			convert_tuples_by_name(RelationGetDescr(leaf_part_rri->ri_RelationDesc),
 								   RelationGetDescr(leaf_part_rri->ri_PartitionRoot));
+		/* First time creating the map for this result relation. */
+		Assert(!leaf_part_rri->ri_ChildToRootMapValid);
+		leaf_part_rri->ri_ChildToRootMapValid = true;
+	}
 
 	/*
 	 * Since we've just initialized this ResultRelInfo, it's not in any list
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index a33423c..145b49b 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -144,10 +144,41 @@ ExecCheckPlanOutput(Relation resultRel, List *targetList)
 }
 
 /*
+ * Initialize ri_returningList and ri_projectReturning for RETURNING
+ */
+static void
+InitReturningProjection(ModifyTableState *mtstate,
+						ResultRelInfo *resultRelInfo)
+{
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	int		whichrel = resultRelInfo - mtstate->resultRelInfo;
+	List	*rlist;
+	TupleTableSlot *slot;
+	ExprContext *econtext;
+
+	Assert(whichrel >= 0 && whichrel < mtstate->mt_nplans);
+	rlist = (List *) list_nth(plan->returningLists, whichrel);
+	slot = mtstate->ps.ps_ResultTupleSlot;
+	Assert(slot != NULL);
+	econtext = mtstate->ps.ps_ExprContext;
+	Assert(econtext != NULL);
+
+	/* Must not do this a second time! */
+	Assert(resultRelInfo->ri_returningList == NIL &&
+		   resultRelInfo->ri_projectReturning == NULL);
+	resultRelInfo->ri_returningList = rlist;
+	resultRelInfo->ri_projectReturning =
+		ExecBuildProjectionInfo(rlist, econtext, slot, &mtstate->ps,
+								resultRelInfo->ri_RelationDesc->rd_att);
+}
+
+/*
  * ExecProcessReturning --- evaluate a RETURNING list
  *
  * resultRelInfo: current result rel
- * tupleSlot: slot holding tuple actually inserted/updated/deleted
+ * tupleSlot: slot holding tuple actually inserted/updated or NULL for delete
+ * tupleid, oldtuple: when called for delete, one of these can be used to
+ * fill the RETURNING slot for the relation
  * planSlot: slot holding tuple returned by top subplan node
  *
  * Note: If tupleSlot is NULL, the FDW should have already provided econtext's
@@ -156,12 +187,50 @@ ExecCheckPlanOutput(Relation resultRel, List *targetList)
  * Returns a slot holding the result tuple
  */
 static TupleTableSlot *
-ExecProcessReturning(ResultRelInfo *resultRelInfo,
+ExecProcessReturning(ModifyTableState *mtstate,
+					 ResultRelInfo *resultRelInfo,
 					 TupleTableSlot *tupleSlot,
+					 ItemPointer tupleid, HeapTuple oldtuple,
 					 TupleTableSlot *planSlot)
 {
-	ProjectionInfo *projectReturning = resultRelInfo->ri_projectReturning;
-	ExprContext *econtext = projectReturning->pi_exprContext;
+	EState *estate = mtstate->ps.state;
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	ProjectionInfo *projectReturning;
+	ExprContext *econtext;
+	bool		clearTupleSlot = false;
+	TupleTableSlot *result;
+
+	if (plan->returningLists == NIL)
+		return NULL;
+
+	if (resultRelInfo->ri_returningList == NIL)
+		InitReturningProjection(mtstate, resultRelInfo);
+
+	projectReturning = resultRelInfo->ri_projectReturning;
+	econtext = projectReturning->pi_exprContext;
+
+	/*
+	 * Fill tupleSlot with provided tuple or after fetching the tuple with
+	 * provided tupleid.
+	 */
+	if (tupleSlot == NULL && resultRelInfo->ri_FdwRoutine == NULL)
+	{
+		/* FDW must have provided a slot containing the deleted row */
+		Assert(resultRelInfo->ri_FdwRoutine == NULL);
+		tupleSlot = ExecGetReturningSlot(estate, resultRelInfo);
+		if (oldtuple != NULL)
+		{
+			ExecForceStoreHeapTuple(oldtuple, tupleSlot, false);
+		}
+		else
+		{
+			if (!table_tuple_fetch_row_version(resultRelInfo->ri_RelationDesc,
+											   tupleid, SnapshotAny,
+											   tupleSlot))
+				elog(ERROR, "failed to fetch deleted tuple for DELETE RETURNING");
+		}
+		clearTupleSlot = true;
+	}
 
 	/* Make tuple and any needed join variables available to ExecProject */
 	if (tupleSlot)
@@ -176,7 +245,362 @@ ExecProcessReturning(ResultRelInfo *resultRelInfo,
 		RelationGetRelid(resultRelInfo->ri_RelationDesc);
 
 	/* Compute the RETURNING expressions */
-	return ExecProject(projectReturning);
+	result = ExecProject(projectReturning);
+
+	if (clearTupleSlot)
+		ExecClearTuple(tupleSlot);
+
+	return result;
+}
+
+/*
+ * Perform WITH CHECK OPTIONS check, if any.
+ */
+static void
+ExecProcessWithCheckOptions(ModifyTableState *mtstate, ResultRelInfo *resultRelInfo,
+							TupleTableSlot *slot, WCOKind wco_kind)
+{
+	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+	EState *estate = mtstate->ps.state;
+
+	if (node->withCheckOptionLists == NIL)
+		return;
+
+	/* Initilize expression state if not already done. */
+	if (resultRelInfo->ri_WithCheckOptions == NIL)
+	{
+		int		whichrel = resultRelInfo - mtstate->resultRelInfo;
+		List   *wcoList;
+		List   *wcoExprs = NIL;
+		ListCell   *ll;
+
+		Assert(whichrel >= 0 && whichrel < mtstate->mt_nplans);
+		wcoList = (List *) list_nth(node->withCheckOptionLists, whichrel);
+		foreach(ll, wcoList)
+		{
+			WithCheckOption *wco = (WithCheckOption *) lfirst(ll);
+			ExprState  *wcoExpr = ExecInitQual((List *) wco->qual,
+											   &mtstate->ps);
+
+			wcoExprs = lappend(wcoExprs, wcoExpr);
+		}
+
+		resultRelInfo->ri_WithCheckOptions = wcoList;
+		resultRelInfo->ri_WithCheckOptionExprs = wcoExprs;
+	}
+
+	/*
+	 * ExecWithCheckOptions() will skip any WCOs which are not of the kind
+	 * we are looking for at this point.
+	 */
+	ExecWithCheckOptions(wco_kind, resultRelInfo, slot, estate);
+}
+
+/*
+ * Return the list of arbiter indexes to be used for ON CONFLICT processing
+ * on given result relation, fetching it from the plan if not already done.
+ */
+static List *
+GetOnConflictArbiterIndexes(ModifyTableState *mtstate,
+							ResultRelInfo *resultRelInfo)
+{
+	if (resultRelInfo->ri_onConflictArbiterIndexes == NIL)
+	{
+		ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+
+		resultRelInfo->ri_onConflictArbiterIndexes = plan->arbiterIndexes;
+	}
+
+	return resultRelInfo->ri_onConflictArbiterIndexes;
+}
+
+/*
+ * Initialize target list, projection and qual for ON CONFLICT DO UPDATE.
+ */
+static void
+InitOnConflictState(ModifyTableState *mtstate,
+					   ResultRelInfo *resultRelInfo)
+{
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	EState	   *estate = mtstate->ps.state;
+	TupleDesc	relationDesc;
+	TupleDesc	tupDesc;
+	ExprContext *econtext;
+
+	/* insert may only have one relation, inheritance is not expanded */
+	Assert(mtstate->mt_nplans == 1);
+
+	/* already exists if created by RETURNING processing above */
+	if (mtstate->ps.ps_ExprContext == NULL)
+		ExecAssignExprContext(estate, &mtstate->ps);
+
+	econtext = mtstate->ps.ps_ExprContext;
+	relationDesc = resultRelInfo->ri_RelationDesc->rd_att;
+
+	/* create state for DO UPDATE SET operation */
+	resultRelInfo->ri_onConflict = makeNode(OnConflictSetState);
+
+	/* initialize slot for the existing tuple */
+	resultRelInfo->ri_onConflict->oc_Existing =
+		table_slot_create(resultRelInfo->ri_RelationDesc,
+						  &mtstate->ps.state->es_tupleTable);
+
+	/*
+	 * Create the tuple slot for the UPDATE SET projection. We want a slot
+	 * of the table's type here, because the slot will be used to insert
+	 * into the table, and for RETURNING processing - which may access
+	 * system attributes.
+	 */
+	tupDesc = ExecTypeFromTL((List *) plan->onConflictSet);
+	resultRelInfo->ri_onConflict->oc_ProjSlot =
+		ExecInitExtraTupleSlot(mtstate->ps.state, tupDesc,
+							   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
+
+	/* build UPDATE SET projection state */
+	resultRelInfo->ri_onConflict->oc_ProjInfo =
+		ExecBuildProjectionInfo(plan->onConflictSet, econtext,
+								resultRelInfo->ri_onConflict->oc_ProjSlot,
+								&mtstate->ps,
+								relationDesc);
+
+	/* initialize state to evaluate the WHERE clause, if any */
+	if (plan->onConflictWhere)
+	{
+		ExprState  *qualexpr;
+
+		qualexpr = ExecInitQual((List *) plan->onConflictWhere,
+								&mtstate->ps);
+		resultRelInfo->ri_onConflict->oc_WhereClause = qualexpr;
+	}
+}
+
+/*
+ * Initialize ri_junkFilter if needed.
+ *
+ * INSERT queries need a filter if there are any junk attrs in the tlist.
+ * UPDATE and DELETE always need a filter, since there's always at least one
+ * junk attribute present --- no need to look first.  Typically, this will be
+ * a 'ctid' or 'wholerow' attribute, but in the case of a foreign data wrapper
+ * it might be a set of junk attributes sufficient to identify the remote row.
+ *
+ * If there are multiple result relations, each one needs its own junk filter.
+ * Note multiple rels are only possible for UPDATE/DELETE, so we can't be
+ * fooled by some needing a filter and some not.
+ *
+ * This is also a convenient place to verify that the output of an INSERT or
+ * UPDATE matches the target table(s).
+ */
+static void
+InitJunkFilter(ModifyTableState *mtstate, ResultRelInfo *resultRelInfo)
+{
+	EState	   *estate = mtstate->ps.state;
+	CmdType		operation = mtstate->operation;
+	Plan	   *subplan = mtstate->mt_plans[mtstate->mt_whichplan]->plan;
+	ListCell   *l;
+	bool		junk_filter_needed = false;
+
+	switch (operation)
+	{
+		case CMD_INSERT:
+			foreach(l, subplan->targetlist)
+			{
+				TargetEntry *tle = (TargetEntry *) lfirst(l);
+
+				if (tle->resjunk)
+				{
+					junk_filter_needed = true;
+					break;
+				}
+			}
+			break;
+		case CMD_UPDATE:
+		case CMD_DELETE:
+			junk_filter_needed = true;
+			break;
+		default:
+			elog(ERROR, "unknown operation");
+			break;
+	}
+
+	if (junk_filter_needed)
+	{
+		JunkFilter *j;
+		TupleTableSlot *junkresslot;
+
+		junkresslot =
+			ExecInitExtraTupleSlot(estate, NULL,
+								   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
+		j = ExecInitJunkFilter(subplan->targetlist, junkresslot);
+
+		if (operation == CMD_UPDATE || operation == CMD_DELETE)
+		{
+			/* For UPDATE/DELETE, find the appropriate junk attr now */
+			char		relkind;
+
+			relkind = resultRelInfo->ri_RelationDesc->rd_rel->relkind;
+			if (relkind == RELKIND_RELATION ||
+				relkind == RELKIND_MATVIEW ||
+				relkind == RELKIND_PARTITIONED_TABLE)
+			{
+				j->jf_junkAttNo = ExecFindJunkAttribute(j, "ctid");
+				if (!AttributeNumberIsValid(j->jf_junkAttNo))
+					elog(ERROR, "could not find junk ctid column");
+			}
+			else if (relkind == RELKIND_FOREIGN_TABLE)
+			{
+				/*
+				 * When there is a row-level trigger, there should be
+				 * a wholerow attribute.
+				 */
+				j->jf_junkAttNo = ExecFindJunkAttribute(j, "wholerow");
+			}
+			else
+			{
+				j->jf_junkAttNo = ExecFindJunkAttribute(j, "wholerow");
+				if (!AttributeNumberIsValid(j->jf_junkAttNo))
+					elog(ERROR, "could not find junk wholerow column");
+			}
+		}
+
+		/* Must not do this a second time! */
+		Assert(resultRelInfo->ri_junkFilter == NULL);
+		resultRelInfo->ri_junkFilter = j;
+		resultRelInfo->ri_junkFilterValid = true;
+	}
+
+	if (operation == CMD_INSERT || operation == CMD_UPDATE)
+		ExecCheckPlanOutput(resultRelInfo->ri_RelationDesc,
+							subplan->targetlist);
+}
+
+/*
+ * Returns the map needed to convert given child relation's tuples to the
+ * root relation's format, possibly initializing if not already done.
+ */
+static TupleConversionMap *
+GetChildToRootMap(ModifyTableState *mtstate, ResultRelInfo *resultRelInfo)
+{
+	if (!resultRelInfo->ri_ChildToRootMapValid)
+	{
+		Relation	relation = resultRelInfo->ri_RelationDesc;
+		Relation	targetRel = mtstate->rootResultRelInfo->ri_RelationDesc;
+
+		resultRelInfo->ri_ChildToRootMap =
+			convert_tuples_by_name(RelationGetDescr(relation),
+								   RelationGetDescr(targetRel));
+		resultRelInfo->ri_ChildToRootMapValid = true;
+	}
+
+	return resultRelInfo->ri_ChildToRootMap;
+}
+
+/*
+ * ExecGetResultRelation
+ *		Returns mtstate->resultRelInfo[whichrel], possibly initializing it
+ *		if being requested for the first time
+ */
+ResultRelInfo *
+ExecGetResultRelation(ModifyTableState *mtstate, int whichrel)
+{
+	EState	   *estate = mtstate->ps.state;
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	Index		rti;
+	ResultRelInfo *resultRelInfo = NULL;
+
+	/*
+	 * Initialized result relations are added to es_result_relations, so check
+	 * there first.  Remember that es_result_relations is indexed by RT index,
+	 * so fetch the relation's RT index from the plan.
+	 */
+	Assert(plan != NULL);
+	Assert(whichrel >= 0 && whichrel < mtstate->mt_nplans);
+	rti = list_nth_int(plan->resultRelations, whichrel);
+	if (estate->es_result_relations)
+		resultRelInfo = estate->es_result_relations[rti - 1];
+
+	/* Nope, so initialize. */
+	if (resultRelInfo == NULL)
+	{
+		int		eflags = estate->es_top_eflags;
+		CmdType	operation = mtstate->operation;
+		MemoryContext oldcxt;
+
+		Assert(whichrel >= 0);
+		resultRelInfo = &mtstate->resultRelInfo[whichrel];
+
+		/* Things built here have to last for the query duration. */
+		oldcxt = MemoryContextSwitchTo(estate->es_query_cxt);
+
+		/*
+		 * Perform InitResultRelInfo() and save the pointer in
+		 * es_result_relations.
+		 */
+		ExecInitResultRelation(estate, resultRelInfo, rti);
+
+		/*
+		 * A few more initializations that are not handled by
+		 * InitResultRelInfo() follow.
+		 */
+
+		/*
+		 * Verify result relation is a valid target for the current operation.
+		 */
+		CheckValidResultRel(resultRelInfo, operation);
+
+		/* Initialize the usesFdwDirectModify flag */
+		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(whichrel,
+															  plan->fdwDirectModifyPlans);
+
+		/* Also let FDWs init themselves for foreign-table result rels */
+		if (!resultRelInfo->ri_usesFdwDirectModify &&
+			resultRelInfo->ri_FdwRoutine != NULL &&
+			resultRelInfo->ri_FdwRoutine->BeginForeignModify != NULL)
+		{
+			List	   *fdw_private = (List *) list_nth(plan->fdwPrivLists,
+														whichrel);
+
+			resultRelInfo->ri_FdwRoutine->BeginForeignModify(mtstate,
+															 resultRelInfo,
+															 fdw_private,
+															 whichrel,
+															 eflags);
+		}
+
+		/*
+		 * If transition tuples will be captured, initialize a map to convert
+		 * child tuples into the format of the table mentioned in the query
+		 * (root relation), because the transition tuple store can only store
+		 * tuples in the root table format.  However for INSERT, the map is
+		 * only initialized for a given partition when the partition itself is
+		 * first initialized by ExecFindPartition.  Also, this map is also
+		 * needed if an UPDATE ends up having to move tuples across
+		 * partitions, because in that case the child tuple to be moved first
+		 * needs to be converted into the root table's format.  In that case,
+		 * we use GetChildToRootMap() to either create one from scratch if
+		 * we didn't already create it here.
+		 *
+		 * Note: We cannot always initialize this map lazily, that is, use
+		 * GetChildToRootMap(), because AfterTriggerSaveEvent(), which needs
+		 * the map, doesn't have access to the "target" relation that is
+		 * needed to create the map.
+		 */
+		if (mtstate->mt_transition_capture && operation != CMD_INSERT)
+		{
+			Relation	relation = resultRelInfo->ri_RelationDesc;
+			Relation	targetRel = mtstate->rootResultRelInfo->ri_RelationDesc;
+
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(relation),
+									   RelationGetDescr(targetRel));
+			/* First time creating the map for this result relation. */
+			Assert(!resultRelInfo->ri_ChildToRootMapValid);
+			resultRelInfo->ri_ChildToRootMapValid = true;
+		}
+
+		MemoryContextSwitchTo(oldcxt);
+	}
+
+	return resultRelInfo;
 }
 
 /*
@@ -398,12 +822,27 @@ ExecInsert(ModifyTableState *mtstate,
 	{
 		ResultRelInfo *partRelInfo;
 
+		/*
+		 * ExecInitPartitionInfo() expects that the root parent's ri_onConflict
+		 * is initialized. XXX maybe it shouldn't?
+		 */
+		if (onconflict != ONCONFLICT_NONE &&
+			resultRelInfo->ri_onConflict == NULL)
+		{
+			(void) GetOnConflictArbiterIndexes(mtstate, resultRelInfo);
+			if (onconflict == ONCONFLICT_UPDATE)
+				InitOnConflictState(mtstate, resultRelInfo);
+		}
+
 		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
 									   resultRelInfo, slot,
 									   &partRelInfo);
 		resultRelInfo = partRelInfo;
 	}
 
+	if (resultRelInfo->ri_IndexRelationDescs == NULL)
+		ExecOpenIndices(resultRelInfo, onconflict != ONCONFLICT_NONE);
+
 	ExecMaterializeSlot(slot);
 
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
@@ -489,12 +928,7 @@ ExecInsert(ModifyTableState *mtstate,
 		wco_kind = (mtstate->operation == CMD_UPDATE) ?
 			WCO_RLS_UPDATE_CHECK : WCO_RLS_INSERT_CHECK;
 
-		/*
-		 * ExecWithCheckOptions() will skip any WCOs which are not of the kind
-		 * we are looking for at this point.
-		 */
-		if (resultRelInfo->ri_WithCheckOptions != NIL)
-			ExecWithCheckOptions(wco_kind, resultRelInfo, slot, estate);
+		ExecProcessWithCheckOptions(mtstate, resultRelInfo, slot, wco_kind);
 
 		/*
 		 * Check the constraints of the tuple.
@@ -521,7 +955,8 @@ ExecInsert(ModifyTableState *mtstate,
 			bool		specConflict;
 			List	   *arbiterIndexes;
 
-			arbiterIndexes = resultRelInfo->ri_onConflictArbiterIndexes;
+			arbiterIndexes = GetOnConflictArbiterIndexes(mtstate,
+														 resultRelInfo);
 
 			/*
 			 * Do a non-conclusive check for conflicts first.
@@ -691,12 +1126,11 @@ ExecInsert(ModifyTableState *mtstate,
 	 * ExecWithCheckOptions() will skip any WCOs which are not of the kind we
 	 * are looking for at this point.
 	 */
-	if (resultRelInfo->ri_WithCheckOptions != NIL)
-		ExecWithCheckOptions(WCO_VIEW_CHECK, resultRelInfo, slot, estate);
+	ExecProcessWithCheckOptions(mtstate, resultRelInfo, slot, WCO_VIEW_CHECK);
 
 	/* Process RETURNING if present */
-	if (resultRelInfo->ri_projectReturning)
-		result = ExecProcessReturning(resultRelInfo, slot, planSlot);
+	result = ExecProcessReturning(mtstate, resultRelInfo, slot, NULL, NULL,
+								  planSlot);
 
 	return result;
 }
@@ -1011,45 +1445,23 @@ ldelete:;
 						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
-	if (processReturning && resultRelInfo->ri_projectReturning)
+	if (processReturning)
 	{
-		/*
-		 * We have to put the target tuple into a slot, which means first we
-		 * gotta fetch it.  We can use the trigger tuple slot.
-		 */
-		TupleTableSlot *rslot;
-
-		if (resultRelInfo->ri_FdwRoutine)
-		{
-			/* FDW must have provided a slot containing the deleted row */
-			Assert(!TupIsNull(slot));
-		}
-		else
-		{
-			slot = ExecGetReturningSlot(estate, resultRelInfo);
-			if (oldtuple != NULL)
-			{
-				ExecForceStoreHeapTuple(oldtuple, slot, false);
-			}
-			else
-			{
-				if (!table_tuple_fetch_row_version(resultRelationDesc, tupleid,
-												   SnapshotAny, slot))
-					elog(ERROR, "failed to fetch deleted tuple for DELETE RETURNING");
-			}
-		}
-
-		rslot = ExecProcessReturning(resultRelInfo, slot, planSlot);
+		TupleTableSlot *rslot = ExecProcessReturning(mtstate, resultRelInfo,
+													 slot, tupleid, oldtuple,
+													 planSlot);
 
 		/*
 		 * Before releasing the target tuple again, make sure rslot has a
 		 * local copy of any pass-by-reference values.
 		 */
-		ExecMaterializeSlot(rslot);
-
-		ExecClearTuple(slot);
-
-		return rslot;
+		if (rslot)
+		{
+			ExecMaterializeSlot(rslot);
+			if (slot)
+				ExecClearTuple(slot);
+			return rslot;
+		}
 	}
 
 	return NULL;
@@ -1082,7 +1494,6 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 						 TupleTableSlot **inserted_tuple)
 {
 	EState	   *estate = mtstate->ps.state;
-	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 	TupleConversionMap *tupconv_map;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
@@ -1101,13 +1512,27 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 				 errmsg("invalid ON UPDATE specification"),
 				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
 
-	/*
-	 * When an UPDATE is run on a leaf partition, we will not have partition
-	 * tuple routing set up.  In that case, fail with partition constraint
-	 * violation error.
-	 */
-	if (proute == NULL)
-		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+	/* Initialize tuple routing info if not already done. */
+	if (mtstate->mt_partition_tuple_routing == NULL)
+	{
+		Relation	targetRel = mtstate->rootResultRelInfo->ri_RelationDesc;
+		MemoryContext	oldcxt;
+
+		/* Things built here have to last for the query duration. */
+		oldcxt = MemoryContextSwitchTo(estate->es_query_cxt);
+
+		mtstate->mt_partition_tuple_routing =
+			ExecSetupPartitionTupleRouting(estate, mtstate, targetRel);
+
+		/*
+		 * Before a partition's tuple can be re-routed, it must first
+		 * be converted to the root's format and we need a slot for
+		 * storing such tuple.
+		 */
+		Assert(mtstate->mt_root_tuple_slot == NULL);
+		mtstate->mt_root_tuple_slot = table_slot_create(targetRel, NULL);
+		MemoryContextSwitchTo(oldcxt);
+	}
 
 	/*
 	 * Row movement, part 1.  Delete the tuple, but skip RETURNING processing.
@@ -1161,7 +1586,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 	 * convert the tuple into root's tuple descriptor if needed, since
 	 * ExecInsert() starts the search from root.
 	 */
-	tupconv_map = resultRelInfo->ri_ChildToRootMap;
+	tupconv_map = GetChildToRootMap(mtstate, resultRelInfo);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
@@ -1226,6 +1651,9 @@ ExecUpdate(ModifyTableState *mtstate,
 	if (IsBootstrapProcessingMode())
 		elog(ERROR, "cannot UPDATE during bootstrap");
 
+	if (resultRelInfo->ri_IndexRelationDescs == NULL)
+		ExecOpenIndices(resultRelInfo, false);
+
 	ExecMaterializeSlot(slot);
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1318,16 +1746,9 @@ lreplace:;
 			resultRelationDesc->rd_rel->relispartition &&
 			!ExecPartitionCheck(resultRelInfo, slot, estate, false);
 
-		if (!partition_constraint_failed &&
-			resultRelInfo->ri_WithCheckOptions != NIL)
-		{
-			/*
-			 * ExecWithCheckOptions() will skip any WCOs which are not of the
-			 * kind we are looking for at this point.
-			 */
-			ExecWithCheckOptions(WCO_RLS_UPDATE_CHECK,
-								 resultRelInfo, slot, estate);
-		}
+		if (!partition_constraint_failed)
+			ExecProcessWithCheckOptions(mtstate, resultRelInfo, slot,
+										WCO_RLS_UPDATE_CHECK);
 
 		/*
 		 * If a partition check failed, try to move the row into the right
@@ -1340,6 +1761,13 @@ lreplace:;
 			bool		retry;
 
 			/*
+			 * When an UPDATE is run directly on a leaf partition, simply fail
+			 * with partition constraint violation error.
+			 */
+			if (resultRelInfo == mtstate->rootResultRelInfo)
+				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+			/*
 			 * ExecCrossPartitionUpdate will first DELETE the row from the
 			 * partition it's currently in and then insert it back into the
 			 * root table, which will re-route it to the correct partition.
@@ -1535,18 +1963,12 @@ lreplace:;
 	 * required to do this after testing all constraints and uniqueness
 	 * violations per the SQL spec, so we do it after actually updating the
 	 * record in the heap and all indexes.
-	 *
-	 * ExecWithCheckOptions() will skip any WCOs which are not of the kind we
-	 * are looking for at this point.
 	 */
-	if (resultRelInfo->ri_WithCheckOptions != NIL)
-		ExecWithCheckOptions(WCO_VIEW_CHECK, resultRelInfo, slot, estate);
+	ExecProcessWithCheckOptions(mtstate, resultRelInfo, slot, WCO_VIEW_CHECK);
 
 	/* Process RETURNING if present */
-	if (resultRelInfo->ri_projectReturning)
-		return ExecProcessReturning(resultRelInfo, slot, planSlot);
-
-	return NULL;
+	return ExecProcessReturning(mtstate, resultRelInfo, slot, NULL, NULL,
+								planSlot);
 }
 
 /*
@@ -1570,10 +1992,10 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 					 bool canSetTag,
 					 TupleTableSlot **returning)
 {
-	ExprContext *econtext = mtstate->ps.ps_ExprContext;
+	ExprContext *econtext;
 	Relation	relation = resultRelInfo->ri_RelationDesc;
-	ExprState  *onConflictSetWhere = resultRelInfo->ri_onConflict->oc_WhereClause;
-	TupleTableSlot *existing = resultRelInfo->ri_onConflict->oc_Existing;
+	ExprState  *onConflictSetWhere;
+	TupleTableSlot *existing;
 	TM_FailureData tmfd;
 	LockTupleMode lockmode;
 	TM_Result	test;
@@ -1581,6 +2003,13 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	TransactionId xmin;
 	bool		isnull;
 
+	if (resultRelInfo->ri_onConflict == NULL)
+		InitOnConflictState(mtstate, resultRelInfo);
+
+	econtext = mtstate->ps.ps_ExprContext;
+	onConflictSetWhere = resultRelInfo->ri_onConflict->oc_WhereClause;
+	existing = resultRelInfo->ri_onConflict->oc_Existing;
+
 	/* Determine lock mode to use */
 	lockmode = ExecUpdateLockMode(estate, resultRelInfo);
 
@@ -1719,27 +2148,23 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 		return true;			/* done with the tuple */
 	}
 
-	if (resultRelInfo->ri_WithCheckOptions != NIL)
-	{
-		/*
-		 * Check target's existing tuple against UPDATE-applicable USING
-		 * security barrier quals (if any), enforced here as RLS checks/WCOs.
-		 *
-		 * The rewriter creates UPDATE RLS checks/WCOs for UPDATE security
-		 * quals, and stores them as WCOs of "kind" WCO_RLS_CONFLICT_CHECK,
-		 * but that's almost the extent of its special handling for ON
-		 * CONFLICT DO UPDATE.
-		 *
-		 * The rewriter will also have associated UPDATE applicable straight
-		 * RLS checks/WCOs for the benefit of the ExecUpdate() call that
-		 * follows.  INSERTs and UPDATEs naturally have mutually exclusive WCO
-		 * kinds, so there is no danger of spurious over-enforcement in the
-		 * INSERT or UPDATE path.
-		 */
-		ExecWithCheckOptions(WCO_RLS_CONFLICT_CHECK, resultRelInfo,
-							 existing,
-							 mtstate->ps.state);
-	}
+	/*
+	 * Check target's existing tuple against UPDATE-applicable USING
+	 * security barrier quals (if any), enforced here as RLS checks/WCOs.
+	 *
+	 * The rewriter creates UPDATE RLS checks/WCOs for UPDATE security
+	 * quals, and stores them as WCOs of "kind" WCO_RLS_CONFLICT_CHECK,
+	 * but that's almost the extent of its special handling for ON
+	 * CONFLICT DO UPDATE.
+	 *
+	 * The rewriter will also have associated UPDATE applicable straight
+	 * RLS checks/WCOs for the benefit of the ExecUpdate() call that
+	 * follows.  INSERTs and UPDATEs naturally have mutually exclusive WCO
+	 * kinds, so there is no danger of spurious over-enforcement in the
+	 * INSERT or UPDATE path.
+	 */
+	ExecProcessWithCheckOptions(mtstate, resultRelInfo, existing,
+								WCO_RLS_CONFLICT_CHECK);
 
 	/* Project the new tuple version */
 	ExecProject(resultRelInfo->ri_onConflict->oc_ProjInfo);
@@ -1929,11 +2354,12 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
+	ModifyTable *plan = (ModifyTable *) node->ps.plan;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *resultRelInfo;
+	ResultRelInfo *resultRelInfo = NULL;
 	PlanState  *subplanstate;
-	JunkFilter *junkfilter;
+	JunkFilter *junkfilter = NULL;
 	TupleTableSlot *slot;
 	TupleTableSlot *planSlot;
 	ItemPointer tupleid;
@@ -1974,9 +2400,7 @@ ExecModifyTable(PlanState *pstate)
 	}
 
 	/* Preload local variables */
-	resultRelInfo = node->resultRelInfo + node->mt_whichplan;
 	subplanstate = node->mt_plans[node->mt_whichplan];
-	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
@@ -2000,17 +2424,37 @@ ExecModifyTable(PlanState *pstate)
 		if (pstate->ps_ExprContext)
 			ResetExprContext(pstate->ps_ExprContext);
 
+		/*
+		 * FDWs that can push down a modify operation would need to see the
+		 * ResultRelInfo, so fetch one if not already done before executing
+		 * the subplan, potentially opening it for the first time.
+		 */
+		if (bms_is_member(node->mt_whichplan, plan->fdwDirectModifyPlans) &&
+			resultRelInfo == NULL)
+		{
+			resultRelInfo = ExecGetResultRelation(node, node->mt_whichplan);
+
+			/*
+			 * Must make sure to initialize the RETURNING projection as well,
+			 * because some FDWs rely on accessing ri_projectReturning to
+			 * set its "scan" tuple to use below for computing the actual
+			 * RETURNING targetlist.
+			 */
+			if (plan->returningLists && resultRelInfo->ri_returningList == NIL)
+				InitReturningProjection(node, resultRelInfo);
+		}
+
 		planSlot = ExecProcNode(subplanstate);
 
 		if (TupIsNull(planSlot))
 		{
-			/* advance to next subplan if any */
+			/* Signal to initialize the next plan's relation. */
+			resultRelInfo = NULL;
+
 			node->mt_whichplan++;
 			if (node->mt_whichplan < node->mt_nplans)
 			{
-				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
-				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				continue;
@@ -2020,8 +2464,25 @@ ExecModifyTable(PlanState *pstate)
 		}
 
 		/*
+		 * Fetch the result relation for the current plan if not already done,
+		 * potentially opening it for the first time.
+		 */
+		if (resultRelInfo == NULL)
+		{
+			resultRelInfo = ExecGetResultRelation(node, node->mt_whichplan);
+			if (!resultRelInfo->ri_junkFilterValid)
+				InitJunkFilter(node, resultRelInfo);
+			junkfilter = resultRelInfo->ri_junkFilter;
+		}
+
+		/*
 		 * Ensure input tuple is the right format for the target relation.
 		 */
+		if (node->mt_scans[node->mt_whichplan] == NULL)
+			node->mt_scans[node->mt_whichplan] =
+				ExecInitExtraTupleSlot(node->ps.state,
+									   ExecGetResultType(subplanstate),
+									   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
 		if (node->mt_scans[node->mt_whichplan]->tts_ops != planSlot->tts_ops)
 		{
 			ExecCopySlot(node->mt_scans[node->mt_whichplan], planSlot);
@@ -2042,7 +2503,8 @@ ExecModifyTable(PlanState *pstate)
 			 * ExecProcessReturning by IterateDirectModify, so no need to
 			 * provide it here.
 			 */
-			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
+			slot = ExecProcessReturning(node, resultRelInfo, NULL, NULL, NULL,
+										planSlot);
 
 			return slot;
 		}
@@ -2175,13 +2637,10 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
-	ListCell   *l,
-			   *l1;
+	ListCell   *l;
 	int			i;
 	Relation	rel;
-	bool		update_tuple_routing_needed = node->partColsUpdated;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2198,7 +2657,20 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->canSetTag = node->canSetTag;
 	mtstate->mt_done = false;
 
+	/*
+	 * call ExecInitNode on each of the plans to be executed and save the
+	 * results into the array "mt_plans".
+	 */
+	mtstate->mt_nplans = nplans;
 	mtstate->mt_plans = (PlanState **) palloc0(sizeof(PlanState *) * nplans);
+	i = 0;
+	foreach(l, node->plans)
+	{
+		subplan = (Plan *) lfirst(l);
+
+		mtstate->mt_plans[i++] = ExecInitNode(subplan, estate, eflags);
+	}
+
 	mtstate->resultRelInfo = (ResultRelInfo *)
 		palloc(nplans * sizeof(ResultRelInfo));
 	mtstate->mt_scans = (TupleTableSlot **) palloc0(sizeof(TupleTableSlot *) * nplans);
@@ -2225,13 +2697,17 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	}
 	else
 	{
-		mtstate->rootResultRelInfo = mtstate->resultRelInfo;
-		ExecInitResultRelation(estate, mtstate->resultRelInfo,
-							   linitial_int(node->resultRelations));
+		/*
+		 * Unlike a partitioned target relation, the target relation in this
+		 * case will be actually used by ExecModifyTable(), so use
+		 * ExecGetResultRelation() to get the ResultRelInfo, because it
+		 * initializes some fields that a bare InitResultRelInfo() doesn't.
+		 */
+		mtstate->rootResultRelInfo = ExecGetResultRelation(mtstate, 0);
+		Assert(mtstate->rootResultRelInfo == mtstate->resultRelInfo);
 	}
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
-	mtstate->mt_nplans = nplans;
 
 	/* set up epqstate with dummy subplan data for the moment */
 	EvalPlanQualInit(&mtstate->mt_epqstate, estate, NULL, NIL, node->epqParam);
@@ -2244,163 +2720,9 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
 		ExecSetupTransitionCaptureState(mtstate, estate);
 
-	/*
-	 * call ExecInitNode on each of the plans to be executed and save the
-	 * results into the array "mt_plans".  This is also a convenient place to
-	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.
-	 */
-	resultRelInfo = mtstate->resultRelInfo;
-	i = 0;
-	forboth(l, node->resultRelations, l1, node->plans)
-	{
-		Index		resultRelation = lfirst_int(l);
-
-		subplan = (Plan *) lfirst(l1);
-
-		/*
-		 * This opens result relation and fills ResultRelInfo. (root relation
-		 * was initialized already.)
-		 */
-		if (resultRelInfo != mtstate->rootResultRelInfo)
-			ExecInitResultRelation(estate, resultRelInfo, resultRelation);
-
-		/* Initialize the usesFdwDirectModify flag */
-		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(i,
-															  node->fdwDirectModifyPlans);
-
-		/*
-		 * Verify result relation is a valid target for the current operation
-		 */
-		CheckValidResultRel(resultRelInfo, operation);
-
-		/*
-		 * If there are indices on the result relation, open them and save
-		 * descriptors in the result relation info, so that we can add new
-		 * index entries for the tuples we add/update.  We need not do this
-		 * for a DELETE, however, since deletion doesn't affect indexes. Also,
-		 * inside an EvalPlanQual operation, the indexes might be open
-		 * already, since we share the resultrel state with the original
-		 * query.
-		 */
-		if (resultRelInfo->ri_RelationDesc->rd_rel->relhasindex &&
-			operation != CMD_DELETE &&
-			resultRelInfo->ri_IndexRelationDescs == NULL)
-			ExecOpenIndices(resultRelInfo,
-							node->onConflictAction != ONCONFLICT_NONE);
-
-		/*
-		 * If this is an UPDATE and a BEFORE UPDATE trigger is present, the
-		 * trigger itself might modify the partition-key values. So arrange
-		 * for tuple routing.
-		 */
-		if (resultRelInfo->ri_TrigDesc &&
-			resultRelInfo->ri_TrigDesc->trig_update_before_row &&
-			operation == CMD_UPDATE)
-			update_tuple_routing_needed = true;
-
-		/* Now init the plan for this result rel */
-		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
-		mtstate->mt_scans[i] =
-			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
-								   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
-
-		/* Also let FDWs init themselves for foreign-table result rels */
-		if (!resultRelInfo->ri_usesFdwDirectModify &&
-			resultRelInfo->ri_FdwRoutine != NULL &&
-			resultRelInfo->ri_FdwRoutine->BeginForeignModify != NULL)
-		{
-			List	   *fdw_private = (List *) list_nth(node->fdwPrivLists, i);
-
-			resultRelInfo->ri_FdwRoutine->BeginForeignModify(mtstate,
-															 resultRelInfo,
-															 fdw_private,
-															 i,
-															 eflags);
-		}
-
-		/*
-		 * If needed, initialize a map to convert tuples in the child format
-		 * to the format of the table mentioned in the query (root relation).
-		 * It's needed for update tuple routing, because the routing starts
-		 * from the root relation.  It's also needed for capturing transition
-		 * tuples, because the transition tuple store can only store tuples in
-		 * the root table format.
-		 *
-		 * For INSERT, the map is only initialized for a given partition when
-		 * the partition itself is first initialized by ExecFindPartition().
-		 */
-		if (update_tuple_routing_needed ||
-			(mtstate->mt_transition_capture &&
-			 mtstate->operation != CMD_INSERT))
-			resultRelInfo->ri_ChildToRootMap =
-				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
-									   RelationGetDescr(mtstate->rootResultRelInfo->ri_RelationDesc));
-		resultRelInfo++;
-		i++;
-	}
-
-	/* Get the target relation */
-	rel = mtstate->rootResultRelInfo->ri_RelationDesc;
-
-	/*
-	 * If it's not a partitioned table after all, UPDATE tuple routing should
-	 * not be attempted.
-	 */
-	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
-		update_tuple_routing_needed = false;
-
-	/*
-	 * Build state for tuple routing if it's an INSERT or if it's an UPDATE of
-	 * partition key.
-	 */
-	if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
-		(operation == CMD_INSERT || update_tuple_routing_needed))
-		mtstate->mt_partition_tuple_routing =
-			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
-
-	/*
-	 * For update row movement we'll need a dedicated slot to store the tuples
-	 * that have been converted from partition format to the root table
-	 * format.
-	 */
-	if (update_tuple_routing_needed)
-		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-
-	/*
-	 * Initialize any WITH CHECK OPTION constraints if needed.
-	 */
-	resultRelInfo = mtstate->resultRelInfo;
-	i = 0;
-	foreach(l, node->withCheckOptionLists)
-	{
-		List	   *wcoList = (List *) lfirst(l);
-		List	   *wcoExprs = NIL;
-		ListCell   *ll;
-
-		foreach(ll, wcoList)
-		{
-			WithCheckOption *wco = (WithCheckOption *) lfirst(ll);
-			ExprState  *wcoExpr = ExecInitQual((List *) wco->qual,
-											   &mtstate->ps);
-
-			wcoExprs = lappend(wcoExprs, wcoExpr);
-		}
-
-		resultRelInfo->ri_WithCheckOptions = wcoList;
-		resultRelInfo->ri_WithCheckOptionExprs = wcoExprs;
-		resultRelInfo++;
-		i++;
-	}
-
-	/*
-	 * Initialize RETURNING projections if needed.
-	 */
+	/* Initialize some global state for RETURNING projections. */
 	if (node->returningLists)
 	{
-		TupleTableSlot *slot;
-		ExprContext *econtext;
-
 		/*
 		 * Initialize result tuple slot and assign its rowtype using the first
 		 * RETURNING list.  We assume the rest will look the same.
@@ -2409,27 +2731,10 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 		/* Set up a slot for the output of the RETURNING projection(s) */
 		ExecInitResultTupleSlotTL(&mtstate->ps, &TTSOpsVirtual);
-		slot = mtstate->ps.ps_ResultTupleSlot;
 
 		/* Need an econtext too */
 		if (mtstate->ps.ps_ExprContext == NULL)
 			ExecAssignExprContext(estate, &mtstate->ps);
-		econtext = mtstate->ps.ps_ExprContext;
-
-		/*
-		 * Build a projection for each result rel.
-		 */
-		resultRelInfo = mtstate->resultRelInfo;
-		foreach(l, node->returningLists)
-		{
-			List	   *rlist = (List *) lfirst(l);
-
-			resultRelInfo->ri_returningList = rlist;
-			resultRelInfo->ri_projectReturning =
-				ExecBuildProjectionInfo(rlist, econtext, slot, &mtstate->ps,
-										resultRelInfo->ri_RelationDesc->rd_att);
-			resultRelInfo++;
-		}
 	}
 	else
 	{
@@ -2443,67 +2748,18 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		mtstate->ps.ps_ExprContext = NULL;
 	}
 
-	/* Set the list of arbiter indexes if needed for ON CONFLICT */
-	resultRelInfo = mtstate->resultRelInfo;
-	if (node->onConflictAction != ONCONFLICT_NONE)
-		resultRelInfo->ri_onConflictArbiterIndexes = node->arbiterIndexes;
+	/* Get the target relation */
+	rel = mtstate->rootResultRelInfo->ri_RelationDesc;
 
 	/*
-	 * If needed, Initialize target list, projection and qual for ON CONFLICT
-	 * DO UPDATE.
+	 * Build state for tuple routing if it's an INSERT.  An UPDATE might need
+	 * it too, but it's initialized only when it actually ends up moving
+	 * tuples between partitions; see ExecCrossPartitionUpdate().
 	 */
-	if (node->onConflictAction == ONCONFLICT_UPDATE)
-	{
-		ExprContext *econtext;
-		TupleDesc	relationDesc;
-		TupleDesc	tupDesc;
-
-		/* insert may only have one plan, inheritance is not expanded */
-		Assert(nplans == 1);
-
-		/* already exists if created by RETURNING processing above */
-		if (mtstate->ps.ps_ExprContext == NULL)
-			ExecAssignExprContext(estate, &mtstate->ps);
-
-		econtext = mtstate->ps.ps_ExprContext;
-		relationDesc = resultRelInfo->ri_RelationDesc->rd_att;
-
-		/* create state for DO UPDATE SET operation */
-		resultRelInfo->ri_onConflict = makeNode(OnConflictSetState);
-
-		/* initialize slot for the existing tuple */
-		resultRelInfo->ri_onConflict->oc_Existing =
-			table_slot_create(resultRelInfo->ri_RelationDesc,
-							  &mtstate->ps.state->es_tupleTable);
-
-		/*
-		 * Create the tuple slot for the UPDATE SET projection. We want a slot
-		 * of the table's type here, because the slot will be used to insert
-		 * into the table, and for RETURNING processing - which may access
-		 * system attributes.
-		 */
-		tupDesc = ExecTypeFromTL((List *) node->onConflictSet);
-		resultRelInfo->ri_onConflict->oc_ProjSlot =
-			ExecInitExtraTupleSlot(mtstate->ps.state, tupDesc,
-								   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
-
-		/* build UPDATE SET projection state */
-		resultRelInfo->ri_onConflict->oc_ProjInfo =
-			ExecBuildProjectionInfo(node->onConflictSet, econtext,
-									resultRelInfo->ri_onConflict->oc_ProjSlot,
-									&mtstate->ps,
-									relationDesc);
-
-		/* initialize state to evaluate the WHERE clause, if any */
-		if (node->onConflictWhere)
-		{
-			ExprState  *qualexpr;
-
-			qualexpr = ExecInitQual((List *) node->onConflictWhere,
-									&mtstate->ps);
-			resultRelInfo->ri_onConflict->oc_WhereClause = qualexpr;
-		}
-	}
+	if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
+		operation == CMD_INSERT)
+		mtstate->mt_partition_tuple_routing =
+			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
 	 * If we have any secondary relations in an UPDATE or DELETE, they need to
@@ -2541,109 +2797,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 						mtstate->mt_arowmarks[0]);
 
 	/*
-	 * Initialize the junk filter(s) if needed.  INSERT queries need a filter
-	 * if there are any junk attrs in the tlist.  UPDATE and DELETE always
-	 * need a filter, since there's always at least one junk attribute present
-	 * --- no need to look first.  Typically, this will be a 'ctid' or
-	 * 'wholerow' attribute, but in the case of a foreign data wrapper it
-	 * might be a set of junk attributes sufficient to identify the remote
-	 * row.
-	 *
-	 * If there are multiple result relations, each one needs its own junk
-	 * filter.  Note multiple rels are only possible for UPDATE/DELETE, so we
-	 * can't be fooled by some needing a filter and some not.
-	 *
-	 * This section of code is also a convenient place to verify that the
-	 * output of an INSERT or UPDATE matches the target table(s).
-	 */
-	{
-		bool		junk_filter_needed = false;
-
-		switch (operation)
-		{
-			case CMD_INSERT:
-				foreach(l, subplan->targetlist)
-				{
-					TargetEntry *tle = (TargetEntry *) lfirst(l);
-
-					if (tle->resjunk)
-					{
-						junk_filter_needed = true;
-						break;
-					}
-				}
-				break;
-			case CMD_UPDATE:
-			case CMD_DELETE:
-				junk_filter_needed = true;
-				break;
-			default:
-				elog(ERROR, "unknown operation");
-				break;
-		}
-
-		if (junk_filter_needed)
-		{
-			resultRelInfo = mtstate->resultRelInfo;
-			for (i = 0; i < nplans; i++)
-			{
-				JunkFilter *j;
-				TupleTableSlot *junkresslot;
-
-				subplan = mtstate->mt_plans[i]->plan;
-				if (operation == CMD_INSERT || operation == CMD_UPDATE)
-					ExecCheckPlanOutput(resultRelInfo->ri_RelationDesc,
-										subplan->targetlist);
-
-				junkresslot =
-					ExecInitExtraTupleSlot(estate, NULL,
-										   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
-				j = ExecInitJunkFilter(subplan->targetlist,
-									   junkresslot);
-
-				if (operation == CMD_UPDATE || operation == CMD_DELETE)
-				{
-					/* For UPDATE/DELETE, find the appropriate junk attr now */
-					char		relkind;
-
-					relkind = resultRelInfo->ri_RelationDesc->rd_rel->relkind;
-					if (relkind == RELKIND_RELATION ||
-						relkind == RELKIND_MATVIEW ||
-						relkind == RELKIND_PARTITIONED_TABLE)
-					{
-						j->jf_junkAttNo = ExecFindJunkAttribute(j, "ctid");
-						if (!AttributeNumberIsValid(j->jf_junkAttNo))
-							elog(ERROR, "could not find junk ctid column");
-					}
-					else if (relkind == RELKIND_FOREIGN_TABLE)
-					{
-						/*
-						 * When there is a row-level trigger, there should be
-						 * a wholerow attribute.
-						 */
-						j->jf_junkAttNo = ExecFindJunkAttribute(j, "wholerow");
-					}
-					else
-					{
-						j->jf_junkAttNo = ExecFindJunkAttribute(j, "wholerow");
-						if (!AttributeNumberIsValid(j->jf_junkAttNo))
-							elog(ERROR, "could not find junk wholerow column");
-					}
-				}
-
-				resultRelInfo->ri_junkFilter = j;
-				resultRelInfo++;
-			}
-		}
-		else
-		{
-			if (operation == CMD_INSERT)
-				ExecCheckPlanOutput(mtstate->resultRelInfo->ri_RelationDesc,
-									subplan->targetlist);
-		}
-	}
-
-	/*
 	 * Lastly, if this is not the primary (canSetTag) ModifyTable node, add it
 	 * to estate->es_auxmodifytables so that it will be run to completion by
 	 * ExecPostprocessPlan.  (It'd actually work fine to add the primary
@@ -2673,20 +2826,6 @@ ExecEndModifyTable(ModifyTableState *node)
 	int			i;
 
 	/*
-	 * Allow any FDWs to shut down
-	 */
-	for (i = 0; i < node->mt_nplans; i++)
-	{
-		ResultRelInfo *resultRelInfo = node->resultRelInfo + i;
-
-		if (!resultRelInfo->ri_usesFdwDirectModify &&
-			resultRelInfo->ri_FdwRoutine != NULL &&
-			resultRelInfo->ri_FdwRoutine->EndForeignModify != NULL)
-			resultRelInfo->ri_FdwRoutine->EndForeignModify(node->ps.state,
-														   resultRelInfo);
-	}
-
-	/*
 	 * Close all the partitioned tables, leaf partitions, and their indices
 	 * and release the slot used for tuple routing, if set.
 	 */
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 46a2dc9..9ae7e40 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -22,5 +22,6 @@ extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
 extern void ExecReScanModifyTable(ModifyTableState *node);
+extern ResultRelInfo *ExecGetResultRelation(ModifyTableState *mtstate, int whichrel);
 
 #endif							/* NODEMODIFYTABLE_H */
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 6c0a7d6..f2f4bed 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -463,6 +463,7 @@ typedef struct ResultRelInfo
 
 	/* for removing junk attributes from tuples */
 	JunkFilter *ri_junkFilter;
+	bool		ri_junkFilterValid;	/* has the filter been initialized? */
 
 	/* list of RETURNING expressions */
 	List	   *ri_returningList;
@@ -497,6 +498,7 @@ typedef struct ResultRelInfo
 	 * transition tuple capture or update partition row movement is active.
 	 */
 	TupleConversionMap *ri_ChildToRootMap;
+	bool		ri_ChildToRootMapValid;	/* has the map been initialized? */
 
 	/* for use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
diff --git a/src/test/regress/expected/insert_conflict.out b/src/test/regress/expected/insert_conflict.out
index ff157ce..74cd7e2 100644
--- a/src/test/regress/expected/insert_conflict.out
+++ b/src/test/regress/expected/insert_conflict.out
@@ -52,10 +52,7 @@ explain (costs off) insert into insertconflicttest values(0, 'Crowberry') on con
    Conflict Arbiter Indexes: op_index_key, collation_index_key, both_index_key
    Conflict Filter: (SubPlan 1)
    ->  Result
-   SubPlan 1
-     ->  Index Only Scan using both_index_expr_key on insertconflicttest ii
-           Index Cond: (key = excluded.key)
-(8 rows)
+(5 rows)
 
 -- Neither collation nor operator class specifications are required --
 -- supplying them merely *limits* matches to indexes with matching opclasses
diff --git a/src/test/regress/expected/updatable_views.out b/src/test/regress/expected/updatable_views.out
index caed1c1..d8d2a3d 100644
--- a/src/test/regress/expected/updatable_views.out
+++ b/src/test/regress/expected/updatable_views.out
@@ -1862,28 +1862,22 @@ UPDATE rw_view1 SET a = a + 5; -- should fail
 ERROR:  new row violates check option for view "rw_view1"
 DETAIL:  Failing row contains (15).
 EXPLAIN (costs off) INSERT INTO rw_view1 VALUES (5);
-                       QUERY PLAN                        
----------------------------------------------------------
+      QUERY PLAN      
+----------------------
  Insert on base_tbl b
    ->  Result
-   SubPlan 1
-     ->  Index Only Scan using ref_tbl_pkey on ref_tbl r
-           Index Cond: (a = b.a)
-(5 rows)
+(2 rows)
 
 EXPLAIN (costs off) UPDATE rw_view1 SET a = a + 5;
-                        QUERY PLAN                         
------------------------------------------------------------
+               QUERY PLAN                
+-----------------------------------------
  Update on base_tbl b
    ->  Hash Join
          Hash Cond: (b.a = r.a)
          ->  Seq Scan on base_tbl b
          ->  Hash
                ->  Seq Scan on ref_tbl r
-   SubPlan 1
-     ->  Index Only Scan using ref_tbl_pkey on ref_tbl r_1
-           Index Cond: (a = b.a)
-(9 rows)
+(6 rows)
 
 DROP TABLE base_tbl, ref_tbl CASCADE;
 NOTICE:  drop cascades to view rw_view1
diff --git a/src/test/regress/expected/update.out b/src/test/regress/expected/update.out
index bf939d7..0ad0d1a 100644
--- a/src/test/regress/expected/update.out
+++ b/src/test/regress/expected/update.out
@@ -341,8 +341,8 @@ DETAIL:  Failing row contains (105, 85, null, b, 15).
 -- fail, no partition key update, so no attempt to move tuple,
 -- but "a = 'a'" violates partition constraint enforced by root partition)
 UPDATE part_b_10_b_20 set a = 'a';
-ERROR:  new row for relation "part_c_1_100" violates partition constraint
-DETAIL:  Failing row contains (null, 1, 96, 12, a).
+ERROR:  new row for relation "part_b_10_b_20" violates partition constraint
+DETAIL:  Failing row contains (null, 96, a, 12, 1).
 -- ok, partition key update, no constraint violation
 UPDATE range_parted set d = d - 10 WHERE d > 10;
 -- ok, no partition key update, no constraint violation
@@ -372,8 +372,8 @@ UPDATE part_b_10_b_20 set c = c + 20 returning c, b, a;
 
 -- fail, row movement happens only within the partition subtree.
 UPDATE part_b_10_b_20 set b = b - 6 WHERE c > 116 returning *;
-ERROR:  new row for relation "part_d_1_15" violates partition constraint
-DETAIL:  Failing row contains (2, 117, 2, b, 7).
+ERROR:  new row for relation "part_b_10_b_20" violates partition constraint
+DETAIL:  Failing row contains (2, 117, b, 7, 2).
 -- ok, row movement, with subset of rows moved into different partition.
 UPDATE range_parted set b = b - 6 WHERE c > 116 returning a, b + c;
  a | ?column? 
@@ -814,8 +814,8 @@ INSERT into sub_parted VALUES (1,2,10);
 -- Test partition constraint violation when intermediate ancestor is used and
 -- constraint is inherited from upper root.
 UPDATE sub_parted set a = 2 WHERE c = 10;
-ERROR:  new row for relation "sub_part2" violates partition constraint
-DETAIL:  Failing row contains (2, 10, 2).
+ERROR:  new row for relation "sub_parted" violates partition constraint
+DETAIL:  Failing row contains (2, 2, 10).
 -- Test update-partition-key, where the unpruned partitions do not have their
 -- partition keys updated.
 SELECT tableoid::regclass::text, * FROM list_parted WHERE a = 2 ORDER BY 1;
-- 
1.8.3.1

From c86c7e7fb5112ae7e95704ae8f687f50f78da29c Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Mon, 19 Oct 2020 17:17:33 +0900
Subject: [PATCH v2 1/2] Call BeginDirectModify from ExecInitModifyTable

This allows ModifyTable to directly install the target foreign
table's ResultRelInfo into the ForeignScanState node that will be
used to perform the "direct modify" operation rather than have
ExecInitForeignScan() do it by getting it via es_result_relations.

This is in preparation of a later commit to make ModifyTable node
initialize ResultRelInfos lazily, whereby accessing a given target
table's ResultRelInfo directly through es_result_relations while
the ModifyTable is executing will become a deprecated pattern.
---
 src/backend/executor/nodeForeignscan.c | 15 ++++-----------
 src/backend/executor/nodeModifyTable.c | 32 +++++++++++++++++++++++---------
 2 files changed, 27 insertions(+), 20 deletions(-)

diff --git a/src/backend/executor/nodeForeignscan.c b/src/backend/executor/nodeForeignscan.c
index 0b20f94..7101c68 100644
--- a/src/backend/executor/nodeForeignscan.c
+++ b/src/backend/executor/nodeForeignscan.c
@@ -215,24 +215,17 @@ ExecInitForeignScan(ForeignScan *node, EState *estate, int eflags)
 	scanstate->fdwroutine = fdwroutine;
 	scanstate->fdw_state = NULL;
 
-	/*
-	 * For the FDW's convenience, look up the modification target relation's.
-	 * ResultRelInfo.
-	 */
-	if (node->resultRelation > 0)
-		scanstate->resultRelInfo = estate->es_result_relations[node->resultRelation - 1];
-
 	/* Initialize any outer plan. */
 	if (outerPlan(node))
 		outerPlanState(scanstate) =
 			ExecInitNode(outerPlan(node), estate, eflags);
 
 	/*
-	 * Tell the FDW to initialize the scan.
+	 * Tell the FDW to initialize the scan.  For modify operations, it's the
+	 * enclosing ModifyTable node's job to call the FDW after setting up the
+	 * target foreign table's ResultRelInfo.
 	 */
-	if (node->operation != CMD_SELECT)
-		fdwroutine->BeginDirectModify(scanstate, eflags);
-	else
+	if (node->operation == CMD_SELECT)
 		fdwroutine->BeginForeignScan(scanstate, eflags);
 
 	return scanstate;
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 29e07b7..05e68ef 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -2306,17 +2306,31 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 								   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
 
 		/* Also let FDWs init themselves for foreign-table result rels */
-		if (!resultRelInfo->ri_usesFdwDirectModify &&
-			resultRelInfo->ri_FdwRoutine != NULL &&
-			resultRelInfo->ri_FdwRoutine->BeginForeignModify != NULL)
+		if (resultRelInfo->ri_FdwRoutine != NULL)
 		{
-			List	   *fdw_private = (List *) list_nth(node->fdwPrivLists, i);
+			if (resultRelInfo->ri_usesFdwDirectModify)
+			{
+				ForeignScanState *fscan = (ForeignScanState *) mtstate->mt_plans[i];
 
-			resultRelInfo->ri_FdwRoutine->BeginForeignModify(mtstate,
-															 resultRelInfo,
-															 fdw_private,
-															 i,
-															 eflags);
+				/*
+				 * For the FDW's convenience, set the ForeignScanState node's
+				 * ResultRelInfo to let the FDW know which result relation it
+				 * is going to work with.
+				 */
+				Assert(IsA(fscan, ForeignScanState));
+				fscan->resultRelInfo = resultRelInfo;
+				resultRelInfo->ri_FdwRoutine->BeginDirectModify(fscan, eflags);
+			}
+			else if (resultRelInfo->ri_FdwRoutine->BeginForeignModify != NULL)
+			{
+				List   *fdw_private = (List *) list_nth(node->fdwPrivLists, i);
+
+				resultRelInfo->ri_FdwRoutine->BeginForeignModify(mtstate,
+																 resultRelInfo,
+																 fdw_private,
+																 i,
+																 eflags);
+			}
 		}
 
 		/*
-- 
1.8.3.1

From 0a19e71adba1e97f5150229e576e3f93eb2db0de Mon Sep 17 00:00:00 2001
From: amitlan <amitlangote09@gmail.com>
Date: Thu, 2 Jul 2020 10:51:45 +0900
Subject: [PATCH v2 2/2] Initialize result relation information lazily

Currently, all elements of the ModifyTableState.resultRelInfo array
are initialized in ExecInitModifyTable(), possibly wastefully,
because only one or a handful of potentially many result relations
appearing in that array may actually have any rows to update or
delete.

This commit refactors all places that directly access the individual
elements of the array to instead go through a lazy-initialization-on-
access function, such that only the elements corresponding to result
relations that are actually operated on are initialized.

Also, extend this lazy initialization approach to some of the
individual fields of ResultRelInfo such that even for the result
relations that are initialized, those fields are only initialized on
first access.  While no performance improvement is to be expected
there, it can lead to a simpler initialization logic of the
ResultRelInfo itself, because the conditions for whether a given
field is needed or not tends to look confusing.  One side-effect
of this is that any "SubPlans" referenced in the expressions of
those fields are also lazily initialized and hence changes the
output of EXPLAIN (without ANALYZE) in some regression tests.

Another unrelated regression test output change is in update.out,
which is caused by deferred initialization of PartitionTupleRouting
for update tuple routing.  Whereas previously a partition constraint
violation error would be reported as occurring on a leaf partition,
due to the aforementioned change, it is now shown as occurring on
the query's target relation, which is valid because it is really
that table's (which is a sub-partitioned table) partition constraint
that is actually violated in the affected test cases.
---
 src/backend/commands/explain.c                |    6 +-
 src/backend/executor/execMain.c               |    7 +
 src/backend/executor/execPartition.c          |  116 ++-
 src/backend/executor/nodeModifyTable.c        | 1123 ++++++++++++++-----------
 src/include/executor/nodeModifyTable.h        |    1 +
 src/include/nodes/execnodes.h                 |    2 +
 src/test/regress/expected/insert_conflict.out |    5 +-
 src/test/regress/expected/updatable_views.out |   18 +-
 src/test/regress/expected/update.out          |   12 +-
 9 files changed, 737 insertions(+), 553 deletions(-)

diff --git a/src/backend/commands/explain.c b/src/backend/commands/explain.c
index 43f9b01..edd79d7 100644
--- a/src/backend/commands/explain.c
+++ b/src/backend/commands/explain.c
@@ -18,7 +18,9 @@
 #include "commands/createas.h"
 #include "commands/defrem.h"
 #include "commands/prepare.h"
+#include "executor/executor.h"
 #include "executor/nodeHash.h"
+#include "executor/nodeModifyTable.h"
 #include "foreign/fdwapi.h"
 #include "jit/jit.h"
 #include "nodes/extensible.h"
@@ -3678,14 +3680,14 @@ show_modifytable_info(ModifyTableState *mtstate, List *ancestors,
 	/* Should we explicitly label target relations? */
 	labeltargets = (mtstate->mt_nplans > 1 ||
 					(mtstate->mt_nplans == 1 &&
-					 mtstate->resultRelInfo->ri_RangeTableIndex != node->nominalRelation));
+					 ExecGetResultRelation(mtstate, 0)->ri_RangeTableIndex != node->nominalRelation));
 
 	if (labeltargets)
 		ExplainOpenGroup("Target Tables", "Target Tables", false, es);
 
 	for (j = 0; j < mtstate->mt_nplans; j++)
 	{
-		ResultRelInfo *resultRelInfo = mtstate->resultRelInfo + j;
+		ResultRelInfo *resultRelInfo = ExecGetResultRelation(mtstate, j);
 		FdwRoutine *fdwroutine = resultRelInfo->ri_FdwRoutine;
 
 		if (labeltargets)
diff --git a/src/backend/executor/execMain.c b/src/backend/executor/execMain.c
index 7179f58..f484e6a 100644
--- a/src/backend/executor/execMain.c
+++ b/src/backend/executor/execMain.c
@@ -1236,6 +1236,7 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
 	resultRelInfo->ri_ConstraintExprs = NULL;
 	resultRelInfo->ri_GeneratedExprs = NULL;
 	resultRelInfo->ri_junkFilter = NULL;
+	resultRelInfo->ri_junkFilterValid = false;;
 	resultRelInfo->ri_projectReturning = NULL;
 	resultRelInfo->ri_onConflictArbiterIndexes = NIL;
 	resultRelInfo->ri_onConflict = NULL;
@@ -1247,6 +1248,7 @@ InitResultRelInfo(ResultRelInfo *resultRelInfo,
 													 * ExecInitRoutingInfo */
 	resultRelInfo->ri_PartitionTupleSlot = NULL;	/* ditto */
 	resultRelInfo->ri_ChildToRootMap = NULL;
+	resultRelInfo->ri_ChildToRootMapValid = false;
 	resultRelInfo->ri_CopyMultiInsertBuffer = NULL;
 }
 
@@ -1440,6 +1442,11 @@ ExecCloseResultRelations(EState *estate)
 		ResultRelInfo *resultRelInfo = lfirst(l);
 
 		ExecCloseIndices(resultRelInfo);
+		if (!resultRelInfo->ri_usesFdwDirectModify &&
+			resultRelInfo->ri_FdwRoutine != NULL &&
+			resultRelInfo->ri_FdwRoutine->EndForeignModify != NULL)
+			resultRelInfo->ri_FdwRoutine->EndForeignModify(estate,
+														   resultRelInfo);
 	}
 
 	/* Close any relations that have been opened by ExecGetTriggerResultRel(). */
diff --git a/src/backend/executor/execPartition.c b/src/backend/executor/execPartition.c
index 86594bd..8265db2 100644
--- a/src/backend/executor/execPartition.c
+++ b/src/backend/executor/execPartition.c
@@ -20,6 +20,7 @@
 #include "catalog/pg_type.h"
 #include "executor/execPartition.h"
 #include "executor/executor.h"
+#include "executor/nodeModifyTable.h"
 #include "foreign/fdwapi.h"
 #include "mb/pg_wchar.h"
 #include "miscadmin.h"
@@ -157,10 +158,11 @@ typedef struct PartitionDispatchData
 typedef struct SubplanResultRelHashElem
 {
 	Oid			relid;			/* hash key -- must be first */
-	ResultRelInfo *rri;
+	int			index;
 } SubplanResultRelHashElem;
 
 
+static ResultRelInfo *ExecLookupUpdateResultRelByOid(ModifyTableState *mtstate, Oid reloid);
 static void ExecHashSubPlanResultRelsByOid(ModifyTableState *mtstate,
 										   PartitionTupleRouting *proute);
 static ResultRelInfo *ExecInitPartitionInfo(ModifyTableState *mtstate,
@@ -218,7 +220,6 @@ ExecSetupPartitionTupleRouting(EState *estate, ModifyTableState *mtstate,
 							   Relation rel)
 {
 	PartitionTupleRouting *proute;
-	ModifyTable *node = mtstate ? (ModifyTable *) mtstate->ps.plan : NULL;
 
 	/*
 	 * Here we attempt to expend as little effort as possible in setting up
@@ -240,17 +241,6 @@ ExecSetupPartitionTupleRouting(EState *estate, ModifyTableState *mtstate,
 	ExecInitPartitionDispatchInfo(estate, proute, RelationGetRelid(rel),
 								  NULL, 0);
 
-	/*
-	 * If performing an UPDATE with tuple routing, we can reuse partition
-	 * sub-plan result rels.  We build a hash table to map the OIDs of
-	 * partitions present in mtstate->resultRelInfo to their ResultRelInfos.
-	 * Every time a tuple is routed to a partition that we've yet to set the
-	 * ResultRelInfo for, before we go to the trouble of making one, we check
-	 * for a pre-made one in the hash table.
-	 */
-	if (node && node->operation == CMD_UPDATE)
-		ExecHashSubPlanResultRelsByOid(mtstate, proute);
-
 	return proute;
 }
 
@@ -350,7 +340,6 @@ ExecFindPartition(ModifyTableState *mtstate,
 		is_leaf = partdesc->is_leaf[partidx];
 		if (is_leaf)
 		{
-
 			/*
 			 * We've reached the leaf -- hurray, we're done.  Look to see if
 			 * we've already got a ResultRelInfo for this partition.
@@ -367,20 +356,19 @@ ExecFindPartition(ModifyTableState *mtstate,
 
 				/*
 				 * We have not yet set up a ResultRelInfo for this partition,
-				 * but if we have a subplan hash table, we might have one
-				 * there.  If not, we'll have to create one.
+				 * but if the partition is also an UPDATE result relation, use
+				 * the one in mtstate->resultRelInfo instead of creating a new
+				 * one with ExecInitPartitionInfo().
 				 */
-				if (proute->subplan_resultrel_htab)
+				if (mtstate->operation == CMD_UPDATE && mtstate->ps.plan)
 				{
 					Oid			partoid = partdesc->oids[partidx];
-					SubplanResultRelHashElem *elem;
 
-					elem = hash_search(proute->subplan_resultrel_htab,
-									   &partoid, HASH_FIND, NULL);
-					if (elem)
+					rri = ExecLookupUpdateResultRelByOid(mtstate, partoid);
+
+					if (rri)
 					{
 						found = true;
-						rri = elem->rri;
 
 						/* Verify this ResultRelInfo allows INSERTs */
 						CheckValidResultRel(rri, CMD_INSERT);
@@ -508,6 +496,41 @@ ExecFindPartition(ModifyTableState *mtstate,
 }
 
 /*
+ * ExecLookupUpdateResultRelByOid
+ * 		If the table with given OID appears in the list of result relations
+ * 		to be updated by the given ModifyTable node, return its
+ * 		ResultRelInfo, NULL otherwise.
+ */
+static ResultRelInfo *
+ExecLookupUpdateResultRelByOid(ModifyTableState *mtstate, Oid reloid)
+{
+	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
+	SubplanResultRelHashElem *elem;
+	ResultRelInfo *result = NULL;
+
+	Assert(proute != NULL);
+	if (proute->subplan_resultrel_htab == NULL)
+		ExecHashSubPlanResultRelsByOid(mtstate, proute);
+
+	elem = hash_search(proute->subplan_resultrel_htab, &reloid,
+					   HASH_FIND, NULL);
+
+	if (elem)
+	{
+		result = ExecGetResultRelation(mtstate, elem->index);
+
+		/*
+		 * This is required in order to convert the partition's tuple to be
+		 * compatible with the root partitioned table's tuple descriptor. When
+		 * generating the per-subplan result rels, this was not set.
+		 */
+		result->ri_PartitionRoot = proute->partition_root;
+	}
+
+	return result;
+}
+
+/*
  * ExecHashSubPlanResultRelsByOid
  *		Build a hash table to allow fast lookups of subplan ResultRelInfos by
  *		partition Oid.  We also populate the subplan ResultRelInfo with an
@@ -517,9 +540,13 @@ static void
 ExecHashSubPlanResultRelsByOid(ModifyTableState *mtstate,
 							   PartitionTupleRouting *proute)
 {
+	EState	   *estate = mtstate->ps.state;
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	ListCell   *l;
 	HASHCTL		ctl;
 	HTAB	   *htab;
 	int			i;
+	MemoryContext oldcxt = MemoryContextSwitchTo(estate->es_query_cxt);
 
 	memset(&ctl, 0, sizeof(ctl));
 	ctl.keysize = sizeof(Oid);
@@ -530,26 +557,26 @@ ExecHashSubPlanResultRelsByOid(ModifyTableState *mtstate,
 					   &ctl, HASH_ELEM | HASH_BLOBS | HASH_CONTEXT);
 	proute->subplan_resultrel_htab = htab;
 
-	/* Hash all subplans by their Oid */
-	for (i = 0; i < mtstate->mt_nplans; i++)
+	/*
+	 * Map each result relation's OID to its ordinal position in
+	 * plan->resultRelations.
+	 */
+	i = 0;
+	foreach(l, plan->resultRelations)
 	{
-		ResultRelInfo *rri = &mtstate->resultRelInfo[i];
+		Index		rti = lfirst_int(l);
+		RangeTblEntry *rte = exec_rt_fetch(rti, estate);
+		Oid			partoid = rte->relid;
 		bool		found;
-		Oid			partoid = RelationGetRelid(rri->ri_RelationDesc);
 		SubplanResultRelHashElem *elem;
 
 		elem = (SubplanResultRelHashElem *)
 			hash_search(htab, &partoid, HASH_ENTER, &found);
 		Assert(!found);
-		elem->rri = rri;
-
-		/*
-		 * This is required in order to convert the partition's tuple to be
-		 * compatible with the root partitioned table's tuple descriptor. When
-		 * generating the per-subplan result rels, this was not set.
-		 */
-		rri->ri_PartitionRoot = proute->partition_root;
+		elem->index = i++;
 	}
+
+	MemoryContextSwitchTo(oldcxt);
 }
 
 /*
@@ -570,7 +597,8 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
 	Relation	rootrel = rootResultRelInfo->ri_RelationDesc,
 				partrel;
-	Relation	firstResultRel = mtstate->resultRelInfo[0].ri_RelationDesc;
+	Relation	firstResultRel = NULL;
+	Index		firstVarno = 0;
 	ResultRelInfo *leaf_part_rri;
 	MemoryContext oldcxt;
 	AttrMap    *part_attmap = NULL;
@@ -606,19 +634,26 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 						(node != NULL &&
 						 node->onConflictAction != ONCONFLICT_NONE));
 
+	if (node)
+	{
+		ResultRelInfo *firstResultRelInfo = ExecGetResultRelation(mtstate, 0);
+
+		firstResultRel = firstResultRelInfo->ri_RelationDesc;
+		firstVarno = firstResultRelInfo->ri_RangeTableIndex;
+	}
+
 	/*
 	 * Build WITH CHECK OPTION constraints for the partition.  Note that we
 	 * didn't build the withCheckOptionList for partitions within the planner,
 	 * but simple translation of varattnos will suffice.  This only occurs for
 	 * the INSERT case or in the case of UPDATE tuple routing where we didn't
-	 * find a result rel to reuse in ExecSetupPartitionTupleRouting().
+	 * find a result rel to reuse.
 	 */
 	if (node && node->withCheckOptionLists != NIL)
 	{
 		List	   *wcoList;
 		List	   *wcoExprs = NIL;
 		ListCell   *ll;
-		int			firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
 
 		/*
 		 * In the case of INSERT on a partitioned table, there is only one
@@ -682,7 +717,6 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 		TupleTableSlot *slot;
 		ExprContext *econtext;
 		List	   *returningList;
-		int			firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
 
 		/* See the comment above for WCO lists. */
 		Assert((node->operation == CMD_INSERT &&
@@ -741,7 +775,6 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 	 */
 	if (node && node->onConflictAction != ONCONFLICT_NONE)
 	{
-		int			firstVarno = mtstate->resultRelInfo[0].ri_RangeTableIndex;
 		TupleDesc	partrelDesc = RelationGetDescr(partrel);
 		ExprContext *econtext = mtstate->ps.ps_ExprContext;
 		ListCell   *lc;
@@ -916,9 +949,14 @@ ExecInitPartitionInfo(ModifyTableState *mtstate, EState *estate,
 	 * from partition's rowtype to the root partition table's.
 	 */
 	if (mtstate->mt_transition_capture || mtstate->mt_oc_transition_capture)
+	{
 		leaf_part_rri->ri_ChildToRootMap =
 			convert_tuples_by_name(RelationGetDescr(leaf_part_rri->ri_RelationDesc),
 								   RelationGetDescr(leaf_part_rri->ri_PartitionRoot));
+		/* First time creating the map for this result relation. */
+		Assert(!leaf_part_rri->ri_ChildToRootMapValid);
+		leaf_part_rri->ri_ChildToRootMapValid = true;
+	}
 
 	/*
 	 * Since we've just initialized this ResultRelInfo, it's not in any list
diff --git a/src/backend/executor/nodeModifyTable.c b/src/backend/executor/nodeModifyTable.c
index 05e68ef..db24dff 100644
--- a/src/backend/executor/nodeModifyTable.c
+++ b/src/backend/executor/nodeModifyTable.c
@@ -144,10 +144,41 @@ ExecCheckPlanOutput(Relation resultRel, List *targetList)
 }
 
 /*
+ * Initialize ri_returningList and ri_projectReturning for RETURNING
+ */
+static void
+InitReturningProjection(ModifyTableState *mtstate,
+						ResultRelInfo *resultRelInfo)
+{
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	int		whichrel = resultRelInfo - mtstate->resultRelInfo;
+	List	*rlist;
+	TupleTableSlot *slot;
+	ExprContext *econtext;
+
+	Assert(whichrel >= 0 && whichrel < mtstate->mt_nplans);
+	rlist = (List *) list_nth(plan->returningLists, whichrel);
+	slot = mtstate->ps.ps_ResultTupleSlot;
+	Assert(slot != NULL);
+	econtext = mtstate->ps.ps_ExprContext;
+	Assert(econtext != NULL);
+
+	/* Must not do this a second time! */
+	Assert(resultRelInfo->ri_returningList == NIL &&
+		   resultRelInfo->ri_projectReturning == NULL);
+	resultRelInfo->ri_returningList = rlist;
+	resultRelInfo->ri_projectReturning =
+		ExecBuildProjectionInfo(rlist, econtext, slot, &mtstate->ps,
+								resultRelInfo->ri_RelationDesc->rd_att);
+}
+
+/*
  * ExecProcessReturning --- evaluate a RETURNING list
  *
  * resultRelInfo: current result rel
- * tupleSlot: slot holding tuple actually inserted/updated/deleted
+ * tupleSlot: slot holding tuple actually inserted/updated or NULL for delete
+ * tupleid, oldtuple: when called for delete, one of these can be used to
+ * fill the RETURNING slot for the relation
  * planSlot: slot holding tuple returned by top subplan node
  *
  * Note: If tupleSlot is NULL, the FDW should have already provided econtext's
@@ -156,12 +187,50 @@ ExecCheckPlanOutput(Relation resultRel, List *targetList)
  * Returns a slot holding the result tuple
  */
 static TupleTableSlot *
-ExecProcessReturning(ResultRelInfo *resultRelInfo,
+ExecProcessReturning(ModifyTableState *mtstate,
+					 ResultRelInfo *resultRelInfo,
 					 TupleTableSlot *tupleSlot,
+					 ItemPointer tupleid, HeapTuple oldtuple,
 					 TupleTableSlot *planSlot)
 {
-	ProjectionInfo *projectReturning = resultRelInfo->ri_projectReturning;
-	ExprContext *econtext = projectReturning->pi_exprContext;
+	EState *estate = mtstate->ps.state;
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	ProjectionInfo *projectReturning;
+	ExprContext *econtext;
+	bool		clearTupleSlot = false;
+	TupleTableSlot *result;
+
+	if (plan->returningLists == NIL)
+		return NULL;
+
+	if (resultRelInfo->ri_returningList == NIL)
+		InitReturningProjection(mtstate, resultRelInfo);
+
+	projectReturning = resultRelInfo->ri_projectReturning;
+	econtext = projectReturning->pi_exprContext;
+
+	/*
+	 * Fill tupleSlot with provided tuple or after fetching the tuple with
+	 * provided tupleid.
+	 */
+	if (tupleSlot == NULL && resultRelInfo->ri_FdwRoutine == NULL)
+	{
+		/* FDW must have provided a slot containing the deleted row */
+		Assert(resultRelInfo->ri_FdwRoutine == NULL);
+		tupleSlot = ExecGetReturningSlot(estate, resultRelInfo);
+		if (oldtuple != NULL)
+		{
+			ExecForceStoreHeapTuple(oldtuple, tupleSlot, false);
+		}
+		else
+		{
+			if (!table_tuple_fetch_row_version(resultRelInfo->ri_RelationDesc,
+											   tupleid, SnapshotAny,
+											   tupleSlot))
+				elog(ERROR, "failed to fetch deleted tuple for DELETE RETURNING");
+		}
+		clearTupleSlot = true;
+	}
 
 	/* Make tuple and any needed join variables available to ExecProject */
 	if (tupleSlot)
@@ -176,7 +245,392 @@ ExecProcessReturning(ResultRelInfo *resultRelInfo,
 		RelationGetRelid(resultRelInfo->ri_RelationDesc);
 
 	/* Compute the RETURNING expressions */
-	return ExecProject(projectReturning);
+	result = ExecProject(projectReturning);
+
+	if (clearTupleSlot)
+		ExecClearTuple(tupleSlot);
+
+	return result;
+}
+
+/*
+ * Perform WITH CHECK OPTIONS check, if any.
+ */
+static void
+ExecProcessWithCheckOptions(ModifyTableState *mtstate, ResultRelInfo *resultRelInfo,
+							TupleTableSlot *slot, WCOKind wco_kind)
+{
+	ModifyTable *node = (ModifyTable *) mtstate->ps.plan;
+	EState *estate = mtstate->ps.state;
+
+	if (node->withCheckOptionLists == NIL)
+		return;
+
+	/* Initilize expression state if not already done. */
+	if (resultRelInfo->ri_WithCheckOptions == NIL)
+	{
+		int		whichrel = resultRelInfo - mtstate->resultRelInfo;
+		List   *wcoList;
+		List   *wcoExprs = NIL;
+		ListCell   *ll;
+
+		Assert(whichrel >= 0 && whichrel < mtstate->mt_nplans);
+		wcoList = (List *) list_nth(node->withCheckOptionLists, whichrel);
+		foreach(ll, wcoList)
+		{
+			WithCheckOption *wco = (WithCheckOption *) lfirst(ll);
+			ExprState  *wcoExpr = ExecInitQual((List *) wco->qual,
+											   &mtstate->ps);
+
+			wcoExprs = lappend(wcoExprs, wcoExpr);
+		}
+
+		resultRelInfo->ri_WithCheckOptions = wcoList;
+		resultRelInfo->ri_WithCheckOptionExprs = wcoExprs;
+	}
+
+	/*
+	 * ExecWithCheckOptions() will skip any WCOs which are not of the kind
+	 * we are looking for at this point.
+	 */
+	ExecWithCheckOptions(wco_kind, resultRelInfo, slot, estate);
+}
+
+/*
+ * Return the list of arbiter indexes to be used for ON CONFLICT processing
+ * on given result relation, fetching it from the plan if not already done.
+ */
+static List *
+GetOnConflictArbiterIndexes(ModifyTableState *mtstate,
+							ResultRelInfo *resultRelInfo)
+{
+	if (resultRelInfo->ri_onConflictArbiterIndexes == NIL)
+	{
+		ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+
+		resultRelInfo->ri_onConflictArbiterIndexes = plan->arbiterIndexes;
+	}
+
+	return resultRelInfo->ri_onConflictArbiterIndexes;
+}
+
+/*
+ * Initialize target list, projection and qual for ON CONFLICT DO UPDATE.
+ */
+static void
+InitOnConflictState(ModifyTableState *mtstate,
+					   ResultRelInfo *resultRelInfo)
+{
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	EState	   *estate = mtstate->ps.state;
+	TupleDesc	relationDesc;
+	TupleDesc	tupDesc;
+	ExprContext *econtext;
+
+	/* insert may only have one relation, inheritance is not expanded */
+	Assert(mtstate->mt_nplans == 1);
+
+	/* already exists if created by RETURNING processing above */
+	if (mtstate->ps.ps_ExprContext == NULL)
+		ExecAssignExprContext(estate, &mtstate->ps);
+
+	econtext = mtstate->ps.ps_ExprContext;
+	relationDesc = resultRelInfo->ri_RelationDesc->rd_att;
+
+	/* create state for DO UPDATE SET operation */
+	resultRelInfo->ri_onConflict = makeNode(OnConflictSetState);
+
+	/* initialize slot for the existing tuple */
+	resultRelInfo->ri_onConflict->oc_Existing =
+		table_slot_create(resultRelInfo->ri_RelationDesc,
+						  &mtstate->ps.state->es_tupleTable);
+
+	/*
+	 * Create the tuple slot for the UPDATE SET projection. We want a slot
+	 * of the table's type here, because the slot will be used to insert
+	 * into the table, and for RETURNING processing - which may access
+	 * system attributes.
+	 */
+	tupDesc = ExecTypeFromTL((List *) plan->onConflictSet);
+	resultRelInfo->ri_onConflict->oc_ProjSlot =
+		ExecInitExtraTupleSlot(mtstate->ps.state, tupDesc,
+							   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
+
+	/* build UPDATE SET projection state */
+	resultRelInfo->ri_onConflict->oc_ProjInfo =
+		ExecBuildProjectionInfo(plan->onConflictSet, econtext,
+								resultRelInfo->ri_onConflict->oc_ProjSlot,
+								&mtstate->ps,
+								relationDesc);
+
+	/* initialize state to evaluate the WHERE clause, if any */
+	if (plan->onConflictWhere)
+	{
+		ExprState  *qualexpr;
+
+		qualexpr = ExecInitQual((List *) plan->onConflictWhere,
+								&mtstate->ps);
+		resultRelInfo->ri_onConflict->oc_WhereClause = qualexpr;
+	}
+}
+
+/*
+ * Initialize ri_junkFilter if needed.
+ *
+ * INSERT queries need a filter if there are any junk attrs in the tlist.
+ * UPDATE and DELETE always need a filter, since there's always at least one
+ * junk attribute present --- no need to look first.  Typically, this will be
+ * a 'ctid' or 'wholerow' attribute, but in the case of a foreign data wrapper
+ * it might be a set of junk attributes sufficient to identify the remote row.
+ *
+ * If there are multiple result relations, each one needs its own junk filter.
+ * Note multiple rels are only possible for UPDATE/DELETE, so we can't be
+ * fooled by some needing a filter and some not.
+ *
+ * This is also a convenient place to verify that the output of an INSERT or
+ * UPDATE matches the target table(s).
+ */
+static void
+InitJunkFilter(ModifyTableState *mtstate, ResultRelInfo *resultRelInfo)
+{
+	EState	   *estate = mtstate->ps.state;
+	CmdType		operation = mtstate->operation;
+	Plan	   *subplan = mtstate->mt_plans[mtstate->mt_whichplan]->plan;
+	ListCell   *l;
+	bool		junk_filter_needed = false;
+
+	switch (operation)
+	{
+		case CMD_INSERT:
+			foreach(l, subplan->targetlist)
+			{
+				TargetEntry *tle = (TargetEntry *) lfirst(l);
+
+				if (tle->resjunk)
+				{
+					junk_filter_needed = true;
+					break;
+				}
+			}
+			break;
+		case CMD_UPDATE:
+		case CMD_DELETE:
+			junk_filter_needed = true;
+			break;
+		default:
+			elog(ERROR, "unknown operation");
+			break;
+	}
+
+	if (junk_filter_needed)
+	{
+		JunkFilter *j;
+		TupleTableSlot *junkresslot;
+
+		junkresslot =
+			ExecInitExtraTupleSlot(estate, NULL,
+								   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
+
+		/*
+		 * For an INSERT or UPDATE, the result tuple must always match
+		 * the target table's descriptor.  For a DELETE, it won't
+		 * (indeed, there's probably no non-junk output columns).
+		 */
+		if (operation == CMD_INSERT || operation == CMD_UPDATE)
+		{
+			ExecCheckPlanOutput(resultRelInfo->ri_RelationDesc,
+								subplan->targetlist);
+			j = ExecInitJunkFilterInsertion(subplan->targetlist,
+											RelationGetDescr(resultRelInfo->ri_RelationDesc),
+											junkresslot);
+		}
+		else
+			j = ExecInitJunkFilter(subplan->targetlist,
+								   junkresslot);
+
+		if (operation == CMD_UPDATE || operation == CMD_DELETE)
+		{
+			/* For UPDATE/DELETE, find the appropriate junk attr now */
+			char		relkind;
+
+			relkind = resultRelInfo->ri_RelationDesc->rd_rel->relkind;
+			if (relkind == RELKIND_RELATION ||
+				relkind == RELKIND_MATVIEW ||
+				relkind == RELKIND_PARTITIONED_TABLE)
+			{
+				j->jf_junkAttNo = ExecFindJunkAttribute(j, "ctid");
+				if (!AttributeNumberIsValid(j->jf_junkAttNo))
+					elog(ERROR, "could not find junk ctid column");
+			}
+			else if (relkind == RELKIND_FOREIGN_TABLE)
+			{
+				/*
+				 * When there is a row-level trigger, there should be
+				 * a wholerow attribute.
+				 */
+				j->jf_junkAttNo = ExecFindJunkAttribute(j, "wholerow");
+			}
+			else
+			{
+				j->jf_junkAttNo = ExecFindJunkAttribute(j, "wholerow");
+				if (!AttributeNumberIsValid(j->jf_junkAttNo))
+					elog(ERROR, "could not find junk wholerow column");
+			}
+		}
+
+		/* Must not do this a second time! */
+		Assert(resultRelInfo->ri_junkFilter == NULL);
+		resultRelInfo->ri_junkFilter = j;
+		resultRelInfo->ri_junkFilterValid = true;
+	}
+
+	if (operation == CMD_INSERT || operation == CMD_UPDATE)
+		ExecCheckPlanOutput(resultRelInfo->ri_RelationDesc,
+							subplan->targetlist);
+}
+
+/*
+ * Returns the map needed to convert given child relation's tuples to the
+ * root relation's format, possibly initializing if not already done.
+ */
+static TupleConversionMap *
+GetChildToRootMap(ModifyTableState *mtstate, ResultRelInfo *resultRelInfo)
+{
+	if (!resultRelInfo->ri_ChildToRootMapValid)
+	{
+		Relation	relation = resultRelInfo->ri_RelationDesc;
+		Relation	targetRel = mtstate->rootResultRelInfo->ri_RelationDesc;
+
+		resultRelInfo->ri_ChildToRootMap =
+			convert_tuples_by_name(RelationGetDescr(relation),
+								   RelationGetDescr(targetRel));
+		resultRelInfo->ri_ChildToRootMapValid = true;
+	}
+
+	return resultRelInfo->ri_ChildToRootMap;
+}
+
+/*
+ * ExecGetResultRelation
+ *		Returns mtstate->resultRelInfo[whichrel], possibly initializing it
+ *		if being requested for the first time
+ */
+ResultRelInfo *
+ExecGetResultRelation(ModifyTableState *mtstate, int whichrel)
+{
+	EState	   *estate = mtstate->ps.state;
+	ModifyTable *plan = (ModifyTable *) mtstate->ps.plan;
+	Index		rti;
+	ResultRelInfo *resultRelInfo = NULL;
+
+	/*
+	 * Initialized result relations are added to es_result_relations, so check
+	 * there first.  Remember that es_result_relations is indexed by RT index,
+	 * so fetch the relation's RT index from the plan.
+	 */
+	Assert(plan != NULL);
+	Assert(whichrel >= 0 && whichrel < mtstate->mt_nplans);
+	rti = list_nth_int(plan->resultRelations, whichrel);
+	if (estate->es_result_relations)
+		resultRelInfo = estate->es_result_relations[rti - 1];
+
+	/* Nope, so initialize. */
+	if (resultRelInfo == NULL)
+	{
+		int		eflags = estate->es_top_eflags;
+		CmdType	operation = mtstate->operation;
+		MemoryContext oldcxt;
+
+		Assert(whichrel >= 0);
+		resultRelInfo = &mtstate->resultRelInfo[whichrel];
+
+		/* Things built here have to last for the query duration. */
+		oldcxt = MemoryContextSwitchTo(estate->es_query_cxt);
+
+		/*
+		 * Perform InitResultRelInfo() and save the pointer in
+		 * es_result_relations.
+		 */
+		ExecInitResultRelation(estate, resultRelInfo, rti);
+
+		/*
+		 * A few more initializations that are not handled by
+		 * InitResultRelInfo() follow.
+		 */
+
+		/*
+		 * Verify result relation is a valid target for the current operation.
+		 */
+		CheckValidResultRel(resultRelInfo, operation);
+
+		/* Initialize the usesFdwDirectModify flag */
+		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(whichrel,
+															  plan->fdwDirectModifyPlans);
+
+		/* Also let FDWs init themselves for foreign-table result rels */
+		if (resultRelInfo->ri_FdwRoutine != NULL)
+		{
+			if (resultRelInfo->ri_usesFdwDirectModify)
+			{
+				ForeignScanState *fscan = (ForeignScanState *) mtstate->mt_plans[whichrel];
+
+				/*
+				 * For the FDW's convenience, set the ForeignScanState node's
+				 * ResultRelInfo to let the FDW know which result relation it
+				 * is going to work with.
+				 */
+				Assert(IsA(fscan, ForeignScanState));
+				fscan->resultRelInfo = resultRelInfo;
+				resultRelInfo->ri_FdwRoutine->BeginDirectModify(fscan, eflags);
+			}
+			else if (resultRelInfo->ri_FdwRoutine->BeginForeignModify != NULL)
+			{
+				List   *fdw_private = (List *) list_nth(plan->fdwPrivLists,
+														whichrel);
+
+				resultRelInfo->ri_FdwRoutine->BeginForeignModify(mtstate,
+																 resultRelInfo,
+																 fdw_private,
+																 whichrel,
+																 eflags);
+			}
+		}
+
+		/*
+		 * If transition tuples will be captured, initialize a map to convert
+		 * child tuples into the format of the table mentioned in the query
+		 * (root relation), because the transition tuple store can only store
+		 * tuples in the root table format.  However for INSERT, the map is
+		 * only initialized for a given partition when the partition itself is
+		 * first initialized by ExecFindPartition.  Also, this map is also
+		 * needed if an UPDATE ends up having to move tuples across
+		 * partitions, because in that case the child tuple to be moved first
+		 * needs to be converted into the root table's format.  In that case,
+		 * we use GetChildToRootMap() to either create one from scratch if
+		 * we didn't already create it here.
+		 *
+		 * Note: We cannot always initialize this map lazily, that is, use
+		 * GetChildToRootMap(), because AfterTriggerSaveEvent(), which needs
+		 * the map, doesn't have access to the "target" relation that is
+		 * needed to create the map.
+		 */
+		if (mtstate->mt_transition_capture && operation != CMD_INSERT)
+		{
+			Relation	relation = resultRelInfo->ri_RelationDesc;
+			Relation	targetRel = mtstate->rootResultRelInfo->ri_RelationDesc;
+
+			resultRelInfo->ri_ChildToRootMap =
+				convert_tuples_by_name(RelationGetDescr(relation),
+									   RelationGetDescr(targetRel));
+			/* First time creating the map for this result relation. */
+			Assert(!resultRelInfo->ri_ChildToRootMapValid);
+			resultRelInfo->ri_ChildToRootMapValid = true;
+		}
+
+		MemoryContextSwitchTo(oldcxt);
+	}
+
+	return resultRelInfo;
 }
 
 /*
@@ -398,12 +852,27 @@ ExecInsert(ModifyTableState *mtstate,
 	{
 		ResultRelInfo *partRelInfo;
 
+		/*
+		 * ExecInitPartitionInfo() expects that the root parent's ri_onConflict
+		 * is initialized. XXX maybe it shouldn't?
+		 */
+		if (onconflict != ONCONFLICT_NONE &&
+			resultRelInfo->ri_onConflict == NULL)
+		{
+			(void) GetOnConflictArbiterIndexes(mtstate, resultRelInfo);
+			if (onconflict == ONCONFLICT_UPDATE)
+				InitOnConflictState(mtstate, resultRelInfo);
+		}
+
 		slot = ExecPrepareTupleRouting(mtstate, estate, proute,
 									   resultRelInfo, slot,
 									   &partRelInfo);
 		resultRelInfo = partRelInfo;
 	}
 
+	if (resultRelInfo->ri_IndexRelationDescs == NULL)
+		ExecOpenIndices(resultRelInfo, onconflict != ONCONFLICT_NONE);
+
 	ExecMaterializeSlot(slot);
 
 	resultRelationDesc = resultRelInfo->ri_RelationDesc;
@@ -489,12 +958,7 @@ ExecInsert(ModifyTableState *mtstate,
 		wco_kind = (mtstate->operation == CMD_UPDATE) ?
 			WCO_RLS_UPDATE_CHECK : WCO_RLS_INSERT_CHECK;
 
-		/*
-		 * ExecWithCheckOptions() will skip any WCOs which are not of the kind
-		 * we are looking for at this point.
-		 */
-		if (resultRelInfo->ri_WithCheckOptions != NIL)
-			ExecWithCheckOptions(wco_kind, resultRelInfo, slot, estate);
+		ExecProcessWithCheckOptions(mtstate, resultRelInfo, slot, wco_kind);
 
 		/*
 		 * Check the constraints of the tuple.
@@ -521,7 +985,8 @@ ExecInsert(ModifyTableState *mtstate,
 			bool		specConflict;
 			List	   *arbiterIndexes;
 
-			arbiterIndexes = resultRelInfo->ri_onConflictArbiterIndexes;
+			arbiterIndexes = GetOnConflictArbiterIndexes(mtstate,
+														 resultRelInfo);
 
 			/*
 			 * Do a non-conclusive check for conflicts first.
@@ -691,12 +1156,11 @@ ExecInsert(ModifyTableState *mtstate,
 	 * ExecWithCheckOptions() will skip any WCOs which are not of the kind we
 	 * are looking for at this point.
 	 */
-	if (resultRelInfo->ri_WithCheckOptions != NIL)
-		ExecWithCheckOptions(WCO_VIEW_CHECK, resultRelInfo, slot, estate);
+	ExecProcessWithCheckOptions(mtstate, resultRelInfo, slot, WCO_VIEW_CHECK);
 
 	/* Process RETURNING if present */
-	if (resultRelInfo->ri_projectReturning)
-		result = ExecProcessReturning(resultRelInfo, slot, planSlot);
+	result = ExecProcessReturning(mtstate, resultRelInfo, slot, NULL, NULL,
+								  planSlot);
 
 	return result;
 }
@@ -1011,45 +1475,23 @@ ldelete:;
 						 ar_delete_trig_tcs);
 
 	/* Process RETURNING if present and if requested */
-	if (processReturning && resultRelInfo->ri_projectReturning)
+	if (processReturning)
 	{
-		/*
-		 * We have to put the target tuple into a slot, which means first we
-		 * gotta fetch it.  We can use the trigger tuple slot.
-		 */
-		TupleTableSlot *rslot;
-
-		if (resultRelInfo->ri_FdwRoutine)
-		{
-			/* FDW must have provided a slot containing the deleted row */
-			Assert(!TupIsNull(slot));
-		}
-		else
-		{
-			slot = ExecGetReturningSlot(estate, resultRelInfo);
-			if (oldtuple != NULL)
-			{
-				ExecForceStoreHeapTuple(oldtuple, slot, false);
-			}
-			else
-			{
-				if (!table_tuple_fetch_row_version(resultRelationDesc, tupleid,
-												   SnapshotAny, slot))
-					elog(ERROR, "failed to fetch deleted tuple for DELETE RETURNING");
-			}
-		}
-
-		rslot = ExecProcessReturning(resultRelInfo, slot, planSlot);
+		TupleTableSlot *rslot = ExecProcessReturning(mtstate, resultRelInfo,
+													 slot, tupleid, oldtuple,
+													 planSlot);
 
 		/*
 		 * Before releasing the target tuple again, make sure rslot has a
 		 * local copy of any pass-by-reference values.
 		 */
-		ExecMaterializeSlot(rslot);
-
-		ExecClearTuple(slot);
-
-		return rslot;
+		if (rslot)
+		{
+			ExecMaterializeSlot(rslot);
+			if (slot)
+				ExecClearTuple(slot);
+			return rslot;
+		}
 	}
 
 	return NULL;
@@ -1082,7 +1524,6 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 						 TupleTableSlot **inserted_tuple)
 {
 	EState	   *estate = mtstate->ps.state;
-	PartitionTupleRouting *proute = mtstate->mt_partition_tuple_routing;
 	TupleConversionMap *tupconv_map;
 	bool		tuple_deleted;
 	TupleTableSlot *epqslot = NULL;
@@ -1101,13 +1542,27 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 				 errmsg("invalid ON UPDATE specification"),
 				 errdetail("The result tuple would appear in a different partition than the original tuple.")));
 
-	/*
-	 * When an UPDATE is run on a leaf partition, we will not have partition
-	 * tuple routing set up.  In that case, fail with partition constraint
-	 * violation error.
-	 */
-	if (proute == NULL)
-		ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+	/* Initialize tuple routing info if not already done. */
+	if (mtstate->mt_partition_tuple_routing == NULL)
+	{
+		Relation	targetRel = mtstate->rootResultRelInfo->ri_RelationDesc;
+		MemoryContext	oldcxt;
+
+		/* Things built here have to last for the query duration. */
+		oldcxt = MemoryContextSwitchTo(estate->es_query_cxt);
+
+		mtstate->mt_partition_tuple_routing =
+			ExecSetupPartitionTupleRouting(estate, mtstate, targetRel);
+
+		/*
+		 * Before a partition's tuple can be re-routed, it must first
+		 * be converted to the root's format and we need a slot for
+		 * storing such tuple.
+		 */
+		Assert(mtstate->mt_root_tuple_slot == NULL);
+		mtstate->mt_root_tuple_slot = table_slot_create(targetRel, NULL);
+		MemoryContextSwitchTo(oldcxt);
+	}
 
 	/*
 	 * Row movement, part 1.  Delete the tuple, but skip RETURNING processing.
@@ -1161,7 +1616,7 @@ ExecCrossPartitionUpdate(ModifyTableState *mtstate,
 	 * convert the tuple into root's tuple descriptor if needed, since
 	 * ExecInsert() starts the search from root.
 	 */
-	tupconv_map = resultRelInfo->ri_ChildToRootMap;
+	tupconv_map = GetChildToRootMap(mtstate, resultRelInfo);
 	if (tupconv_map != NULL)
 		slot = execute_attr_map_slot(tupconv_map->attrMap,
 									 slot,
@@ -1226,6 +1681,9 @@ ExecUpdate(ModifyTableState *mtstate,
 	if (IsBootstrapProcessingMode())
 		elog(ERROR, "cannot UPDATE during bootstrap");
 
+	if (resultRelInfo->ri_IndexRelationDescs == NULL)
+		ExecOpenIndices(resultRelInfo, false);
+
 	ExecMaterializeSlot(slot);
 
 	/* BEFORE ROW UPDATE Triggers */
@@ -1318,16 +1776,9 @@ lreplace:;
 			resultRelationDesc->rd_rel->relispartition &&
 			!ExecPartitionCheck(resultRelInfo, slot, estate, false);
 
-		if (!partition_constraint_failed &&
-			resultRelInfo->ri_WithCheckOptions != NIL)
-		{
-			/*
-			 * ExecWithCheckOptions() will skip any WCOs which are not of the
-			 * kind we are looking for at this point.
-			 */
-			ExecWithCheckOptions(WCO_RLS_UPDATE_CHECK,
-								 resultRelInfo, slot, estate);
-		}
+		if (!partition_constraint_failed)
+			ExecProcessWithCheckOptions(mtstate, resultRelInfo, slot,
+										WCO_RLS_UPDATE_CHECK);
 
 		/*
 		 * If a partition check failed, try to move the row into the right
@@ -1340,6 +1791,13 @@ lreplace:;
 			bool		retry;
 
 			/*
+			 * When an UPDATE is run directly on a leaf partition, simply fail
+			 * with partition constraint violation error.
+			 */
+			if (resultRelInfo == mtstate->rootResultRelInfo)
+				ExecPartitionCheckEmitError(resultRelInfo, slot, estate);
+
+			/*
 			 * ExecCrossPartitionUpdate will first DELETE the row from the
 			 * partition it's currently in and then insert it back into the
 			 * root table, which will re-route it to the correct partition.
@@ -1535,18 +1993,12 @@ lreplace:;
 	 * required to do this after testing all constraints and uniqueness
 	 * violations per the SQL spec, so we do it after actually updating the
 	 * record in the heap and all indexes.
-	 *
-	 * ExecWithCheckOptions() will skip any WCOs which are not of the kind we
-	 * are looking for at this point.
 	 */
-	if (resultRelInfo->ri_WithCheckOptions != NIL)
-		ExecWithCheckOptions(WCO_VIEW_CHECK, resultRelInfo, slot, estate);
+	ExecProcessWithCheckOptions(mtstate, resultRelInfo, slot, WCO_VIEW_CHECK);
 
 	/* Process RETURNING if present */
-	if (resultRelInfo->ri_projectReturning)
-		return ExecProcessReturning(resultRelInfo, slot, planSlot);
-
-	return NULL;
+	return ExecProcessReturning(mtstate, resultRelInfo, slot, NULL, NULL,
+								planSlot);
 }
 
 /*
@@ -1570,10 +2022,10 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 					 bool canSetTag,
 					 TupleTableSlot **returning)
 {
-	ExprContext *econtext = mtstate->ps.ps_ExprContext;
+	ExprContext *econtext;
 	Relation	relation = resultRelInfo->ri_RelationDesc;
-	ExprState  *onConflictSetWhere = resultRelInfo->ri_onConflict->oc_WhereClause;
-	TupleTableSlot *existing = resultRelInfo->ri_onConflict->oc_Existing;
+	ExprState  *onConflictSetWhere;
+	TupleTableSlot *existing;
 	TM_FailureData tmfd;
 	LockTupleMode lockmode;
 	TM_Result	test;
@@ -1581,6 +2033,13 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 	TransactionId xmin;
 	bool		isnull;
 
+	if (resultRelInfo->ri_onConflict == NULL)
+		InitOnConflictState(mtstate, resultRelInfo);
+
+	econtext = mtstate->ps.ps_ExprContext;
+	onConflictSetWhere = resultRelInfo->ri_onConflict->oc_WhereClause;
+	existing = resultRelInfo->ri_onConflict->oc_Existing;
+
 	/* Determine lock mode to use */
 	lockmode = ExecUpdateLockMode(estate, resultRelInfo);
 
@@ -1719,27 +2178,23 @@ ExecOnConflictUpdate(ModifyTableState *mtstate,
 		return true;			/* done with the tuple */
 	}
 
-	if (resultRelInfo->ri_WithCheckOptions != NIL)
-	{
-		/*
-		 * Check target's existing tuple against UPDATE-applicable USING
-		 * security barrier quals (if any), enforced here as RLS checks/WCOs.
-		 *
-		 * The rewriter creates UPDATE RLS checks/WCOs for UPDATE security
-		 * quals, and stores them as WCOs of "kind" WCO_RLS_CONFLICT_CHECK,
-		 * but that's almost the extent of its special handling for ON
-		 * CONFLICT DO UPDATE.
-		 *
-		 * The rewriter will also have associated UPDATE applicable straight
-		 * RLS checks/WCOs for the benefit of the ExecUpdate() call that
-		 * follows.  INSERTs and UPDATEs naturally have mutually exclusive WCO
-		 * kinds, so there is no danger of spurious over-enforcement in the
-		 * INSERT or UPDATE path.
-		 */
-		ExecWithCheckOptions(WCO_RLS_CONFLICT_CHECK, resultRelInfo,
-							 existing,
-							 mtstate->ps.state);
-	}
+	/*
+	 * Check target's existing tuple against UPDATE-applicable USING
+	 * security barrier quals (if any), enforced here as RLS checks/WCOs.
+	 *
+	 * The rewriter creates UPDATE RLS checks/WCOs for UPDATE security
+	 * quals, and stores them as WCOs of "kind" WCO_RLS_CONFLICT_CHECK,
+	 * but that's almost the extent of its special handling for ON
+	 * CONFLICT DO UPDATE.
+	 *
+	 * The rewriter will also have associated UPDATE applicable straight
+	 * RLS checks/WCOs for the benefit of the ExecUpdate() call that
+	 * follows.  INSERTs and UPDATEs naturally have mutually exclusive WCO
+	 * kinds, so there is no danger of spurious over-enforcement in the
+	 * INSERT or UPDATE path.
+	 */
+	ExecProcessWithCheckOptions(mtstate, resultRelInfo, existing,
+								WCO_RLS_CONFLICT_CHECK);
 
 	/* Project the new tuple version */
 	ExecProject(resultRelInfo->ri_onConflict->oc_ProjInfo);
@@ -1929,11 +2384,12 @@ static TupleTableSlot *
 ExecModifyTable(PlanState *pstate)
 {
 	ModifyTableState *node = castNode(ModifyTableState, pstate);
+	ModifyTable *plan = (ModifyTable *) node->ps.plan;
 	EState	   *estate = node->ps.state;
 	CmdType		operation = node->operation;
-	ResultRelInfo *resultRelInfo;
+	ResultRelInfo *resultRelInfo = NULL;
 	PlanState  *subplanstate;
-	JunkFilter *junkfilter;
+	JunkFilter *junkfilter = NULL;
 	TupleTableSlot *slot;
 	TupleTableSlot *planSlot;
 	ItemPointer tupleid;
@@ -1974,9 +2430,7 @@ ExecModifyTable(PlanState *pstate)
 	}
 
 	/* Preload local variables */
-	resultRelInfo = node->resultRelInfo + node->mt_whichplan;
 	subplanstate = node->mt_plans[node->mt_whichplan];
-	junkfilter = resultRelInfo->ri_junkFilter;
 
 	/*
 	 * Fetch rows from subplan(s), and execute the required table modification
@@ -2000,17 +2454,37 @@ ExecModifyTable(PlanState *pstate)
 		if (pstate->ps_ExprContext)
 			ResetExprContext(pstate->ps_ExprContext);
 
+		/*
+		 * FDWs that can push down a modify operation would need to see the
+		 * ResultRelInfo, so fetch one if not already done before executing
+		 * the subplan, potentially opening it for the first time.
+		 */
+		if (bms_is_member(node->mt_whichplan, plan->fdwDirectModifyPlans) &&
+			resultRelInfo == NULL)
+		{
+			resultRelInfo = ExecGetResultRelation(node, node->mt_whichplan);
+
+			/*
+			 * Must make sure to initialize the RETURNING projection as well,
+			 * because some FDWs rely on accessing ri_projectReturning to
+			 * set its "scan" tuple to use below for computing the actual
+			 * RETURNING targetlist.
+			 */
+			if (plan->returningLists && resultRelInfo->ri_returningList == NIL)
+				InitReturningProjection(node, resultRelInfo);
+		}
+
 		planSlot = ExecProcNode(subplanstate);
 
 		if (TupIsNull(planSlot))
 		{
-			/* advance to next subplan if any */
+			/* Signal to initialize the next plan's relation. */
+			resultRelInfo = NULL;
+
 			node->mt_whichplan++;
 			if (node->mt_whichplan < node->mt_nplans)
 			{
-				resultRelInfo++;
 				subplanstate = node->mt_plans[node->mt_whichplan];
-				junkfilter = resultRelInfo->ri_junkFilter;
 				EvalPlanQualSetPlan(&node->mt_epqstate, subplanstate->plan,
 									node->mt_arowmarks[node->mt_whichplan]);
 				continue;
@@ -2020,8 +2494,25 @@ ExecModifyTable(PlanState *pstate)
 		}
 
 		/*
+		 * Fetch the result relation for the current plan if not already done,
+		 * potentially opening it for the first time.
+		 */
+		if (resultRelInfo == NULL)
+		{
+			resultRelInfo = ExecGetResultRelation(node, node->mt_whichplan);
+			if (!resultRelInfo->ri_junkFilterValid)
+				InitJunkFilter(node, resultRelInfo);
+			junkfilter = resultRelInfo->ri_junkFilter;
+		}
+
+		/*
 		 * Ensure input tuple is the right format for the target relation.
 		 */
+		if (node->mt_scans[node->mt_whichplan] == NULL)
+			node->mt_scans[node->mt_whichplan] =
+				ExecInitExtraTupleSlot(node->ps.state,
+									   ExecGetResultType(subplanstate),
+									   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
 		if (node->mt_scans[node->mt_whichplan]->tts_ops != planSlot->tts_ops)
 		{
 			ExecCopySlot(node->mt_scans[node->mt_whichplan], planSlot);
@@ -2042,7 +2533,8 @@ ExecModifyTable(PlanState *pstate)
 			 * ExecProcessReturning by IterateDirectModify, so no need to
 			 * provide it here.
 			 */
-			slot = ExecProcessReturning(resultRelInfo, NULL, planSlot);
+			slot = ExecProcessReturning(node, resultRelInfo, NULL, NULL, NULL,
+										planSlot);
 
 			return slot;
 		}
@@ -2175,13 +2667,10 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	ModifyTableState *mtstate;
 	CmdType		operation = node->operation;
 	int			nplans = list_length(node->plans);
-	ResultRelInfo *resultRelInfo;
 	Plan	   *subplan;
-	ListCell   *l,
-			   *l1;
+	ListCell   *l;
 	int			i;
 	Relation	rel;
-	bool		update_tuple_routing_needed = node->partColsUpdated;
 
 	/* check for unsupported flags */
 	Assert(!(eflags & (EXEC_FLAG_BACKWARD | EXEC_FLAG_MARK)));
@@ -2198,7 +2687,20 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	mtstate->canSetTag = node->canSetTag;
 	mtstate->mt_done = false;
 
+	/*
+	 * call ExecInitNode on each of the plans to be executed and save the
+	 * results into the array "mt_plans".
+	 */
+	mtstate->mt_nplans = nplans;
 	mtstate->mt_plans = (PlanState **) palloc0(sizeof(PlanState *) * nplans);
+	i = 0;
+	foreach(l, node->plans)
+	{
+		subplan = (Plan *) lfirst(l);
+
+		mtstate->mt_plans[i++] = ExecInitNode(subplan, estate, eflags);
+	}
+
 	mtstate->resultRelInfo = (ResultRelInfo *)
 		palloc(nplans * sizeof(ResultRelInfo));
 	mtstate->mt_scans = (TupleTableSlot **) palloc0(sizeof(TupleTableSlot *) * nplans);
@@ -2225,13 +2727,17 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	}
 	else
 	{
-		mtstate->rootResultRelInfo = mtstate->resultRelInfo;
-		ExecInitResultRelation(estate, mtstate->resultRelInfo,
-							   linitial_int(node->resultRelations));
+		/*
+		 * Unlike a partitioned target relation, the target relation in this
+		 * case will be actually used by ExecModifyTable(), so use
+		 * ExecGetResultRelation() to get the ResultRelInfo, because it
+		 * initializes some fields that a bare InitResultRelInfo() doesn't.
+		 */
+		mtstate->rootResultRelInfo = ExecGetResultRelation(mtstate, 0);
+		Assert(mtstate->rootResultRelInfo == mtstate->resultRelInfo);
 	}
 
 	mtstate->mt_arowmarks = (List **) palloc0(sizeof(List *) * nplans);
-	mtstate->mt_nplans = nplans;
 
 	/* set up epqstate with dummy subplan data for the moment */
 	EvalPlanQualInit(&mtstate->mt_epqstate, estate, NULL, NIL, node->epqParam);
@@ -2244,177 +2750,9 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 	if (!(eflags & EXEC_FLAG_EXPLAIN_ONLY))
 		ExecSetupTransitionCaptureState(mtstate, estate);
 
-	/*
-	 * call ExecInitNode on each of the plans to be executed and save the
-	 * results into the array "mt_plans".  This is also a convenient place to
-	 * verify that the proposed target relations are valid and open their
-	 * indexes for insertion of new index entries.
-	 */
-	resultRelInfo = mtstate->resultRelInfo;
-	i = 0;
-	forboth(l, node->resultRelations, l1, node->plans)
-	{
-		Index		resultRelation = lfirst_int(l);
-
-		subplan = (Plan *) lfirst(l1);
-
-		/*
-		 * This opens result relation and fills ResultRelInfo. (root relation
-		 * was initialized already.)
-		 */
-		if (resultRelInfo != mtstate->rootResultRelInfo)
-			ExecInitResultRelation(estate, resultRelInfo, resultRelation);
-
-		/* Initialize the usesFdwDirectModify flag */
-		resultRelInfo->ri_usesFdwDirectModify = bms_is_member(i,
-															  node->fdwDirectModifyPlans);
-
-		/*
-		 * Verify result relation is a valid target for the current operation
-		 */
-		CheckValidResultRel(resultRelInfo, operation);
-
-		/*
-		 * If there are indices on the result relation, open them and save
-		 * descriptors in the result relation info, so that we can add new
-		 * index entries for the tuples we add/update.  We need not do this
-		 * for a DELETE, however, since deletion doesn't affect indexes. Also,
-		 * inside an EvalPlanQual operation, the indexes might be open
-		 * already, since we share the resultrel state with the original
-		 * query.
-		 */
-		if (resultRelInfo->ri_RelationDesc->rd_rel->relhasindex &&
-			operation != CMD_DELETE &&
-			resultRelInfo->ri_IndexRelationDescs == NULL)
-			ExecOpenIndices(resultRelInfo,
-							node->onConflictAction != ONCONFLICT_NONE);
-
-		/*
-		 * If this is an UPDATE and a BEFORE UPDATE trigger is present, the
-		 * trigger itself might modify the partition-key values. So arrange
-		 * for tuple routing.
-		 */
-		if (resultRelInfo->ri_TrigDesc &&
-			resultRelInfo->ri_TrigDesc->trig_update_before_row &&
-			operation == CMD_UPDATE)
-			update_tuple_routing_needed = true;
-
-		/* Now init the plan for this result rel */
-		mtstate->mt_plans[i] = ExecInitNode(subplan, estate, eflags);
-		mtstate->mt_scans[i] =
-			ExecInitExtraTupleSlot(mtstate->ps.state, ExecGetResultType(mtstate->mt_plans[i]),
-								   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
-
-		/* Also let FDWs init themselves for foreign-table result rels */
-		if (resultRelInfo->ri_FdwRoutine != NULL)
-		{
-			if (resultRelInfo->ri_usesFdwDirectModify)
-			{
-				ForeignScanState *fscan = (ForeignScanState *) mtstate->mt_plans[i];
-
-				/*
-				 * For the FDW's convenience, set the ForeignScanState node's
-				 * ResultRelInfo to let the FDW know which result relation it
-				 * is going to work with.
-				 */
-				Assert(IsA(fscan, ForeignScanState));
-				fscan->resultRelInfo = resultRelInfo;
-				resultRelInfo->ri_FdwRoutine->BeginDirectModify(fscan, eflags);
-			}
-			else if (resultRelInfo->ri_FdwRoutine->BeginForeignModify != NULL)
-			{
-				List   *fdw_private = (List *) list_nth(node->fdwPrivLists, i);
-
-				resultRelInfo->ri_FdwRoutine->BeginForeignModify(mtstate,
-																 resultRelInfo,
-																 fdw_private,
-																 i,
-																 eflags);
-			}
-		}
-
-		/*
-		 * If needed, initialize a map to convert tuples in the child format
-		 * to the format of the table mentioned in the query (root relation).
-		 * It's needed for update tuple routing, because the routing starts
-		 * from the root relation.  It's also needed for capturing transition
-		 * tuples, because the transition tuple store can only store tuples in
-		 * the root table format.
-		 *
-		 * For INSERT, the map is only initialized for a given partition when
-		 * the partition itself is first initialized by ExecFindPartition().
-		 */
-		if (update_tuple_routing_needed ||
-			(mtstate->mt_transition_capture &&
-			 mtstate->operation != CMD_INSERT))
-			resultRelInfo->ri_ChildToRootMap =
-				convert_tuples_by_name(RelationGetDescr(resultRelInfo->ri_RelationDesc),
-									   RelationGetDescr(mtstate->rootResultRelInfo->ri_RelationDesc));
-		resultRelInfo++;
-		i++;
-	}
-
-	/* Get the target relation */
-	rel = mtstate->rootResultRelInfo->ri_RelationDesc;
-
-	/*
-	 * If it's not a partitioned table after all, UPDATE tuple routing should
-	 * not be attempted.
-	 */
-	if (rel->rd_rel->relkind != RELKIND_PARTITIONED_TABLE)
-		update_tuple_routing_needed = false;
-
-	/*
-	 * Build state for tuple routing if it's an INSERT or if it's an UPDATE of
-	 * partition key.
-	 */
-	if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
-		(operation == CMD_INSERT || update_tuple_routing_needed))
-		mtstate->mt_partition_tuple_routing =
-			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
-
-	/*
-	 * For update row movement we'll need a dedicated slot to store the tuples
-	 * that have been converted from partition format to the root table
-	 * format.
-	 */
-	if (update_tuple_routing_needed)
-		mtstate->mt_root_tuple_slot = table_slot_create(rel, NULL);
-
-	/*
-	 * Initialize any WITH CHECK OPTION constraints if needed.
-	 */
-	resultRelInfo = mtstate->resultRelInfo;
-	i = 0;
-	foreach(l, node->withCheckOptionLists)
-	{
-		List	   *wcoList = (List *) lfirst(l);
-		List	   *wcoExprs = NIL;
-		ListCell   *ll;
-
-		foreach(ll, wcoList)
-		{
-			WithCheckOption *wco = (WithCheckOption *) lfirst(ll);
-			ExprState  *wcoExpr = ExecInitQual((List *) wco->qual,
-											   &mtstate->ps);
-
-			wcoExprs = lappend(wcoExprs, wcoExpr);
-		}
-
-		resultRelInfo->ri_WithCheckOptions = wcoList;
-		resultRelInfo->ri_WithCheckOptionExprs = wcoExprs;
-		resultRelInfo++;
-		i++;
-	}
-
-	/*
-	 * Initialize RETURNING projections if needed.
-	 */
+	/* Initialize some global state for RETURNING projections. */
 	if (node->returningLists)
 	{
-		TupleTableSlot *slot;
-		ExprContext *econtext;
-
 		/*
 		 * Initialize result tuple slot and assign its rowtype using the first
 		 * RETURNING list.  We assume the rest will look the same.
@@ -2423,27 +2761,10 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 
 		/* Set up a slot for the output of the RETURNING projection(s) */
 		ExecInitResultTupleSlotTL(&mtstate->ps, &TTSOpsVirtual);
-		slot = mtstate->ps.ps_ResultTupleSlot;
 
 		/* Need an econtext too */
 		if (mtstate->ps.ps_ExprContext == NULL)
 			ExecAssignExprContext(estate, &mtstate->ps);
-		econtext = mtstate->ps.ps_ExprContext;
-
-		/*
-		 * Build a projection for each result rel.
-		 */
-		resultRelInfo = mtstate->resultRelInfo;
-		foreach(l, node->returningLists)
-		{
-			List	   *rlist = (List *) lfirst(l);
-
-			resultRelInfo->ri_returningList = rlist;
-			resultRelInfo->ri_projectReturning =
-				ExecBuildProjectionInfo(rlist, econtext, slot, &mtstate->ps,
-										resultRelInfo->ri_RelationDesc->rd_att);
-			resultRelInfo++;
-		}
 	}
 	else
 	{
@@ -2457,67 +2778,18 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 		mtstate->ps.ps_ExprContext = NULL;
 	}
 
-	/* Set the list of arbiter indexes if needed for ON CONFLICT */
-	resultRelInfo = mtstate->resultRelInfo;
-	if (node->onConflictAction != ONCONFLICT_NONE)
-		resultRelInfo->ri_onConflictArbiterIndexes = node->arbiterIndexes;
+	/* Get the target relation */
+	rel = mtstate->rootResultRelInfo->ri_RelationDesc;
 
 	/*
-	 * If needed, Initialize target list, projection and qual for ON CONFLICT
-	 * DO UPDATE.
+	 * Build state for tuple routing if it's an INSERT.  An UPDATE might need
+	 * it too, but it's initialized only when it actually ends up moving
+	 * tuples between partitions; see ExecCrossPartitionUpdate().
 	 */
-	if (node->onConflictAction == ONCONFLICT_UPDATE)
-	{
-		ExprContext *econtext;
-		TupleDesc	relationDesc;
-		TupleDesc	tupDesc;
-
-		/* insert may only have one plan, inheritance is not expanded */
-		Assert(nplans == 1);
-
-		/* already exists if created by RETURNING processing above */
-		if (mtstate->ps.ps_ExprContext == NULL)
-			ExecAssignExprContext(estate, &mtstate->ps);
-
-		econtext = mtstate->ps.ps_ExprContext;
-		relationDesc = resultRelInfo->ri_RelationDesc->rd_att;
-
-		/* create state for DO UPDATE SET operation */
-		resultRelInfo->ri_onConflict = makeNode(OnConflictSetState);
-
-		/* initialize slot for the existing tuple */
-		resultRelInfo->ri_onConflict->oc_Existing =
-			table_slot_create(resultRelInfo->ri_RelationDesc,
-							  &mtstate->ps.state->es_tupleTable);
-
-		/*
-		 * Create the tuple slot for the UPDATE SET projection. We want a slot
-		 * of the table's type here, because the slot will be used to insert
-		 * into the table, and for RETURNING processing - which may access
-		 * system attributes.
-		 */
-		tupDesc = ExecTypeFromTL((List *) node->onConflictSet);
-		resultRelInfo->ri_onConflict->oc_ProjSlot =
-			ExecInitExtraTupleSlot(mtstate->ps.state, tupDesc,
-								   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
-
-		/* build UPDATE SET projection state */
-		resultRelInfo->ri_onConflict->oc_ProjInfo =
-			ExecBuildProjectionInfo(node->onConflictSet, econtext,
-									resultRelInfo->ri_onConflict->oc_ProjSlot,
-									&mtstate->ps,
-									relationDesc);
-
-		/* initialize state to evaluate the WHERE clause, if any */
-		if (node->onConflictWhere)
-		{
-			ExprState  *qualexpr;
-
-			qualexpr = ExecInitQual((List *) node->onConflictWhere,
-									&mtstate->ps);
-			resultRelInfo->ri_onConflict->oc_WhereClause = qualexpr;
-		}
-	}
+	if (rel->rd_rel->relkind == RELKIND_PARTITIONED_TABLE &&
+		operation == CMD_INSERT)
+		mtstate->mt_partition_tuple_routing =
+			ExecSetupPartitionTupleRouting(estate, mtstate, rel);
 
 	/*
 	 * If we have any secondary relations in an UPDATE or DELETE, they need to
@@ -2555,121 +2827,6 @@ ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags)
 						mtstate->mt_arowmarks[0]);
 
 	/*
-	 * Initialize the junk filter(s) if needed.  INSERT queries need a filter
-	 * if there are any junk attrs in the tlist.  UPDATE and DELETE always
-	 * need a filter, since there's always at least one junk attribute present
-	 * --- no need to look first.  Typically, this will be a 'ctid' or
-	 * 'wholerow' attribute, but in the case of a foreign data wrapper it
-	 * might be a set of junk attributes sufficient to identify the remote
-	 * row.
-	 *
-	 * If there are multiple result relations, each one needs its own junk
-	 * filter.  Note multiple rels are only possible for UPDATE/DELETE, so we
-	 * can't be fooled by some needing a filter and some not.
-	 *
-	 * This section of code is also a convenient place to verify that the
-	 * output of an INSERT or UPDATE matches the target table(s).
-	 */
-	{
-		bool		junk_filter_needed = false;
-
-		switch (operation)
-		{
-			case CMD_INSERT:
-				foreach(l, subplan->targetlist)
-				{
-					TargetEntry *tle = (TargetEntry *) lfirst(l);
-
-					if (tle->resjunk)
-					{
-						junk_filter_needed = true;
-						break;
-					}
-				}
-				break;
-			case CMD_UPDATE:
-			case CMD_DELETE:
-				junk_filter_needed = true;
-				break;
-			default:
-				elog(ERROR, "unknown operation");
-				break;
-		}
-
-		if (junk_filter_needed)
-		{
-			resultRelInfo = mtstate->resultRelInfo;
-			for (i = 0; i < nplans; i++)
-			{
-				JunkFilter *j;
-				TupleTableSlot *junkresslot;
-
-				subplan = mtstate->mt_plans[i]->plan;
-
-				junkresslot =
-					ExecInitExtraTupleSlot(estate, NULL,
-										   table_slot_callbacks(resultRelInfo->ri_RelationDesc));
-
-				/*
-				 * For an INSERT or UPDATE, the result tuple must always match
-				 * the target table's descriptor.  For a DELETE, it won't
-				 * (indeed, there's probably no non-junk output columns).
-				 */
-				if (operation == CMD_INSERT || operation == CMD_UPDATE)
-				{
-					ExecCheckPlanOutput(resultRelInfo->ri_RelationDesc,
-										subplan->targetlist);
-					j = ExecInitJunkFilterInsertion(subplan->targetlist,
-													RelationGetDescr(resultRelInfo->ri_RelationDesc),
-													junkresslot);
-				}
-				else
-					j = ExecInitJunkFilter(subplan->targetlist,
-										   junkresslot);
-
-				if (operation == CMD_UPDATE || operation == CMD_DELETE)
-				{
-					/* For UPDATE/DELETE, find the appropriate junk attr now */
-					char		relkind;
-
-					relkind = resultRelInfo->ri_RelationDesc->rd_rel->relkind;
-					if (relkind == RELKIND_RELATION ||
-						relkind == RELKIND_MATVIEW ||
-						relkind == RELKIND_PARTITIONED_TABLE)
-					{
-						j->jf_junkAttNo = ExecFindJunkAttribute(j, "ctid");
-						if (!AttributeNumberIsValid(j->jf_junkAttNo))
-							elog(ERROR, "could not find junk ctid column");
-					}
-					else if (relkind == RELKIND_FOREIGN_TABLE)
-					{
-						/*
-						 * When there is a row-level trigger, there should be
-						 * a wholerow attribute.
-						 */
-						j->jf_junkAttNo = ExecFindJunkAttribute(j, "wholerow");
-					}
-					else
-					{
-						j->jf_junkAttNo = ExecFindJunkAttribute(j, "wholerow");
-						if (!AttributeNumberIsValid(j->jf_junkAttNo))
-							elog(ERROR, "could not find junk wholerow column");
-					}
-				}
-
-				resultRelInfo->ri_junkFilter = j;
-				resultRelInfo++;
-			}
-		}
-		else
-		{
-			if (operation == CMD_INSERT)
-				ExecCheckPlanOutput(mtstate->resultRelInfo->ri_RelationDesc,
-									subplan->targetlist);
-		}
-	}
-
-	/*
 	 * Lastly, if this is not the primary (canSetTag) ModifyTable node, add it
 	 * to estate->es_auxmodifytables so that it will be run to completion by
 	 * ExecPostprocessPlan.  (It'd actually work fine to add the primary
@@ -2699,20 +2856,6 @@ ExecEndModifyTable(ModifyTableState *node)
 	int			i;
 
 	/*
-	 * Allow any FDWs to shut down
-	 */
-	for (i = 0; i < node->mt_nplans; i++)
-	{
-		ResultRelInfo *resultRelInfo = node->resultRelInfo + i;
-
-		if (!resultRelInfo->ri_usesFdwDirectModify &&
-			resultRelInfo->ri_FdwRoutine != NULL &&
-			resultRelInfo->ri_FdwRoutine->EndForeignModify != NULL)
-			resultRelInfo->ri_FdwRoutine->EndForeignModify(node->ps.state,
-														   resultRelInfo);
-	}
-
-	/*
 	 * Close all the partitioned tables, leaf partitions, and their indices
 	 * and release the slot used for tuple routing, if set.
 	 */
diff --git a/src/include/executor/nodeModifyTable.h b/src/include/executor/nodeModifyTable.h
index 46a2dc9..9ae7e40 100644
--- a/src/include/executor/nodeModifyTable.h
+++ b/src/include/executor/nodeModifyTable.h
@@ -22,5 +22,6 @@ extern void ExecComputeStoredGenerated(ResultRelInfo *resultRelInfo,
 extern ModifyTableState *ExecInitModifyTable(ModifyTable *node, EState *estate, int eflags);
 extern void ExecEndModifyTable(ModifyTableState *node);
 extern void ExecReScanModifyTable(ModifyTableState *node);
+extern ResultRelInfo *ExecGetResultRelation(ModifyTableState *mtstate, int whichrel);
 
 #endif							/* NODEMODIFYTABLE_H */
diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h
index 6c0a7d6..f2f4bed 100644
--- a/src/include/nodes/execnodes.h
+++ b/src/include/nodes/execnodes.h
@@ -463,6 +463,7 @@ typedef struct ResultRelInfo
 
 	/* for removing junk attributes from tuples */
 	JunkFilter *ri_junkFilter;
+	bool		ri_junkFilterValid;	/* has the filter been initialized? */
 
 	/* list of RETURNING expressions */
 	List	   *ri_returningList;
@@ -497,6 +498,7 @@ typedef struct ResultRelInfo
 	 * transition tuple capture or update partition row movement is active.
 	 */
 	TupleConversionMap *ri_ChildToRootMap;
+	bool		ri_ChildToRootMapValid;	/* has the map been initialized? */
 
 	/* for use by copy.c when performing multi-inserts */
 	struct CopyMultiInsertBuffer *ri_CopyMultiInsertBuffer;
diff --git a/src/test/regress/expected/insert_conflict.out b/src/test/regress/expected/insert_conflict.out
index ff157ce..74cd7e2 100644
--- a/src/test/regress/expected/insert_conflict.out
+++ b/src/test/regress/expected/insert_conflict.out
@@ -52,10 +52,7 @@ explain (costs off) insert into insertconflicttest values(0, 'Crowberry') on con
    Conflict Arbiter Indexes: op_index_key, collation_index_key, both_index_key
    Conflict Filter: (SubPlan 1)
    ->  Result
-   SubPlan 1
-     ->  Index Only Scan using both_index_expr_key on insertconflicttest ii
-           Index Cond: (key = excluded.key)
-(8 rows)
+(5 rows)
 
 -- Neither collation nor operator class specifications are required --
 -- supplying them merely *limits* matches to indexes with matching opclasses
diff --git a/src/test/regress/expected/updatable_views.out b/src/test/regress/expected/updatable_views.out
index caed1c1..d8d2a3d 100644
--- a/src/test/regress/expected/updatable_views.out
+++ b/src/test/regress/expected/updatable_views.out
@@ -1862,28 +1862,22 @@ UPDATE rw_view1 SET a = a + 5; -- should fail
 ERROR:  new row violates check option for view "rw_view1"
 DETAIL:  Failing row contains (15).
 EXPLAIN (costs off) INSERT INTO rw_view1 VALUES (5);
-                       QUERY PLAN                        
----------------------------------------------------------
+      QUERY PLAN      
+----------------------
  Insert on base_tbl b
    ->  Result
-   SubPlan 1
-     ->  Index Only Scan using ref_tbl_pkey on ref_tbl r
-           Index Cond: (a = b.a)
-(5 rows)
+(2 rows)
 
 EXPLAIN (costs off) UPDATE rw_view1 SET a = a + 5;
-                        QUERY PLAN                         
------------------------------------------------------------
+               QUERY PLAN                
+-----------------------------------------
  Update on base_tbl b
    ->  Hash Join
          Hash Cond: (b.a = r.a)
          ->  Seq Scan on base_tbl b
          ->  Hash
                ->  Seq Scan on ref_tbl r
-   SubPlan 1
-     ->  Index Only Scan using ref_tbl_pkey on ref_tbl r_1
-           Index Cond: (a = b.a)
-(9 rows)
+(6 rows)
 
 DROP TABLE base_tbl, ref_tbl CASCADE;
 NOTICE:  drop cascades to view rw_view1
diff --git a/src/test/regress/expected/update.out b/src/test/regress/expected/update.out
index bf939d7..0ad0d1a 100644
--- a/src/test/regress/expected/update.out
+++ b/src/test/regress/expected/update.out
@@ -341,8 +341,8 @@ DETAIL:  Failing row contains (105, 85, null, b, 15).
 -- fail, no partition key update, so no attempt to move tuple,
 -- but "a = 'a'" violates partition constraint enforced by root partition)
 UPDATE part_b_10_b_20 set a = 'a';
-ERROR:  new row for relation "part_c_1_100" violates partition constraint
-DETAIL:  Failing row contains (null, 1, 96, 12, a).
+ERROR:  new row for relation "part_b_10_b_20" violates partition constraint
+DETAIL:  Failing row contains (null, 96, a, 12, 1).
 -- ok, partition key update, no constraint violation
 UPDATE range_parted set d = d - 10 WHERE d > 10;
 -- ok, no partition key update, no constraint violation
@@ -372,8 +372,8 @@ UPDATE part_b_10_b_20 set c = c + 20 returning c, b, a;
 
 -- fail, row movement happens only within the partition subtree.
 UPDATE part_b_10_b_20 set b = b - 6 WHERE c > 116 returning *;
-ERROR:  new row for relation "part_d_1_15" violates partition constraint
-DETAIL:  Failing row contains (2, 117, 2, b, 7).
+ERROR:  new row for relation "part_b_10_b_20" violates partition constraint
+DETAIL:  Failing row contains (2, 117, b, 7, 2).
 -- ok, row movement, with subset of rows moved into different partition.
 UPDATE range_parted set b = b - 6 WHERE c > 116 returning a, b + c;
  a | ?column? 
@@ -814,8 +814,8 @@ INSERT into sub_parted VALUES (1,2,10);
 -- Test partition constraint violation when intermediate ancestor is used and
 -- constraint is inherited from upper root.
 UPDATE sub_parted set a = 2 WHERE c = 10;
-ERROR:  new row for relation "sub_part2" violates partition constraint
-DETAIL:  Failing row contains (2, 10, 2).
+ERROR:  new row for relation "sub_parted" violates partition constraint
+DETAIL:  Failing row contains (2, 2, 10).
 -- Test update-partition-key, where the unpruned partitions do not have their
 -- partition keys updated.
 SELECT tableoid::regclass::text, * FROM list_parted WHERE a = 2 ORDER BY 1;
-- 
1.8.3.1