Fix ORCA EXISTS subquery planning with LIMIT/OFFSET

src/backend/gporca/libgpopt/include/gpopt/xforms/CXformApply2Join.h

@@ -49,6 +49,21 @@ class CXformApply2Join : public CXformExploration
 
 		COperator::EOperatorId op_id = pexprApply->Pop()->Eopid();
 
+		// An outer Apply introduced below a grouping operation for EXISTS or
+		// a quantified subquery can have a multi-row inner input. If
+		// decorrelation fails, it cannot be implemented as a scalar SubPlan:
+		// the grouping above it does not enforce scalar cardinality below it.
+		// Keep scalar correlated execution only when the inner input is
+		// proven to return at most one row. SubqueryUnnest also generates a
+		// correlated alternative directly from the original subquery.
+		if (COperator::EopLogicalLeftOuterApply == op_id &&
+			COperator::EopScalarSubquery ==
+				CLogicalApply::PopConvert(pexprApply->Pop())->EopidOriginSubq() &&
+			1 < (*pexprApply)[1]->DeriveMaxCard().Ull())
+		{
+			return false;
+		}
+
 		// consider only Inner/Outer/Left (Anti) Semi Apply here,
 		// correlated left anti semi apply (with ALL/NOT-IN semantics) can only be generated by SubqueryHandler
 		return COperator::EopLogicalInnerApply == op_id ||

src/backend/optimizer/plan/orca.c

@@ -34,6 +34,7 @@
 #include "optimizer/planner.h"
 #include "optimizer/tlist.h"
 #include "optimizer/transform.h"
+#include "optimizer/walkers.h"
 #include "parser/parse_collate.h"
 #include "parser/parsetree.h"
 #include "rewrite/rewriteManip.h"
@@ -52,6 +53,7 @@ extern PlannedStmt * GPOPTOptimizedPlan(Query *parse, bool *had_unexpected_failu
 static Plan *remove_redundant_results(PlannerInfo *root, Plan *plan);
 static Node *remove_redundant_results_mutator(Node *node, void *);
 static bool can_replace_tlist(Plan *plan);
+static bool preserves_subplans(List *old_tlist, List *new_tlist);
 static Node *push_down_expr_mutator(Node *node, List *child_tlist);
 
 /*
@@ -453,11 +455,12 @@ remove_redundant_results_mutator(Node *node, void *ctx)
 			!expression_returns_set((Node *) result_plan->plan.targetlist) &&
 			can_replace_tlist(child_plan))
 		{
-			List	   *tlist = result_plan->plan.targetlist;
+			List	   *tlist = copyObject(result_plan->plan.targetlist);
 			ListCell   *lc;
 
 			child_plan = (Plan *)
 				remove_redundant_results_mutator((Node *) child_plan, ctx);
+			result_plan->plan.lefttree = child_plan;
 
 			foreach(lc, tlist)
 			{
@@ -467,6 +470,24 @@ remove_redundant_results_mutator(Node *node, void *ctx)
 															child_plan->targetlist);
 			}
 
+			/*
+			 * The slice table has already been built. Dropping a SubPlan
+			 * reference here can leave its Motion slices without an executable
+			 * plan tree. Keep the Result unless all child SubPlans survive.
+			 */
+			if (!preserves_subplans(child_plan->targetlist, tlist))
+			{
+				/*
+				 * The child was already processed above. Process the Result's
+				 * target list here so that SubPlans retained by the Result are
+				 * processed too, without descending into the child a second time.
+				 */
+				result_plan->plan.targetlist = (List *)
+					remove_redundant_results_mutator(
+						(Node *) result_plan->plan.targetlist, ctx);
+				return node;
+			}
+
 			child_plan->targetlist = tlist;
 			child_plan->flow = result_plan->plan.flow;
 
@@ -480,6 +501,34 @@ remove_redundant_results_mutator(Node *node, void *ctx)
 							 true);
 }
 
+/* Allow projection folding when it retains every referenced SubPlan. */
+static bool
+preserves_subplans(List *old_tlist, List *new_tlist)
+{
+	List	   *old_subplans;
+	List	   *new_subplans;
+	ListCell   *lc;
+	Bitmapset  *remaining = NULL;
+	bool		preserved;
+
+	if (!contain_subplans((Node *) old_tlist))
+		return true;
+
+	old_subplans = extract_nodes_expression((Node *) old_tlist, T_SubPlan, false);
+	new_subplans = new_tlist ?
+		extract_nodes_expression((Node *) new_tlist, T_SubPlan, false) : NIL;
+	foreach(lc, old_subplans)
+		remaining = bms_add_member(remaining, ((SubPlan *) lfirst(lc))->plan_id);
+	foreach(lc, new_subplans)
+		remaining = bms_del_member(remaining, ((SubPlan *) lfirst(lc))->plan_id);
+
+	preserved = bms_is_empty(remaining);
+	bms_free(remaining);
+	list_free(old_subplans);
+	list_free(new_subplans);
+	return preserved;
+}
+
 /*
  * Can the target list of a Plan node safely be replaced?
  */
@@ -541,20 +590,23 @@ push_down_expr_mutator(Node *node, List *child_tlist)
 		{
 			TargetEntry *child_tle = (TargetEntry *)
 				list_nth(child_tlist, var->varattno - 1);
+			Expr	   *expr = copyObject(child_tle->expr);
+
+			/* Do not change the child if the proposed replacement is rejected. */
 			// The const expr pertatining to a column in a child result node
 			// has const.consttypmod set as default value.
 			// correct typmod can be found at var.vartypmod.
 			// const.consttypmod value needs to be fixed before replacing var with const.
-			if (IsA(child_tle->expr, Const))
+			if (IsA(expr, Const))
 			{
-				((Const *) child_tle->expr)->consttypmod = ((Var *) node)->vartypmod;
+				((Const *) expr)->consttypmod = var->vartypmod;
 			}
-			else if (IsA(child_tle->expr, Var))
+			else if (IsA(expr, Var))
 			{
-				((Var *) child_tle->expr)->vartypmod = ((Var *) node)->vartypmod;
+				((Var *) expr)->vartypmod = var->vartypmod;
 			}
 
-			return (Node *) child_tle->expr;
+			return (Node *) expr;
 		}
 	}
 	return expression_tree_mutator(node, push_down_expr_mutator, child_tlist);