Repair rare failure of MULTIEXPR_SUBLINK subplans in inherited updates.

Prior to v14, if we have a MULTIEXPR SubPlan (that is, use of the syntax
UPDATE ... SET (c1, ...) = (SELECT ...)) in an UPDATE with an inherited
or partitioned target table, inheritance_planner() will clone the
targetlist and therefore also the MULTIEXPR SubPlan and the Param nodes
referencing it for each child target table.  Up to now, we've allowed
all the clones to share the underlying subplan as well as the output
parameter IDs -- that is, the runtime ParamExecData slots.  That
technique is borrowed from the far older code that supports initplans,
and it works okay in that case because the cloned SubPlan nodes are
essentially identical.  So it doesn't matter which one of the clones
the shared ParamExecData.execPlan field might point to.

However, this fails to hold for MULTIEXPR SubPlans, because they can
have nonempty "args" lists (values to be passed into the subplan), and
those lists could get mutated to different states in the various clones.
In the submitted reproducer, as well as the test case added here, one
clone contains Vars with varno OUTER_VAR where another has INNER_VAR,
because the child tables are respectively on the outer or inner side of
the join.  Sharing the execPlan pointer can result in trying to evaluate
an args list that doesn't match the local execution state, with mayhem
ensuing.  The result often is to trigger consistency checks in the
executor, but I believe this could end in a crash or incorrect updates.

To fix, assign new Param IDs to each of the cloned SubPlans, so that
they don't share ParamExecData slots at runtime.  It still seems fine
for the clones to share the underlying subplan, and extra ParamExecData
slots are cheap enough that this fix shouldn't cost much.

This has been busted since we invented MULTIEXPR SubPlans in 9.5.
Probably the lack of previous reports is because query plans in which
the different clones of a MULTIEXPR mutate to effectively-different
states are pretty rare.  There's no issue in v14 and later, because
without inheritance_planner() there's never a reason to clone
MULTIEXPR SubPlans.

Per bug #17596 from Andre Lin.  Patch v10-v13 only.

Discussion: https://postgr.es/m/17596-c5357f61427a81dc@postgresql.org

1 files changed, 95 insertions, 0 deletions

diff --git a/src/backend/optimizer/plan/subselect.c b/src/backend/optimizer/plan/subselect.c
index 528e2a36244..4c6f84a6755 100644
--- a/src/backend/optimizer/plan/subselect.c
+++ b/src/backend/optimizer/plan/subselect.c
@@ -851,6 +851,101 @@ hash_ok_operator(OpExpr *expr)
 	}
 }
 
+/*
+ * SS_make_multiexprs_unique
+ *
+ * After cloning an UPDATE targetlist that contains MULTIEXPR_SUBLINK
+ * SubPlans, inheritance_planner() must call this to assign new, unique Param
+ * IDs to the cloned MULTIEXPR_SUBLINKs' output parameters.  See notes in
+ * ExecScanSubPlan.
+ */
+void
+SS_make_multiexprs_unique(PlannerInfo *root, PlannerInfo *subroot)
+{
+	List	   *new_multiexpr_params = NIL;
+	int			offset;
+	ListCell   *lc;
+
+	/*
+	 * Find MULTIEXPR SubPlans in the cloned query.  We need only look at the
+	 * top level of the targetlist.
+	 */
+	foreach(lc, subroot->parse->targetList)
+	{
+		TargetEntry *tent = (TargetEntry *) lfirst(lc);
+		SubPlan    *splan;
+		Plan	   *plan;
+		List	   *params;
+
+		if (!IsA(tent->expr, SubPlan))
+			continue;
+		splan = (SubPlan *) tent->expr;
+		if (splan->subLinkType != MULTIEXPR_SUBLINK)
+			continue;
+
+		/* Found one, get the associated subplan */
+		plan = (Plan *) list_nth(root->glob->subplans, splan->plan_id - 1);
+
+		/*
+		 * Generate new PARAM_EXEC Param nodes, and overwrite splan->setParam
+		 * with their IDs.  This is just like what build_subplan did when it
+		 * made the SubPlan node we're cloning.  But because the param IDs are
+		 * assigned globally, we'll get new IDs.  (We assume here that the
+		 * subroot's tlist is a clone we can scribble on.)
+		 */
+		params = generate_subquery_params(root,
+										  plan->targetlist,
+										  &splan->setParam);
+
+		/*
+		 * We will append the replacement-Params lists to
+		 * root->multiexpr_params, but for the moment just make a local list.
+		 * Since we lack easy access here to the original subLinkId, we have
+		 * to fall back on the slightly shaky assumption that the MULTIEXPR
+		 * SubPlans appear in the targetlist in subLinkId order.  This should
+		 * be safe enough given the way that the parser builds the targetlist
+		 * today.  I wouldn't want to rely on it going forward, but since this
+		 * code has a limited lifespan it should be fine.  We can partially
+		 * protect against problems with assertions below.
+		 */
+		new_multiexpr_params = lappend(new_multiexpr_params, params);
+	}
+
+	/*
+	 * Now we must find the Param nodes that reference the MULTIEXPR outputs
+	 * and update their sublink IDs so they'll reference the new outputs.
+	 * Fortunately, those too must be at top level of the cloned targetlist.
+	 */
+	offset = list_length(root->multiexpr_params);
+
+	foreach(lc, subroot->parse->targetList)
+	{
+		TargetEntry *tent = (TargetEntry *) lfirst(lc);
+		Param	   *p;
+		int			subqueryid;
+		int			colno;
+
+		if (!IsA(tent->expr, Param))
+			continue;
+		p = (Param *) tent->expr;
+		if (p->paramkind != PARAM_MULTIEXPR)
+			continue;
+		subqueryid = p->paramid >> 16;
+		colno = p->paramid & 0xFFFF;
+		Assert(subqueryid > 0 &&
+			   subqueryid <= list_length(new_multiexpr_params));
+		Assert(colno > 0 &&
+			   colno <= list_length((List *) list_nth(new_multiexpr_params,
+													  subqueryid - 1)));
+		subqueryid += offset;
+		p->paramid = (subqueryid << 16) + colno;
+	}
+
+	/* Finally, attach new replacement lists to the global list */
+	root->multiexpr_params = list_concat(root->multiexpr_params,
+										 new_multiexpr_params);
+}
+
 
 /*
  * SS_process_ctes: process a query's WITH list