Don't allow LIMIT/OFFSET clause within sub-selects to be pushed to workers.

Allowing sub-select containing LIMIT/OFFSET in workers can lead to
inconsistent results at the top-level as there is no guarantee that the
row order will be fully deterministic.  The fix is to prohibit pushing
LIMIT/OFFSET within sub-selects to workers.

Reported-by: Andrew Fletcher
Bug: 15324
Author: Amit Kapila
Reviewed-by: Dilip Kumar
Backpatch-through: 9.6
Discussion: https://postgr.es/m/153417684333.10284.11356259990921828616@wrigleys.postgresql.org

Author: Amit Kapila

src/backend/optimizer/path/allpaths.c

@@ -585,7 +585,20 @@ set_rel_consider_parallel(PlannerInfo *root, RelOptInfo *rel,
 			 * the SubqueryScanPath as not parallel-safe.  (Note that
 			 * set_subquery_pathlist() might push some of these quals down
 			 * into the subquery itself, but that doesn't change anything.)
+			 *
+			 * We can't push sub-select containing LIMIT/OFFSET to workers as
+			 * there is no guarantee that the row order will be fully
+			 * deterministic, and applying LIMIT/OFFSET will lead to
+			 * inconsistent results at the top-level.  (In some cases, where
+			 * the result is ordered, we could relax this restriction.  But it
+			 * doesn't currently seem worth expending extra effort to do so.)
 			 */
+			{
+				Query	   *subquery = castNode(Query, rte->subquery);
+
+				if (limit_needed(subquery))
+					return;
+			}
 			break;
 
 		case RTE_JOIN:

src/backend/optimizer/plan/planner.c

@@ -98,7 +98,6 @@ static void preprocess_rowmarks(PlannerInfo *root);
 static double preprocess_limit(PlannerInfo *root,
 				 double tuple_fraction,
 				 int64 *offset_est, int64 *count_est);
-static bool limit_needed(Query *parse);
 static void remove_useless_groupby_columns(PlannerInfo *root);
 static List *preprocess_groupclause(PlannerInfo *root, List *force);
 static List *extract_rollup_sets(List *groupingSets);
@@ -2492,7 +2491,7 @@ preprocess_limit(PlannerInfo *root, double tuple_fraction,
  * a key distinction: here we need hard constants in OFFSET/LIMIT, whereas
  * in preprocess_limit it's good enough to consider estimated values.
  */
-static bool
+bool
 limit_needed(Query *parse)
 {
 	Node	   *node;

src/include/optimizer/planner.h

@@ -46,6 +46,8 @@ extern bool is_dummy_plan(Plan *plan);
 extern RowMarkType select_rowmark_type(RangeTblEntry *rte,
 					LockClauseStrength strength);
 
+extern bool limit_needed(Query *parse);
+
 extern void mark_partial_aggref(Aggref *agg, AggSplit aggsplit);
 
 extern Path *get_cheapest_fractional_path(RelOptInfo *rel,

src/test/regress/expected/select_parallel.out

@@ -205,6 +205,25 @@ explain (costs off, verbose)
                            Output: b.unique1
 (15 rows)
 
+-- LIMIT/OFFSET within sub-selects can't be pushed to workers.
+explain (costs off)
+  select * from tenk1 a where two in
+    (select two from tenk1 b where stringu1 like '%AAAA' limit 3);
+                          QUERY PLAN                           
+---------------------------------------------------------------
+ Hash Semi Join
+   Hash Cond: (a.two = b.two)
+   ->  Gather
+         Workers Planned: 4
+         ->  Parallel Seq Scan on tenk1 a
+   ->  Hash
+         ->  Limit
+               ->  Gather
+                     Workers Planned: 4
+                     ->  Parallel Seq Scan on tenk1 b
+                           Filter: (stringu1 ~~ '%AAAA'::text)
+(11 rows)
+
 explain (costs off)
   select stringu1::int2 from tenk1 where unique1 = 1;
                   QUERY PLAN                   

src/test/regress/sql/select_parallel.sql

@@ -88,6 +88,11 @@ explain (costs off, verbose)
   select count(*) from tenk1 a where (unique1, two) in
     (select unique1, row_number() over() from tenk1 b);
 
+-- LIMIT/OFFSET within sub-selects can't be pushed to workers.
+explain (costs off)
+  select * from tenk1 a where two in
+    (select two from tenk1 b where stringu1 like '%AAAA' limit 3);
+
 explain (costs off)
   select stringu1::int2 from tenk1 where unique1 = 1;