Make 'rows' estimate more accurate for plans that fetch only a few rows.

In commit c5f6dbbe, we changed the row and cost estimates on plan nodes
to represent per-segment costs. That made some estimates worse, because
the effects of the estimate "clamping" compounds. Per my comment on the
PR back then:

> One interesting effect of this change, that explains many of the
> plan changes: If you have a table with very few rows, or e.g. a qual
> like id = 123 that matches exactly one row, the Seq/Index Scan on it
> will be marked with rows=1. It now means that we estimate that every
> segment returns one row, although in reality, only one of them will
> return a row, and the rest will return nothing. That's because the
> row count estimates are "clamped" in the planner to at least
> 1. That's not a big deal on its own, but if you then have e.g. a
> Gather Motion on top of the Scan, the planner will estimate that the
> Gather Motion returns as many rows as there are segments. If you
> have e.g. 100 segments, that's relatively a big discrepancy, with
> 100 rows vs 1. I don't think that's a big problem in practice, I
> don't think most plans are very sensitive to that kind of a
> misestimate. What do you think?
>
> If we wanted to fix that, perhaps we should stop "clamping" the
> estimates to 1. I don't think there's any fundamental reason we need
> to do it. Perhaps clamp down to 1 / numsegments instead.

But I came up with a less intrusive idea, implemented in this commit:
Most Motion nodes have a "parent" RelOptInfo, and the RelOptInfo
contains an estimate of the total number of rows, before dividing it
with the number of segments or clamping. So if the row estimate we get
from the subpath seems clamped to 1.0, we look at the row estimate on
the underlying RelOptInfo instead, and use that if it's smaller. That
makes the row count estimates better for plans that fetch a single row
or a few rows, same as they were before commit c5f6dbbe. Not all
RelOptInfos have a row count estimate, and the subpaths estimate is
more accurate if the number of rows produced by the path differs from
the number of rows in the underlying relation, e.g.  because of a
ProjectSet node, so we still prefer the subpath's estimate if it
doesn't seem clamped.
Reviewed-by: NZhenghua Lyu <zlv@pivotal.io>

src/backend/cdb/cdbpath.c

@@ -77,18 +77,50 @@ cdbpath_cost_motion(PlannerInfo *root, CdbMotionPath *motionpath)
 	double		sendrows;
 	double		send_segments;
 	double		recv_segments;
+	double		total_rows;
+
+	if (CdbPathLocus_IsPartitioned(motionpath->path.locus))
+		recv_segments = CdbPathLocus_NumSegments(motionpath->path.locus);
+	else
+		recv_segments = 1;
 
 	if (CdbPathLocus_IsPartitioned(subpath->locus))
 		send_segments = CdbPathLocus_NumSegments(subpath->locus);
 	else
 		send_segments = 1;
 
-	if (CdbPathLocus_IsPartitioned(motionpath->path.locus))
-		recv_segments = CdbPathLocus_NumSegments(motionpath->path.locus);
-	else
-		recv_segments = 1;
+	/*
+	 * Estimate the total number of rows being sent.
+	 *
+	 * The base estimate is computed by multiplying the subpath's rows with
+	 * the number of sending segments. But in some cases, that leads to too
+	 * large estimates, if the subpath's estimate was "clamped" to 1 row. The
+	 * typical example is a single-row select like "SELECT * FROM table WHERE
+	 * key = 123. The Scan on the table returns only one row, on one segment,
+	 * and the estimate on the Scan node is 1 row. If you have e.g. 3
+	 * segments, and we just multiplied the subpath's row estimate by 3, we
+	 * would estimate that the Gather returns 3 rows, even though there is
+	 * only one matching row in the table. Using the 'rows' estimate on the
+	 * RelOptInfo is more accurate in such cases. To correct that, if the
+	 * subpath's estimate is 1 row, but the underlying relation's estimate is
+	 * smaller, use the underlying relation's estimate.
+	 *
+	 * We don't always use the relation's estimate, because there might be
+	 * nodes like ProjectSet or Limit in the subpath, in which case the
+	 * subpath's estimate is more accurate. Also, the relation might not have
+	 * a valid 'rows' estimate; upper rels, for example, do not. So check for
+	 * that too.
+	 */
+	total_rows = subpath->rows * send_segments;
+	if (subpath->rows == 1.0 &&
+		motionpath->path.parent->rows > 0 &&
+		motionpath->path.parent->rows < total_rows)
+	{
+		/* use the RelOptInfo's estimate */
+		total_rows = motionpath->path.parent->rows;
+	}
 
-	motionpath->path.rows = clamp_row_est(subpath->rows * (send_segments / recv_segments));
+	motionpath->path.rows = clamp_row_est(total_rows / recv_segments);
 
 	cost_per_row = (gp_motion_cost_per_row > 0.0)
 		? gp_motion_cost_per_row

src/test/fsync/expected/bgwriter_checkpoint.out

@@ -19,14 +19,20 @@
 -- s/num times hit:\'[4-9]\'/num times hit:\'greater_than_two\'/
 -- end_matchsubs
 begin;
-create function num_dirty(relid oid) returns bigint as
+create function num_dirty_on_qes(relid oid) returns setof bigint as
 $$
-   select count(*) from dirty_buffers()
+declare
+  rfnode oid;
+  result int4;
+begin
+   select relfilenode into rfnode from pg_class where oid=$1;
+
+   select count(*) into result from dirty_buffers_on_qes()
      as (tablespace oid, database oid, relfilenode oid, block int)
-     where relfilenode in
-           (select relfilenode from pg_class
-            where oid=$1);
-$$ language sql;
+     where relfilenode = rfnode;
+   return next result;
+end
+$$ language plpgsql execute on all segments;
 -- Wait until number of dirty buffers for the specified relfiles drops
 -- to 0 or timeout occurs.  Returns false if timeout occurred.
 create function wait_for_bgwriter(
@@ -40,7 +46,7 @@ declare
 begin
    i := 0;
    loop
-      select sum(num_dirty($1)) into d from gp_dist_random('gp_id');
+      select sum(nd) into d from num_dirty_on_qes($1) nd;
       if (d = 0) then
          return true;
       end if;
@@ -103,16 +109,16 @@ from gp_segment_configuration where role = 'p' and content > -1;
 (3 rows)
 
 -- Ensure no buffers are dirty before we start.
-select * from dirty_buffers()
+select * from dirty_buffers_on_qd()
  as (tablespace oid, database oid, relfilenode oid, block int);
  tablespace | database | relfilenode | block 
 ------------+----------+-------------+-------
 (0 rows)
 
-select dirty_buffers() from gp_dist_random('gp_id')
+select * from dirty_buffers_on_qes()
  as (tablespace oid, database oid, relfilenode oid, block int);
- dirty_buffers 
----------------
+ tablespace | database | relfilenode | block 
+------------+----------+-------------+-------
 (0 rows)
 
 -- Make buffers dirty.  At least two relfiles must be sync'ed during
@@ -120,15 +126,13 @@ select dirty_buffers() from gp_dist_random('gp_id')
 insert into fsync_test1 select i, i from generate_series(1,1000)i;
 delete from fsync_test2;
 -- Should return at least one dirty buffer.
-select sum(num_dirty('fsync_test1'::regclass)) > 0 as passed
- from gp_dist_random('gp_id');
+select sum(nd) > 0 as passed from num_dirty_on_qes('fsync_test1'::regclass) nd;
  passed 
 --------
  t
 (1 row)
 
-select sum(num_dirty('fsync_test2'::regclass)) > 0 as passed
- from gp_dist_random('gp_id');
+select sum(nd) > 0 as passed from num_dirty_on_qes('fsync_test2'::regclass) nd;
  passed 
 --------
  t
@@ -192,10 +196,10 @@ from gp_segment_configuration where role = 'p' and content > -1;
 
 checkpoint;
 -- There should be no dirty buffers after checkpoint.
-select dirty_buffers() from gp_dist_random('gp_id')
+select * from dirty_buffers_on_qes()
  as (tablespace oid, database oid, relfilenode oid, block int);
- dirty_buffers 
----------------
+ tablespace | database | relfilenode | block 
+------------+----------+-------------+-------
 (0 rows)
 
 -- Validate that the number of files fsync'ed by checkpointer is at

src/test/fsync/input/setup.source

-create or replace function dirty_buffers() returns setof record as
+create or replace function dirty_buffers_on_qd() returns setof record as
  '@abs_builddir@/fsync_helper@DLSUFFIX@', 'dirty_buffers'
-  LANGUAGE C VOLATILE STRICT NO SQL;
+  LANGUAGE C VOLATILE STRICT NO SQL EXECUTE ON MASTER;
+
+create or replace function dirty_buffers_on_qes() returns setof record as
+ '@abs_builddir@/fsync_helper@DLSUFFIX@', 'dirty_buffers'
+  LANGUAGE C VOLATILE STRICT NO SQL EXECUTE ON ALL SEGMENTS;

src/test/fsync/output/setup.source

-create or replace function dirty_buffers() returns setof record as
+create or replace function dirty_buffers_on_qd() returns setof record as
  '@abs_builddir@/fsync_helper@DLSUFFIX@', 'dirty_buffers'
-  LANGUAGE C VOLATILE STRICT NO SQL;
+  LANGUAGE C VOLATILE STRICT NO SQL EXECUTE ON MASTER;
+create or replace function dirty_buffers_on_qes() returns setof record as
+ '@abs_builddir@/fsync_helper@DLSUFFIX@', 'dirty_buffers'
+  LANGUAGE C VOLATILE STRICT NO SQL EXECUTE ON ALL SEGMENTS;

src/test/fsync/sql/bgwriter_checkpoint.sql

@@ -19,14 +19,20 @@
 -- s/num times hit:\'[4-9]\'/num times hit:\'greater_than_two\'/
 -- end_matchsubs
 begin;
-create function num_dirty(relid oid) returns bigint as
+create function num_dirty_on_qes(relid oid) returns setof bigint as
 $$
-   select count(*) from dirty_buffers()
+declare
+  rfnode oid;
+  result int4;
+begin
+   select relfilenode into rfnode from pg_class where oid=$1;
+
+   select count(*) into result from dirty_buffers_on_qes()
      as (tablespace oid, database oid, relfilenode oid, block int)
-     where relfilenode in
-           (select relfilenode from pg_class
-            where oid=$1);
-$$ language sql;
+     where relfilenode = rfnode;
+   return next result;
+end
+$$ language plpgsql execute on all segments;
 
 -- Wait until number of dirty buffers for the specified relfiles drops
 -- to 0 or timeout occurs.  Returns false if timeout occurred.
@@ -41,7 +47,7 @@ declare
 begin
    i := 0;
    loop
-      select sum(num_dirty($1)) into d from gp_dist_random('gp_id');
+      select sum(nd) into d from num_dirty_on_qes($1) nd;
       if (d = 0) then
          return true;
       end if;
@@ -84,9 +90,9 @@ select gp_inject_fault_infinite('fault_in_background_writer_main', 'suspend', db
 from gp_segment_configuration where role = 'p' and content > -1;
 
 -- Ensure no buffers are dirty before we start.
-select * from dirty_buffers()
+select * from dirty_buffers_on_qd()
  as (tablespace oid, database oid, relfilenode oid, block int);
-select dirty_buffers() from gp_dist_random('gp_id')
+select * from dirty_buffers_on_qes()
  as (tablespace oid, database oid, relfilenode oid, block int);
 
 -- Make buffers dirty.  At least two relfiles must be sync'ed during
@@ -95,10 +101,8 @@ insert into fsync_test1 select i, i from generate_series(1,1000)i;
 delete from fsync_test2;
 
 -- Should return at least one dirty buffer.
-select sum(num_dirty('fsync_test1'::regclass)) > 0 as passed
- from gp_dist_random('gp_id');
-select sum(num_dirty('fsync_test2'::regclass)) > 0 as passed
- from gp_dist_random('gp_id');
+select sum(nd) > 0 as passed from num_dirty_on_qes('fsync_test1'::regclass) nd;
+select sum(nd) > 0 as passed from num_dirty_on_qes('fsync_test2'::regclass) nd;
 
 -- Flush all dirty pages by BgBufferSync()
 select gp_inject_fault_infinite('bg_buffer_sync_default_logic', 'skip', dbid)
@@ -126,7 +130,7 @@ from gp_segment_configuration where role = 'p' and content > -1;
 checkpoint;
 
 -- There should be no dirty buffers after checkpoint.
-select dirty_buffers() from gp_dist_random('gp_id')
+select * from dirty_buffers_on_qes()
  as (tablespace oid, database oid, relfilenode oid, block int);
 
 -- Validate that the number of files fsync'ed by checkpointer is at

src/test/heap_checksum/expected/heap_checksum_corruption.out

@@ -81,7 +81,7 @@ RETURNS integer as $$
       f.close()
 
   return 0
-$$ LANGUAGE plpython3u;
+$$ LANGUAGE plpython3u EXECUTE ON ALL SEGMENTS;
 CREATE OR REPLACE FUNCTION invalidate_buffers_for_rel(tablename text) RETURNS BOOL AS
 $$
 DECLARE
@@ -136,7 +136,7 @@ select invalidate_buffers_for_rel('corrupt_table') from gp_dist_random('gp_id');
 (3 rows)
 
 -- Verify corruption on heap table
-select SUM(corrupt_file(get_relation_path('corrupt_table'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_relation_path('corrupt_table'), -50));
 INFO:  corrupting file base/81967/81931 at -50  (seg0 slice1 127.0.0.1:25432 pid=27287)
 INFO:  corrupting file base/81967/81931 at -50  (seg1 slice1 127.0.0.1:25433 pid=27288)
 INFO:  corrupting file base/81967/81931 at -50  (seg2 slice1 127.0.0.1:25434 pid=27289)
@@ -164,7 +164,7 @@ select invalidate_buffers_for_rel(get_toast_name('corrupt_toast_table')) from gp
 (3 rows)
 
 -- Verify corruption on toast table
-select SUM(corrupt_file(get_toast_path('corrupt_toast_table'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_toast_path('corrupt_toast_table'), -50));
 INFO:  corrupting file base/81967/81933 at -50  (seg0 slice1 127.0.0.1:25432 pid=27287)
 INFO:  corrupting file base/81967/81933 at -50  (seg1 slice1 127.0.0.1:25433 pid=27288)
 INFO:  corrupting file base/81967/81933 at -50  (seg2 slice1 127.0.0.1:25434 pid=27289)
@@ -193,7 +193,7 @@ select invalidate_buffers_for_rel('btree_index') from gp_dist_random('gp_id');
 (3 rows)
 
 -- Verify corruption on Btree index
-select SUM(corrupt_file(get_index_path('corrupt_btree_index'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_index_path('corrupt_btree_index'), -50));
 INFO:  corrupting file base/81967/81936 at -50  (seg0 slice1 127.0.0.1:25432 pid=27287)
 INFO:  corrupting file base/81967/81936 at -50  (seg1 slice1 127.0.0.1:25433 pid=27288)
 INFO:  corrupting file base/81967/81936 at -50  (seg2 slice1 127.0.0.1:25434 pid=27289)
@@ -222,7 +222,7 @@ select invalidate_buffers_for_rel('bitmap_index') from gp_dist_random('gp_id');
 (3 rows)
 
 -- Verify corruption on Bitmap index 
-select SUM(corrupt_file(get_index_path('corrupt_bitmap_index'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_index_path('corrupt_bitmap_index'), -50));
 INFO:  corrupting file base/81967/81938 at -50  (seg0 slice1 127.0.0.1:25432 pid=27287)
 INFO:  corrupting file base/81967/81938 at -50  (seg1 slice1 127.0.0.1:25433 pid=27288)
 INFO:  corrupting file base/81967/81938 at -50  (seg2 slice1 127.0.0.1:25434 pid=27289)
@@ -264,7 +264,7 @@ select count(*) from mark_buffer_dirty_hint;
 create table flush_xlog_page_to_disk (c int);
 NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'c' as the Greenplum Database data distribution key for this table.
 -- corrupt the page on disk
-select SUM(corrupt_file(get_relation_path('mark_buffer_dirty_hint'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_relation_path('mark_buffer_dirty_hint'), -50));
 INFO:  corrupting file base/81967/81939 at -50  (seg0 slice1 127.0.0.1:25432 pid=27287)
 INFO:  corrupting file base/81967/81939 at -50  (seg1 slice1 127.0.0.1:25433 pid=27288)
 INFO:  corrupting file base/81967/81939 at -50  (seg2 slice1 127.0.0.1:25434 pid=27289)

src/test/heap_checksum/sql/heap_checksum_corruption.sql

@@ -91,7 +91,7 @@ RETURNS integer as $$
       f.close()
 
   return 0
-$$ LANGUAGE plpython3u;
+$$ LANGUAGE plpython3u EXECUTE ON ALL SEGMENTS;
 
 CREATE OR REPLACE FUNCTION invalidate_buffers_for_rel(tablename text) RETURNS BOOL AS
 $$
@@ -122,7 +122,7 @@ insert into corrupt_table select i from generate_series(1, 10) i;
 checkpoint;
 select invalidate_buffers_for_rel('corrupt_table') from gp_dist_random('gp_id');
 -- Verify corruption on heap table
-select SUM(corrupt_file(get_relation_path('corrupt_table'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_relation_path('corrupt_table'), -50));
 SELECT COUNT(*) FROM corrupt_table;
 
 -- Corrupt a heap table, with toast table 
@@ -131,7 +131,7 @@ insert into corrupt_toast_table select i, ("decode"(repeat('a',3000000),'escape'
 checkpoint;
 select invalidate_buffers_for_rel(get_toast_name('corrupt_toast_table')) from gp_dist_random('gp_id');
 -- Verify corruption on toast table
-select SUM(corrupt_file(get_toast_path('corrupt_toast_table'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_toast_path('corrupt_toast_table'), -50));
 SELECT md5(comment::text) FROM corrupt_toast_table;
 
 -- Corrupt a Btree Index
@@ -141,7 +141,7 @@ insert into corrupt_btree_index select i, 'a' from generate_series(1, 10) i;
 checkpoint;
 select invalidate_buffers_for_rel('btree_index') from gp_dist_random('gp_id');
 -- Verify corruption on Btree index
-select SUM(corrupt_file(get_index_path('corrupt_btree_index'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_index_path('corrupt_btree_index'), -50));
 insert into corrupt_btree_index select i, 'a' from generate_series(1, 10) i; -- insert will trigger scan of the index
 
 -- Corrupt a Bitmap Index 
@@ -151,7 +151,7 @@ insert into corrupt_bitmap_index select i, 'a' from generate_series(1, 10) i;
 checkpoint;
 select invalidate_buffers_for_rel('bitmap_index') from gp_dist_random('gp_id');
 -- Verify corruption on Bitmap index 
-select SUM(corrupt_file(get_index_path('corrupt_bitmap_index'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_index_path('corrupt_bitmap_index'), -50));
 insert into corrupt_bitmap_index select i, 'a' from generate_series(1, 10) i; -- insert will trigger scan of the index
 
 -- Test make sure full page image is captured in XLOG if hint bit change is the first change after checkpoint
@@ -168,7 +168,7 @@ select count(*) from mark_buffer_dirty_hint;
 -- using a DML to trigger the XLogFlush() to have the backup block written
 create table flush_xlog_page_to_disk (c int);
 -- corrupt the page on disk
-select SUM(corrupt_file(get_relation_path('mark_buffer_dirty_hint'), -50)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_relation_path('mark_buffer_dirty_hint'), -50));
 -- invalidate buffer and confirm data is corrupted
 select invalidate_buffers_for_rel('mark_buffer_dirty_hint') from gp_dist_random('gp_id');
 select count(*) from mark_buffer_dirty_hint;

src/test/regress/expected/ao_checksum_corruption.out

@@ -63,7 +63,7 @@ RETURNS integer as $$
       f.close()
 
   return 0
-$$ LANGUAGE plpython3u;
+$$ LANGUAGE plpython3u EXECUTE ON ALL SEGMENTS;
 -- Corrupt a file by replacing the last occurrence of 'str' within the file
 -- with 'replacement'
 CREATE FUNCTION corrupt_file(data_file text, target bytea, replacement bytea)
@@ -90,11 +90,11 @@ RETURNS integer as $$
       f.close()
 
   return 0
-$$ LANGUAGE plpython3u;
+$$ LANGUAGE plpython3u EXECUTE ON ALL SEGMENTS;
 -- Large content, corrupt block header
 create table corrupt_header_large_co(comment bytea ) with (appendonly=true, orientation=column, checksum=true) DISTRIBUTED RANDOMLY;
 insert into corrupt_header_large_co select ("decode"(repeat('a',33554432),'escape')) from generate_series(1,8)  ;
-select SUM(corrupt_file(get_aoseg1_path('corrupt_header_large_co'), 8)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_header_large_co'), 8));
 INFO:  corrupting file base/107062/107070.1 at 8  (seg0 slice1 127.0.0.1:40000 pid=28137)
 INFO:  corrupting file base/107062/107070.1 at 8  (seg1 slice1 127.0.0.1:40001 pid=28138)
 INFO:  corrupting file base/107062/107070.1 at 8  (seg2 slice1 127.0.0.1:40002 pid=28139)
@@ -108,7 +108,7 @@ ERROR:  header checksum does not match, expected 0xBA9CC649 and found 0x0C997AE7
 -- Large content, corrupt content
 create table corrupt_content_large_co(comment bytea ) with (appendonly=true, orientation=column, checksum=true) DISTRIBUTED RANDOMLY;
 insert into corrupt_content_large_co select ("decode"(repeat('a',33554432),'escape')) from generate_series(1,8)  ;
-select SUM(corrupt_file(get_aoseg1_path('corrupt_content_large_co'), -3)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_content_large_co'), -3));
 INFO:  corrupting file base/107062/107080.1 at -3  (seg0 slice1 127.0.0.1:40000 pid=28137)
 INFO:  corrupting file base/107062/107080.1 at -3  (seg1 slice1 127.0.0.1:40001 pid=28138)
 INFO:  corrupting file base/107062/107080.1 at -3  (seg2 slice1 127.0.0.1:40002 pid=28139)
@@ -123,7 +123,7 @@ ERROR:  block checksum does not match, expected 0x9C02F450 and found 0x0A78638C
 create table corrupt_header_small_co(a int) with (appendonly=true, orientation=column, checksum=true);
 NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a' as the Greenplum Database data distribution key for this table.
 insert into corrupt_header_small_co values (1),(1),(1),(-1),(1),(1),(1),(2),(2),(2),(2),(2),(2),(2),(33),(3),(3),(3),(1),(8),(19),(20),(31),(32),(33),(34),(5),(5),(5),(5),(5),(6),(6),(6),(6),(6),(6),(7),(7),(7),(7),(7),(7),(7),(7), (null),(7),(7),(7),(null),(8),(8),(8),(8),(8),(8),(4),(4),(null),(4),(17),(17),(17),(null),(null),(null);
-select SUM(corrupt_file(get_aoseg1_path('corrupt_header_small_co'), 8)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_header_small_co'), 8));
 INFO:  corrupting file base/107062/107090.1 at 8  (seg0 slice1 127.0.0.1:40000 pid=28137)
 INFO:  corrupting file base/107062/107090.1 at 8  (seg1 slice1 127.0.0.1:40001 pid=28138)
 INFO:  corrupting file base/107062/107090.1 at 8  (seg2 slice1 127.0.0.1:40002 pid=28139)
@@ -138,7 +138,7 @@ ERROR:  header checksum does not match, expected 0xAA39D941 and found 0xE8FB379B
 create table corrupt_content_small_co(a int) with (appendonly=true, orientation=column, checksum=true);
 NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a' as the Greenplum Database data distribution key for this table.
 insert into corrupt_content_small_co values (1),(1),(1),(-1),(1),(1),(1),(2),(2),(2),(2),(2),(2),(2),(33),(3),(3),(3),(1),(8),(19),(20),(31),(32),(33),(34),(5),(5),(5),(5),(5),(6),(6),(6),(6),(6),(6),(7),(7),(7),(7),(7),(7),(7),(7), (null),(7),(7),(7),(null),(8),(8),(8),(8),(8),(8),(4),(4),(null),(4),(17),(17),(17),(null),(null),(null);
-select SUM(corrupt_file(get_aoseg1_path('corrupt_content_small_co'), -3)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_content_small_co'), -3));
 INFO:  corrupting file base/107062/107097.1 at -3  (seg0 slice1 127.0.0.1:40000 pid=28137)
 INFO:  corrupting file base/107062/107097.1 at -3  (seg1 slice1 127.0.0.1:40001 pid=28138)
 INFO:  corrupting file base/107062/107097.1 at -3  (seg2 slice1 127.0.0.1:40002 pid=28139)
@@ -153,7 +153,7 @@ ERROR:  block checksum does not match, expected 0xA3CAEC7A and found 0x35B07BA6
 create table corrupt_header_small_ao(a int) with (appendonly=true, orientation=row, checksum=true);
 NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a' as the Greenplum Database data distribution key for this table.
 insert into corrupt_header_small_ao values (1),(1),(1),(-1),(1),(1),(1),(2),(2),(2),(2),(2),(2),(2),(33),(3),(3),(3),(1),(8),(19),(20),(31),(32),(33),(34),(5),(5),(5),(5),(5),(6),(6),(6),(6),(6),(6),(7),(7),(7),(7),(7),(7),(7),(7), (null),(7),(7),(7),(null),(8),(8),(8),(8),(8),(8),(4),(4),(null),(4),(17),(17),(17),(null),(null),(null);
-select SUM(corrupt_file(get_aoseg1_path('corrupt_header_small_ao'), 8)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_header_small_ao'), 8));
 INFO:  corrupting file base/107062/107104.1 at 8  (seg0 slice1 127.0.0.1:40000 pid=28137)
 INFO:  corrupting file base/107062/107104.1 at 8  (seg1 slice1 127.0.0.1:40001 pid=28138)
 INFO:  corrupting file base/107062/107104.1 at 8  (seg2 slice1 127.0.0.1:40002 pid=28139)
@@ -168,7 +168,7 @@ ERROR:  header checksum does not match, expected 0x413AF15D and found 0x805DA822
 create table corrupt_content_small_ao(a int) with (appendonly=true, orientation=row, checksum=true);
 NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 'a' as the Greenplum Database data distribution key for this table.
 insert into corrupt_content_small_ao values (1),(1),(1),(-1),(1),(1),(1),(2),(2),(2),(2),(2),(2),(2),(33),(3),(3),(3),(1),(8),(19),(20),(31),(32),(33),(34),(5),(5),(5),(5),(5),(6),(6),(6),(6),(6),(6),(7),(7),(7),(7),(7),(7),(7),(7), (null),(7),(7),(7),(null),(8),(8),(8),(8),(8),(8),(4),(4),(null),(4),(17),(17),(17),(null),(null),(null);
-select SUM(corrupt_file(get_aoseg1_path('corrupt_content_small_ao'), -3)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_content_small_ao'), -3));
 INFO:  corrupting file base/107062/107111.1 at -3  (seg0 slice1 127.0.0.1:40000 pid=28137)
 INFO:  corrupting file base/107062/107111.1 at -3  (seg1 slice1 127.0.0.1:40001 pid=28138)
 INFO:  corrupting file base/107062/107111.1 at -3  (seg2 slice1 127.0.0.1:40002 pid=28139)
@@ -185,7 +185,7 @@ NOTICE:  Table doesn't have 'DISTRIBUTED BY' clause -- Using column named 't' as
 insert into appendonly_verify_block_checksums_co
   select 'abcdefghijlmnopqrstuvxyz' from generate_series(1, 5);
 -- Corrupt the table by flip the 'xyz' on the last row with ###
-select SUM(corrupt_file(get_aoseg1_path('appendonly_verify_block_checksums_co'), 'xyz', '###')) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('appendonly_verify_block_checksums_co'), 'xyz', '###'));
 INFO:  skipping non-existent file base/107062/107118.1  (seg0 slice1 127.0.0.1:40000 pid=28137)
 INFO:  skipping non-existent file base/107062/107118.1  (seg1 slice1 127.0.0.1:40001 pid=28138)
 INFO:  corrupting file base/107062/107118.1 by replacing <type 'str'> with ###  (seg2 slice1 127.0.0.1:40002 pid=28139)

src/test/regress/expected/bfv_partition_plans.out

@@ -1179,24 +1179,23 @@ analyze r_part;
 explain select * from r_part r1, r_part r2 where r1.a=1; -- should eliminate partitions in the r1 copy of r_part
                                            QUERY PLAN                                            
 -------------------------------------------------------------------------------------------------
- Gather Motion 3:1  (slice1; segments: 3)  (cost=10000000000.00..10000000010.72 rows=9 width=16)
-   ->  Nested Loop  (cost=10000000000.00..10000000010.54 rows=3 width=16)
-         ->  Broadcast Motion 1:3  (slice2; segments: 1)  (cost=0.00..1.05 rows=3 width=8)
+ Gather Motion 3:1  (slice1; segments: 3)  (cost=10000000000.00..10000000010.38 rows=9 width=16)
+   ->  Nested Loop  (cost=10000000000.00..10000000010.26 rows=3 width=16)
+         ->  Broadcast Motion 1:3  (slice2; segments: 1)  (cost=0.00..1.03 rows=1 width=8)
                ->  Seq Scan on r_part_1_prt_1 r1  (cost=0.00..1.01 rows=1 width=8)
                      Filter: (a = 1)
-         ->  Materialize  (cost=0.00..9.17 rows=3 width=8)
-               ->  Append  (cost=0.00..9.12 rows=3 width=8)
-                     ->  Seq Scan on r_part_1_prt_1 r2  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_2 r2_1  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_3 r2_2  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_4 r2_3  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_5 r2_4  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_6 r2_5  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_7 r2_6  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_8 r2_7  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_9 r2_8  (cost=0.00..1.00 rows=1 width=8)
+         ->  Append  (cost=0.00..9.13 rows=9 width=8)
+               ->  Seq Scan on r_part_1_prt_1 r2  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_2 r2_1  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_3 r2_2  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_4 r2_3  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_5 r2_4  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_6 r2_5  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_7 r2_6  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_8 r2_7  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_9 r2_8  (cost=0.00..1.01 rows=1 width=8)
  Optimizer: Postgres query optimizer
-(17 rows)
+(16 rows)
 
 -- the numbers in the filter should be both on segment 0
 explain select * from r_part where a in (7,8); -- should eliminate partitions

src/test/regress/expected/bfv_partition_plans_optimizer.out

@@ -1183,27 +1183,23 @@ analyze r_part;
 explain select * from r_part r1, r_part r2 where r1.a=1; -- should eliminate partitions in the r1 copy of r_part
                                            QUERY PLAN                                            
 -------------------------------------------------------------------------------------------------
- Gather Motion 3:1  (slice1; segments: 3)  (cost=10000000000.00..10000000010.72 rows=9 width=16)
-   ->  Nested Loop  (cost=10000000000.00..10000000010.54 rows=3 width=16)
-         ->  Broadcast Motion 1:3  (slice2; segments: 1)  (cost=0.00..1.05 rows=3 width=8)
+ Gather Motion 3:1  (slice1; segments: 3)  (cost=10000000000.00..10000000010.38 rows=9 width=16)
+   ->  Nested Loop  (cost=10000000000.00..10000000010.26 rows=3 width=16)
+         ->  Broadcast Motion 1:3  (slice2; segments: 1)  (cost=0.00..1.03 rows=1 width=8)
                ->  Seq Scan on r_part_1_prt_1 r1  (cost=0.00..1.01 rows=1 width=8)
                      Filter: (a = 1)
-         ->  Materialize  (cost=0.00..9.17 rows=3 width=8)
-               ->  Append  (cost=0.00..9.12 rows=3 width=8)
-                     ->  Seq Scan on r_part_1_prt_1 r2  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_2 r2_1  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_3 r2_2  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_4 r2_3  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_5 r2_4  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_6 r2_5  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_7 r2_6  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_8 r2_7  (cost=0.00..1.01 rows=1 width=8)
-                     ->  Seq Scan on r_part_1_prt_9 r2_8  (cost=0.00..1.00 rows=1 width=8)
+         ->  Append  (cost=0.00..9.13 rows=9 width=8)
+               ->  Seq Scan on r_part_1_prt_1 r2  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_2 r2_1  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_3 r2_2  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_4 r2_3  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_5 r2_4  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_6 r2_5  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_7 r2_6  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_8 r2_7  (cost=0.00..1.01 rows=1 width=8)
+               ->  Seq Scan on r_part_1_prt_9 r2_8  (cost=0.00..1.01 rows=1 width=8)
  Optimizer: Postgres query optimizer
- JIT:
-   Functions: 5
-   Options: Inlining true, Optimization true, Expressions true, Deforming true
-(20 rows)
+(16 rows)
 
 -- the numbers in the filter should be both on segment 0
 explain select * from r_part where a in (7,8); -- should eliminate partitions

src/test/regress/expected/dpe.out

@@ -857,35 +857,32 @@ reset enable_nestloop;
 --
 -- one of the joined tables can be used for partition elimination, the other can not
 explain (costs off, timing off, summary off, analyze) select * from t, t1, pt where t1.t2 = t.t2 and t1.tid = ptid;
-                                                    QUERY PLAN                                                    
-------------------------------------------------------------------------------------------------------------------
+                                            QUERY PLAN                                             
+---------------------------------------------------------------------------------------------------
  Gather Motion 3:1  (slice1; segments: 3) (actual rows=36 loops=1)
    ->  Hash Join (actual rows=16 loops=1)
-         Hash Cond: (pt_1_prt_2.ptid = t1.tid)
-         ->  Append (actual rows=8 loops=1)
-               Partition Selectors: $0
-               ->  Seq Scan on pt_1_prt_2 (never executed)
-               ->  Seq Scan on pt_1_prt_3 (actual rows=3 loops=1)
-               ->  Seq Scan on pt_1_prt_4 (actual rows=5 loops=1)
-               ->  Seq Scan on pt_1_prt_5 (never executed)
-               ->  Seq Scan on pt_1_prt_6 (never executed)
-               ->  Seq Scan on pt_1_prt_junk_data (never executed)
-         ->  Hash (actual rows=4 loops=1)
+         Hash Cond: (t1.t2 = t.t2)
+         ->  Hash Join (actual rows=8 loops=1)
+               Hash Cond: (pt_1_prt_2.ptid = t1.tid)
+               ->  Append (actual rows=8 loops=1)
+                     Partition Selectors: $0
+                     ->  Seq Scan on pt_1_prt_2 (never executed)
+                     ->  Seq Scan on pt_1_prt_3 (actual rows=3 loops=1)
+                     ->  Seq Scan on pt_1_prt_4 (actual rows=5 loops=1)
+                     ->  Seq Scan on pt_1_prt_5 (never executed)
+                     ->  Seq Scan on pt_1_prt_6 (never executed)
+                     ->  Seq Scan on pt_1_prt_junk_data (never executed)
+               ->  Hash (actual rows=2 loops=1)
+                     Buckets: 131072  Batches: 1  Memory Usage: 1025kB
+                     ->  Partition Selector (selector id: $0) (actual rows=2 loops=1)
+                           ->  Broadcast Motion 3:3  (slice2; segments: 3) (actual rows=2 loops=1)
+                                 ->  Seq Scan on t1 (actual rows=1 loops=1)
+         ->  Hash (actual rows=2 loops=1)
                Buckets: 131072  Batches: 1  Memory Usage: 1025kB
-               ->  Partition Selector (selector id: $0) (actual rows=4 loops=1)
-                     ->  Broadcast Motion 3:3  (slice2; segments: 3) (actual rows=4 loops=1)
-                           ->  Hash Join (actual rows=4 loops=1)
-                                 Hash Cond: (t.t2 = t1.t2)
-                                 ->  Redistribute Motion 3:3  (slice3; segments: 3) (actual rows=2 loops=1)
-                                       Hash Key: t.t2
-                                       ->  Seq Scan on t (actual rows=2 loops=1)
-                                 ->  Hash (actual rows=2 loops=1)
-                                       Buckets: 131072  Batches: 1  Memory Usage: 1025kB
-                                       ->  Redistribute Motion 3:3  (slice4; segments: 3) (actual rows=2 loops=1)
-                                             Hash Key: t1.t2
-                                             ->  Seq Scan on t1 (actual rows=1 loops=1)
+               ->  Broadcast Motion 3:3  (slice3; segments: 3) (actual rows=2 loops=1)
+                     ->  Seq Scan on t (actual rows=2 loops=1)
  Optimizer: Postgres query optimizer
-(26 rows)
+(23 rows)
 
 select * from t, t1, pt where t1.t2 = t.t2 and t1.tid = ptid;
  dist | tid |   t1   | t2  | dist | tid |   t1   | t2  | dist |   pt1   |  pt2  |    pt3    | ptid 

src/test/regress/expected/dpe_optimizer.out

@@ -857,35 +857,32 @@ reset enable_nestloop;
 --
 -- one of the joined tables can be used for partition elimination, the other can not
 explain (costs off, timing off, summary off, analyze) select * from t, t1, pt where t1.t2 = t.t2 and t1.tid = ptid;
-                                                    QUERY PLAN                                                    
-------------------------------------------------------------------------------------------------------------------
+                                            QUERY PLAN                                             
+---------------------------------------------------------------------------------------------------
  Gather Motion 3:1  (slice1; segments: 3) (actual rows=36 loops=1)
    ->  Hash Join (actual rows=16 loops=1)
-         Hash Cond: (pt_1_prt_2.ptid = t1.tid)
-         ->  Append (actual rows=8 loops=1)
-               Partition Selectors: $0
-               ->  Seq Scan on pt_1_prt_2 (never executed)
-               ->  Seq Scan on pt_1_prt_3 (actual rows=3 loops=1)
-               ->  Seq Scan on pt_1_prt_4 (actual rows=5 loops=1)
-               ->  Seq Scan on pt_1_prt_5 (never executed)
-               ->  Seq Scan on pt_1_prt_6 (never executed)
-               ->  Seq Scan on pt_1_prt_junk_data (never executed)
-         ->  Hash (actual rows=4 loops=1)
+         Hash Cond: (t1.t2 = t.t2)
+         ->  Hash Join (actual rows=8 loops=1)
+               Hash Cond: (pt_1_prt_2.ptid = t1.tid)
+               ->  Append (actual rows=8 loops=1)
+                     Partition Selectors: $0
+                     ->  Seq Scan on pt_1_prt_2 (never executed)
+                     ->  Seq Scan on pt_1_prt_3 (actual rows=3 loops=1)
+                     ->  Seq Scan on pt_1_prt_4 (actual rows=5 loops=1)
+                     ->  Seq Scan on pt_1_prt_5 (never executed)
+                     ->  Seq Scan on pt_1_prt_6 (never executed)
+                     ->  Seq Scan on pt_1_prt_junk_data (never executed)
+               ->  Hash (actual rows=2 loops=1)
+                     Buckets: 131072  Batches: 1  Memory Usage: 1025kB
+                     ->  Partition Selector (selector id: $0) (actual rows=2 loops=1)
+                           ->  Broadcast Motion 3:3  (slice2; segments: 3) (actual rows=2 loops=1)
+                                 ->  Seq Scan on t1 (actual rows=1 loops=1)
+         ->  Hash (actual rows=2 loops=1)
                Buckets: 131072  Batches: 1  Memory Usage: 1025kB
-               ->  Partition Selector (selector id: $0) (actual rows=4 loops=1)
-                     ->  Broadcast Motion 3:3  (slice2; segments: 3) (actual rows=4 loops=1)
-                           ->  Hash Join (actual rows=4 loops=1)
-                                 Hash Cond: (t.t2 = t1.t2)
-                                 ->  Redistribute Motion 3:3  (slice3; segments: 3) (actual rows=2 loops=1)
-                                       Hash Key: t.t2
-                                       ->  Seq Scan on t (actual rows=2 loops=1)
-                                 ->  Hash (actual rows=2 loops=1)
-                                       Buckets: 131072  Batches: 1  Memory Usage: 1025kB
-                                       ->  Redistribute Motion 3:3  (slice4; segments: 3) (actual rows=2 loops=1)
-                                             Hash Key: t1.t2
-                                             ->  Seq Scan on t1 (actual rows=1 loops=1)
+               ->  Broadcast Motion 3:3  (slice3; segments: 3) (actual rows=2 loops=1)
+                     ->  Seq Scan on t (actual rows=2 loops=1)
  Optimizer: Postgres query optimizer
-(26 rows)
+(23 rows)
 
 select * from t, t1, pt where t1.t2 = t.t2 and t1.tid = ptid;
  dist | tid |   t1   | t2  | dist | tid |   t1   | t2  | dist |   pt1   |  pt2  |    pt3    | ptid 

src/test/regress/expected/equivclass.out

@@ -95,6 +95,7 @@ create table ec0 (ff int8 primary key, f1 int8, f2 int8);
 create table ec1 (ff int8 primary key, f1 int8alias1, f2 int8alias2);
 create table ec2 (xf int8 primary key, x1 int8alias1, x2 int8alias2);
 -- for the moment we only want to look at nestloop plans
+set enable_nestloop = on;
 set enable_hashjoin = off;
 set enable_mergejoin = off;
 --
@@ -147,18 +148,17 @@ explain (costs off)
 
 explain (costs off)
   select * from ec1, ec2 where ff = x1 and ff = '42'::int8;
-                                     QUERY PLAN                                      
--------------------------------------------------------------------------------------
+                                  QUERY PLAN                                   
+-------------------------------------------------------------------------------
  Gather Motion 3:1  (slice1; segments: 3)
    ->  Nested Loop
          Join Filter: (ec1.ff = ec2.x1)
+         ->  Broadcast Motion 1:3  (slice2; segments: 1)
+               ->  Index Scan using ec1_pkey on ec1
+                     Index Cond: ((ff = '42'::bigint) AND (ff = '42'::bigint))
          ->  Seq Scan on ec2
-         ->  Materialize
-               ->  Broadcast Motion 1:3  (slice2; segments: 1)
-                     ->  Index Scan using ec1_pkey on ec1
-                           Index Cond: ((ff = '42'::bigint) AND (ff = '42'::bigint))
  Optimizer: Postgres query optimizer
-(9 rows)
+(8 rows)
 
 explain (costs off)
   select * from ec1, ec2 where ff = x1 and ff = '42'::int8alias1;
@@ -169,11 +169,10 @@ explain (costs off)
          ->  Broadcast Motion 3:3  (slice2; segments: 3)
                ->  Index Scan using ec1_pkey on ec1
                      Index Cond: (ff = '42'::int8alias1)
-         ->  Materialize
-               ->  Seq Scan on ec2
-                     Filter: (x1 = '42'::int8alias1)
+         ->  Seq Scan on ec2
+               Filter: (x1 = '42'::int8alias1)
  Optimizer: Postgres query optimizer
-(9 rows)
+(8 rows)
 
 explain (costs off)
   select * from ec1, ec2 where ff = x1 and '42'::int8 = x1;
@@ -185,11 +184,10 @@ explain (costs off)
          ->  Broadcast Motion 1:3  (slice2; segments: 1)
                ->  Index Scan using ec1_pkey on ec1
                      Index Cond: (ff = '42'::bigint)
-         ->  Materialize
-               ->  Seq Scan on ec2
-                     Filter: ('42'::bigint = x1)
+         ->  Seq Scan on ec2
+               Filter: ('42'::bigint = x1)
  Optimizer: Postgres query optimizer
-(10 rows)
+(9 rows)
 
 explain (costs off)
   select * from ec1, ec2 where ff = x1 and x1 = '42'::int8alias1;
@@ -200,11 +198,10 @@ explain (costs off)
          ->  Broadcast Motion 3:3  (slice2; segments: 3)
                ->  Index Scan using ec1_pkey on ec1
                      Index Cond: (ff = '42'::int8alias1)
-         ->  Materialize
-               ->  Seq Scan on ec2
-                     Filter: (x1 = '42'::int8alias1)
+         ->  Seq Scan on ec2
+               Filter: (x1 = '42'::int8alias1)
  Optimizer: Postgres query optimizer
-(9 rows)
+(8 rows)
 
 explain (costs off)
   select * from ec1, ec2 where ff = x1 and x1 = '42'::int8alias2;
@@ -298,22 +295,28 @@ explain (costs off)
      union all
      select ff + 4 as x from ec1) as ss2
   where ss1.x = ec1.f1 and ss1.x = ss2.x and ec1.ff = 42::int8;
-                                QUERY PLAN                                 
----------------------------------------------------------------------------
+                                 QUERY PLAN                                  
+-----------------------------------------------------------------------------
  Gather Motion 3:1  (slice1; segments: 3)
    ->  Nested Loop
          ->  Broadcast Motion 3:3  (slice2; segments: 3)
                ->  Nested Loop
-                     Join Filter: ((((ec1_1.ff + 2) + 1)) = ec1.f1)
-                     ->  Merge Append
-                           Sort Key: (((ec1_1.ff + 2) + 1))
-                           ->  Index Scan using ec1_expr2 on ec1 ec1_1
-                           ->  Index Scan using ec1_expr3 on ec1 ec1_2
-                           ->  Index Scan using ec1_expr4 on ec1 ec1_3
-                     ->  Materialize
-                           ->  Broadcast Motion 1:3  (slice3; segments: 1)
-                                 ->  Index Scan using ec1_pkey on ec1
-                                       Index Cond: (ff = '42'::bigint)
+                     ->  Broadcast Motion 1:3  (slice3; segments: 1)
+                           ->  Index Scan using ec1_pkey on ec1
+                                 Index Cond: (ff = '42'::bigint)
+                     ->  Append
+                           ->  Bitmap Heap Scan on ec1 ec1_1
+                                 Recheck Cond: (((ff + 2) + 1) = ec1.f1)
+                                 ->  Bitmap Index Scan on ec1_expr2
+                                       Index Cond: (((ff + 2) + 1) = ec1.f1)
+                           ->  Bitmap Heap Scan on ec1 ec1_2
+                                 Recheck Cond: (((ff + 3) + 1) = ec1.f1)
+                                 ->  Bitmap Index Scan on ec1_expr3
+                                       Index Cond: (((ff + 3) + 1) = ec1.f1)
+                           ->  Bitmap Heap Scan on ec1 ec1_3
+                                 Recheck Cond: ((ff + 4) = ec1.f1)
+                                 ->  Bitmap Index Scan on ec1_expr4
+                                       Index Cond: ((ff + 4) = ec1.f1)
          ->  Append
                ->  Index Scan using ec1_expr2 on ec1 ec1_4
                      Index Cond: (((ff + 2) + 1) = (((ec1_1.ff + 2) + 1)))
@@ -322,7 +325,7 @@ explain (costs off)
                ->  Index Scan using ec1_expr4 on ec1 ec1_6
                      Index Cond: ((ff + 4) = (((ec1_1.ff + 2) + 1)))
  Optimizer: Postgres query optimizer
-(22 rows)
+(28 rows)
 
 -- let's try that as a mergejoin
 set enable_mergejoin = on;

src/test/regress/expected/equivclass_optimizer.out

@@ -95,6 +95,7 @@ create table ec0 (ff int8 primary key, f1 int8, f2 int8);
 create table ec1 (ff int8 primary key, f1 int8alias1, f2 int8alias2);
 create table ec2 (xf int8 primary key, x1 int8alias1, x2 int8alias2);
 -- for the moment we only want to look at nestloop plans
+set enable_nestloop = on;
 set enable_hashjoin = off;
 set enable_mergejoin = off;
 --
@@ -168,11 +169,10 @@ explain (costs off)
          ->  Broadcast Motion 3:3  (slice2; segments: 3)
                ->  Index Scan using ec1_pkey on ec1
                      Index Cond: (ff = '42'::int8alias1)
-         ->  Materialize
-               ->  Seq Scan on ec2
-                     Filter: (x1 = '42'::int8alias1)
+         ->  Seq Scan on ec2
+               Filter: (x1 = '42'::int8alias1)
  Optimizer: Postgres query optimizer
-(9 rows)
+(8 rows)
 
 explain (costs off)
   select * from ec1, ec2 where ff = x1 and '42'::int8 = x1;

src/test/regress/expected/gporca.out

@@ -12225,20 +12225,20 @@ where out.b in (select coalesce(tcorr2.a, 99)
                 from tcorr1 left outer join tcorr2 on tcorr1.a=tcorr2.a+out.a);
                                                    QUERY PLAN                                                   
 ----------------------------------------------------------------------------------------------------------------
- Nested Loop Semi Join  (cost=10000000001.10..10000000003.36 rows=3 width=8)
-   ->  Gather Motion 3:1  (slice1; segments: 3)  (cost=0.00..1.05 rows=3 width=8)
+ Nested Loop Semi Join  (cost=10000000001.05..10000000003.21 rows=3 width=8)
+   ->  Gather Motion 3:1  (slice1; segments: 3)  (cost=0.00..1.03 rows=1 width=8)
          ->  Seq Scan on tcorr1 "out"  (cost=0.00..1.01 rows=1 width=8)
-   ->  Materialize  (cost=1.10..2.28 rows=1 width=4)
-         ->  Subquery Scan on "ANY_subquery"  (cost=1.10..2.27 rows=1 width=4)
+   ->  Materialize  (cost=1.05..2.17 rows=1 width=4)
+         ->  Subquery Scan on "ANY_subquery"  (cost=1.05..2.16 rows=1 width=4)
                Filter: ("out".b = "ANY_subquery"."coalesce")
-               ->  Hash Left Join  (cost=1.10..2.23 rows=3 width=4)
+               ->  Hash Left Join  (cost=1.05..2.12 rows=3 width=4)
                      Hash Cond: (tcorr1.a = (tcorr2.a + "out".a))
-                     ->  Materialize  (cost=0.00..1.06 rows=3 width=4)
-                           ->  Gather Motion 3:1  (slice2; segments: 3)  (cost=0.00..1.05 rows=3 width=4)
+                     ->  Materialize  (cost=0.00..1.03 rows=1 width=4)
+                           ->  Gather Motion 3:1  (slice2; segments: 3)  (cost=0.00..1.03 rows=1 width=4)
                                  ->  Seq Scan on tcorr1  (cost=0.00..1.01 rows=1 width=4)
-                     ->  Hash  (cost=1.06..1.06 rows=3 width=4)
-                           ->  Materialize  (cost=0.00..1.06 rows=3 width=4)
-                                 ->  Gather Motion 3:1  (slice3; segments: 3)  (cost=0.00..1.05 rows=3 width=4)
+                     ->  Hash  (cost=1.03..1.03 rows=1 width=4)
+                           ->  Materialize  (cost=0.00..1.03 rows=1 width=4)
+                                 ->  Gather Motion 3:1  (slice3; segments: 3)  (cost=0.00..1.03 rows=1 width=4)
                                        ->  Seq Scan on tcorr2  (cost=0.00..1.01 rows=1 width=4)
  Optimizer: Postgres query optimizer
 (16 rows)
@@ -12261,15 +12261,15 @@ where out.b in (select max(tcorr2.b + out.b - 1)
                 where tcorr2.a=out.a);
                                                   QUERY PLAN                                                   
 ---------------------------------------------------------------------------------------------------------------
- Gather Motion 3:1  (slice1; segments: 3)  (cost=0.00..1.05 rows=3 width=8)
+ Gather Motion 3:1  (slice1; segments: 3)  (cost=0.00..1.03 rows=1 width=8)
    ->  Seq Scan on tcorr1 "out"  (cost=0.00..1.01 rows=1 width=8)
          Filter: (SubPlan 1)
          SubPlan 1
-           ->  Aggregate  (cost=1.12..1.13 rows=1 width=4)
-                 ->  Result  (cost=0.00..1.09 rows=3 width=4)
+           ->  Aggregate  (cost=1.05..1.06 rows=1 width=4)
+                 ->  Result  (cost=0.00..1.04 rows=1 width=4)
                        Filter: (tcorr2.a = "out".a)
-                       ->  Materialize  (cost=0.00..1.06 rows=3 width=8)
-                             ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.05 rows=3 width=8)
+                       ->  Materialize  (cost=0.00..1.03 rows=1 width=8)
+                             ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.03 rows=1 width=8)
                                    ->  Seq Scan on tcorr2  (cost=0.00..1.01 rows=1 width=8)
  Optimizer: Postgres query optimizer
 (11 rows)
@@ -12321,20 +12321,20 @@ where out.b in (select coalesce(tcorr2.a, 99)
                 from tcorr1 left outer join tcorr2 on tcorr1.a=tcorr2.a+out.a);
                                                    QUERY PLAN                                                   
 ----------------------------------------------------------------------------------------------------------------
- Nested Loop Semi Join  (cost=10000000001.10..10000000003.36 rows=3 width=8)
-   ->  Gather Motion 3:1  (slice1; segments: 3)  (cost=0.00..1.05 rows=3 width=8)
+ Nested Loop Semi Join  (cost=10000000001.05..10000000003.21 rows=3 width=8)
+   ->  Gather Motion 3:1  (slice1; segments: 3)  (cost=0.00..1.03 rows=1 width=8)
          ->  Seq Scan on tcorr1 "out"  (cost=0.00..1.01 rows=1 width=8)
-   ->  Materialize  (cost=1.10..2.28 rows=1 width=4)
-         ->  Subquery Scan on "ANY_subquery"  (cost=1.10..2.27 rows=1 width=4)
+   ->  Materialize  (cost=1.05..2.17 rows=1 width=4)
+         ->  Subquery Scan on "ANY_subquery"  (cost=1.05..2.16 rows=1 width=4)
                Filter: ("out".b = "ANY_subquery"."coalesce")
-               ->  Hash Left Join  (cost=1.10..2.23 rows=3 width=4)
+               ->  Hash Left Join  (cost=1.05..2.12 rows=3 width=4)
                      Hash Cond: (tcorr1.a = (tcorr2.a + "out".a))
-                     ->  Materialize  (cost=0.00..1.06 rows=3 width=4)
-                           ->  Gather Motion 3:1  (slice2; segments: 3)  (cost=0.00..1.05 rows=3 width=4)
+                     ->  Materialize  (cost=0.00..1.03 rows=1 width=4)
+                           ->  Gather Motion 3:1  (slice2; segments: 3)  (cost=0.00..1.03 rows=1 width=4)
                                  ->  Seq Scan on tcorr1  (cost=0.00..1.01 rows=1 width=4)
-                     ->  Hash  (cost=1.06..1.06 rows=3 width=4)
-                           ->  Materialize  (cost=0.00..1.06 rows=3 width=4)
-                                 ->  Gather Motion 3:1  (slice3; segments: 3)  (cost=0.00..1.05 rows=3 width=4)
+                     ->  Hash  (cost=1.03..1.03 rows=1 width=4)
+                           ->  Materialize  (cost=0.00..1.03 rows=1 width=4)
+                                 ->  Gather Motion 3:1  (slice3; segments: 3)  (cost=0.00..1.03 rows=1 width=4)
                                        ->  Seq Scan on tcorr2  (cost=0.00..1.01 rows=1 width=4)
  Optimizer: Postgres query optimizer
 (16 rows)
@@ -12357,15 +12357,15 @@ where out.b in (select max(tcorr2.b + out.b - 1)
                 where tcorr2.a=out.a);
                                                   QUERY PLAN                                                   
 ---------------------------------------------------------------------------------------------------------------
- Gather Motion 3:1  (slice1; segments: 3)  (cost=0.00..1.05 rows=3 width=8)
+ Gather Motion 3:1  (slice1; segments: 3)  (cost=0.00..1.03 rows=1 width=8)
    ->  Seq Scan on tcorr1 "out"  (cost=0.00..1.01 rows=1 width=8)
          Filter: (SubPlan 1)
          SubPlan 1
-           ->  Aggregate  (cost=1.12..1.13 rows=1 width=4)
-                 ->  Result  (cost=0.00..1.09 rows=3 width=4)
+           ->  Aggregate  (cost=1.05..1.06 rows=1 width=4)
+                 ->  Result  (cost=0.00..1.04 rows=1 width=4)
                        Filter: (tcorr2.a = "out".a)
-                       ->  Materialize  (cost=0.00..1.06 rows=3 width=8)
-                             ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.05 rows=3 width=8)
+                       ->  Materialize  (cost=0.00..1.03 rows=1 width=8)
+                             ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.03 rows=1 width=8)
                                    ->  Seq Scan on tcorr2  (cost=0.00..1.01 rows=1 width=8)
  Optimizer: Postgres query optimizer
 (11 rows)

src/test/regress/expected/join.out

@@ -3216,25 +3216,23 @@ explain (costs off)
 select * from
   tenk1, int8_tbl a, int8_tbl b
 where thousand = a.q1 and tenthous = b.q1 and a.q2 = 1 and b.q2 = 2;
-                            QUERY PLAN                             
--------------------------------------------------------------------
+                                   QUERY PLAN                                    
+---------------------------------------------------------------------------------
  Gather Motion 3:1  (slice1; segments: 3)
    ->  Nested Loop
-         Join Filter: (tenk1.tenthous = b.q1)
-         ->  Nested Loop
-               ->  Broadcast Motion 3:3  (slice2; segments: 3)
-                     ->  Seq Scan on int8_tbl a
-                           Filter: (q2 = 1)
-               ->  Bitmap Heap Scan on tenk1
-                     Recheck Cond: (thousand = a.q1)
-                     ->  Bitmap Index Scan on tenk1_thous_tenthous
-                           Index Cond: (thousand = a.q1)
+         ->  Broadcast Motion 3:3  (slice2; segments: 3)
+               ->  Seq Scan on int8_tbl b
+                     Filter: (q2 = 2)
          ->  Materialize
-               ->  Broadcast Motion 3:3  (slice3; segments: 3)
-                     ->  Seq Scan on int8_tbl b
-                           Filter: (q2 = 2)
+               ->  Nested Loop
+                     ->  Materialize
+                           ->  Broadcast Motion 3:3  (slice3; segments: 3)
+                                 ->  Seq Scan on int8_tbl a
+                                       Filter: (q2 = 1)
+                     ->  Index Scan using tenk1_thous_tenthous on tenk1
+                           Index Cond: ((thousand = a.q1) AND (tenthous = b.q1))
  Optimizer: Postgres query optimizer
-(16 rows)
+(14 rows)
 
 reset enable_nestloop;
 --
@@ -6502,38 +6500,38 @@ from onek t1, tenk1 t2
 where exists (select 1 from tenk1 t3
               where t3.thousand = t1.unique1 and t3.tenthous = t2.hundred)
       and t1.unique1 < 1;
-                                                 QUERY PLAN                                                  
--------------------------------------------------------------------------------------------------------------
+                                                      QUERY PLAN                                                      
+----------------------------------------------------------------------------------------------------------------------
  Gather Motion 3:1  (slice1; segments: 3)
    Output: t1.unique1, t2.hundred
    ->  Nested Loop
          Output: t1.unique1, t2.hundred
          ->  Broadcast Motion 3:3  (slice2; segments: 3)
                Output: t1.unique1, t3.tenthous
-               ->  Hash Join
+               ->  Nested Loop
                      Output: t1.unique1, t3.tenthous
-                     Hash Cond: (t1.unique1 = t3.thousand)
-                     ->  Broadcast Motion 3:3  (slice3; segments: 3)
-                           Output: t1.unique1
-                           ->  Index Only Scan using onek_unique1 on public.onek t1
-                                 Output: t1.unique1
-                                 Index Cond: (t1.unique1 < 1)
-                     ->  Hash
+                     Join Filter: (t1.unique1 = t3.thousand)
+                     ->  HashAggregate
                            Output: t3.thousand, t3.tenthous
-                           ->  HashAggregate
+                           Group Key: t3.thousand, t3.tenthous
+                           ->  Redistribute Motion 3:3  (slice3; segments: 3)
                                  Output: t3.thousand, t3.tenthous
-                                 Group Key: t3.thousand, t3.tenthous
-                                 ->  Redistribute Motion 3:3  (slice4; segments: 3)
+                                 Hash Key: t3.thousand, t3.tenthous
+                                 ->  Index Only Scan using tenk1_thous_tenthous on public.tenk1 t3
                                        Output: t3.thousand, t3.tenthous
-                                       Hash Key: t3.thousand, t3.tenthous
-                                       ->  Index Only Scan using tenk1_thous_tenthous on public.tenk1 t3
-                                             Output: t3.thousand, t3.tenthous
-                                             Index Cond: (t3.thousand < 1)
+                                       Index Cond: (t3.thousand < 1)
+                     ->  Materialize
+                           Output: t1.unique1
+                           ->  Broadcast Motion 3:3  (slice4; segments: 3)
+                                 Output: t1.unique1
+                                 ->  Index Only Scan using onek_unique1 on public.onek t1
+                                       Output: t1.unique1
+                                       Index Cond: (t1.unique1 < 1)
          ->  Index Only Scan using tenk1_hundred on public.tenk1 t2
                Output: t2.hundred
                Index Cond: (t2.hundred = t3.tenthous)
  Optimizer: Postgres query optimizer
- Settings: enable_bitmapscan=off, enable_mergejoin=on, enable_nestloop=on, enable_seqscan=off, optimizer=off
+ Settings: enable_bitmapscan=off, enable_mergejoin=on, enable_nestloop=on, enable_seqscan=off, jit=off, optimizer=off
 (30 rows)
 
 -- ... unless it actually is unique

src/test/regress/expected/qp_correlated_query.out

@@ -377,38 +377,37 @@ select A.j from A, B, C where A.j = (select C.j from C where C.j = A.j and C.i i
 explain select A.i, B.i, C.j from A, B, C where A.j = (select C.j from C where C.j = A.j and C.i not in (select B.i from B where C.i = B.i and B.i !=10)) order by A.i, B.i, C.j limit 10;
                                                                               QUERY PLAN                                                                               
 -----------------------------------------------------------------------------------------------------------------------------------------------------------------------
- Limit  (cost=20000000100.29..20000000100.52 rows=10 width=12)
-   ->  Gather Motion 3:1  (slice3; segments: 3)  (cost=20000000100.29..20000000100.52 rows=10 width=12)
+ Limit  (cost=20000000003.86..20000000004.00 rows=10 width=12)
+   ->  Gather Motion 3:1  (slice1; segments: 3)  (cost=20000000003.86..20000000004.28 rows=30 width=12)
          Merge Key: a.i, b.i, c.j
-         ->  Limit  (cost=20000000100.29..20000000100.32 rows=4 width=12)
-               ->  Sort  (cost=20000000100.29..20000000100.43 rows=18 width=12)
+         ->  Limit  (cost=20000000003.86..20000000003.88 rows=10 width=12)
+               ->  Sort  (cost=20000000003.85..20000000003.90 rows=18 width=12)
                      Sort Key: a.i, b.i, c.j
-                     ->  Nested Loop  (cost=20000000000.00..20000000099.13 rows=18 width=12)
-                           ->  Nested Loop  (cost=10000000000.00..10000000093.81 rows=2 width=8)
-                                 ->  Broadcast Motion 3:3  (slice1; segments: 3)  (cost=0.00..90.51 rows=1 width=4)
-                                       ->  Seq Scan on a  (cost=0.00..90.47 rows=1 width=4)
+                     ->  Nested Loop  (cost=20000000000.00..20000000003.48 rows=18 width=12)
+                           ->  Nested Loop  (cost=10000000000.00..10000000002.08 rows=2 width=8)
+                                 ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.04 rows=1 width=4)
+                                       ->  Seq Scan on a  (cost=0.00..1.02 rows=1 width=4)
                                              Filter: (j = (SubPlan 2))
                                              SubPlan 2
-                                               ->  Result  (cost=0.00..17.68 rows=5 width=4)
+                                               ->  Result  (cost=0.00..1.16 rows=4 width=4)
                                                      Filter: (c_1.j = a.j)
-                                                     ->  Materialize  (cost=0.00..17.68 rows=5 width=4)
-                                                           ->  Broadcast Motion 3:3  (slice4; segments: 3)  (cost=0.00..17.66 rows=2 width=4)
-                                                                 ->  Seq Scan on c c_1  (cost=0.00..17.57 rows=2 width=4)
+                                                     ->  Materialize  (cost=0.00..1.11 rows=4 width=4)
+                                                           ->  Broadcast Motion 3:3  (slice3; segments: 3)  (cost=0.00..1.09 rows=4 width=4)
+                                                                 ->  Seq Scan on c c_1  (cost=0.00..1.03 rows=2 width=4)
                                                                        Filter: (SubPlan 1)
                                                                        SubPlan 1
-                                                                         ->  Result  (cost=0.00..3.20 rows=5 width=4)
+                                                                         ->  Result  (cost=0.00..1.17 rows=5 width=4)
                                                                                Filter: (c_1.i = b_1.i)
-                                                                               ->  Materialize  (cost=0.00..3.20 rows=5 width=4)
-                                                                                     ->  Broadcast Motion 3:3  (slice5; segments: 3)  (cost=0.00..3.17 rows=2 width=4)
-                                                                                           ->  Seq Scan on b b_1  (cost=0.00..3.08 rows=2 width=4)
+                                                                               ->  Materialize  (cost=0.00..1.12 rows=5 width=4)
+                                                                                     ->  Broadcast Motion 3:3  (slice4; segments: 3)  (cost=0.00..1.09 rows=5 width=4)
+                                                                                           ->  Seq Scan on b b_1  (cost=0.00..1.02 rows=2 width=4)
                                                                                                  Filter: (i <> 10)
-                                 ->  Materialize  (cost=0.00..3.09 rows=2 width=4)
-                                       ->  Seq Scan on b  (cost=0.00..3.06 rows=2 width=4)
-                           ->  Materialize  (cost=0.00..3.58 rows=9 width=4)
-                                 ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..3.45 rows=9 width=4)
-                                       ->  Seq Scan on c  (cost=0.00..3.09 rows=3 width=4)
+                                 ->  Seq Scan on b  (cost=0.00..1.02 rows=2 width=4)
+                           ->  Materialize  (cost=0.00..1.19 rows=9 width=4)
+                                 ->  Broadcast Motion 3:3  (slice5; segments: 3)  (cost=0.00..1.15 rows=9 width=4)
+                                       ->  Seq Scan on c  (cost=0.00..1.03 rows=3 width=4)
  Optimizer: Postgres query optimizer
-(31 rows)
+(30 rows)
 
 select A.i, B.i, C.j from A, B, C where A.j = (select C.j from C where C.j = A.j and C.i not in (select B.i from B where C.i = B.i and B.i !=10)) order by A.i, B.i, C.j limit 10;
  i | i  | j  
@@ -597,37 +596,36 @@ ERROR:  correlated subquery with skip-level correlations is not supported
 explain select A.i, B.i, C.j from A, B, C where A.j = (select C.j from C where C.j = A.j and C.i = any (select B.i from B where C.i = B.i and B.i !=10)) order by A.i, B.i, C.j limit 10;
                                                                     QUERY PLAN                                                                     
 ---------------------------------------------------------------------------------------------------------------------------------------------------
- Limit  (cost=20000000045.25..20000000045.47 rows=10 width=12)
-   ->  Gather Motion 3:1  (slice3; segments: 3)  (cost=20000000045.25..20000000045.47 rows=10 width=12)
+ Limit  (cost=20000000003.86..20000000004.00 rows=10 width=12)
+   ->  Gather Motion 3:1  (slice1; segments: 3)  (cost=20000000003.86..20000000004.28 rows=30 width=12)
          Merge Key: a.i, b.i, c.j
-         ->  Limit  (cost=20000000045.25..20000000045.27 rows=4 width=12)
-               ->  Sort  (cost=20000000045.25..20000000045.38 rows=18 width=12)
+         ->  Limit  (cost=20000000003.86..20000000003.88 rows=10 width=12)
+               ->  Sort  (cost=20000000003.85..20000000003.90 rows=18 width=12)
                      Sort Key: a.i, b.i, c.j
-                     ->  Nested Loop  (cost=20000000000.00..20000000044.08 rows=18 width=12)
-                           ->  Nested Loop  (cost=10000000000.00..10000000038.76 rows=2 width=8)
-                                 ->  Broadcast Motion 3:3  (slice1; segments: 3)  (cost=0.00..35.46 rows=1 width=4)
-                                       ->  Seq Scan on a  (cost=0.00..35.42 rows=1 width=4)
+                     ->  Nested Loop  (cost=20000000000.00..20000000003.48 rows=18 width=12)
+                           ->  Nested Loop  (cost=10000000000.00..10000000002.08 rows=2 width=8)
+                                 ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.04 rows=1 width=4)
+                                       ->  Seq Scan on a  (cost=0.00..1.02 rows=1 width=4)
                                              Filter: (j = (SubPlan 1))
                                              SubPlan 1
-                                               ->  Hash Semi Join  (cost=3.26..6.67 rows=6 width=4)
+                                               ->  Hash Semi Join  (cost=1.18..2.55 rows=5 width=4)
                                                      Hash Cond: (c_1.i = b_1.i)
-                                                     ->  Result  (cost=0.00..3.31 rows=9 width=8)
+                                                     ->  Result  (cost=0.00..1.28 rows=9 width=8)
                                                            Filter: (c_1.j = a.j)
-                                                           ->  Materialize  (cost=0.00..3.31 rows=9 width=8)
-                                                                 ->  Broadcast Motion 3:3  (slice4; segments: 3)  (cost=0.00..3.27 rows=3 width=8)
-                                                                       ->  Seq Scan on c c_1  (cost=0.00..3.09 rows=3 width=8)
-                                                     ->  Hash  (cost=3.20..3.20 rows=2 width=4)
-                                                           ->  Materialize  (cost=0.00..3.20 rows=5 width=4)
-                                                                 ->  Broadcast Motion 3:3  (slice5; segments: 3)  (cost=0.00..3.17 rows=2 width=4)
-                                                                       ->  Seq Scan on b b_1  (cost=0.00..3.08 rows=2 width=4)
+                                                           ->  Materialize  (cost=0.00..1.19 rows=9 width=8)
+                                                                 ->  Broadcast Motion 3:3  (slice3; segments: 3)  (cost=0.00..1.15 rows=9 width=8)
+                                                                       ->  Seq Scan on c c_1  (cost=0.00..1.03 rows=3 width=8)
+                                                     ->  Hash  (cost=1.12..1.12 rows=5 width=4)
+                                                           ->  Materialize  (cost=0.00..1.12 rows=5 width=4)
+                                                                 ->  Broadcast Motion 3:3  (slice4; segments: 3)  (cost=0.00..1.09 rows=5 width=4)
+                                                                       ->  Seq Scan on b b_1  (cost=0.00..1.02 rows=2 width=4)
                                                                              Filter: (i <> 10)
-                                 ->  Materialize  (cost=0.00..3.09 rows=2 width=4)
-                                       ->  Seq Scan on b  (cost=0.00..3.06 rows=2 width=4)
-                           ->  Materialize  (cost=0.00..3.58 rows=9 width=4)
-                                 ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..3.45 rows=9 width=4)
-                                       ->  Seq Scan on c  (cost=0.00..3.09 rows=3 width=4)
+                                 ->  Seq Scan on b  (cost=0.00..1.02 rows=2 width=4)
+                           ->  Materialize  (cost=0.00..1.19 rows=9 width=4)
+                                 ->  Broadcast Motion 3:3  (slice5; segments: 3)  (cost=0.00..1.15 rows=9 width=4)
+                                       ->  Seq Scan on c  (cost=0.00..1.03 rows=3 width=4)
  Optimizer: Postgres query optimizer
-(30 rows)
+(29 rows)
 
 select A.i, B.i, C.j from A, B, C where A.j = (select C.j from C where C.j = A.j and C.i = any (select B.i from B where C.i = B.i and B.i !=10)) order by A.i, B.i, C.j limit 10;
  i  | i  |  j  

src/test/regress/expected/qp_subquery.out

@@ -945,18 +945,17 @@ select t1.a, t2.b from t1 inner join t2 on  (t1.a=t2.a) where ((t1.a,t2.b) not i
 explain select t1.a, t2.b from t1, t2 where t1.a=t2.a or ((t1.a,t2.b) not in (select i1.a,i1.b from i1));
                                            QUERY PLAN                                           
 ------------------------------------------------------------------------------------------------
- Gather Motion 3:1  (slice1; segments: 3)  (cost=10000000000.00..10000000002.15 rows=3 width=8)
-   ->  Nested Loop  (cost=10000000000.00..10000000002.10 rows=1 width=8)
+ Gather Motion 3:1  (slice1; segments: 3)  (cost=10000000000.00..10000000002.10 rows=3 width=8)
+   ->  Nested Loop  (cost=10000000000.00..10000000002.05 rows=1 width=8)
          Join Filter: ((t1.a = t2.a) OR (NOT (hashed SubPlan 1)))
-         ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.05 rows=3 width=4)
+         ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.03 rows=1 width=4)
                ->  Seq Scan on t1  (cost=0.00..1.01 rows=1 width=4)
-         ->  Materialize  (cost=0.00..1.01 rows=1 width=8)
-               ->  Seq Scan on t2  (cost=0.00..1.01 rows=1 width=8)
+         ->  Seq Scan on t2  (cost=0.00..1.01 rows=1 width=8)
          SubPlan 1
            ->  Broadcast Motion 3:3  (slice3; segments: 3)  (cost=0.00..1.01 rows=1 width=8)
                  ->  Seq Scan on i1  (cost=0.00..1.01 rows=1 width=8)
  Optimizer: Postgres query optimizer
-(11 rows)
+(10 rows)
 
 --
 -- not in subquery involving vars from different rels with left join. 

src/test/regress/expected/qp_subquery_optimizer.out

@@ -943,18 +943,17 @@ select t1.a, t2.b from t1 inner join t2 on  (t1.a=t2.a) where ((t1.a,t2.b) not i
 explain select t1.a, t2.b from t1, t2 where t1.a=t2.a or ((t1.a,t2.b) not in (select i1.a,i1.b from i1));
                                            QUERY PLAN                                           
 ------------------------------------------------------------------------------------------------
- Gather Motion 3:1  (slice1; segments: 3)  (cost=10000000001.01..10000000003.14 rows=4 width=8)
-   ->  Nested Loop  (cost=10000000001.01..10000000003.14 rows=2 width=8)
+ Gather Motion 3:1  (slice1; segments: 3)  (cost=10000000000.00..10000000002.10 rows=3 width=8)
+   ->  Nested Loop  (cost=10000000000.00..10000000002.05 rows=1 width=8)
          Join Filter: ((t1.a = t2.a) OR (NOT (hashed SubPlan 1)))
-         ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.05 rows=1 width=4)
+         ->  Broadcast Motion 3:3  (slice2; segments: 3)  (cost=0.00..1.03 rows=1 width=4)
                ->  Seq Scan on t1  (cost=0.00..1.01 rows=1 width=4)
-         ->  Materialize  (cost=0.00..1.01 rows=1 width=8)
-               ->  Seq Scan on t2  (cost=0.00..1.01 rows=1 width=8)
+         ->  Seq Scan on t2  (cost=0.00..1.01 rows=1 width=8)
          SubPlan 1
            ->  Broadcast Motion 3:3  (slice3; segments: 3)  (cost=0.00..1.01 rows=1 width=8)
                  ->  Seq Scan on i1  (cost=0.00..1.01 rows=1 width=8)
  Optimizer: Postgres query optimizer
-(11 rows)
+(10 rows)
 
 --
 -- not in subquery involving vars from different rels with left join. 

src/test/regress/expected/rpt.out

@@ -925,17 +925,17 @@ explain (costs off) insert into t_replicate_volatile select nextval('seq_for_ins
 
 -- update & delete
 explain (costs off) update t_replicate_volatile set a = 1 where b > random();
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) update t_replicate_volatile set a = 1 from t_replicate_volatile x where x.a + random() = t_replicate_volatile.b;
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) update t_replicate_volatile set a = 1 from t_hashdist x where x.a + random() = t_replicate_volatile.b;
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) delete from t_replicate_volatile where a < random();
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) delete from t_replicate_volatile using t_replicate_volatile x where t_replicate_volatile.a + x.b < random();
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) update t_replicate_volatile set a = random();
-ERROR:  could not devise a plan. (cdbpath.c:2408)
+ERROR:  could not devise a plan. (cdbpath.c:2440)
 -- limit
 explain (costs off) insert into t_replicate_volatile select * from t_replicate_volatile limit random();
                                    QUERY PLAN                                    

src/test/regress/expected/rpt_optimizer.out

@@ -905,20 +905,20 @@ explain (costs off) insert into t_replicate_volatile select nextval('seq_for_ins
 
 -- update & delete
 explain (costs off) update t_replicate_volatile set a = 1 where b > random();
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) update t_replicate_volatile set a = 1 from t_replicate_volatile x where x.a + random() = t_replicate_volatile.b;
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) update t_replicate_volatile set a = 1 from t_hashdist x where x.a + random() = t_replicate_volatile.b;
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) delete from t_replicate_volatile where a < random();
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) delete from t_replicate_volatile using t_replicate_volatile x where t_replicate_volatile.a + x.b < random();
-ERROR:  could not devise a plan (cdbpath.c:2545)
+ERROR:  could not devise a plan (cdbpath.c:2577)
 explain (costs off) update t_replicate_volatile set a = random();
-ERROR:  could not devise a plan. (cdbpath.c:2408)
+ERROR:  could not devise a plan. (cdbpath.c:2440)
 -- limit
 explain (costs off) insert into t_replicate_volatile select * from t_replicate_volatile limit random();
-                                      QUERY PLAN                                       
+                                   QUERY PLAN                                    
 ---------------------------------------------------------------------------------------
  Insert on t_replicate_volatile
    ->  Result

src/test/regress/expected/stats_ext.out

@@ -362,7 +362,7 @@ SELECT * FROM check_estimated_rows('SELECT * FROM functional_dependencies WHERE
 SELECT * FROM check_estimated_rows('SELECT * FROM functional_dependencies WHERE a = 1 AND b = ''1'' AND c = 1');
  estimated | actual 
 -----------+--------
-         3 |      1
+         1 |      1
 (1 row)
 
 -- create statistics
@@ -377,7 +377,7 @@ SELECT * FROM check_estimated_rows('SELECT * FROM functional_dependencies WHERE
 SELECT * FROM check_estimated_rows('SELECT * FROM functional_dependencies WHERE a = 1 AND b = ''1'' AND c = 1');
  estimated | actual 
 -----------+--------
-         3 |      1
+         1 |      1
 (1 row)
 
 -- a => b, a => c, b => c
@@ -389,13 +389,13 @@ ANALYZE functional_dependencies;
 SELECT * FROM check_estimated_rows('SELECT * FROM functional_dependencies WHERE a = 1 AND b = ''1''');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM functional_dependencies WHERE a = 1 AND b = ''1'' AND c = 1');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 -- create statistics
@@ -451,7 +451,7 @@ SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a = 1 AND b =
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a = 1 AND b = ''1'' AND c = 1');
  estimated | actual 
 -----------+--------
-         3 |      1
+         1 |      1
 (1 row)
 
 -- create statistics
@@ -466,7 +466,7 @@ SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a = 1 AND b =
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a = 1 AND b = ''1'' AND c = 1');
  estimated | actual 
 -----------+--------
-         3 |      1
+         1 |      1
 (1 row)
 
 -- 100 distinct combinations, all in the MCV list
@@ -478,37 +478,37 @@ ANALYZE mcv_lists;
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a = 1 AND b = ''1''');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a < 1 AND b < ''1''');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a <= 0 AND b <= ''0''');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a = 1 AND b = ''1'' AND c = 1');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a < 5 AND b < ''1'' AND c < 5');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a <= 4 AND b <= ''0'' AND c <= 4');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 -- create statistics
@@ -566,7 +566,7 @@ ALTER TABLE mcv_lists ALTER COLUMN c TYPE numeric;
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a = 1 AND b = ''1''');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 ANALYZE mcv_lists;
@@ -590,13 +590,13 @@ ANALYZE mcv_lists;
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a IS NULL AND b IS NULL');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists WHERE a IS NULL AND b IS NULL AND c IS NULL');
  estimated | actual 
 -----------+--------
-         3 |     50
+         1 |     50
 (1 row)
 
 -- create statistics
@@ -690,19 +690,19 @@ SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists_bool WHERE a AND b A
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists_bool WHERE NOT a AND b AND c');
  estimated | actual 
 -----------+--------
-         3 |      0
+         1 |      0
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists_bool WHERE NOT a AND NOT b AND c');
  estimated | actual 
 -----------+--------
-         3 |      0
+         1 |      0
 (1 row)
 
 SELECT * FROM check_estimated_rows('SELECT * FROM mcv_lists_bool WHERE NOT a AND b AND NOT c');
  estimated | actual 
 -----------+--------
-         3 |      0
+         1 |      0
 (1 row)
 
 -- Permission tests. Users should not be able to see specific data values in

src/test/regress/sql/ao_checksum_corruption.sql

@@ -69,7 +69,7 @@ RETURNS integer as $$
       f.close()
 
   return 0
-$$ LANGUAGE plpython3u;
+$$ LANGUAGE plpython3u EXECUTE ON ALL SEGMENTS;
 
 -- Corrupt a file by replacing the last occurrence of 'str' within the file
 -- with 'replacement'
@@ -97,48 +97,48 @@ RETURNS integer as $$
       f.close()
 
   return 0
-$$ LANGUAGE plpython3u;
+$$ LANGUAGE plpython3u EXECUTE ON ALL SEGMENTS;
 
 
 -- Large content, corrupt block header
 create table corrupt_header_large_co(comment bytea ) with (appendonly=true, orientation=column, checksum=true) DISTRIBUTED RANDOMLY;
 insert into corrupt_header_large_co select ("decode"(repeat('a',33554432),'escape')) from generate_series(1,8)  ;
-select SUM(corrupt_file(get_aoseg1_path('corrupt_header_large_co'), 8)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_header_large_co'), 8));
 
 SELECT COUNT(*) FROM corrupt_header_large_co;
 
 -- Large content, corrupt content
 create table corrupt_content_large_co(comment bytea ) with (appendonly=true, orientation=column, checksum=true) DISTRIBUTED RANDOMLY;
 insert into corrupt_content_large_co select ("decode"(repeat('a',33554432),'escape')) from generate_series(1,8)  ;
-select SUM(corrupt_file(get_aoseg1_path('corrupt_content_large_co'), -3)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_content_large_co'), -3));
 
 SELECT COUNT(*) FROM corrupt_content_large_co;
 
 -- Small content, corrupt block header
 create table corrupt_header_small_co(a int) with (appendonly=true, orientation=column, checksum=true);
 insert into corrupt_header_small_co values (1),(1),(1),(-1),(1),(1),(1),(2),(2),(2),(2),(2),(2),(2),(33),(3),(3),(3),(1),(8),(19),(20),(31),(32),(33),(34),(5),(5),(5),(5),(5),(6),(6),(6),(6),(6),(6),(7),(7),(7),(7),(7),(7),(7),(7), (null),(7),(7),(7),(null),(8),(8),(8),(8),(8),(8),(4),(4),(null),(4),(17),(17),(17),(null),(null),(null);
-select SUM(corrupt_file(get_aoseg1_path('corrupt_header_small_co'), 8)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_header_small_co'), 8));
 
 SELECT COUNT(*) FROM corrupt_header_small_co;
 
 -- Small content, corrupt content
 create table corrupt_content_small_co(a int) with (appendonly=true, orientation=column, checksum=true);
 insert into corrupt_content_small_co values (1),(1),(1),(-1),(1),(1),(1),(2),(2),(2),(2),(2),(2),(2),(33),(3),(3),(3),(1),(8),(19),(20),(31),(32),(33),(34),(5),(5),(5),(5),(5),(6),(6),(6),(6),(6),(6),(7),(7),(7),(7),(7),(7),(7),(7), (null),(7),(7),(7),(null),(8),(8),(8),(8),(8),(8),(4),(4),(null),(4),(17),(17),(17),(null),(null),(null);
-select SUM(corrupt_file(get_aoseg1_path('corrupt_content_small_co'), -3)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_content_small_co'), -3));
 
 SELECT COUNT(*) FROM corrupt_content_small_co;
 
 -- Row-oriented, Small content, corrupt block header
 create table corrupt_header_small_ao(a int) with (appendonly=true, orientation=row, checksum=true);
 insert into corrupt_header_small_ao values (1),(1),(1),(-1),(1),(1),(1),(2),(2),(2),(2),(2),(2),(2),(33),(3),(3),(3),(1),(8),(19),(20),(31),(32),(33),(34),(5),(5),(5),(5),(5),(6),(6),(6),(6),(6),(6),(7),(7),(7),(7),(7),(7),(7),(7), (null),(7),(7),(7),(null),(8),(8),(8),(8),(8),(8),(4),(4),(null),(4),(17),(17),(17),(null),(null),(null);
-select SUM(corrupt_file(get_aoseg1_path('corrupt_header_small_ao'), 8)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_header_small_ao'), 8));
 
 SELECT COUNT(*) FROM corrupt_header_small_ao;
 
 -- Row-oriented, Small content, corrupt content
 create table corrupt_content_small_ao(a int) with (appendonly=true, orientation=row, checksum=true);
 insert into corrupt_content_small_ao values (1),(1),(1),(-1),(1),(1),(1),(2),(2),(2),(2),(2),(2),(2),(33),(3),(3),(3),(1),(8),(19),(20),(31),(32),(33),(34),(5),(5),(5),(5),(5),(6),(6),(6),(6),(6),(6),(7),(7),(7),(7),(7),(7),(7),(7), (null),(7),(7),(7),(null),(8),(8),(8),(8),(8),(8),(4),(4),(null),(4),(17),(17),(17),(null),(null),(null);
-select SUM(corrupt_file(get_aoseg1_path('corrupt_content_small_ao'), -3)) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('corrupt_content_small_ao'), -3));
 
 SELECT COUNT(*) FROM corrupt_content_small_ao;
 
@@ -149,7 +149,7 @@ insert into appendonly_verify_block_checksums_co
   select 'abcdefghijlmnopqrstuvxyz' from generate_series(1, 5);
 
 -- Corrupt the table by flip the 'xyz' on the last row with ###
-select SUM(corrupt_file(get_aoseg1_path('appendonly_verify_block_checksums_co'), 'xyz', '###')) from gp_dist_random('gp_id');
+select SUM(corrupt_file(get_aoseg1_path('appendonly_verify_block_checksums_co'), 'xyz', '###'));
 
 -- Fails, checksum is wrong.
 SELECT * FROM appendonly_verify_block_checksums_co;

src/test/regress/sql/equivclass.sql

@@ -103,6 +103,7 @@ create table ec1 (ff int8 primary key, f1 int8alias1, f2 int8alias2);
 create table ec2 (xf int8 primary key, x1 int8alias1, x2 int8alias2);
 
 -- for the moment we only want to look at nestloop plans
+set enable_nestloop = on;
 set enable_hashjoin = off;
 set enable_mergejoin = off;