Multiple TOAST tables support for TOAST API

TOAST mechanics supposes that all TOASTed columns of single table use one TOAST
table entity for TOASTed records. This TOAST table has strictly defined structure.
This is fine for reference TOAST mechanics but TOAST API supposes that custom
Toasters could use TOAST tables with different structure along with different
columns in a table could have different Toasters assigned. This requires that
regular table should have possibility to access several (different) TOAST tables.

Using Toasters with TOAST tables structure different from regular (generic) on
one table leads to segmentation fault. This patch is adressed to solve this problem.

This patch provides multiple TOAST tables support. Multiple TOAST tables support
consists of 2 new pg_class columns - reltoasterids and reltoastrelids. These
two columns contain arrays of Toaster OIDs and TOAST relation OIDs used by table
columns. New pg_class columns affect a lot of source code.

Author: Teodor Sigaev <teodor@sigaev.ru>
Author: Oleg Bartunov <obartunov@postgrespro.ru>
Author: Nikita Glukhov <n.gluhov@postgrespro.ru>
Author: Nikita Malakhov <n.malakhov@postgrespro.ru>

Author: Nikita Malakhov

contrib/amcheck/t/001_verify_heapam.pl

@@ -97,7 +97,7 @@ sub get_toast_for
 		SELECT 'pg_toast.' || t.relname
 			FROM pg_catalog.pg_class c, pg_catalog.pg_class t
 			WHERE c.relname = '$relname'
-			  AND c.reltoastrelid = t.oid));
+			  AND t.oid = ANY(c.reltoastrelids)));
 }
 
 # (Re)create and populate a test table of the given name.

contrib/amcheck/verify_heapam.c

@@ -114,7 +114,13 @@ typedef struct HeapCheckContext
 	Relation	toast_rel;
 	Relation   *toast_indexes;
 	Relation	valid_toast_index;
-	int			num_toast_indexes;
+   struct ToastRelContext
+   {
+      Relation	toast_rel;
+      Relation	*toast_indexes;
+      Relation	valid_toast_index;
+      int		num_toast_indexes;
+   } *toast_rels;
 
 	/* Values for iterating over pages in the relation */
 	BlockNumber blkno;
@@ -154,6 +160,7 @@ typedef struct HeapCheckContext
 /* Internal implementation */
 static void check_tuple(HeapCheckContext *ctx);
 static void check_toast_tuple(HeapTuple toasttup, HeapCheckContext *ctx,
+							  struct ToastRelContext *toast_rel,
 							  ToastedAttribute *ta, int32 *expected_chunk_seq,
 							  uint32 extsize);
 
@@ -364,29 +371,48 @@ verify_heapam(PG_FUNCTION_ARGS)
 		last_block = (BlockNumber) lb;
 	}
 
+	/* XXX spaces */
 	/* Optionally open the toast relation, if any. */
-	if (ctx.rel->rd_rel->reltoastrelid && check_toast)
+	if (ctx.rel->rd_ntoasters > 0 && check_toast)
 	{
-		int			offset;
-
-		/* Main relation has associated toast relation */
-		ctx.toast_rel = table_open(ctx.rel->rd_rel->reltoastrelid,
-								   AccessShareLock);
-		offset = toast_open_indexes(ctx.toast_rel,
-									AccessShareLock,
-									&(ctx.toast_indexes),
-									&(ctx.num_toast_indexes));
-		ctx.valid_toast_index = ctx.toast_indexes[offset];
+		ctx.toast_rels = palloc(sizeof(*ctx.toast_rels) * ctx.rel->rd_ntoasters);
+
+      for (int i = 0; i < ctx.rel->rd_ntoasters; i++)
+      {
+         Oid toastrelid = ctx.rel->rd_toastrelids[i];
+
+         if (OidIsValid(toastrelid))
+         {
+            int                     offset;
+
+            /* Main relation has associated toast relation */
+            ctx.toast_rels[i].toast_rel =
+               table_open(toastrelid, AccessShareLock);
+
+            offset = toast_open_indexes(ctx.toast_rels[i].toast_rel,
+                                        AccessShareLock,
+                                        &(ctx.toast_rels[i].toast_indexes),
+                                        &(ctx.toast_rels[i].num_toast_indexes));
+
+            ctx.toast_rels[i].valid_toast_index =
+               ctx.toast_rels[i].toast_indexes[offset];
+         }
+         else
+         {
+            ctx.toast_rels[i].toast_rel = NULL;
+            ctx.toast_rels[i].toast_indexes = NULL;
+            ctx.toast_rels[i].num_toast_indexes = 0;
+            ctx.toast_rels[i].valid_toast_index = 0;
+         }
+      }
 	}
 	else
 	{
 		/*
 		 * Main relation has no associated toast relation, or we're
 		 * intentionally skipping it.
 		 */
-		ctx.toast_rel = NULL;
-		ctx.toast_indexes = NULL;
-		ctx.num_toast_indexes = 0;
+		ctx.toast_rels = NULL;
 	}
 
 	update_cached_xid_range(&ctx);
@@ -535,12 +561,21 @@ verify_heapam(PG_FUNCTION_ARGS)
 		ReleaseBuffer(vmbuffer);
 
 	/* Close the associated toast table and indexes, if any. */
-	if (ctx.toast_indexes)
-		toast_close_indexes(ctx.toast_indexes, ctx.num_toast_indexes,
-							AccessShareLock);
-	if (ctx.toast_rel)
-		table_close(ctx.toast_rel, AccessShareLock);
-
+   if (ctx.toast_rels)
+   {
+      for (int i = 0; i < ctx.rel->rd_ntoasters; i++)
+      {
+         if (ctx.toast_rels[i].toast_indexes)
+            toast_close_indexes(ctx.toast_rels[i].toast_indexes,
+               ctx.toast_rels[i].num_toast_indexes,
+               AccessShareLock);
+
+         if (ctx.toast_rels[i].toast_rel)
+            table_close(ctx.toast_rels[i].toast_rel, AccessShareLock);
+      }
+
+      pfree(ctx.toast_rels);
+   }
 	/* Close the main relation */
 	relation_close(ctx.rel, AccessShareLock);
 
@@ -1157,6 +1192,7 @@ check_tuple_visibility(HeapCheckContext *ctx)
  */
 static void
 check_toast_tuple(HeapTuple toasttup, HeapCheckContext *ctx,
+				  struct ToastRelContext *toast_rel,
 				  ToastedAttribute *ta, int32 *expected_chunk_seq,
 				  uint32 extsize)
 {
@@ -1169,7 +1205,7 @@ check_toast_tuple(HeapTuple toasttup, HeapCheckContext *ctx,
 
 	/* Sanity-check the sequence number. */
 	chunk_seq = DatumGetInt32(fastgetattr(toasttup, 2,
-										  ctx->toast_rel->rd_att, &isnull));
+										  toast_rel->toast_rel->rd_att, &isnull));
 	if (isnull)
 	{
 		report_toast_corruption(ctx, ta,
@@ -1189,7 +1225,7 @@ check_toast_tuple(HeapTuple toasttup, HeapCheckContext *ctx,
 
 	/* Sanity-check the chunk data. */
 	chunk = DatumGetPointer(fastgetattr(toasttup, 3,
-										ctx->toast_rel->rd_att, &isnull));
+										toast_rel->toast_rel->rd_att, &isnull));
 	if (isnull)
 	{
 		report_toast_corruption(ctx, ta,
@@ -1328,7 +1364,7 @@ check_tuple_attribute(HeapCheckContext *ctx)
 	{
 		uint8		va_tag = VARTAG_EXTERNAL(tp + ctx->offset);
 
-		if (va_tag != VARTAG_ONDISK)
+		if (va_tag != VARTAG_ONDISK && va_tag != VARTAG_CUSTOM)
 		{
 			report_corruption(ctx,
 							  psprintf("toasted attribute has unexpected TOAST tag %u",
@@ -1422,7 +1458,7 @@ check_tuple_attribute(HeapCheckContext *ctx)
 	}
 
 	/* The relation better have a toast table */
-	if (!ctx->rel->rd_rel->reltoastrelid)
+	if (ctx->rel->rd_ntoasters <= 0)
 	{
 		report_corruption(ctx,
 						  psprintf("toast value %u is external but relation has no toast relation",
@@ -1431,7 +1467,7 @@ check_tuple_attribute(HeapCheckContext *ctx)
 	}
 
 	/* If we were told to skip toast checking, then we're done. */
-	if (ctx->toast_rel == NULL)
+	if (ctx->toast_rels == NULL)
 		return true;
 
 	/*
@@ -1472,6 +1508,7 @@ check_toasted_attribute(HeapCheckContext *ctx, ToastedAttribute *ta)
 	uint32		extsize;
 	int32		expected_chunk_seq = 0;
 	int32		last_chunk_seq;
+	struct ToastRelContext *toast_rel = NULL;
 
 	extsize = VARATT_EXTERNAL_GET_EXTSIZE(ta->toast_pointer);
 	last_chunk_seq = (extsize - 1) / TOAST_MAX_CHUNK_SIZE;
@@ -1484,13 +1521,27 @@ check_toasted_attribute(HeapCheckContext *ctx, ToastedAttribute *ta)
 				BTEqualStrategyNumber, F_OIDEQ,
 				ObjectIdGetDatum(ta->toast_pointer.va_valueid));
 
+   for (int i = 0; i < ctx->rel->rd_ntoasters; i++)
+   {
+      if (ta->toast_pointer.va_toastrelid == ctx->rel->rd_toastrelids[i])
+      {
+         toast_rel = &ctx->toast_rels[i];
+         break;
+      }
+   }
+
+   if (!toast_rel)
+      report_corruption(ctx,
+                        psprintf("toast pointer's va_toastrelid %u does not contain in reltoastrelids",
+                        ta->toast_pointer.va_toastrelid));
+
 	/*
 	 * Check if any chunks for this toasted object exist in the toast table,
 	 * accessible via the index.
 	 */
 	init_toast_snapshot(&SnapshotToast);
-	toastscan = systable_beginscan_ordered(ctx->toast_rel,
-										   ctx->valid_toast_index,
+	toastscan = systable_beginscan_ordered(toast_rel->toast_rel,
+										   toast_rel->valid_toast_index,
 										   &SnapshotToast, 1,
 										   &toastkey);
 	found_toasttup = false;
@@ -1499,7 +1550,7 @@ check_toasted_attribute(HeapCheckContext *ctx, ToastedAttribute *ta)
 									 ForwardScanDirection)) != NULL)
 	{
 		found_toasttup = true;
-		check_toast_tuple(toasttup, ctx, ta, &expected_chunk_seq, extsize);
+		check_toast_tuple(toasttup, ctx, toast_rel, ta, &expected_chunk_seq, extsize);
 	}
 	systable_endscan_ordered(toastscan);
 

contrib/pg_visibility/expected/pg_visibility.out

@@ -11,22 +11,22 @@ SAVEPOINT q; SELECT * FROM pg_visibility_map(:oid); ROLLBACK TO q;
 ERROR:  XX000
 -- ERROR:  could not open relation with OID 16xxx
 SAVEPOINT q; SELECT 1; ROLLBACK TO q;
- ?column? 
+ ?column?
 ----------
         1
 (1 row)
 
 SAVEPOINT q; SELECT 1; ROLLBACK TO q;
- ?column? 
+ ?column?
 ----------
         1
 (1 row)
 
 SELECT pg_relation_size(:oid), pg_relation_filepath(:oid),
   has_table_privilege(:oid, 'SELECT');
- pg_relation_size | pg_relation_filepath | has_table_privilege 
+ pg_relation_size | pg_relation_filepath | has_table_privilege
 ------------------+----------------------+---------------------
-                  |                      | 
+                  |                      |
 (1 row)
 
 SELECT * FROM pg_visibility_map(:oid);
@@ -132,42 +132,42 @@ alter table regular_table alter column b set storage external;
 insert into regular_table values (1, repeat('one', 1000)), (2, repeat('two', 1000));
 vacuum (disable_page_skipping) regular_table;
 select count(*) > 0 from pg_visibility('regular_table');
- ?column? 
+ ?column?
 ----------
  t
 (1 row)
 
-select count(*) > 0 from pg_visibility((select reltoastrelid from pg_class where relname = 'regular_table'));
- ?column? 
+select count(*) > 0 from pg_visibility((select reltoastrelids[1] from pg_class where relname = 'regular_table'));
+ ?column?
 ----------
  t
 (1 row)
 
 truncate regular_table;
 select count(*) > 0 from pg_visibility('regular_table');
- ?column? 
+ ?column?
 ----------
  f
 (1 row)
 
-select count(*) > 0 from pg_visibility((select reltoastrelid from pg_class where relname = 'regular_table'));
- ?column? 
+select count(*) > 0 from pg_visibility((select reltoastrelids[1] from pg_class where relname = 'regular_table'));
+ ?column?
 ----------
  f
 (1 row)
 
 create materialized view matview_visibility_test as select * from regular_table;
 vacuum (disable_page_skipping) matview_visibility_test;
 select count(*) > 0 from pg_visibility('matview_visibility_test');
- ?column? 
+ ?column?
 ----------
  f
 (1 row)
 
 insert into regular_table values (1), (2);
 refresh materialized view matview_visibility_test;
 select count(*) > 0 from pg_visibility('matview_visibility_test');
- ?column? 
+ ?column?
 ----------
  t
 (1 row)
@@ -176,32 +176,32 @@ select count(*) > 0 from pg_visibility('matview_visibility_test');
 insert into test_partition values (1);
 vacuum (disable_page_skipping) test_partition;
 select count(*) > 0 from pg_visibility('test_partition', 0);
- ?column? 
+ ?column?
 ----------
  t
 (1 row)
 
 select count(*) > 0 from pg_visibility_map('test_partition');
- ?column? 
+ ?column?
 ----------
  t
 (1 row)
 
 select count(*) > 0 from pg_visibility_map_summary('test_partition');
- ?column? 
+ ?column?
 ----------
  t
 (1 row)
 
 select * from pg_check_frozen('test_partition'); -- hopefully none
- t_ctid 
+ t_ctid
 --------
 (0 rows)
 
 select pg_truncate_visibility_map('test_partition');
- pg_truncate_visibility_map 
+ pg_truncate_visibility_map
 ----------------------------
- 
+
 (1 row)
 
 -- test copy freeze
@@ -213,15 +213,15 @@ truncate copyfreeze;
 copy copyfreeze from stdin freeze;
 commit;
 select * from pg_visibility_map('copyfreeze');
- blkno | all_visible | all_frozen 
+ blkno | all_visible | all_frozen
 -------+-------------+------------
      0 | t           | t
      1 | t           | t
      2 | t           | t
 (3 rows)
 
 select * from pg_check_frozen('copyfreeze');
- t_ctid 
+ t_ctid
 --------
 (0 rows)
 
@@ -235,15 +235,15 @@ copy copyfreeze from stdin;
 copy copyfreeze from stdin freeze;
 commit;
 select * from pg_visibility_map('copyfreeze');
- blkno | all_visible | all_frozen 
+ blkno | all_visible | all_frozen
 -------+-------------+------------
      0 | f           | f
      1 | f           | f
      2 | t           | t
 (3 rows)
 
 select * from pg_check_frozen('copyfreeze');
- t_ctid 
+ t_ctid
 --------
 (0 rows)
 
@@ -255,15 +255,15 @@ copy copyfreeze from stdin;
 copy copyfreeze from stdin freeze;
 commit;
 select * from pg_visibility_map('copyfreeze');
- blkno | all_visible | all_frozen 
+ blkno | all_visible | all_frozen
 -------+-------------+------------
      0 | t           | t
      1 | f           | f
      2 | t           | t
 (3 rows)
 
 select * from pg_check_frozen('copyfreeze');
- t_ctid 
+ t_ctid
 --------
 (0 rows)