Horton

Re: BitmapHeapScan streaming read user and prelim refactoring

Richard Guo <guofenglinux@gmail.com>

View thread View on postgr.es
From: Richard Guo <guofenglinux@gmail.com>
To: Melanie Plageman <melanieplageman@gmail.com>
Cc: Tomas Vondra <tomas@vondra.me>, Heikki Linnakangas <hlinnaka@iki.fi>, Thomas Munro <thomas.munro@gmail.com>, Andres Freund <andres@anarazel.de>, Pg Hackers <pgsql-hackers@postgresql.org>, Nazir Bilal Yavuz <byavuz81@gmail.com>
Date: 2024-12-19T09:15:10Z
Lists: pgsql-hackers 
On Sat, Oct 19, 2024 at 5:48 AM Melanie Plageman
<melanieplageman@gmail.com> wrote:
> I plan to commit 0002 and 0003 next week. I'm interested if you think
> 0001 is correct.
> I may also commit 0004-0006 as I feel they are ready too.

I noticed an oversight on master, which I think was introduced by the
0005 patch.  In ExecReScanBitmapHeapScan:

-      if (scan->st.bitmap.rs_shared_iterator)
-      {
-              tbm_end_shared_iterate(scan->st.bitmap.rs_shared_iterator);
-              scan->st.bitmap.rs_shared_iterator = NULL;
-      }
-
-      if (scan->st.bitmap.rs_iterator)
-      {
-              tbm_end_private_iterate(scan->st.bitmap.rs_iterator);
-              scan->st.bitmap.rs_iterator = NULL;
-      }
+      tbm_end_iterate(&scan->st.rs_tbmiterator);

I don't think it's safe to call tbm_end_iterate directly on
scan->st.rs_tbmiterator, as it may have already been cleaned up by
this point.

Here is a query that triggers a SIGSEGV fault on master.

create table t (a int);
insert into t values (1), (2);
create index on t (a);

set enable_seqscan to off;
set enable_indexscan to off;

select * from t t1 left join t t2 on t1.a in (select a from t t3 where
t2.a > 1);


I think we need to check whether rs_tbmiterator is NULL before calling
tbm_end_iterate on it, like below.

--- a/src/backend/executor/nodeBitmapHeapscan.c
+++ b/src/backend/executor/nodeBitmapHeapscan.c
@@ -572,9 +572,11 @@ ExecReScanBitmapHeapScan(BitmapHeapScanState *node)
    if (scan)
    {
        /*
-        * End iteration on iterators saved in scan descriptor.
+        * End iteration on iterators saved in scan descriptor, if they
+        * haven't already been cleaned up.
         */
-       tbm_end_iterate(&scan->st.rs_tbmiterator);
+       if (!tbm_exhausted(&scan->st.rs_tbmiterator))
+           tbm_end_iterate(&scan->st.rs_tbmiterator);

        /* rescan to release any page pin */
        table_rescan(node->ss.ss_currentScanDesc, NULL);

Thanks
Richard

Commits

  1. Fix bitmapheapscan incorrect recheck of NULL tuples

  2. Increase default maintenance_io_concurrency to 16

  3. Separate TBM[Shared|Private]Iterator and TBMIterateResult

  4. Improve read_stream.c advice for dense streams.

  5. Increase default effective_io_concurrency to 16

  6. Delay extraction of TIDBitmap per page offsets

  7. Add lossy indicator to TBMIterateResult

  8. Move BitmapTableScan per-scan setup into a helper

  9. Add and use BitmapHeapScanDescData struct

  10. Fix bitmap table scan crash on iterator release

  11. Bitmap Table Scans use unified TBMIterator

  12. Add common interface for TBMIterators

  13. Make table_scan_bitmap_next_block() async-friendly

  14. Move EXPLAIN counter increment to heapam_scan_bitmap_next_block

  15. Refactor tidstore.c iterator buffering.

  16. BitmapHeapScan: Remove incorrect assert and reset field

  17. Change BitmapAdjustPrefetchIterator to accept BlockNumber

  18. BitmapHeapScan: Use correct recheck flag for skip_fetch

  19. BitmapHeapScan: Push skip_fetch optimization into table AM

  20. BitmapHeapScan: postpone setting can_skip_fetch

  21. BitmapHeapScan: begin scan after bitmap creation

  22. Fix EXPLAIN Bitmap heap scan to count pages with no visible tuples

  23. Remove redundant snapshot copying from parallel leader to workers

  24. Remove some obsolete smgrcloseall() calls.

  25. Remove the "snapshot too old" feature.

  26. Compute XID horizon for page level index vacuum on primary.