Horton

Re: BUG #17949: Adding an index introduces serialisation anomalies.

Thomas Munro <thomas.munro@gmail.com>

View thread View on postgr.es
From: Thomas Munro <thomas.munro@gmail.com>
To: Heikki Linnakangas <hlinnaka@iki.fi>
Cc: Dmitry Dolgov <9erthalion6@gmail.com>, artem.anisimov.255@gmail.com, pgsql-bugs@lists.postgresql.org, Teodor Sigaev <teodor@sigaev.ru>
Date: 2023-07-03T22:02:03Z
Lists: pgsql-bugs 
On Wed, Jun 28, 2023 at 3:18 PM Thomas Munro <thomas.munro@gmail.com> wrote:
> I'll push these in a couple of days if there are no further comments.

Done.

Thanks Artem, Dmitry and Heikki.

I wonder how we might be more systematic about this.  There are some
general principles that were not respected here, but I'm not sure if
they're even written down let alone defended with code.  Something to
think about.

It's not great to add a new use of BufferGetBlockNumber() (in terms of
false sharing just to get a value that we must have had moment earlier
in order to pin the buffer), but we do that all the time.  That seems
like a micro-optimisation worth looking into in some systematic way
across all AMs.

Commits

  1. Fix race in SSI interaction with bitmap heap scan.

  2. Fix race in SSI interaction with gin fast path.

  3. Fix race in SSI interaction with empty btrees.

  4. Re-think predicate locking on GIN indexes.