Re: pgsql: Fix remaining race condition with CLOG truncation and LISTEN/NOT - Mailing list pgsql-committers

From Michael Paquier
Subject Re: pgsql: Fix remaining race condition with CLOG truncation and LISTEN/NOT
Date
Msg-id aUNA4l9w1Flr9Y2z@paquier.xyz
Whole thread Raw
In response to pgsql: Fix remaining race condition with CLOG truncation and LISTEN/NOT  (Heikki Linnakangas <heikki.linnakangas@iki.fi>)
List pgsql-committers
Hi Heikki,

On Wed, Nov 12, 2025 at 07:16:38PM +0000, Heikki Linnakangas wrote:
> Fix remaining race condition with CLOG truncation and LISTEN/NOTIFY
>
> Previous commit fixed a bug where VACUUM would truncate the CLOG
> that's still needed to check the commit status of XIDs in the async
> notify queue, but as mentioned in the commit message, it wasn't a full
> fix. If a backend is executing asyncQueueReadAllNotifications() and
> has just made a local copy of an async SLRU page which contains old
> XIDs, vacuum can concurrently truncate the CLOG covering those XIDs,
> and the backend still gets an error when it calls
> TransactionIdDidCommit() on those XIDs in the local copy. This commit
> fixes that race condition.
>
> To fix, hold the SLRU bank lock across the TransactionIdDidCommit()
> calls in NOTIFY processing.

While reading through this set of changes, the addition of this commit
across the v14~v16 range has made me smile:

+    * Now that we have let go of the SLRU bank lock, send the notifications
+    * to our backend

SLRU bank locks have been introduced in v17 by Alvaro in 53c2a97a9266.
Just noticed while going through.
--
Michael

Attachment

pgsql-committers by date:

Previous
From: Michael Paquier
Date:
Subject: pgsql: Fix const correctness in pgstat data serialization callbacks
Next
From: Michael Paquier
Date:
Subject: pgsql: btree_gist: Fix memory allocation formula