Horton

Re: Replace known_assigned_xids_lck by memory barrier

Nathan Bossart <nathandbossart@gmail.com>

View thread View on postgr.es
From: Nathan Bossart <nathandbossart@gmail.com>
To: Michail Nikolaev <michail.nikolaev@gmail.com>
Cc: PostgreSQL Hackers <pgsql-hackers@postgresql.org>, Tom Lane <tgl@sss.pgh.pa.us>
Date: 2023-08-16T20:07:15Z
Lists: pgsql-hackers

Attachments

On Wed, Aug 16, 2023 at 09:29:10PM +0200, Michail Nikolaev wrote:
> As answer: probably we need to change
> "If we know that we're holding ProcArrayLock exclusively, we don't
> need the read barrier."
> to
> "If we're removing xid, we don't need the read barrier because only
> the startup process can remove and add xids to KnownAssignedXids"

Ah, that explains it.  v5 of the patch is attached.

-- 
Nathan Bossart
Amazon Web Services: https://aws.amazon.com

Commits

  1. Replace known_assigned_xids_lck with memory barriers.

  2. Replace the KnownAssignedXids hash table with a sorted-array data structure,