Feedback-ID: i0fe9450f:Fastmail
Date: Mon, 10 Aug 2026 08:42:06 +0900
From: Michael Paquier <michael@paquier.xyz>
To: Niall Newman <nn@turacolabs.com>
Cc: pgsql-bugs@lists.postgresql.org
Subject: Re: PostgreSQL 18.4 backend SIGSEGV in pgstat_gc_entry_refs() after
 caught DSM attach error
Message-ID: <ankQTuCD9Gcq8wXn@paquier.xyz>
References: <2FDAA194-9CF3-4FD7-A450-F1A4BEB125F6@turacolabs.com>
 <anV8ftNMW3xyKSMG@paquier.xyz>
MIME-Version: 1.0
Content-Type: multipart/signed; micalg=pgp-sha512;
	protocol="application/pgp-signature"; boundary="KB9QQpTSIaf40ku/"
Content-Disposition: inline
In-Reply-To: <anV8ftNMW3xyKSMG@paquier.xyz>
Archived-At: 
 <https://www.postgresql.org/message-id/ankQTuCD9Gcq8wXn%40paquier.xyz>
Precedence: bulk

--KB9QQpTSIaf40ku/
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Fri, Aug 07, 2026 at 03:34:38PM +0900, Michael Paquier wrote:
> 5) While on it, I think that we could do something about the
> dshash_find_or_insert() in pgstat_get_entry_ref(), where we could
> clean up the local reference if dshash_find_or_insert() returns NULL.
> That's content only worth on HEAD, as the problem is unlikely going to
> happen in practice, and that's only a local reference.
>=20
> To summarize, I have fixed 2) with 4069df21beb8 backpatched down to
> v15, and did bf80a4c2d238 for 1) on HEAD.  3) is not necessary.  4)
> should have its own discussion, as it's a broader change impacting
> anything that uses DSA/DSM.  And I am planning to apply the attached
> for 5) only on HEAD as a follow-up improvement.

A consequence of 5) that I have missed during my initial lookup is
that it could be possible to finish with a NULL pointer dereference if
a concurrent backend has the idea to bump the refcount of the entry
that has a stale NULL shared_entry (due to entry reinit, for one, or
even a drop/create).  A OOM-ed backend calling pgstat_gc_entry_refs()
would be in trouble.

It's the first time I've heard about dshash_find_or_insert_extended().
That's kind of nice, perhaps it would make sense to extend its use in
other areas of the code..
--
Michael

--KB9QQpTSIaf40ku/
Content-Type: application/pgp-signature; name=signature.asc

-----BEGIN PGP SIGNATURE-----

iQIzBAEBCgAdFiEEG72nH6vTowiyblFKnvQgOdbyQH0FAmp5EE4ACgkQnvQgOdby
QH2JPw/+OZDjmC7PTfWUkb07WpSDs8yDrE1MZYurLMdl2T+kiSLP9xlm3XxBIAVy
ji4PSPJw35UtjlmUWHa2JOzOGxWGdopizoU4ISBFHwb6pmlIKb1hj2oXXTtfeDXR
wDUJZglAiAar9UiKgl7MVV9NFcgMb8gsG8XMtkeILuS4nQeWg6XS4GPmq6jcKcD2
qeDIed6nMGQsjlX/p1Nrw9PMu8F/4tRLia2m88Ige5buolHLXXNZujJ9J5uue0XZ
IFOe3pfIu+VVX3I8gSbvSle3lnBCZpghr3ACUUVuwsjg62ukZdHKLiOQXQqnX5Sq
qT0L6MbcoOaBEsUrFYsoX6WjK1QJ0gnw7z7Suil/B2785UTPw04c0F75zZWdSiUS
rTRLB08mmmgfOxvOe7dbt7sIcWo0HBS/iQyHDF7+fP8htx5xAXwROwh7MWmKa4RV
l4f/oN17zBHcL0SWDoWOD7gghRwGMhzY8yY3Hpfq/36OvE617sDX9XQJa/+LwUZ+
FtgJjSe3DmDpChumgG7iJgVCzXoPWZU4DfQh4/K3dBm0PUqx/7SWakJXO+aJTwc8
HK8n4QOXqdeyvEnQTArqTAdgtnzGP5RQORs6Y+SpXHXiwgc0+W4tUjnAlTfm0uhZ
orUJ1ceUROaqD6VZDr2foa5k0N0WUAy2WDQEzb/VQj2U2Zw7ao4=
=jksH
-----END PGP SIGNATURE-----

--KB9QQpTSIaf40ku/--