Re: Use-after-free issue in postgres_fdw

Etsuro Fujita <etsuro.fujita@gmail.com>

From: Etsuro Fujita <etsuro.fujita@gmail.com>
To: Matheus Alcantara <matheusssilv97@gmail.com>
Cc: PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>, Michael Paquier <michael@paquier.xyz>
Date: 2026-03-21T11:44:47Z
Lists: pgsql-hackers
Hi Matheus,

On Sat, Mar 21, 2026 at 1:00 AM Matheus Alcantara
<matheusssilv97@gmail.com> wrote:
> On 19/03/26 11:56, Etsuro Fujita wrote:
> > I think the root cause is that it is too early to free the PGconn in
> > pgfdw_reject_incomplete_xact_state_change() even if the connection is
> > in a state where we cannot use it any further; I think we should delay
> > that until abort cleanup (ie, pgfdw_xact_callback()).  Attached is a
> > patch for that.
>
> IIUC, with this patch, disconnect_pg_server() will be called at
> pgfdw_xact_callback() via pgfdw_reset_xact_state() when the top level
> transaction is rollback right?

That is right.

> I've tested and it seems to fix the issue, when "close c1;" is
> executed "conn" points to a valid connection pointer and this
> connection is properly disconnected when the top level transactions is
> rollback.
>
> This issue is reproducible on v14, so I think that we need a back-port.

Yes, I am thinking of back-patching the fix to all supported versions,
if no objections.

Thanks for testing/reviewing the patch!

Best regards,
Etsuro Fujita



Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →
  1. postgres_fdw: Fix handling of abort-cleanup-failed connections.