Re: Serverside SNI support in libpq

From: Daniel Gustafsson <daniel(at)yesql(dot)se>
To: Noah Misch <noah(at)leadboat(dot)com>
Cc: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Zsolt Parragi <zsolt(dot)parragi(at)percona(dot)com>, Michael Banck <mbanck(at)gmx(dot)net>, Jacob Champion <jacob(dot)champion(at)enterprisedb(dot)com>, Jelte Fennema-Nio <postgres(at)jeltef(dot)nl>, Heikki Linnakangas <hlinnaka(at)iki(dot)fi>, "li(dot)evan(dot)chao" <li(dot)evan(dot)chao(at)gmail(dot)com>, Michael Paquier <michael(at)paquier(dot)xyz>, Andres Freund <andres(at)anarazel(dot)de>, Pgsql Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: Serverside SNI support in libpq
Date: 2026-09-22 09:25:39
Message-ID: 5DE15F1A-3D87-4DD1-904E-8DA7FC9D102A@yesql.se
Views: Whole Thread | Raw Message | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> On 20 Sep 2026, at 23:16, Noah Misch <noah(at)leadboat(dot)com> wrote:

> security@ received a report about a NULL pointer SIGSEGV in $SUBJECT. I'm
> attaching the reporter's materials.

When reloading SSL configuration fails, we retain the previous config, but the
ssl_sni GUC was changed which could lead to new connections trying to access
SSL config which didn't exist. The proposed fix is to allow mismatched SSL
config and ssl_sni settings which I think is more confusing than treating
ssl_sni like the rest of the SSL config and retaining it's value on config
reload failure. The attached v2 does that, adds a WARNING (on top of the
already present logged warning) and expands the tests from the original
submission.

--
Daniel Gustafsson

Attachment Content-Type Size
v2-0001-Keep-current-ssl_sni-setting-on-SSL-config-reload.patch application/octet-stream 6.8 KB

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message vignesh C 2026-09-22 09:33:39 Re: sequencesync worker race with REFRESH SEQUENCES
Previous Message Michael Paquier 2026-09-22 09:24:36 Re: Add a permission check to pg_stat_get_backend_subxact()