Re: md5 issues Postgres14 on OL7 - Mailing list pgsql-general

From Christoph Moench-Tegeder
Subject Re: md5 issues Postgres14 on OL7
Date
Msg-id YdS5NMkQ5/TIDSBe@elch.exwg.net
Whole thread Raw
In response to Re: md5 issues Postgres14 on OL7  (Michael Paquier <michael@paquier.xyz>)
List pgsql-general
## Michael Paquier (michael@paquier.xyz):

> On Mon, Dec 20, 2021 at 03:22:31PM +0100, Christoph Moench-Tegeder wrote:
> > Active FIPS mode (/proc/sys/crypto/fips_enabled => 1) on the server does
> > produce this behaviour.
> 
> Most likely, this is a build linked with OpenSSL?  The way MD5 hashes
> are computed in Postgres has largely changed in 14, and the code has
> been refactored so as we rely on the EVP APIs from OpenSSL when
> building with --with-ssl=openssl, having as direct consequence to
> allocate a bit more memory every time a hash is computed.

You can reproduce that behaviour with the PGDG-RPMs on CentOS 7.
Enable FIPS-mode, reboot, and immediately md5() fails. The PGDG-RPMS
are built with openssl ("--with-openssl" in pg_config output),
as of course you need SSL today.

"Supports FIPS mode" is one of the selling points for your cryptohash
patches in the Release Notes, and that means no md5 when FIPS is
enforced (I think FIPS is a little too strict in this regard, as
people do invent horrid workarounds, which does not really improve
matters; but that's another can of worms).

Anyway, it's not the memory, but "out of memory" is all PostgreSQL
reports when anything in the hashing operations returns a failure.

Regards,
Christoph

-- 
Spare Space



pgsql-general by date:

Previous
From: Chocimier
Date:
Subject: Re: As a Linux distro, how to package multiple postgres major versions?
Next
From: "Theodore M Rolle, Jr."
Date:
Subject: Re: Tab-completion error...?