Re: [kitten] krb5 gss_pseudo_random implementation/spec variance

Jeffrey Hutzelman <jhutz@cmu.edu> Wed, 11 December 2013 22:17 UTC

Message-ID: <1386800244.25257.10.camel@destiny.pc.cs.cmu.edu>
From: Jeffrey Hutzelman <jhutz@cmu.edu>
To: kitten@ietf.org
Date: Wed, 11 Dec 2013 17:17:24 -0500
In-Reply-To: <3541_1386796395_rBBLDESk020467_x7d61qv852r.fsf@equal-rites.mit.edu>
References: <3541_1386796395_rBBLDESk020467_x7d61qv852r.fsf@equal-rites.mit.edu>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Cc: jhutz@cmu.edu
Subject: Re: [kitten] krb5 gss_pseudo_random implementation/spec variance
Precedence: list

On Wed, 2013-12-11 at 16:12 -0500, Greg Hudson wrote:

> The MIT krb5 implementation (present since release 1.8 in early 2010)
> starts at 00 00 00 00 instead of 00 00 00 01.
> 
> The Heimdal implementation (present since release 1.4 in late 2010) also
> starts at 00 00 00 00.  Heimdal encoded the counter in little-endian,
> but this was changed to big-endian on master on 2013-10-30.  Because
> both implementations start at zero, the difference in endianness
> doesn't cause a discrepancy in the first block of output.

... and this is why things like this ought to have test vectors.
Which, of course, only helps you if the test vectors weren't generated
by running an incorrect implementation.

-- Jeff

[kitten] krb5 gss_pseudo_random implementation/sp… Greg Hudson
Re: [kitten] krb5 gss_pseudo_random implementatio… Jeffrey Hutzelman
Re: [kitten] krb5 gss_pseudo_random implementatio… Benjamin Kaduk
Re: [kitten] krb5 gss_pseudo_random implementatio… Nico Williams
Re: [kitten] krb5 gss_pseudo_random implementatio… Greg Hudson