Re: [secdir] SECDIR Review of draft-ietf-qresync-rfc5162bis-10

[Jeffrey Hutzelman <[e-mail]>]

On Thu, 2014-02-13 at 21:20 -0500, Phillip Hallam-Baker wrote:

> There is a problem in that it does not state what the attack model is. It
> seems as if the attack model is limited to a passive attack.

Not at all.  It's just a 10+ year old document that doesn't spell things
out very well.

> If there is an active MITM attack, SSL will only provide protection if the
> server certificate is authenticated. Otherwise, passing the username and
> password enclair is problematic.

Indeed.  Section 11.1 goes into a fair amount of detail about verifying
the server hostname found in the certificate, but says nothing about
validation of the certificate itself.  This is an omission which I like
to think the IETF has been more careful about in more recent documents.

At the time, I think it was somehow assumed that if you used TLS then
_of course_ you would do certificate validation, and in fact probably
your TLS library would do it for you.  Again, these days I like to think
we know better.

In any case, if you think it's worth spending time on a better treatment
of security considerations for IMAP, feel free.  I have no time for
that, and Tero is breathing down my neck about old reviews I still
haven't done. :-(


-- Jeff