On Sun, Oct 30, 2011 at 8:40 PM, Nico Williams <[email protected]> wrote:
> To be fair krb5 pre-auth was never an issue before because a) we
> didn't have initial credential acquisition interfaces for GSS, b) we
> didn't have IAKERB.  It's likely that this will never be a real issue.
>  Indeed, what I'd want to do as a client app is specify things like
> "don't use weak authentication methods" and enctypes/cipher suites.  I
> don't think I'd care to choose "PKINIT with user certs on a smartcard"
> vs. "PKINIT with SACRED instead of smartcards" vs. FAST armored
> PA-ENC-TIMESTAMP.  So, I give, thanks for the analogy.

This is what had me nervous Sam, and I now agree that I shouldn't look
at GSS-EAP through the same prism when it comes to two-level
negotiation.

What I want is probably best left as cred options (and possibly a
req_flag, but I want to avoid adding req_flags whenever we can, since
that's a very limited namespace) for expressing the app policies I
mentioned above.  Cred options on the default credential can be had by
acquiring a credential for desired_name = GSS_C_NO_NAME and setting
cred options on that, which is how we'd avoid having to add req_flags.

Nico
--
_______________________________________________
abfab mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/abfab

Reply via email to