[krbdev.mit.edu #6835] Re: accept_sec_context RFC4121 support bug in 1.8.3

[Derrick Brashear via RT]

On Tue, Dec 7, 2010 at 5:15 PM, Tom Yu via RT <rt-comment at krbdev.mit.edu> wrote:
> "Derrick Brashear via RT" <rt-comment at krbdev.mit.edu> writes:
>
>> Ah. You may disregard this, though the code should perhaps be
>> commented. RFC 4757 in defining rc4-hmac-exp explicitly
>> codifies the old token format, while not referring to RFC 4121.
>
> What should the comment say?  Should it mention that RFC 4121
> accidentally omitted rc4-hmac-exp from the list of "not-newer"?

That would be my suggestion. I'd be less worried if there hadn't been deviances
from what is specified in versions of both MIT Kerberos and Heimdal.

> I think there would need to be comments both in accept_sec_context.c
> and util_crypt.c, because some of the RFC 4121 vs RFC 1964 selection
> logic moved to util_crypt.c.

That seems reasonable.




-- 
Derrick