Re: [sasl] Mechanism Negotiation and Channel Binding with SCRAM

[Nicolas Williams <[e-mail]>]

On Wed, Mar 03, 2010 at 06:03:23PM +0100, Simon Josefsson wrote:
> Nicolas Williams <[e-mail]> writes:
> 
> > On Tue, Mar 02, 2010 at 12:27:42PM +0000, Alexey Melnikov wrote:
> >> I can be easily confused by declarative language, so I would prefer c). 
> >> But I can live with b).
> >
> > Ok.  For GS2 then I propose the following changes (if you agree I'll
> > make the same changes for SCRAM and post them):
> >
> > -   o  If the client negotiates mechanisms, then clients MUST select the
> > -      PLUS-variant if offered by the server.  Otherwise (the client does
> > -      not negotiate mechanisms), if the client has no prior knowledge
> > -      about mechanisms supported by the server and wasn't explicitly
> > -      configured to use a particular variant of the GS2 mechanism, then
> > -      it MUST select only non-PLUS version of the GS2 mechanism.
> > +   o  Clients that support mechanism negotiation and channel binding
> > +      MUST use a "p" gs2-cb-flag when the server offers the PLUS-variant
> > +      of the desired GS2 mechanism.
> 
> Fine.
> 
> > +   o  Clients that do not support mechanism negotiation but do support
> > +      channel binding MUST use a "p" gs2-cb-flag and SHOULD use the non-
> > +      PLUS variant of the desired GS2 mechanism.
> 
> This is an unrelated change, and I don't like it -- why should, for
> example, clients that _requires_ channel bindings have to select the
> non-PLUS name?
> 
> This change actually _introduce_ the same kind of problem that we wanted
> to fix!  Now the problem is the reverse -- with the above text, clients
> that ONLY support channel binding-enabled GS2/SCRAM will need logic to
> deal with the non-PLUS name.  This is the reverse problem as Tobias
> noticed: clients that DO NOT support channel bindings shouldn't have to
> worry about the PLUS name.

OK.

> >     o  If the client does not support channel binding, then it MUST use a
> >        "n" gs2-cb-flag.
> >
> >     o  If the client supports channel binding and the server does not
> >        appear to (i.e., the client did not see the -PLUS name), then the
> > -      client MUST either fail authentication or it MUST chose the non-
> > -      PLUS mechanism name and use a "y" gs2-cb-flag.
> > +      client SHOULD either fail authentication immediately or choose
> > +      the non-PLUS mechanism name and use a "y" gs2-cb-flag.
> 
> Why SHOULD here?

Because the client could also continue and use a "p" gs2-cb-flag.  Ah!
What we must say is that the client MUST NOT use a "n" gs2-cb-flag --
the other options are all equally reasonable.

Nico
--