On Fri, 2008-10-24 at 14:46 -0400, Dale Worley wrote:
> On Fri, 2008-10-24 at 12:03 -0400, Arjun Nair wrote:
> > - Change this, so that all out-of-dialog REQUESTs are challenged, and
> > in-dialog REQUESTs go through unchallenged 
> > - With an exception for REGISTER, as sipXregistrar is in a better
> > position to authenticate them. So, no REGISTERs are challenged.
> > - And an exception for OPTIONS, as they are great for debugging, so,
> > we will not require any authentication from them either.
> > 
> > - We use the presence of To-tags to test whether a request is in
> > dialog or not
> > 
> > Any comments on this?
> 
> That sounds quite reasonable.
> 
> > Also, we couldn't come to a definite conclusion for REFERs. Should we
> > authenticate in-dialog REFERs?
> 
> Authenticating in-dialog REFERs is part of how we make consultative
> transfers work.  What is needed for the cons trans mechanism?

No exception to the general in-dialog case is needed for REFER, though.
The authproxy path will generate the challenges for them if needed.
Just treat them like any other in-dialog request and don't challenge to
add PAI.

> >  How about out-of dialog REFERs?  
> 
> I'm not convinced that anyone *uses* out-of-dialog REFERs, despite the
> number of times I have seen mechanisms proposed in the IETF that use
> them.  But it's clear that if someone uses o.o.d. REFER for something,
> the mechanism should require authentication (otherwise it would be quite
> a security problem).  So I vote for authenticating o.o.d. REFER.

Agreed.  I think we should have a short list of methods that allowed
out-of-dialog (OPTIONS, at least) without challenge.  All others should
be challenged if they have on of our identities in the From header.



_______________________________________________
sipx-dev mailing list
[email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-dev
Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-dev

Reply via email to