[OpenSER-Devel] Re: [Devel] Re: qop support and CSeq increasing in UAC Module

[Ovidiu Sas]

Hi Bogdan,

I think the dialog would be a good place to handle cseq.  I raised
this issue again after I saw your changes in the dialog module :)
We could store the outgoing cseq in the dialog and for every outgoing
request, replace the cseq from the request with the one stored on the
dialog and viceversa for replies.


Regards,
Ovidiu Sas

On 7/17/07, Bogdan-Andrei Iancu <bogdan at voice-system.ro> wrote:
> Hi Ovidiu,
>
> well, the things are a bit more complicated than they seam. If in case
> of FROM URI it is just information that does not change across the
> dialog, for CSEQ, its value may change - and I'm not talking about the
> increasing value for each sequential request, but about the fact you can
> apply a new cseq change by yourself within the dialog : like doing
> authentication for a re-INVITE.
> ....and the information about the original cseq is kept in RR which may
> not be changed across the dialog.
>
> most probably, we will need the dialog module as support...still
> evaluating to be honest...
>
> Regards,
> Bogdan
>
> Ovidiu Sas wrote:
> > Hi all,
> >
> > Are there any news with respect to increasing CSeq feature?
> > I search the bug list but I could not find one related to to this issue.
> >
> >
> > Regards,
> > Ovidiu Sas
> >
> >
> > On 1/10/07, Andreas Heise <aheise at gmx.de> wrote:
> >> Hi Bogdan,
> >>
> >> yes with the "fix" suggested by Thomas works with some proxies....
> >>
> >> I hope that real qop support and CSeq feature will be define as todo for
> >> the next roadmap instead of nice to have, because uac_auth is really
> >> needed
> >> often.
> >>
> >> greetings from Berlin,
> >> Andreas
> >>
> >>
> >> Bogdan-Andrei Iancu schrieb:
> >> > Hi Andreas,
> >> >
> >> > yes, that is true - is a bit of complicated to calculate the response
> >> > when qop is on. Have you tried the approach suggested by Thomas? to
> >> > completely ignore the qop from challenge and to send the response as
> >> > if no qop was received......
> >> >
> >> > regards,
> >> > bogdan
> >> >
> >> > Andreas Heise wrote:
> >> >
> >> >>
> >> >> I've make some tests to deeper understand qop, for this I've modify
> >> >> the auth_hdr.c to
> >> >> ignore the qop and add a static qop=auth to the outgoing
> >> >> authorization header, but this
> >> >> way wasn't successful.
> >> >> After review of RFC 2617 I found that in case of qop cnonce and
> >> >> nonce-count (nc) MUST
> >> >> send by the client so it's not so easy as expected.
> >> >>
> >> >> Andreas
> >> >>
> >> >>
> >> >> Andreas Heise schrieb:
> >> >>
> >> >>>
> >> >>> Hi Bogdan,
> >> >>>
> >> >>> I think for the qop it could solved very quickly  if only  the auth
> >> >>> method will be implemented first,
> >> >>> because the Digest for auth should be the same as without qop
> >> >>> (unspecified) see page 14
> >> >>> in the following example....
> >> >>>
> >> >>>
> >> >>>
> >> http://www.softarmor.com/wgdb/docs/draft-smith-sip-auth-examples-00.txt
> >> >>>
> >> >>> so it should work as follow....
> >> >>>
> >> >>> if qop is detected, parse qop value if "auth" is allowed, if yes
> >> >>> send Authenticated Request as without qop,
> >> >>> but add qop="auth" or qop=auth ?!  ........if *only* auth-int is
> >> >>> allowed goto error as before ;-(
> >> >>>
> >> >>> This should make a lot of users happy......
> >> >>>
> >> >>> Am I right? Is it possible for somebody to do this soon ;-)
> >> >>>
> >> >>> regards,
> >> >>> Andreas
> >> >>>
> >> >>> Bogdan-Andrei Iancu
> >> >>> Thu, 04 Jan 2007 08:14:51 -0800
> >> >>>
> >> >>> Hi Klaus, Hi Stefano,
> >> >>>
> >> >>>
> >> >>> fixes to these limitations are scheduled to be done, but I'm afraid
> >> >>> they will not be ready for the next release - more time consuming
> >> >>> things are on the roll and I personally prefer to finish them before
> >> >>> attacking something new...
> >> >>>
> >> >>> Regards,
> >> >>> Bogdan
> >> >>>
> >> >>> Klaus Darilion wrote:
> >> >>>
> >> >>>
> >> >>>    Stefano Capitanio wrote:
> >> >>>
> >> >>>            Hi,
> >> >>>
> >> >>>                  i would like to ask if the two known limitations of
> >> >>> UAC Module
> >> >>>        (authentication does not support qop; CSeq not increased
> >> during
> >> >>>        authentication) will be resolved in future releases and if
> >> >>>        anyone has developed a patch or is working on it.
> >> >>>        I wonder if the dialog module can be used to keep track of
> >> >>> the CSeq
> >> >>>    difference between the SIPclient-->openser and
> >> >>>    openser--->upstreamSIPcomponent and modify all messages on the
> >> fly?
> >> >>>    Or another idea - would i be possible to store the Cseq
> >> difference
> >> >>>    in the record route parameter of uac module too?
> >> >>>
> >> >>>    regards
> >> >>>    klaus
> >> >>>
>
>