On Wed, Jan 18, 2006 at 12:45:43PM +0000, Benny Prijono wrote: > > Even when the Via header is corrupted, sometimes we are still able to > send a response back. Section 18.2.2 of RFC 3261 has some guidelines: > > 1) If the request is coming from a reliable transport (TCP, SCTP, > TLS), send the response using the same connection. > > 2) If it's coming from UDP, then perhaps we can scan the top most Via > header for ";rport" string. If this string is present, then we can > send the response to the source address/port of the packet.
Agreed. The Route, Response-Route headers don't apply if the TCP/SCTP/TLS connection associated with the request is still available. In addition to scanning for the rport parameter in the Via when the request is received over UDP I would suggest sending it back to the IP/port in the udp header (i.e. what you would get back from recvfrom()). > Although that still leaves the question about what Via header would be > sent in the response. If we send the original malformed Via header > as it is, will the upstream proxy/UAC be able to decode them? The upstream proxy/UAC "generated" the malformed request, so it will have to also deal with the consequences of that malformed request. There is no onus on the UAS to protect the UAC from its own stupidity. David > > -benny > > > > [EMAIL PROTECTED] wrote: > > Hi Manju, > > > > Routing Header means only Via header?? > > > > If there is an error in the Route, Record-Route or contact, then UA will be > > able to send 4xx response. If there is an error in the Via header, then UA > > may not be able to send back the response. So in this case it can discard > > the message. > > > > Regards, > > Thangarajan. > > FlextronicsSoftware > > Worlds largest SIP stack vendor > > > > > _______________________________________________ > Sip-implementors mailing list > [email protected] > https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors -- [EMAIL PROTECTED] | phone: 902-832-2649 fax: 902-832-1015 "Complex problems have simple, neat and wrong solutions" - H. L. Menken _______________________________________________ Sip-implementors mailing list [email protected] https://lists.cs.columbia.edu/cucslists/listinfo/sip-implementors
