[asterisk-dev] Handling 482 Loop Detected responses

[Terry Wilson]

I have noticed an issue where if we make an outbound SIP call which is responded to with a 482 Loop Detected, instead of failing the call and letting the dialplan handle the error, we instead set chan->call_forward to a Local channel at the callers context with the extension that they were dialing. I can't think of a reason why we would want to do this. It makes it impossible to handle the error in any meaningful way. It appears Asterisk has done this since r3662:

$ svn log -r 3662
------------------------------------------------------------------------
r3662 | markster | 2004-08-26 20:16:16 -0700 (Thu, 26 Aug 2004) | 2 lines

When detecting a hairpin, redirect to the appropriate local extension (bug #1974)


(bug #1974 doesn't seem to be related, btw)

The code also has the note:
                    /* \note SIP is incapable of performing a hairpin call, which
                    is yet another failure of not having a layer 2 (again, YAY
                     IETF for thinking ahead).  So we treat this as a call
                     forward and hope we end up at the right place... */

which is clearly incorrect as RFC 3261 talks about how to handle spiralin, but if a *loop* (not a spiral) is detected by the destination, it is an error condition and I think should be treated as such. It is always "fun" changing behavior that "has always been there" but in my opinion this is just clearly wrong behavior.  Opinions?

Terry