[asterisk-dev] [Code Review]: Add a test for verifying when username information can be leaked via differing nat settings

[mjordan]

> On Nov. 21, 2011, 2:11 p.m., wdoekes wrote:
> > > logger.warn("Inavlid Response: %s\n" % (err,))
> > 
> > Still one inavlid response left ;)
> > 
> > 
> > > Wouldn't this be super(SIPNatTest, self).__init__()
> > 
> > Right. That's what I meant :)
> > 
> > 
> > > I suppose we could return False and retry if it turns out that Asterisk
> > > is inconsistent somehow responding to our UpdateConfig. What would you
> > > like to see here?
> > 
> > Hm.. I specifically didn't like that the global reactor that was called
> > in main (run()) is now referenced deep inside this class. But when
> > googling around for it a bit, there isn't a canonical better pattern.
> > 
> > Personally, I would add a finish-callback to SIPNatTest init:
> >   test = SIPNatTest(reactor.stop)
> > so one could call:
> >   self.finish()
> > where it now says:
> >   reactor.stop()
> > 
> > But I guess it's fine as it is now, too.

As a random comment, the base class TestCase has a stop_reactor() method that is supposed to be the mechanism used to stop the twisted reactor.  That being said, all it does is log that the reactor is stopping.  There really isn't much to do, since we don't know if the test is being stopped due to an error or because it actually is finished - that processing happens after the twisted reactor exits.  Because its rather trivial, its not something that's terribly important if its enforced.


- mjordan


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviewboard.asterisk.org/r/1590/#review4840
-----------------------------------------------------------


On Nov. 21, 2011, 10:40 a.m., Terry Wilson wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviewboard.asterisk.org/r/1590/
> -----------------------------------------------------------
> 
> (Updated Nov. 21, 2011, 10:40 a.m.)
> 
> 
> Review request for Asterisk Developers, Paul Belanger and mjordan.
> 
> 
> Summary
> -------
> 
> This test loops through all combinations of nat=route/no (force_rport/no in 1.8+) for [general] and [peer] in sip.conf, whether or not the incoming request specified ;rport in the Via, and whether request was sent from the same port as the one listed in the Via, for both a valid and an invalid user. The test records when the behavior differs and matches against when we expect that behavior to differ for each version of Asterisk (1.4, 1.6.2, 1.8, 10, trunk). It also uses a Quine-McClusky minimization solver to reduce the list of failures to a function describing the failures. For example:
> On 1.4:
>     (((NOT port_matches_via) AND peer_nat AND (NOT general_nat)) OR ((NOT port_matches_via) AND (NOT peer_nat) AND general_nat))
> and 1.8:
>     (((NOT port_matches_via) AND (NOT rport_specified) AND (NOT peer_nat) AND general_nat) OR ((NOT port_matches_via) AND (NOT rport_specified) AND peer_nat AND (NOT general_nat)))
> 
> As you see, there is an extra term in 1.8 because in 1.8 we honor the ;rport in the Via no matter what, so differing behavior only occurs when a request comes in *not* specifying ;rport. There is no equivalent for nat=route or nat=never in 1.8+ (these values were originally added to work around a bug in old Uniden phone firmware).
> 
> This patch also adds a very small SyncAMI class for synchronously sending AMI requests (over HTTP because it is much simpler to do via HTTP's built-in request/response architecture) and getting back a response. It also modifies the default manager.conf and http.conf files to support the new API. SyncAMI makes it much easier to do things like update a config and reload it over AMI as you don't have to split up what you are doing into a lot of callback functions and you really do want to block execution until they are complete. Normally, I would have done a separate review for SyncAMI, but as this patch requires its use I've left them together for the review.
> 
> 
> Diffs
> -----
> 
>   /asterisk/trunk/lib/python/qm.py PRE-CREATION 
>   /asterisk/trunk/configs/manager.conf 2750 
>   /asterisk/trunk/lib/python/asterisk/syncami.py PRE-CREATION 
>   /asterisk/trunk/configs/branch-1.4/manager.conf 2750 
>   /asterisk/trunk/configs/http.conf PRE-CREATION 
>   /asterisk/trunk/tests/channels/SIP/nat_supertest/configs/ast1/extensions.conf PRE-CREATION 
>   /asterisk/trunk/tests/channels/SIP/nat_supertest/configs/ast1/sip.conf PRE-CREATION 
>   /asterisk/trunk/tests/channels/SIP/nat_supertest/run-test PRE-CREATION 
>   /asterisk/trunk/tests/channels/SIP/nat_supertest/sipp/register.xml PRE-CREATION 
>   /asterisk/trunk/tests/channels/SIP/nat_supertest/test-config.yaml PRE-CREATION 
>   /asterisk/trunk/tests/channels/SIP/tests.yaml 2750 
> 
> Diff: https://reviewboard.asterisk.org/r/1590/diff
> 
> 
> Testing
> -------
> 
> Tested against 1.4 and 1.8 and have gotten the values I expect back. The nat options for 1.4 and 1.6.2; and 1.8, 10, and trunk are the same, so I expect similar behavior there.
> 
> 
> Thanks,
> 
> Terry
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.digium.com/pipermail/asterisk-dev/attachments/20111121/83310972/attachment-0001.htm>