[tor-bugs] #2511 [Tor Client]: Tor will use an unconfigured bridge if it was a configured bridge last time you ran Tor

[Tor Bug Tracker & Wiki]

#2511: Tor will use an unconfigured bridge if it was a configured bridge last time
you ran Tor
-------------------------+--------------------------------------------------
  Reporter:  arma        |              Owner:                    
      Type:  defect      |             Status:  needs_review      
  Priority:  normal      |          Milestone:  Tor: 0.2.2.x-final
 Component:  Tor Client  |            Version:                    
  Keywords:              |             Parent:                    
    Points:              |   Actualpointsdone:                    
Pointsdone:              |       Actualpoints:                    
-------------------------+--------------------------------------------------

Comment(by arma):

 Replying to [comment:5 nickm]:
 > Will this break bridge authorities?

 Yes! Good catch. We'll want to either check if UseBridges, or if
 authdir_mode_bridge(). I don't have an opinion on which one we should
 check.

 > I personally don't like the approach of mucking up
 router_add_to_routerlist even more: it's too hairy by half as is, and we
 should be looking for ways to make it dumber, not smarter.  Option 3 (or a
 variant option 1 where instead of calling the bridge Down we call it
 "Invalid" or something) seems much better.

 Hm. Well, Option 3 was "do this patch and also add another patch
 elsewhere", so I don't see how that will simplify
 router_add_to_routerlist(). The real issue is that we have a growing set
 of reasons why we wouldn't want to keep a router desc that we just learned
 about. One way to simplify is to stop having so many reasons, but I don't
 see a good way around that. What are some other ways to simplify? We could
 break the checks out into several functions based on topic. Is there
 another place in the code that we should be doing these checks instead?
 Hm.

 Specifically, if we go with option 1 and we want to handle bug2510, we're
 going to need some messier special-casing either in
 router_add_to_routerlist() or elsewhere that says "if we have a bridge
 descriptor and we just got a descriptor that's the same as it but actually
 we got it from this other address then we want to drop the old one so we
 don't drop the new one before we can rewrite it". Maybe that means we
 should be thinking about bug 2510 in parallel to this one, not after.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/2511#comment:7>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online