[tor-bugs] #13151 [Tor]: OR address is in host order in INTRODUCE2 cell

Tor Bug Tracker & Wiki blackhole at torproject.org
Sun Sep 14 01:57:03 UTC 2014


#13151: OR address is in host order in INTRODUCE2 cell
------------------------+--------------------------------
     Reporter:  asn     |      Owner:
         Type:  defect  |     Status:  needs_review
     Priority:  major   |  Milestone:  Tor: 0.2.5.x-final
    Component:  Tor     |    Version:
   Resolution:          |   Keywords:  tor-hs tor-client
Actual Points:          |  Parent ID:
       Points:          |
------------------------+--------------------------------

Comment (by cypherpunks):

 Replying to [comment:2 arma]:
 > I believe the net effect is that the extend request from the hs's
 circuit to the client's chosen rp will succeed if there's already a tls
 connection open between them, and will fail otherwise because the extend
 request will head off to the wrong address.
 >
 > So that means that some rendezvous attempts by hidden services will
 fail. Good thing we allow
 > {{{
 > #define MAX_REND_FAILURES 8
 > }}}
 > tries (and it was even higher up until #4241) -- I guess statistically
 the odds are pretty good. Still, we could cut down on variance in time-
 until-success by making the first try actually work.

 I don't see how the above analysis could be correct, because if it was I'd
 be seeing a lot more hidden service failures. I regularly start a new tor
 instance for a short-lived hidden service (using txtorcon) and then
 connect to it immediately. Sometimes it does take a couple tries to
 connect, but not usually. Having just started, and being not used for
 anything else, the HS tor should not be connected to very many nodes... so
 if a client has only 8 chances to pick one of those nodes as the RP it
 seems like rendezvouses should be failing much more often than not.

 Also, by "tls connection" did you actually mean circuit? My understanding
 is that tor only has tls connections to the immediate next hop in the
 circuit, which (if I'm understanding arma's comment correctly, which I
 must not be) would mean the HS client would need to be picking the HS's
 guard as its RP for the rendezvous to work. Obviously that can't be right!

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/13151#comment:5>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online


More information about the tor-bugs mailing list