[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [tor-bugs] #8387 [Tor]: Unbuilt one-hop circuits sometimes hang around forever
#8387: Unbuilt one-hop circuits sometimes hang around forever
------------------------+---------------------------------------------------
Reporter: mikeperry | Owner: mikeperry
Type: defect | Status: assigned
Priority: normal | Milestone: Tor: 0.2.4.x-final
Component: Tor | Version:
Keywords: tor-client | Parent:
Points: | Actualpoints:
------------------------+---------------------------------------------------
Comment(by mikeperry):
So the specific change from #3443 that I think is likely the culprit here
is the check TO_ORIGIN_CIRCUIT(victim)->cpath->state == CPATH_STATE_CLOSED
at the top of the circuit culling loop of circuit_expire_building(). I
added that check to allow circuits to exist until we sent their create
cell, with the expectation that the orconn pruning code would eventually
clean them up.
It looks like run_connection_housekeeping() is the function that should
prune connections in this case... Unless these connections haven't set
their type to CONN_TYPE_DIR, it looks very much like they should still be
getting closed there... Hrmm..
I like Nick's idea of adding a function to tell us if we have any stale
circuits and/or connections in normal usage, but I am not sure I could
craft such a check myself to both cover all the cases of 'stale' and not
introduce a bunch of false positive warns.. Probably 0.2.5.x material for
that reason, unless Nick is more confident about exactly what checks we
should do in such a case.
In fact, it looks like run_connection_housekeeping() has one instance of
such a check for wedged orconns (but not dirconns.. it seems to just have
code to close those), but I've never seen its warns emitted for this bug.
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/8387#comment:6>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs