[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]
Re: [tor-bugs] #6386 [Firefox Patch Issues]: Deadlock/hang on TBB New Identity
#6386: Deadlock/hang on TBB New Identity
----------------------------------+-----------------------------------------
Reporter: mikeperry | Owner: mikeperry
Type: defect | Status: new
Priority: major | Milestone:
Component: Firefox Patch Issues | Version:
Keywords: | Parent:
Points: | Actualpoints:
----------------------------------+-----------------------------------------
Comment(by mikeperry):
Aha: I found https://hg.mozilla.org/mozilla-central/rev/38fcd9047d39 while
digging through the cache service code. Apparently, it was added to the
cache shutdown syncing code because of a crash bug:
https://bugzilla.mozilla.org/show_bug.cgi?id=673543. However, looking at
the codepaths, the fix for that crash would *also* prevent the release of
our patch condition's lock by nsCacheService::ProcessRequest(). That code
releases the lock we were holding and then waits for *another* condition
using the same lock. This could definitely cause our condition to never
get properly signaled if the timing was just right, causing deadlock.
So it's possible that their crash fix will fix this issue too. I'm going
to apply it to our #5715 patch and I guess we'll just wait another couple
months to see if this race resurfaces, at which point I suppose I could
dig deeper into #7637. But at a glance, #7637 will require way more
mechanics than this simple fix, and it might re-introduce the evercookie
race. So here's hoping this is it!
--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/6386#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