[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-bugs] #14815 [Tor]: use-after-free in cpuworker_onion_handshake_replyfn()

#14815: use-after-free in cpuworker_onion_handshake_replyfn()
------------------------+--------------------------------
     Reporter:  arma    |      Owner:
         Type:  defect  |     Status:  new
     Priority:  normal  |  Milestone:  Tor: 0.2.6.x-final
    Component:  Tor     |    Version:
   Resolution:          |   Keywords:
Actual Points:          |  Parent ID:
       Points:          |
------------------------+--------------------------------

Comment (by Sebastian):

 Here's my analysis:

 Something calls cpuworker_cancel_circ_handshake(), which calls
 workqueue_entry_cancel(). But that doesn't cancel the job, because it's
 currently running. Then cpuworker_cancel_circ_handshake() does
 circ->workqueue_entry = NULL;

 Now the second rolls over, and we hunt dead circs.
 circuit_close_all_marked() checks if a circ has a workqueue_entry != NULL,
 and if so, it doesn't free the circ - but if the workqueue_entry == NULL,
 then it goes ahead and frees the circ. Now the job finishes but the circ
 is already freed, and boom.

 I currently think moving the circ->workqueue_entry = NULL into the if
 (job) inside cpuworker_cancel_circ_handshake() is the correct fix, because
 we also do that in cpuworker_onion_handshake_replyfn().

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/14815#comment:2>
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