[tor-bugs] #23676 [Core Tor/Tor]: kist on 0.3.2.1-alpha-dev beats its head against a wall trying to flush a conn that's closed

[Tor Bug Tracker & Wiki]

#23676: kist on 0.3.2.1-alpha-dev beats its head against a wall trying to flush a
conn that's closed
----------------------------------------+----------------------------------
 Reporter:  arma                        |          Owner:  (none)
     Type:  defect                      |         Status:  needs_review
 Priority:  High                        |      Milestone:  Tor:
                                        |  0.3.2.x-final
Component:  Core Tor/Tor                |        Version:
 Severity:  Normal                      |     Resolution:
 Keywords:  regression, cpu, tor-sched  |  Actual Points:
Parent ID:                              |         Points:
 Reviewer:                              |        Sponsor:
----------------------------------------+----------------------------------

Comment (by pastly):

 I fear my discussion with dgoulet today on IRC may have fallen into the
 trap of assuming things about where cells are and how many of them are in
 various places. I think we really need that patch that logs more info
 about cell counts in values queues/bufs when we hit this case.

 That said, I believe the branch provided above (`bug23676_032_02` on
 ​https://github.com/pastly/public-tor.git) is a fix for the problem.

 There's concern that `channel_more_to_flush` is checking the wrong queue.
 David is working on that as I understand it. It might deserve its own
 ticket. This function has been doing this for a long time, so a change
 should be tested on a relay or two.

 I also understand David to be working on changing some logic in
 `channel_flush_some_cells`, basic on logic in our IRC discussion (that I'd
 like to see verified, see first paragraph).

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/23676#comment:10>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online