[tor-bugs] #6341 [Tor Relay]: connection_or_flush_from_first_active_circuit() does wrong thing when ewma_enabled (was: connection_or_flush_from_first_active_circuit() wants refactoring)

[Tor Bug Tracker & Wiki]

#6341: connection_or_flush_from_first_active_circuit() does wrong thing when
ewma_enabled
-----------------------+----------------------------------------------------
 Reporter:  arma       |          Owner:     
     Type:  defect     |         Status:  new
 Priority:  normal     |      Milestone:     
Component:  Tor Relay  |        Version:     
 Keywords:             |         Parent:     
   Points:             |   Actualpoints:     
-----------------------+----------------------------------------------------

Comment(by arma):

 We *are* triggering the "if" when ewma is on. We just shouldn't be.
 Specifically, in the case where circ is based on
 conn->active_circuit_pqueue, it's very likely that circ !=
 conn->active_circuits, and thus we're skipping checks like "Is the cell
 queue low enough to unblock all the streams that are waiting" and "Did we
 just run out of cells on this circuit's queue?".

 We can't just remove the "if" clause, since then we'd run lines like
 "conn->active_circuits = *next_circ_on_conn_p(circ, conn);" which should
 only happen when !ewma_enabled.

 I'm beginning to think this is a "0.2.3 milestone" thing.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/6341#comment:3>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online