[tor-bugs] #12890 [Tor]: Design and implement optimizations for socket write limits

[Tor Bug Tracker & Wiki]

#12890: Design and implement optimizations for socket write limits
-----------------------------+--------------------------------
     Reporter:  robgjansen   |      Owner:  robgjansen
         Type:  enhancement  |     Status:  new
     Priority:  normal       |  Milestone:  Tor: 0.2.6.x-final
    Component:  Tor          |    Version:
   Resolution:               |   Keywords:  tor-relay
Actual Points:               |  Parent ID:  #12541
       Points:               |
-----------------------------+--------------------------------

Comment (by dgoulet):

 Looking at nickm's branch and the remaining tasks (comment10), seem
 something is missing from the paper
 (http://www.robgjansen.com/publications/kist-sec2014.pdf).

 Algorithm 2 of the paper shows (line 4 and 8) that tor should do a lookup
 on the connection for available write data (raw or not) before doing the
 adjustment. However, I don't see that right now in the code, is it still
 needed?

 I think this might be needed else we might get into a situation where tor
 iterates every X seconds on the sockets and adjust their size even though
 it's not needed which might deteriorate performance?

 For instance, I did a quick kernel trace of tor bootstrapping with the
 kist branch and you can see that there are bunch of get/setsockopt() where
 one of them (fd = 6) actually never writes anything but still gets
 adjusted.

 https://paste.debian.net/plainh/0ed81334

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/12890#comment:16>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online