[tor-bugs] #1795 [Tor Relay]: Prop 174: Optimistic Data for Tor: Server Side

[Tor Bug Tracker & Wiki]

#1795: Prop 174: Optimistic Data for Tor: Server Side
-------------------------+--------------------------------------------------
 Reporter:  nickm        |       Owner:              
     Type:  enhancement  |      Status:  needs_review
 Priority:  normal       |   Milestone:              
Component:  Tor Relay    |     Version:              
 Keywords:  prop174      |      Parent:              
-------------------------+--------------------------------------------------

Comment(by iang):

 You're totally right about checking for CONN_TYPE_EXIT, of course.  I must
 have been confused by a quick glance at or.h and seeing that the EXIT and
 AP states were disjoint, and not noticing that the others are not.

 Checking for write_event in chunk 1 is also a plausible solution; in chunk
 2, I think the state check (suitably corrected to also check for
 CONN_TYPE_EXIT) is the right choice.

 I don't think we want to move code into
 connection_edge_process_relay_cell_not_open(), since the patch causes
 control to skip that call, and fall through to the RELAY_COMMAND_DATA
 handling below.  All that stuff would have to be copied into
 connection_edge_process_relay_cell_not_open(), which seems poor.

 Agreed on the debugging code; sorry that slipped in.

 Isn't it the case that when the cells get *sent* to the server, it will
 trigger the check to see if SENDMEs should be sent?  SENDMEs are sent when
 the buffer empties, not when it fills, right?  I expect that not wrapping
 these calls would have the same (i.e. no) effect, but it seems cleaner to
 me to not check if you should send a SENDME if you know you won't.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/1795#comment:4>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online