[tor-bugs] #6465 [Tor Relay]: Build abstraction layer around TLS

[Tor Bug Tracker & Wiki]

#6465: Build abstraction layer around TLS
-----------------------+----------------------------------------------------
 Reporter:  andrea     |          Owner:  andrea            
     Type:  project    |         Status:  needs_review      
 Priority:  major      |      Milestone:  Tor: 0.2.4.x-final
Component:  Tor Relay  |        Version:  Tor: unspecified  
 Keywords:             |         Parent:                    
   Points:             |   Actualpoints:                    
-----------------------+----------------------------------------------------

Comment(by andrea):

 Responses to part 2 (points I generally agree with/am fine with going
 ahead and adjusting without debate):

 > Last general thing: Some of these functions should probably take const
 > channel_t*, not channel_t*.

 Okay.

 > A general thing: this branch doesn't actually compile for me. I get lots
 > of warnings like this:
 >
 > src/or/channel.c:1595: warning: format %lu expects type long unsigned
 int,
 > but argument 7 has type uint64_t
 >
 > Please try building on a more recent GCC and/or a 64-bit platform and/or
 a
 > 32-bit platform and/or with a recent clang -- whatever you haven't tried
 yet.

 Yeah, that should change.

 > Typo in comment in channel.h : "see thw channel_t"

 Yeah, I obviously meant 'thaw channel_t' :)

 > Should channel_more_to_flush() check outgoing_queue rather than
 cell_queue?
 > Should cell_queue be renamed incoming_queue for consistency? I think it
 > should.

 Yeah, yes to both I think.  Good catch with channel_more_to_flush().

 > In channel_process_incoming, do we want to process that queue in-order?
 It
 > seems that doing the DEL_CURRENT approach is likely to have weird
 > consequences. Perhaps it would be better to extract all the members into
 a
 > separate list (smartlist_add_all(), smartlist_clear()), then process the
 > separate list, then free it.

 Yeah, I think that's probably better.

 > It seems like channel_process_cells, channel_queue_cell, and
 > channel_queue_var_cell have some common code that should be extracted.

 Yeah, probably.

 > The first part of channel_free_all() seems redundant with
 > channel_run_cleanup()

 Yeah, that can change.

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/6465#comment:27>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online