[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 nickm):

 PART 3:

 Woo. 29% done.

 General stuff: I'm perpetually terrified of breaking the handshake in a
 way to allow us to count as authenticated, or to process cells we
 shouldn't, or to send data we shouldn't, without actually completing the
 TLS handshake and verifying the other party with the Tor handshake.  I'm
 also perpetually afraid of breaking the v2 or v1 TLS handshakes and not
 noticing because I only tested master against master.


  * Weird thing: connection_or.c in your branch is 2322 lines long.  In
 master, it's 2290 lines long. I would have expected it to get shorter as
 code moved out.
  * The BASE_CHAN_TO_TLS(c) and TLS_CHAN_TO_BASE(c) macros are dicey; look
 at how connection_t and circuit_t handle these for safety.
  * The new log_debug in connection_mark_for_close makes me wonder: can we
 be checking for this case and warning/asserting if it hits?  If the new
 rule is "never call connection_mark_for_close on an or_conn directly",
 that's going to be fragile.
  * We'd better audit to make sure there are no more conn->state changes in
 connection_or.c
  * Where did that big block of code in connection_or_set_state_open go?

-- 
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/6465#comment:23>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online