[tor-bugs] #20269 [Core Tor/Tor]: bridge users ignore their cached consensus file on startup

[Tor Bug Tracker & Wiki]

#20269: bridge users ignore their cached consensus file on startup
-------------------------------------------------+-------------------------
 Reporter:  arma                                 |          Owner:  arma
     Type:  defect                               |         Status:
                                                 |  needs_revision
 Priority:  Medium                               |      Milestone:  Tor:
                                                 |  0.3.0.x-final
Component:  Core Tor/Tor                         |        Version:
 Severity:  Normal                               |     Resolution:
 Keywords:  029-proposed, nickm-                 |  Actual Points:
  deferred-20161017, review-group-11             |
Parent ID:                                       |         Points:
 Reviewer:  nickm                                |        Sponsor:
-------------------------------------------------+-------------------------

Comment (by arma):

 Replying to [comment:10 teor]:
 > Replying to [comment:9 nickm]:
 > > Okay, but what will actually happen if a new client (with this patch)
 tries to run with an 0.2.2.x bridge?  That will break, right?
 > >
 > > If so, are we okay with that breaking?
 >
 > Yes, but we should log a warning to the client saying the bridge is too
 old.
 > (We should reject the bridge's descriptor in 0.2.9 anyway, as it doesn't
 have an ntor key.)

 It's worse than that: Tor 0.3.0.0 doesn't finish its TLS handshake with
 the 0.2.2.x bridge.

 I ran an 0.2.2.39 bridge and an 0.3.0.0 client.

 Here's what my (modern) client says:
 {{{
 Nov 13 01:09:16.852 [warn] Problem bootstrapping. Stuck at 10%: Finishing
 handshake with directory server. (IOERROR; IOERROR; count 1;
 recommendation warn; host 0000000000000000000000000000000000000000 at
 128.31.0.39:9005)
 Nov 13 01:09:16.852 [warn] 1 connections have failed:
 Nov 13 01:09:16.852 [warn]  1 connections died in state handshaking (Tor,
 v3 handshake) with SSL state SSL negotiation finished successfully in OPEN
 Nov 13 01:09:17.398 [warn] Problem bootstrapping. Stuck at 10%: Finishing
 handshake with directory server. (IOERROR; IOERROR; count 2;
 recommendation warn; host 0000000000000000000000000000000000000000 at
 128.31.0.39:9005)
 Nov 13 01:09:17.398 [warn] 2 connections have failed:
 Nov 13 01:09:17.398 [warn]  2 connections died in state handshaking (Tor,
 v3 handshake) with SSL state SSL negotiation finished successfully in OPEN
 }}}

 And here's what the (ancient) bridge says:
 {{{
 Nov 13 01:09:17.362 [debug] tor_tls_handshake(): Completed V2 TLS
 handshake with client; waiting for renegotiation.
 Nov 13 01:09:17.362 [debug] connection_tls_continue_handshake(): Done with
 initial SSL handshake (server-side). Expecting renegotiation.
 Nov 13 01:09:17.386 [debug] conn_read_callback(): socket 121 wants to
 read.
 Nov 13 01:09:17.386 [debug] connection_read_to_buf(): 121: starting,
 inbuf_datalen 0 (0 pending in tls object). at_most 16384.
 Nov 13 01:09:17.386 [debug] connection_read_to_buf(): After TLS read of 9:
 510 read, 1179 written
 Nov 13 01:09:17.386 [info] connection_or_process_inbuf(): Accumulated too
 much data (9 bytes) on nonopen OR connection from a.b.c.d:43856 in state
 waiting for renegotiation (TLS); closing.
 Nov 13 01:09:17.386 [debug] conn_close_if_marked(): Cleaning up connection
 (fd 121).
 }}}

 So unless we want to put some energy into figuring out how to resume
 supporting 0.2.2.x bridges and relays (in which case we should open a
 separate ticket for that), I suggest we merge this one and call it done.

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/20269#comment:17>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online