Opened 6 weeks ago

Last modified 6 weeks ago

#24903 reopened defect

Bug: Line unexpectedly reached at pathbias_should_count at src/or/circpathbias.c:372

Reported by: starlight Owned by:
Priority: High Milestone: Tor: 0.3.3.x-final
Component: Core Tor/Tor Version: Tor:
Severity: Normal Keywords: tor-assert-nonfatal, regression?
Cc: Actual Points:
Parent ID: #24907 Points: 1
Reviewer: Sponsor:


Experienced bug-check while testing an explict exit with:

setconf __DisablePredictedCircuits=1
setconf ExcludeExitNodes={c1},{c2},{c3}
setconf SocksPort="x.x.x.x:x NoIsolateSOCKSAuth NoIsolateClientProtocol NoIsolateClientAddr NoIsolateDestAddr NoIsolateDestPort"
setconf ExitNodes=X
Bug: Line unexpectedly reached at pathbias_should_count at src/or/circpathbias.c:372.
Stack trace: (on Tor 9e8b762fcecfece6)

log_backtrace                 src/common/backtrace.c:108
tor_bug_occurred_             src/common/util_bug.c:118
pathbias_should_count        *src/or/circpathbias.c:372
pathbias_count_build_attempt  src/or/circpathbias.c:418
circuit_finish_handshake      src/or/circuitbuild.c:1479
command_process_cell          src/or/command.c:209
channel_tls_handle_cell       src/or/channeltls.c:1152
event_base_loop               event.c:1373
do_main_loop                  src/or/main.c:2637
tor_main                      src/or/main.c:3780
main                          src/or/tor_main.c:35

Child Tickets

Change History (3)

comment:1 Changed 6 weeks ago by teor

Keywords: tor-assert-nonfatal regression? added
Milestone: Tor: 0.3.3.x-final
Points: 1
Priority: MediumHigh

comment:2 Changed 6 weeks ago by starlight

Resolution: not a bug
Status: newclosed

Looked at the code and this is a sanity-check complaining it saw a 1-hop circuit and ignored it. As part of the testing I was doing I manually extended a 1-hop circuit, which appears to have triggered this message.

I take it that 1-hop circuits are now so vigorously prevented that one cannot even use them for testing. The configuration for permitting them has been eliminated. This is annoying; single-hop circuit are handy for testing and only an expert who knows what they are doing can create them. Should be allowed.

comment:3 Changed 6 weeks ago by teor

Parent ID: #24907
Resolution: not a bug
Status: closedreopened

This is a bug, because we shouldn't log that message when the controller does something permitted by the spec.

We broke RefuseUnknownExits on exits in #24907, when we deprecated the AllowSingleHopExits client option. And yes, we did lock that down because we don't recommend that anyone uses it in production.

For testing, please open another ticket to have it restored if you really need it.

Note: See TracTickets for help on using tickets.