Opened 7 years ago

Closed 6 years ago

#8196 closed defect (fixed)

pathbias_should_count(): Bug: One-hop circuit has length 1.

Reported by: arma Owned by: mikeperry
Priority: Medium Milestone: Tor: 0.2.4.x-final
Component: Core Tor/Tor Version:
Severity: Keywords: tor-client
Cc: mikeperry Actual Points:
Parent ID: Points:
Reviewer: Sponsor:


$ telnet localhost 9051
Trying ::1...
Connected to localhost.
Escape character is '^]'.
250 OK
extendcircuit 0 moria1
250 EXTENDED 124


Feb 09 14:50:44.000 [notice] pathbias_should_count(): Bug: One-hop circuit has length 1. Path state is new. Circuit is a General-purpose client currently doing handshakes. [2 similar message(s) suppressed in last 600 seconds]

Child Tickets

Change History (7)

comment:1 Changed 7 years ago by nickm

Cc: mikeperry added

comment:2 Changed 7 years ago by mikeperry

We could demote the log message here, but then if you try to extend the circuit again, you'll just hit the other log message in pathbias_should_count() that alerts you to the fact that it had ignored the circuit previously (for being one hop), but then isn't.

Should we maybe give controller-created/manipulated circuits a special flag for stuff like this? We do have CIRCUIT_PURPOSE_CONTROLLER (which is checked by pathbias_should_count()), but no one likes using it (including me) because it changes too much about how Tor actually manages your circuits...

comment:3 Changed 7 years ago by mikeperry

Oh, the shouldcount-enforcement log message is LD_BUG@info. I suppose we could just make this one LD_BUG@info too, if we don't like the magic controller flag idea.

comment:4 Changed 7 years ago by nickm

Owner: set to mikeperry
Status: newassigned

comment:5 Changed 6 years ago by mikeperry

Status: assignedneeds_review

Ok, added a fix to mikeperry/0.2.4-lognoise-fixes. I simply demoted the log message. Creating a flag to allow us to do something special for controller circuits would just end up enabling feature creep to cause the flag to create the same problems that CIRCUIT_PURPOSE_CONTROLLER has right now.

comment:6 Changed 6 years ago by nickm

Looks ok. Pending the rest of that branch being ready (see #8230), this is mergeable (after rebase and autosquash).

comment:7 Changed 6 years ago by nickm

Resolution: fixed
Status: needs_reviewclosed

merged; thanks!

Note: See TracTickets for help on using tickets.