[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [tor-bugs] #23581 [Core Tor/Tor]: Die more helpfully if Schedulers option isn't compatible with platform



#23581: Die more helpfully if Schedulers option isn't compatible with platform
--------------------------+------------------------------------
 Reporter:  pastly        |          Owner:  (none)
     Type:  defect        |         Status:  new
 Priority:  Medium        |      Milestone:  Tor: 0.3.2.x-final
Component:  Core Tor/Tor  |        Version:
 Severity:  Normal        |     Resolution:
 Keywords:  tor-sched     |  Actual Points:
Parent ID:                |         Points:
 Reviewer:                |        Sponsor:
--------------------------+------------------------------------

Comment (by pastly):

 If we've made it to runtime, we must have a `the_scheduler`. Tor shouldn't
 just stop.

 I don't like (2) because I don't like fallback strategies that violate the
 Schedulers torrc option. We already do this in one place now[0], and I
 think that one place is a necessary evil.

 What if we implicitly add "Vanilla" to the end of any Schedulers torrc
 option that doesn't already have "Vanilla" in it somewhere? Then we'd
 always have a lowest common denominator and guaranteed-to-run-on-your-
 system option to fallback to if worst comes to worst.

 [0] KIST is working as usual, then boom support in the kernel drops out
 from under us. We log a notice about the sudden issue, but continue to
 call ourselves "KIST" (not "KISTLite"). We can do this fallback even if we
 don't have KISTLite in Schedulers

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/23581#comment:3>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online
_______________________________________________
tor-bugs mailing list
tor-bugs@xxxxxxxxxxxxxxxxxxxx
https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-bugs