[tor-bugs] #15435 [Tor]: Tor should not close stdin on child processes.

[Tor Bug Tracker & Wiki]

#15435: Tor should not close stdin on child processes.
---------------------------+-----------------------------------------------
     Reporter:  yawning    |      Owner:
         Type:             |     Status:  needs_review
  enhancement              |  Milestone:  Tor: 0.2.7.x-final
     Priority:  normal     |    Version:  Tor: unspecified
    Component:  Tor        |   Keywords:  tor-pt, needs-spec-patch pt-wants
   Resolution:             |  Parent ID:
Actual Points:             |
       Points:             |
---------------------------+-----------------------------------------------
Changes (by yawning):

 * cc: asn (added)


Comment:

 Replying to [comment:12 dcf]:
 > Yes, assuming it's implementable. You don't want to spec something that
 turns out to be awkward to implement.

 Ok.  pyptlib does zero termination handling, so all that changes there is
 a method that wraps getenv.  The obfsproxy side is trivial, since twisted
 makes monitoring for events on file descriptors easy
 (twisted.internet.stdio).

 If asn doesn't want to do the changes I will (Though #15471 might be good
 enough given our Lenooks monoculture).

 > Removing SIGINT from pt-spec sounds good to me, that always seemed more
 trouble than it was worth. If all you have to worry about is SIGINT and
 eof, then that's probably simple enough to implement by hand.

 Even with SIGINT in the picture, I don't think an event that should be
 treated as "run whatever cleanup is needed, terminate now" is that hard to
 implement.  The only reason why I do the "sigterm myself" dance in
 obsf4proxy is because I didn't feel like passing a channel around (I'll
 probably change this actually).

 There's another side to this that still needs writing (and probably a
 separate ticket) that changes the pt teardown process to be multi-stage
 (close stdin, give the child sometime to exit, follow up with more
 destructive methods to kill the child off).  As it stands right now, the
 main benefit to implementing what is specified is that PTs will always
 clean up properly if tor crashes for any reason.

 (Yes, I will change the PT teardown to be multi-stage, provided that this
 ticket happens.)

--
Ticket URL: <https://trac.torproject.org/projects/tor/ticket/15435#comment:13>
Tor Bug Tracker & Wiki <https://trac.torproject.org/>
The Tor Project: anonymity online