DragonFly bugs List (threaded) for 2009-05
DragonFly BSD
DragonFly bugs List (threaded) for 2009-05
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

suspend signal race?


From: YONETANI Tomokazu <qhwt+dfly@xxxxxxxxxx>
Date: Tue, 19 May 2009 14:57:41 +0900

Hi.
I noticed there's still a weird "race" with regard to suspend signal
(ctrl+Z), that I used to observe back in 1.8-RELEASE days.  It's 100%
reproducible on -DEVELOPMENT or 2.2-RELEASE.  What I did was basically
this, followed by ctrl+Z:

  $ su root -c 'vi /usr/pkg/etc/mk.conf'
  [1] + Suspended (signal)	su root -c vi /usr/pkg/etc/mk.conf
  $ fg

The result depends on root's login shell; if it's set to /bin/csh
(the default), the suspended process silently vanishes (killed?).
  $ fg

  Suspended
  $ pgrep -u root vi


If the root's login shell is set to /bin/sh, the vi session won't
resume and gets stuck there (pressing ctrl+T shows that the process
is in [stop] state).  If I send SIGCONT to vi, the vi session resumes
and I can continue to work on it.

The behavior on csh is quite undesirable, as you may lose the suspended
process (I think I've seen that before, but didn't care at that time).
It seems to occur for commands other than vi:
  $ su root -c 'cat'			# affected
  $ su root -c 'seq 1 10000'		# affected
  $ su root -c 'seq 1 10000 | tee a'	# can't be interrupted

Any ideas on how to resolve this issue?



[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]