1.5.21 -- XEmacs becomes defunct while running?

[Christopher Faylor]

On Tue, Aug 08, 2006 at 03:18:45PM -0700, Ed Hirgelt wrote:
>I realize that this isn't the best of all possible bug reports.  But
>it is the most information I have been able to get in the past few
>weeks.

No, but I can't think of much more data I could ask for to help narrow
this down.

>I've been plagued with a problem running XEmacs.  After a random
>amount of time, running subprocesses fail with not a pipe. I have
>never been able to find anything out about it. However, today I was
>looking at pstree and noticed at some point that XEmacs dropped off of
>the output.
>
>Here's the sequence of ps before and after.  Note that I continued to
>edit with XEmacs even after ps said it was defunct.
>
>First, everything is more or less normal
>
>hirged 548 $ ps
>     PID    PPID    PGID     WINPID  TTY  UID    STIME COMMAND
>    3056       1    3056       2328  con 112363 14:42:39 /usr/bin/sh
>    4004    3056    3056       3868  con 112363 14:42:52 /usr/bin/sh
>    1664    4004    3056       2908  con 112363 14:42:52 
>    /usr/X11R6/bin/xinit
>    2780    1664    2780       1404  con 112363 14:42:53 /usr/X11R6/bin/XWin
>    3268    1664    3268       3340  con 112363 14:43:00 /usr/bin/sh
>    3712    3268    3268       3744  con 112363 14:43:01 /usr/bin/rxvt
>I    3776    3712    3776       3708    0 112363 14:43:03 /usr/bin/bash
>    3028    2552    2552       3028    2 112363 14:51:35 <defunct>
>    2552    3028    2552        680    2 112363 14:44:02 /usr/bin/bash
>     172    2552     172       1124    2 112363 14:45:24 /usr/bin/bash
>    1984     172     172       1744    2 112363 14:45:25 /c/Program
>Files/Microsoft Office/OFFICE11/OUTLOOK
>    2516    2552    2516       3788    2 112363 14:46:20 /usr/bin/bash
>    3456    2516    2516        704    2 112363 14:46:20 /c/Program
>Files/Mozilla Firefox/firefox
>    3980       1    3576       3876    2 112363 14:56:38
>/usr/bin/xemacs-21.4.19
>    1416    3980    1416       1904    1 112363 14:56:56
>/usr/lib/xemacs-21.4.19/i686-pc-cygwin/gnuserv
>    1020    2552    1020       2172    2 112363 14:59:32 /usr/bin/ps
>hirged 549 $ ls /proc
>1416  172   2084  2552  3028  3268  3712  3980  cpuinfo  meminfo
>registry  stat    version
>1664  1984  2516  2780  3056  3456  3776  4004  loadavg  partitions
>self      uptime
>
>Process 3980 is my XEmacs.  A little later:
>
>hirged 550 $ ps
>     PID    PPID    PGID     WINPID  TTY  UID    STIME COMMAND
>    3056       1    3056       2328  con 112363 14:42:39 /usr/bin/sh
>    4004    3056    3056       3868  con 112363 14:42:52 /usr/bin/sh
>    1664    4004    3056       2908  con 112363 14:42:52 
>    /usr/X11R6/bin/xinit
>    2780    1664    2780       1404  con 112363 14:42:53 /usr/X11R6/bin/XWin
>    3268    1664    3268       3340  con 112363 14:43:00 /usr/bin/sh
>    3712    3268    3268       3744  con 112363 14:43:01 /usr/bin/rxvt
>I    3776    3712    3776       3708    0 112363 14:43:03 /usr/bin/bash
>    3028    2552    2552       3028    2 112363 14:51:35 <defunct>
>    2552    3028    2552        680    2 112363 14:44:02 /usr/bin/bash
>     172    2552     172       1124    2 112363 14:45:24 /usr/bin/bash
>    1984     172     172       1744    2 112363 14:45:25 /c/Program
>Files/Microsoft Office/OFFICE11/OUTLOOK
>    2516    2552    2516       3788    2 112363 14:46:20 /usr/bin/bash
>    3456    2516    2516        704    2 112363 14:46:20 /c/Program
>Files/Mozilla Firefox/firefox
>    3980    2812    2812       3600    3 112363 15:00:41 <defunct>
>    1416    3980    1416       1904    1 112363 14:56:56
>/usr/lib/xemacs-21.4.19/i686-pc-cygwin/gnuserv
>    3280    2552    3280       1772    2 112363 15:00:48 /usr/bin/ps
>
>You'll notice that 3980 is listed as defunct now.  At this point any
>attempt to run a subprocess from xemacs fails. My last attempt
>resulted in Opening pty or pipe: Invalid argument.
>
>Exiting XEmacs and restarting works for a while but the problem
>recurs. Nothing terribly intersesting is happening on the box other
>than my running XEmacs, Eclipse, and Outlook.
>
>Perhaps related, rxvt sometimes simply disappears out from under me
>after a command, typically a bash script exits. No dump, no nothing.

I have a vague idea what could be causing this and ran some tests
yesterday to no avail, of course.  Corinna is also running tests.  I
suspect that somehow a pid is being reused, which means that somehow a
handle is being closed which shouldn't be closed.

I'm going to add more debugging to cygwin and generate a snapshot
with the debugging turned on.  I'll post a notice about the snapshot
when that happens, so please give it a try.  If there are problems
there should be an error message which will provide clues about what
is going on.

cgf

--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/