openarena crashed with SIGSEGV in pa_pdispatch_run()

the retracing seems to take ages.. here it is:

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xaa1c9b70 (LWP 8709)]
0x02474f40 in ?? ()
(gdb) bt
#0 0x02474f40 in ?? ()
#1 0x0053d6a9 in pa_command_request (pd=0x9d09918, command=61, tag=4294967295, t=0x9d09218, userdata=0x9cfb140) at pulse/stream.c:749
#2 0x01f4240c in pa_pdispatch_run (pd=0x9d09918, packet=0x9d14428, creds=0x9cfcfc8, userdata=0x9cfb140) at pulsecore/pdispatch.c:331
#3 0x00520319 in pstream_packet_callback (p=0x9cfcf00, packet=0x9d14428, creds=0x9cfcfc8, userdata=0x9cfb140) at pulse/context.c:346
#4 0x01f46cb8 in do_read (p=0x9cfcf00) at pulsecore/pstream.c:816
#5 do_something (p=0x9cfcf00) at pulsecore/pstream.c:184
#6 0x01f31d5e in callback (m=0x9cfb070, e=0x9d098e8, fd=35, f=PA_IO_EVENT_INPUT, userdata=0x9cfae20) at pulsecore/iochannel.c:119
#7 0x0053050b in dispatch_pollfds (m=0x9cfb030) at pulse/mainloop.c:688
#8 pa_mainloop_dispatch (m=0x9cfb030) at pulse/mainloop.c:933
#9 0x00530a21 in pa_mainloop_iterate (m=0x9cfb030, block=1, retval=0x0) at pulse/mainloop.c:964
#10 0x00530ae4 in pa_mainloop_run (m=0x9cfb030, retval=0x0) at pulse/mainloop.c:979
#11 0x005422b3 in thread (userdata=0x9cf0860) at pulse/thread-mainloop.c:94
#12 0x01f57045 in internal_thread_func (userdata=0x9cfd350) at pulsecore/thread-posix.c:83
#13 0x005ef985 in start_thread () from /lib/libpthread.so.0
#14 0x0047e1ee in clone () from /lib/libc.so.6

Just saw a libjack0 update.
The crash now looks like:

StacktraceTop:
 ?? () from /usr/lib/alsa-lib/libasound_module_pcm_pulse.so
 run_action (pd=0xa5dcdc8, r=0x24, command=2, ts=0xad014a00)
 pa_pdispatch_run (pd=0xa5ddcc8, packet=0xad01e818,
 pstream_packet_callback (p=0xa5ddb80, packet=0xad01e818,
 do_read (p=0xa5ddb80) at pulsecore/pstream.c:816

any update? it's still crashing.
Should i re-assign to libasound2-plugins or to openal-soft?

also crashes on lucid.

pulseaudio 1:0.9.22~0.9.21+stable-queue-32-g8478-0ubuntu14
libsdl1.2debian-pulseaudio 1.2.14-4ubuntu1.1

same stack as in #2

Workaround:
openarena +set s_alCapture 0

@Yorirou: thanks! you made my day.

not sure what the root cause is but I totally lost faith in the ubuntu pulseaudio package broken for so long.

Seems http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=589896 is also covering the same issue so maybe there's hope for a real fix.

I just tried this on a current live maverick. OpenArena works well there.

On my installed system (upgraded at least from karmic to lucid to maverick) I also have this segfault.
openarena +set s_alCapture 0 workaround works there for me also.

I suspect this is the same bug as <http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=589896>, and I suspect it isn't really openarena's fault (it might be openal-soft, or even pulseaudio). Setting s_alCapture to 0 by default would be a simple hack around it.

Daniel, if you've successfully backported upstream's patch or used my workaround from Debian bug 589896, could you please add a pointer to your backport either here or on that Debian bug, so I can encourage the Debian alsa-lib maintainers to do the same? I've looked in the Code and Overview tabs but couldn't find any reference to recent changes.

(Debian is in freeze now, so we'd prefer a minimal backport if possible.)

Hi Simon, I've only committed the wholesale backport of current git HEAD to the alsa-lib bzr branch here; it still needs to be split into manageable chunks. I'll work on that this weekend, and then I'll push it to pkg-alsa SVN as well.

alsa-lib (1.0.23-1ubuntu2) maverick; urgency=low

  * Apply upstream git changesets aadcbab through 72c7260ce
    and rediff debian/patches/relibtoolise.patch
    - correctly free dl handles, fixes LP: #552411, #584393
      (Closes: #589896)

 -- Daniel T Chen <email address hidden> Sun, 29 Aug 2010 13:13:19 -0400