Ticket #47 (new defect)

Opened 2 years ago

Last modified 1 year ago

Crash with Skype OSS redirect over network on thin client

Reported by: StevenMcCoy Assigned to: lennart
Priority: high Milestone:
Component: daemon Severity: major
Keywords: Cc:

Description

v0.9.5-4 port of Debian testing packages to Ubuntu 6.10.

Crash in the daemon. Not a very helpful backtrace, let me know what more I can do.

Here's the setup:

Skype >> padap >> libpulse >> network >> pulseaudio >> ALSA >> Yealink USB phone

Music like Rhythmnbox works fine, with Skype you hear the initial rings and then the daemon dumps.

For whatever fantastic reason 0.9.5-3 on Ubuntu 6.10-beta worked fine but also dumps on 6.10-release, and hence I rebuilt with the newer upstream packages.

#0 0xb7f6985d in _dl_rtld_di_serinfo () from /lib/ld-linux.so.2 #1 0xb7f69c47 in _dl_rtld_di_serinfo () from /lib/ld-linux.so.2 #2 0xb7f6d942 in _dl_rtld_di_serinfo () from /lib/ld-linux.so.2 #3 0xb7f72d40 in _dl_make_stack_executable () from /lib/ld-linux.so.2 #4 0xb798f522 in snd_mixer_simple_none_register ()

from /usr/lib/libasound.so.2

#5 0xb799132a in snd_mixer_simple_none_register ()

from /usr/lib/libasound.so.2

#6 0xb798c082 in snd_mixer_selem_set_capture_volume ()

from /usr/lib/libasound.so.2

#7 0xb794f9ef in module_alsa_source_LTX_painit ()

from /usr/lib/pulse-0.9/modules/module-alsa-source.so

#8 0xb7f4f59c in pa_source_set_volume () from /usr/lib/libpulsecore.so.2 #9 0xb7a39c6f in ?? () from /usr/lib/pulse-0.9/modules/libprotocol-native.so #10 0x0806c058 in ?? () #11 0x00000001 in ?? () #12 0xbff23eb0 in ?? () #13 0xb7f42e59 in pa_memblockq_push () from /usr/lib/libpulsecore.so.2 #14 0xb7a2a0ec in pa_pdispatch_run ()

from /usr/lib/pulse-0.9/modules/libpdispatch.so

#15 0xb7a3cb8d in ?? () from /usr/lib/pulse-0.9/modules/libprotocol-native.so #16 0x0806b7c8 in ?? () #17 0x08076a00 in ?? () #18 0x00000000 in ?? ()

Attachments

core.3601.gz (172.0 kB) - added by StevenMcCoy on 12/12/06 11:04:02.
core

Change History

12/12/06 11:02:14 changed by StevenMcCoy

Oh awesome, cannot edit or preview the initial entry. 

#0  0xb7f6985d in _dl_rtld_di_serinfo () from /lib/ld-linux.so.2
#1  0xb7f69c47 in _dl_rtld_di_serinfo () from /lib/ld-linux.so.2
#2  0xb7f6d942 in _dl_rtld_di_serinfo () from /lib/ld-linux.so.2
#3  0xb7f72d40 in _dl_make_stack_executable () from /lib/ld-linux.so.2
#4  0xb798f522 in snd_mixer_simple_none_register ()
   from /usr/lib/libasound.so.2
#5  0xb799132a in snd_mixer_simple_none_register ()
   from /usr/lib/libasound.so.2
#6  0xb798c082 in snd_mixer_selem_set_capture_volume ()
   from /usr/lib/libasound.so.2
#7  0xb794f9ef in module_alsa_source_LTX_pa__init ()
   from /usr/lib/pulse-0.9/modules/module-alsa-source.so
#8  0xb7f4f59c in pa_source_set_volume () from /usr/lib/libpulsecore.so.2
#9  0xb7a39c6f in ?? () from /usr/lib/pulse-0.9/modules/libprotocol-native.so
#10 0x0806c058 in ?? ()
#11 0x00000001 in ?? ()
#12 0xbff23eb0 in ?? ()
#13 0xb7f42e59 in pa_memblockq_push () from /usr/lib/libpulsecore.so.2
#14 0xb7a2a0ec in pa_pdispatch_run ()
   from /usr/lib/pulse-0.9/modules/libpdispatch.so
#15 0xb7a3cb8d in ?? () from /usr/lib/pulse-0.9/modules/libprotocol-native.so
#16 0x0806b7c8 in ?? ()
#17 0x08076a00 in ?? ()
#18 0x00000000 in ?? ()

12/12/06 11:04:02 changed by StevenMcCoy

  • attachment core.3601.gz added.

core

12/12/06 13:39:34 changed by StevenMcCoy

  • priority changed from highest to high.
  • severity changed from blocker to major.

Turning up the tracing with --log-level hides the problem, therefore its most likely a special situation with resources.

I have rebuilt the package libraries and binaries without the -O2 optimization to find the debug symbols of where the crash occurs. Its more difficult to reproduce with a debug pulseaudio binary but the crash isn't there its in libprotocol-native and with a debug version of that I haven't been able to reproduce yet.

The situation is probably due to the creation and use of the source channel from the thin client machine to the application on the server. With the debug versions it works but the quality is really bad.

03/02/07 11:22:38 changed by ossman

Can you build it with -g at least so that we get better symbols in the stack trace?

10/23/07 15:33:51 changed by Marco

I'm with the same problem, on thin client, but using the debian etch, xmms and others players works perfectly, but skype I hear the initial rings and stop it in the same moment. I see that the load of processor go to top when it occurs.

12/10/07 16:18:46 changed by phocean

Same problem here, on Debian Testing and all versions of Skype (including the latest beta).