After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 628415 - [audioconvert] segfault when converting bitwidth on Mac OS X
[audioconvert] segfault when converting bitwidth on Mac OS X
Status: RESOLVED NOTGNOME
Product: GStreamer
Classification: Platform
Component: gst-plugins-base
0.10.30
Other Mac OS
: Normal major
: NONE
Assigned To: GStreamer Maintainers
GStreamer Maintainers
: 635700 (view as bug list)
Depends on:
Blocks:
 
 
Reported: 2010-08-31 12:47 UTC by Adrian Sampson
Modified: 2011-05-02 20:51 UTC
See Also:
GNOME target: ---
GNOME version: ---



Description Adrian Sampson 2010-08-31 12:47:30 UTC
Using Gstreamer and gst-plugins-base 0.10.30, the following invocation crashes:

gst-launch audiotestsrc ! audioconvert ! capsfilter caps=audio/x-raw-int,width=16 ! fakesink
Setting pipeline to PAUSED ...
Pipeline is PREROLLING ...
Caught SIGSEGV accessing address 0x5d94a8
  • #0 select\$DARWIN_EXTSN
  • #1 g_poll
  • #2 g_main_context_iterate
  • #3 g_main_loop_run
  • #4 gst_bus_poll
  • #5 dyld_stub_g_strerror
  • #6 _print
  • #7 main

Changing the requested width from 16 to 32 (what's natively supplied by audiotestsrc on my system) works fine. The same line also works with Gstreamer 0.10.29.

This is Mac OS X v10.6.4 (64-bit). Both versions of Gstreamer were installed with MacPorts.
Comment 1 Sebastian Dröge (slomo) 2010-08-31 12:49:59 UTC
Could you get a real backtrace of all threads that were running at that time? This is only the main thread, which didn't do anything other than waiting for events at the time of the crash.
Comment 2 Adrian Sampson 2010-08-31 13:13:36 UTC
Er, yes. Sorry about that. I can try to recompile with debug symbols if that might help.

(gdb) thread apply all bt

Thread 2 (process 1775)

  • #0 __semwait_signal
  • #1 nanosleep
  • #2 g_usleep
  • #3 find_highest_version
  • #4 <signal handler called>

Comment 3 Sebastian Dröge (slomo) 2010-08-31 15:27:38 UTC
Yes, better try again with debug symbols. The symbols below the "signal handler called" are the interesting ones and completely missing here :)
Comment 4 Adrian Sampson 2010-09-12 00:23:42 UTC
Sorry for the radio silence -- I'm having a hard time building gstreamer such that gdb shows me anything more than the above. The configure script seems to turn "debugging" on by default, and even adding "-g" to the CFLAGS doesn't seem to have any effect. Is there some resource you can point me to that shows how to build with the necessary debug symbols?
Comment 5 Akhil Laddha 2011-01-21 06:01:16 UTC
Sebastian, can you please help the reporter with regards to comment#4 ?
Comment 6 Edward Hervey 2011-01-21 10:00:54 UTC
works fine here:
* macosx 10.5 64bit
* current git of everything.
Comment 7 Roberto Huelga 2011-01-25 08:52:19 UTC
I'm having the exact same problem with
Mac OsX 10.6.6
GStreamer 0.10.32
Installed with macport

When I run audioconvert element I get SIGSEGV

$ gst-launch audiotestsrc ! audioconvert ! capsfilter caps=audio/x-raw-int,width=16 ! fakesink
Setting pipeline to PAUSED ...
Pipeline is PREROLLING ...
Caught SIGSEGV accessing address 0x7c1c5e

Thread 3 (process 68825)

  • #0 __semwait_signal
  • #1 nanosleep
  • #2 g_usleep
  • #3 main

Comment 8 Edward Hervey 2011-01-25 10:28:35 UTC
I have this feeling this is the same bug as #615357 (see comment 15 there).
Comment 9 Sebastian Dröge (slomo) 2011-03-14 17:32:44 UTC
*** Bug 635700 has been marked as a duplicate of this bug. ***
Comment 10 Sebastian Dröge (slomo) 2011-03-14 17:36:23 UTC
Looks different than bug #615357 for me, this one here has segfaults while the other one doesn't. Still, the backtraces here are far from useful :(
Comment 11 Viktor Gal 2011-04-22 14:04:56 UTC
i've just compiled the HEAD of the git repo of gstreamer and with the patch committed there by Tim-Philipp Müller (bug #615357 comment 31) i'm getting still getting the same error. :(

(In reply to comment #10)
> Looks different than bug #615357 for me, this one here has segfaults while the
> other one doesn't. Still, the backtraces here are far from useful :(
Comment 12 Tim-Philipp Müller 2011-04-23 10:18:00 UTC
Does this work:

$ ORC_CODE=backup gst-launch audiotestsrc ! audioconvert ! capsfilter caps=audio/x-raw-int,width=16 ! fakesink

?

There's also this orc bug that manifests itself on 64-bit OSX: https://bugs.freedesktop.org/show_bug.cgi?id=33836
Comment 13 Viktor Gal 2011-04-23 10:45:44 UTC
yep it works just fine! i've now tested all the commands/pipelines that was crashing my gstreamer with 'ORC_CODE=backup' flag and all of them ran without any problems.

so it seems, that orc is generating some bad code there.... :(
at least now i can use gstreamer, until the bug is being fixed... 

thnx a lot!

(In reply to comment #12)
> Does this work:
> 
> $ ORC_CODE=backup gst-launch audiotestsrc ! audioconvert ! capsfilter
> caps=audio/x-raw-int,width=16 ! fakesink
> 
> ?
> 
> There's also this orc bug that manifests itself on 64-bit OSX:
> https://bugs.freedesktop.org/show_bug.cgi?id=33836
Comment 14 Tim-Philipp Müller 2011-04-23 11:01:01 UTC
Ok, thanks for confirming. It's an orc bug then.
Comment 15 David Schleef 2011-04-23 11:21:45 UTC
What version of Orc?
Comment 16 Viktor Gal 2011-04-24 11:27:36 UTC
It's a problem of macports, they haven't realised that 4 days after releasing orc 0.4.11 there you've released 0.4.13 that should fix all the bugs in 0.4.11. I've upgraded my local orc to 0.4.13 and it runs without any problems.

(In reply to comment #15)
> What version of Orc?
Comment 17 David Schleef 2011-05-02 20:51:37 UTC
This is fixed in Orc-0.4.14.