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 719398 - gst_discoverer_info_get_tags fails to work under valgrind (in some cases)
gst_discoverer_info_get_tags fails to work under valgrind (in some cases)
Status: RESOLVED NOTABUG
Product: GStreamer
Classification: Platform
Component: gstreamer (core)
1.2.1
Other Linux
: Normal normal
: git master
Assigned To: GStreamer Maintainers
GStreamer Maintainers
Depends on:
Blocks:
 
 
Reported: 2013-11-27 09:04 UTC by Dirk-Jan C. Binnema
Modified: 2013-12-05 23:35 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
test program (1.10 KB, text/plain)
2013-11-27 10:08 UTC, Dirk-Jan C. Binnema
Details

Description Dirk-Jan C. Binnema 2013-11-27 09:04:47 UTC
I've noticed some difference in behavior between valgrind and non-valgrind runs of the gst_discoverer_info_get_tags, which may point at a problem.

I've isolated the issue into the attached test-program (run like './test-disco <path-to-some-mp3>'). It runs fine normally.

However,
  valgrind --tool=memcheck --leak-check=full --show-possibly-lost=no --leak-resolution=med --track-origins=yes  ./test-disco foo.mp3

*usually* (80% or so) fails (with an assert because 'tags' fails); when I leave out the '--track-origins=yes', it *usually* works.
(using Valgrind-3.9.0 on Fedora 20, 64-bit).
Comment 1 Tim-Philipp Müller 2013-11-27 09:54:33 UTC
Could you attach the test program ?
Comment 2 Dirk-Jan C. Binnema 2013-11-27 10:08:18 UTC
Created attachment 262917 [details]
test program
Comment 3 Dirk-Jan C. Binnema 2013-11-27 10:08:56 UTC
(In reply to comment #1)
> Could you attach the test program ?

Oh, I had thought I'd done so -- but here it is.
Comment 4 Dirk-Jan C. Binnema 2013-12-03 22:59:27 UTC
(moving it back to unconfirmed)
Comment 5 Olivier Crête 2013-12-05 23:27:04 UTC
It's because you set the timeout to one second.. In the call to gst_discoverer_new(), replace GST_SECOND with 60*GST_SECOND. Running things under valgrind is much slower.
Comment 6 Dirk-Jan C. Binnema 2013-12-05 23:35:43 UTC
Ah yes, that makes sense. Sorry for the noise.