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 704048 - [abrt] tracker-0.16.0-1.fc19: g_wakeup_new: Process /usr/libexec/tracker-extract was killed by signal 5 (SIGTRAP)
[abrt] tracker-0.16.0-1.fc19: g_wakeup_new: Process /usr/libexec/tracker-extr...
Status: RESOLVED FIXED
Product: tracker
Classification: Core
Component: Extractor
0.16.x
Other Linux
: Normal normal
: ---
Assigned To: tracker-extractor
: 702592 (view as bug list)
Depends on:
Blocks:
 
 
Reported: 2013-07-11 18:37 UTC by Michal Nowak
Modified: 2013-07-30 10:52 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
tracker-0.16.1 backtrace (67.06 KB, text/plain)
2013-07-13 09:27 UTC, Michal Nowak
Details

Description Michal Nowak 2013-07-11 18:37:04 UTC
From https://bugzilla.redhat.com/show_bug.cgi?id=959020:

Version-Release number of selected component:
tracker-0.16.0-1.fc19

Additional info:
backtrace_rating: 4
cmdline:        /usr/libexec/tracker-extract
crash_function: g_wakeup_new
executable:     /usr/libexec/tracker-extract
kernel:         3.9.0-301.fc19.x86_64
runlevel:       N 5
uid:            1000
ureports_counter: 1
xsession_errors: 

Truncated backtrace:
Thread no. 1 (10 frames)
 #2 g_wakeup_new at gwakeup.c:163
 #3 g_cancellable_make_pollfd at gcancellable.c:406
 #4 socket_source_new at gsocket.c:3265
 #5 g_socket_create_source at gsocket.c:3330
 #6 _g_socket_read_with_control_messages at gdbusprivate.c:201
 #7 _g_dbus_worker_do_read_unlocked at gdbusprivate.c:861
 #8 _g_dbus_worker_do_read_cb at gdbusprivate.c:810
 #9 g_simple_async_result_complete at gsimpleasyncresult.c:777
 #10 complete_in_idle_cb at gsimpleasyncresult.c:789
 #15 gdbus_shared_thread_func at gdbusprivate.c:278


See also: https://retrace.fedoraproject.org/faf/reports/84633/
Comment 1 Martyn Russell 2013-07-12 10:44:00 UTC
Hello Michal, are you able to try with tracker 0.16.1 and tell me if you can reproduce this? Thanks.
Comment 2 Michal Nowak 2013-07-12 13:52:21 UTC
Hi Martyn, that report in comment #0 is not mine but from the original Fedora reporter. However I was affected by that bug also, thought with 0.16.1 from Fedora 19. So, yes, I am able to reproduce it with latest stable tracker. I'll try to find the problem-triggering file if you want me to. Anyway, thanks for looking into this.
Comment 3 Michal Nowak 2013-07-13 09:27:30 UTC
Created attachment 249062 [details]
tracker-0.16.1 backtrace

This is my original backtrace, can't reproduce it at the moment.
Comment 4 Michal Nowak 2013-07-13 10:34:57 UTC
Now got the crash again but on another file. However tracker-extract won't crash on that file if run separately...
Comment 5 Matthias Clasen 2013-07-16 16:45:02 UTC
Clearly some sort of fd leak; look at the open_fds list attached to the downstream bug
Comment 6 Martyn Russell 2013-07-19 19:42:59 UTC
Thanks for the tip Matthias. The extractor still has some 100 file descriptors in use on average (here), but that's better than the massive leak I found in the GStreamer extractor upon failure conditions.

This will be fixed in 0.16.2 and 0.17.0.

commit f696dcd00af962e4864ca38365af4f245752a5d5
Author: Martyn Russell <martyn@lanedo.com>
Date:   Fri Jul 19 20:36:43 2013 +0100

    tracker-extract-gstreamer: Don't leak file descriptors & use up all FDs
    
    The GStreamer extractor was not calling all the shutdown functions internally
    to clean up streams in failure conditions.
    
    https://bugzilla.gnome.org/show_bug.cgi?id=704048
    https://bugzilla.redhat.com/show_bug.cgi?id=741479
Comment 7 Martyn Russell 2013-07-19 19:50:09 UTC
*** Bug 702592 has been marked as a duplicate of this bug. ***
Comment 8 Pacho Ramos 2013-07-28 19:41:56 UTC
Looks like it's only committed to master and, then, wouldn't be included in 1.16.2 :/

Thanks!
Comment 9 Martyn Russell 2013-07-30 10:52:07 UTC
(In reply to comment #8)
> Looks like it's only committed to master and, then, wouldn't be included in
> 1.16.2 :/
> 
> Thanks!

I merge all commits necessary from master to tracker-0.16 branch when doing the release - this will certainly be in there.