GNOME Bugzilla – Bug 162184
[PATCH] [mad] crash playing a mp3 file
Last modified: 2005-01-19 11:01:24 UTC
Distribution: Fedora Core release 3 (Heidelberg) Package: gnome-utils Severity: normal Version: GNOME2.8.0 0.8.8 Gnome-Distributor: Red Hat, Inc Synopsis: when with Rhytmbox i play an mp3, it crash... Bugzilla-Product: gnome-utils Bugzilla-Component: gdict Bugzilla-Version: 0.8.8 BugBuddy-GnomeVersion: 2.0 (2.8.0) Description: Description of the crash: Steps to reproduce the crash: 1. open rhythmbox 2. play any mp3 files 3. Expected Results: How often does this happen? Additional Information: Debugging Information: Backtrace was generated from '/usr/bin/rhythmbox' (no debugging symbols found)...Using host libthread_db library "/lib/tls/libthread_db.so.1". (no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...(no debugging symbols found)...[Thread debugging using libthread_db enabled] [New Thread -151151872 (LWP 23941)] [New Thread 19176368 (LWP 23947)] [Thread debugging using libthread_db enabled] [New Thread -151151872 (LWP 23941)] [New Thread 19176368 (LWP 23947)] [Thread debugging using libthread_db enabled] [New Thread -151151872 (LWP 23941)] [New Thread 19176368 (LWP 23947)] [New Thread 55585712 (LWP 23945)] [New Thread 45095856 (LWP 23944)] [New Thread 34606000 (LWP 23943)] [New Thread 109579184 (LWP 23942)] 0x003087a2 in _dl_sysinfo_int80 () from /lib/ld-linux.so.2
+ Trace 53854
Thread 2 (Thread 19176368 (LWP 23947))
------- Bug moved to this database by unknown@bugzilla.gnome.org 2004-12-24 14:20 ------- Unknown version 0.8.8 in product gnome-utils. Setting version to "unspecified". Unknown platform unknown. Setting to default platform "Other". Unknown milestone "unknown" in product "gnome-utils". Setting to default milestone for this product, '---' The original reporter of this bug does not have an account here. Reassigning to the person who moved it here, unknown@bugzilla.gnome.org. Previous reporter was simo.ragnetti@gmail.com. Setting to default status "UNCONFIRMED". Setting qa contact to the default for this product. This bug either had no qa contact or an invalid one.
I belive from the stack trace this should be assigned to gstreamer...let me know if I was wrong.
Oops, that was supposed to be version 0.8.7 (the version that comes with Fedora 3), not 0.6.0...
Created attachment 35197 [details] [review] patch A potential patch. I'm not checking this in, because it's not clear to me why mad->rate would be used.
This can happen indeed when we haven't synced yet. The above patch is not right. The correct thing is to return out of the chain function until mad->rate is set (see _caps_reset()).
*** Bug 162760 has been marked as a duplicate of this bug. ***
Can any of the people who can reproduce this please provide me with a GST_DEBUG=mad:5 log? Set the environment variable GST_DEBUG to "mad:5" and run whatever command you did to get this crash, and redirect the output to some file and attach the file here. Thanks.
Created attachment 36165 [details] Output of gdb amarokapp with env. var. GST_DEBUG="mad:5"
Created attachment 36171 [details] [review] possible fix I think this is the correct patch. If I read the log correctly, negotiation (during crossfading) fails, and that causes a subsequent SIGFPE. That's actually an amarok bug, but we should not crash. Rather, we should safely error out, like this patch tries to do. Please test and see if this patch fixes the crashes and returns a proper error instead.
I was receiving "Illegal instruction" from the gst-plugins-mad 0.8.5, and upgrading to gst-plugins-mad 0.8.7 seems to remedy my trouble. The userland application that makes use of gst-plugins-mad for me is Rhythmbox, on Gentoo Linux distro.
Francois, if you have some time, please test this. Also for the others. I'm marking this fixed, I just applied the patch. Please let me know if, after using this patch, you still do get SIGFPEs, and I'll continue debugging this.