A rebuild of gstreamer0.10-plugins-ugly has fixed the problem on a i586 system, let's push it to updates_testing.

How to test : while one can try to play an mp3 file through clementine or a radio through radiotray, here is a direct gstreamer test :

gst-launch-0.10 filesrc location=somefile.mp3 ! mad ! alsasink

with the bug you get an error and no sound, with the updated packages you can listen the file.

Assignee: lists.jjorge => qa-bugs

Suggested advisory:
========================

Updated gstreamer0.10-plugins-ugly package fix a regression in mp3 playing

========================

Updated packages in core/updates_testing:
========================
gstreamer0.10-plugins-ugly-0.10.19-14.1.mga5.x86_64.rpm
gstreamer0.10-sid-0.10.19-14.1.mga5.x86_64.rpm
gstreamer0.10-a52dec-0.10.19-14.1.mga5.x86_64.rpm
gstreamer0.10-mpeg-0.10.19-14.1.mga5.x86_64.rpm
gstreamer0.10-cdio-0.10.19-14.1.mga5.x86_64.rpm
gstreamer0.10-twolame-0.10.19-14.1.mga5.x86_64.rpm

Source RPMs: 
gstreamer0.10-plugins-ugly-0.10.19-14.1.mga5.src.rpm

Tainted:
========

gstreamer0.10-plugins-ugly-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-plugins-ugly-debuginfo-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-lame-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-x264-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-amrnb-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-amrwbdec-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-sid-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-a52dec-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-mpeg-0.10.19-14.1.mga5.tainted.x86_64.rpm
gstreamer0.10-cdio-0.10.19-14.1.mga5.tainted.x86_64.rpm 

Source RPMs: 
gstreamer0.10-plugins-ugly-0.10.19-14.1.mga5.tainted.src.rpm

Status: NEW => ASSIGNED

Here it works (local storage, nas via nfs, and with the gstreamer test). Thanks

CC: (none) => fred.thuillier

Let's slow down for a minute.

Shlomi reported some regressions in playback with an mp3 file with the updated libmad.  So, if we can't get a resolution to that issue, we might just be better off reverting the libmad update.  Otherwise, there's more that needs to be rebuilt than just gstreamer0.10-plugins-ugly, as mplayer/mencoder and sox are also now broken, which means probably everything using it is (also including k3b, audacity, normalize, cdrdao and several other things).

In fact, we might just want to revert the libmad update for now and deal with this once we have a better idea of how to move forward.

CC: (none) => luigiwalser, shlomif
Whiteboard: has_procedure => has_procedure feedback

(In reply to David Walser from comment #5)
> Shlomi reported some regressions in playback with an mp3 file with the

With which software? I am trying to see which ones are affected : 

 - audacity works both MP3 import and export
 - mplayer stucks
 - k3b crashes

Yeah, we could push an updated libmad package that reverts the security fix for now unless vlc guys find out fast a way to fix libmad without breaking abi

CC: (none) => tmb

(In reply to José Jorge from comment #6)
> (In reply to David Walser from comment #5)
> > Shlomi reported some regressions in playback with an mp3 file with the
> 
> With which software? I am trying to see which ones are affected : 
> 

See https://trac.videolan.org/vlc/ticket/17052 .

>  - audacity works both MP3 import and export
>  - mplayer stucks
>  - k3b crashes

(In reply to José Jorge from comment #6)
> (In reply to David Walser from comment #5)
> > Shlomi reported some regressions in playback with an mp3 file with the
> 
> With which software? I am trying to see which ones are affected : 
> 
>  - audacity works both MP3 import and export
>  - mplayer stucks

mplayer and mpv work mostly fine here with some .mp3 I tried. Can you upload an offending mp3?


>  - k3b crashes

(In reply to Shlomi Fish from comment #9)
> mplayer and mpv work mostly fine here with some .mp3 I tried. Can you upload
> an offending mp3?

Maybe your mplayer is not using libmad?

I get this on any mp3 file :

==========================================================================
Trying to force audio codec driver family libmad...
Opening audio decoder: [libmad] libmad mpeg audio decoder

and it stays forever here until Ctrl + C.

José Jorge:

I am getting this:

shlomif@telaviv1:~/Download/Video$ mplayer scomber_-_Face_in_the_Crowd.mp3 
MPlayer 1.2.1-2.mga6.tainted-5.3.1 (C) 2000-2016 MPlayer Team
mplayer: could not connect to socket
mplayer: No such file or directory
Failed to open LIRC support. You will not be able to use your remote control.

Playing scomber_-_Face_in_the_Crowd.mp3.
libavformat version 56.40.101 (external)
Audio only file format detected.
Clip info:
 Title: Face in the Crowd (ft. Copperh
 Artist: Scomber
 Album: ccMixter
 Year: 2016
 Comment: 
 Genre: Unknown
Load subtitles in ./
==========================================================================
Trying to force audio codec driver family libmad...
Opening audio decoder: [libmad] libmad mpeg audio decoder
AUDIO: 44100 Hz, 2 ch, s16le, 258.2 kbit/18.30% (ratio: 32276->176400)
Selected audio codec: [mad] afm: libmad (libMAD MPEG layer 1-2-3)
==========================================================================
AO: [pulse] Init failed: Connection refused
Failed to initialize audio driver 'pulse'
AO: [alsa] 48000Hz 2ch s16le (2 bytes per sample)
Video: no video
Starting playback...
A:  21.0 (21.0) of 221.0 (03:41.0)  0.6% 

Exiting... (Quit)
shlomif@telaviv1:~/Download/Video$ 


shlomif@telaviv1:/home/music/Music/shlomif-Download$ grep mad /proc/`pgrep mplayer`/maps
7f8597ef0000-7f8597f0e000 r-xp 00000000 08:09 4063673                    /usr/lib64/libmad.so.0.2.1
7f8597f0e000-7f859810e000 ---p 0001e000 08:09 4063673                    /usr/lib64/libmad.so.0.2.1
7f859810e000-7f859810f000 r--p 0001e000 08:09 4063673                    /usr/lib64/libmad.so.0.2.1
7f859810f000-7f8598110000 rw-p 0001f000 08:09 4063673                    /usr/lib64/libmad.so.0.2.1
shlomif@telaviv1:/home/music/Music/shlomif-Download$ 


============

so it seems mplayer is using libmad. Can you try in a new UNIX user account?

I confirm that mplayer freezes with the updated libmad.

(In reply to David Walser from comment #12)
> I confirm that mplayer freezes with the updated libmad.

is it on i586 or x86-64? I'm on x86-64.

(In reply to Shlomi Fish from comment #13)
> (In reply to David Walser from comment #12)
> > I confirm that mplayer freezes with the updated libmad.
> 
> is it on i586 or x86-64? I'm on x86-64.

i586.

I get the freezes in mplayer also, when trying to play an MP3 file, or a video with the MP3 codec for the audio.

Downgrading libmad0 to the version in "Core Release" fixed the issue.

This is on Mageia 5, i586.

CC: (none) => jan-bugs

So what? Shall I push a libmad unpatched? We have much users that use i586!

(In reply to José Jorge from comment #16)
> So what? Shall I push a libmad unpatched? We have much users that use i586!

Yes, unless someone can fix the patch to not break the original ABI.  There was some discussion about how to do that in the QA meeting yesterday, but I don't understand ABIs enough to help with that.

(In reply to David Walser from comment #17)
> (In reply to José Jorge from comment #16)
> > So what? Shall I push a libmad unpatched? We have much users that use i586!
> 
> Yes, unless someone can fix the patch to not break the original ABI.  There
> was some discussion about how to do that in the QA meeting yesterday, but I
> don't understand ABIs enough to help with that.

Please try again with rel 20 of libmad which I just submitted to http://pkgsubmit.mageia.org/ .

(In reply to Shlomi Fish from comment #18)
> (In reply to David Walser from comment #17)
> > (In reply to José Jorge from comment #16)
> > > So what? Shall I push a libmad unpatched? We have much users that use i586!
> > 
> > Yes, unless someone can fix the patch to not break the original ABI.  There
> > was some discussion about how to do that in the QA meeting yesterday, but I
> > don't understand ABIs enough to help with that.
> 
> Please try again with rel 20 of libmad which I just submitted to
> http://pkgsubmit.mageia.org/ .

You'll need to build it in Mageia 5 updates_testing for most of us to be able to try it.

Also, please remember when you rename a patch to use svn mv, so that we can easily see the changes in the patch in svnweb (as well as retaining history).

(In reply to David Walser from comment #19)
> (In reply to Shlomi Fish from comment #18)
> > (In reply to David Walser from comment #17)
> > > (In reply to José Jorge from comment #16)
> > > > So what? Shall I push a libmad unpatched? We have much users that use i586!
> > > 
> > > Yes, unless someone can fix the patch to not break the original ABI.  There
> > > was some discussion about how to do that in the QA meeting yesterday, but I
> > > don't understand ABIs enough to help with that.
> > 
> > Please try again with rel 20 of libmad which I just submitted to
> > http://pkgsubmit.mageia.org/ .
> 
> You'll need to build it in Mageia 5 updates_testing for most of us to be
> able to try it.
> 

Done.

(In reply to Shlomi Fish from comment #20)
> > > 
> > > Please try again with rel 17 of libmad which I just submitted to
> > > http://pkgsubmit.mageia.org/ .
> > 

This fixes the bug for mplayer and clementine (gstreamer). So I assign it to qa :


- please remove from updates_testing gstreamer packages of comment3
- please test libmad update
 
Suggested advisory:
========================

Updated libmad package fix a regression in mp3 playing on 32 bit systems

========================

Updated packages in core/updates_testing:

lib64mad-0.15.1b-17.1.mga5
lib64mad-devel-0.15.1b-17.1.mga5.x86_64.rpm

libmad-0.15.1b-17.1.mga5
libmad-devel-0.15.1b-17.1.mga5.i586.rpm

SRPM :
mad-0.15.1b-17.1.mga5.src.rpm

Thanks Shlomi!

Amazingly, even VLC works without needed a rebuild.  Confirmed mplayer and sox are OK now.

Sysadmins, please remove the gstreamer0.10-plugins-ugly rebuilds (core and tainted) from updates_testing.

CC: (none) => sysadmin-bugs

Shlomi, I see you added further fixes to mad in Cauldron.  I'm guessing we need those for the Mageia 5 build as well?

(In reply to David Walser from comment #24)
> Shlomi, I see you added further fixes to mad in Cauldron.  I'm guessing we
> need those for the Mageia 5 build as well?

I think so - yes.

(In reply to Shlomi Fish from comment #25)
> (In reply to David Walser from comment #24)
> > Shlomi, I see you added further fixes to mad in Cauldron.  I'm guessing we
> > need those for the Mageia 5 build as well?
> 
> I think so - yes.

please test  mad-0.15.1b-17.2.mga5 in mga5 core/updates_testing.

(In reply to Shlomi Fish from comment #26)
> please test  mad-0.15.1b-17.2.mga5 in mga5 core/updates_testing.

Tested again for 32bit. Both mplayer and gstreamer are ok.

Tested on x86_64. Previous version did some glitches in the sound of some mp3 files through clementine. Now it does not anymore. Seems that we can validate this update!

Whiteboard: has_procedure MGA5-32-OK => has_procedure MGA5-32-OK MGA5-64-OK

Suggested advisory:
========================

Updated mad package fix a regression in mp3 playing through gstreamer, mplayer and several other software using this library.

This update fixes crashes in 32 bit and sound glitches in 64 bit.

========================

Updated packages in core/updates_testing:
========================
libmad0-0.15.1b-17.2.mga5.i586.rpm
libmad0-devel-0.15.1b-17.2.mga5.i586.rpm

lib64mad0-0.15.1b-17.2.mga5.x86_64.rpm
lib64mad0-devel-0.15.1b-17.2.mga5.x86_64.rpm

Source RPMs: 
mad-0.15.1b-17.2.mga5.src.rpm

Please sysadmins, don't forget to remove the gstreamer0.10-plugins-ugly rebuilds of comment 3 (core and tainted) from updates_testing.

Keywords: (none) => validated_update

Thanks everybody. Advisory uploaded.

Whiteboard: has_procedure MGA5-32-OK MGA5-64-OK => has_procedure advisory MGA5-32-OK MGA5-64-OK

An update for this issue has been pushed to the Mageia Updates repository.

http://advisories.mageia.org/MGAA-2016-0094.html

Status: ASSIGNED => RESOLVED
Resolution: (none) => FIXED