Bug 14230

Summary:	kwin crashes when activating desktop effects
Product:	Mageia	Reporter:	Jean-Christophe Cardot <mageia>
Component:	RPM Packages	Assignee:	QA Team <qa-bugs>
Status:	RESOLVED FIXED	QA Contact:
Severity:	normal
Priority:	High	CC:	mageia, ottoleipala1, rim33377, rverschelde, stormi-mageia, sysadmin-bugs, tmb, wassi
Version:	4	Keywords:	validated_update
Target Milestone:	---
Hardware:	x86_64
OS:	Linux
See Also:	https://launchpad.net/bugs/1299499 https://bugs.freedesktop.org/show_bug.cgi?id=75723
Whiteboard:	MGA4-64-OK MGA4-32-OK advisory
Source RPM:	mesa	CVE:
Status comment:

Description Jean-Christophe Cardot 2014-10-04 12:44:24 CEST

Since last kernel update (3.12.25 -> 3.14.18), the desktop effects cannot be activated anymore in kwin, else it crashes.

Seems like we run into this bug: https://bugs.launchpad.net/ubuntu/+source/mesa/+bug/1299499

Reproducible: always (just press Alt-Shift-F12 to have kwin crash).

----------

Kwin output:

[fred@localhost ~]$ kwin --replace
QCoreApplication::sendPostedEvents: Cannot send posted events for objects in
another thread
QCoreApplication::sendPostedEvents: Cannot send posted events for objects in
another thread
QCoreApplication::sendPostedEvents: Cannot send posted events for objects in
another thread
QCoreApplication::sendPostedEvents: Cannot send posted events for objects in
another thread
OpenGL vendor string:                   Intel Open Source Technology Center
OpenGL renderer string:                 Mesa DRI Intel(R) G45/G43 
OpenGL version string:                  2.1 Mesa 10.0.5
OpenGL shading language version string: 1.20
Driver:                                 Intel
GPU class:                              i965
OpenGL version:                         2.1
GLSL version:                           1.20
Mesa version:                           10.0.5
X server version:                       1.14.5
Linux kernel version:                   3.14.18
Direct rendering:                       yes
Requires strict binding:                no
GLSL shaders:                           yes
Texture NPOT support:                   yes
Virtual Machine:                        no
kwin: brw_reset.cÂ :43Â :Â brw_get_graphics_reset_status:  l'assertion
Â«Â brw->hw_ctx != ((void *)0)Â Â» a échoué.
Application::crashHandler() called with signal 6; recent crashes: 1
KCrash: Application 'kwin' crashing...
KCrash: Attempting to start /usr/lib64/kde4/libexec/drkonqi from kdeinit
sock_file=/home/fred/.kde4/socket-localhost.localdomain/kdeinit4__0
QCoreApplication::sendPostedEvents: Cannot send posted events for objects in
another thread
QCoreApplication::sendPostedEvents: Cannot send posted events for objects in
another thread
QCoreApplication::sendPostedEvents: Cannot send posted events for objects in
another thread
QCoreApplication::sendPostedEvents: Cannot send posted events for objects in
another thread

[2]+  Stoppé                 kwin --replace

----------

Kwin stack trace:
 
Application: KWin (kwin), signal: Aborted
Using host libthread_db library "/lib64/libthread_db.so.1".
gobject.pyc: gdb was not built with custom backtrace support, disabling.
[Current thread is 1 (Thread 0x7fad59b6f780 (LWP 11183))]

Thread 3 (Thread 0x7fad3b308700 (LWP 11197)):
#0  0x00007fad53b100d8 in pthread_cond_timedwait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0
#1  0x00007fad53d9c3e4 in QWaitCondition::wait(QMutex*, unsigned long) () from /lib64/libQtCore.so.4
#2  0x00007fad53d8ffca in QThreadPoolThread::run() () from /lib64/libQtCore.so.4
#3  0x00007fad53d9bf1f in QThreadPrivate::start(void*) () from /lib64/libQtCore.so.4
#4  0x00007fad53b0bfab in start_thread () from /lib64/libpthread.so.0
#5  0x00007fad5934ba9d in clone () from /lib64/libc.so.6

Thread 2 (Thread 0x7fad3a60b700 (LWP 11223)):
#0  0x00007fad53b100d8 in pthread_cond_timedwait@@GLIBC_2.3.2 () from /lib64/libpthread.so.0
#1  0x00007fad53d9c3e4 in QWaitCondition::wait(QMutex*, unsigned long) () from /lib64/libQtCore.so.4
#2  0x00007fad53d8ffca in QThreadPoolThread::run() () from /lib64/libQtCore.so.4
#3  0x00007fad53d9bf1f in QThreadPrivate::start(void*) () from /lib64/libQtCore.so.4
#4  0x00007fad53b0bfab in start_thread () from /lib64/libpthread.so.0
#5  0x00007fad5934ba9d in clone () from /lib64/libc.so.6

Thread 1 (Thread 0x7fad59b6f780 (LWP 11183)):
[KCrash Handler]
#5  0x00007fad59290969 in raise () from /lib64/libc.so.6
#6  0x00007fad59291f98 in abort () from /lib64/libc.so.6
#7  0x00007fad59289846 in __assert_fail_base () from /lib64/libc.so.6
#8  0x00007fad592898f2 in __assert_fail () from /lib64/libc.so.6
#9  0x00007fad38f3ee25 in brw_get_graphics_reset_status () from /usr/lib64/dri/i965_dri.so
#10 0x00007fad38cd2e68 in _mesa_GetGraphicsResetStatusARB () from /usr/lib64/dri/i965_dri.so
#11 0x00007fad596eaf50 in KWin::SceneOpenGL::paint(QRegion, QList<KWin::Toplevel*>) () from /lib64/libkdeinit4_kwin.so
#12 0x00007fad596d0783 in KWin::Compositor::performCompositing() () from /lib64/libkdeinit4_kwin.so
#13 0x00007fad596d1af9 in KWin::Compositor::slotCompositingOptionsInitialized() () from /lib64/libkdeinit4_kwin.so
#14 0x00007fad5965a965 in KWin::Compositor::qt_static_metacall(QObject*, QMetaObject::Call, int, void**) [clone .part.87] () from /lib64/libkdeinit4_kwin.so
#15 0x00007fad53eb22e8 in QMetaObject::activate(QObject*, QMetaObject const*, int, void**) () from /lib64/libQtCore.so.4
#16 0x00007fad53d8c9f8 in QFutureWatcherBase::event(QEvent*) () from /lib64/libQtCore.so.4
#17 0x00007fad5302444c in QApplicationPrivate::notify_helper(QObject*, QEvent*) () from /lib64/libQtGui.so.4
#18 0x00007fad5302a9cd in QApplication::notify(QObject*, QEvent*) () from /lib64/libQtGui.so.4
#19 0x00007fad57f9318a in KApplication::notify(QObject*, QEvent*) () from /lib64/libkdeui.so.5
#20 0x00007fad53e9e6ad in QCoreApplication::notifyInternal(QObject*, QEvent*) () from /lib64/libQtCore.so.4
#21 0x00007fad53ea157f in QCoreApplicationPrivate::sendPostedEvents(QObject*, int, QThreadData*) () from /lib64/libQtCore.so.4
#22 0x00007fad530c163c in QEventDispatcherX11::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () from /lib64/libQtGui.so.4
#23 0x00007fad53e9d27f in QEventLoop::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () from /lib64/libQtCore.so.4
#24 0x00007fad53e9d575 in QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) () from /lib64/libQtCore.so.4
#25 0x00007fad53ea27fb in QCoreApplication::exec() () from /lib64/libQtCore.so.4
#26 0x00007fad5968ae16 in kdemain () from /lib64/libkdeinit4_kwin.so
#27 0x00007fad5927dc85 in __libc_start_main () from /lib64/libc.so.6
#28 0x000000000040074e in _start ()

----------

It seems like the problem has been corrected in mesa 10.1 but I could not install (and less test) mesa 10.2.7 which is present in the testing medias. I did it with difficulty (upgraded mesa and all related packages), and help from #mageia-fr (thx leuhmanu & Akien), but at the end nothing using open GL worked. So I downgraded everything to go back to "standard" mga4 packages.

This is not very severe: I can leave without desktop effects ;)

Comment 1 Jean-Christophe Cardot 2014-10-04 12:53:04 CEST

Found & tested a workaround in https://bugs.kde.org/show_bug.cgi?id=332784

run this in a konsole:
KWIN_OPENGL_INTERFACE="egl" kwin --replace &

and activate desktop effects (Alt-Shift-F12)

by FabiB 2014-03-31 11:02:34 UTC
As little dirty workaround - you can use the egl backend. wich wont crash here: 
KWIN_OPENGL_INTERFACE="egl" kwin --replace& ##to try it (make sure you enable effects again)
or
mkdir $(kde4-config --localprefix)/env echo 'export KWIN_OPENGL_INTERFACE="egl"' > $(kde4-config --localprefix)/env/kwinegl.sh chmod +x $(kde4-config --localprefix)/env/kwinegl.sh
to make it permanent for your session

Comment 2 Jean-Christophe Cardot 2014-10-04 12:55:12 CEST

sorry for the bad copy/paste, the commands must read:
$ mkdir $(kde4-config --localprefix)/env
$ echo 'export KWIN_OPENGL_INTERFACE="egl"' > $(kde4-config --localprefix)/env/kwinegl.sh
$ chmod +x $(kde4-config --localprefix)/env/kwinegl.sh

David Walser 2014-10-08 01:34:27 CEST

CC: (none) => tmb

Sander Lepik 2014-10-08 13:01:57 CEST

Priority: Normal => High
CC: (none) => mageia

Comment 3 Sander Lepik 2014-10-08 14:23:16 CEST

http://cgit.freedesktop.org/mesa/mesa/patch/?id=0380ec467d78f40b5c8134158ca48b4c5378b282

I'll try to submit patched packages.

Source RPM: (none) => mesa

Comment 4 Sander Lepik 2014-10-08 14:34:44 CEST

*** Bug 14231 has been marked as a duplicate of this bug. ***

CC: (none) => rim33377

Sander Lepik 2014-10-08 14:41:33 CEST

See Also: (none) => https://launchpad.net/bugs/1299499, https://bugs.freedesktop.org/show_bug.cgi?id=75723

Comment 5 Sander Lepik 2014-10-08 16:11:44 CEST

I have uploaded a patched packages for Mageia 4.

I can confirm that it fixes the problem on one of my laptop that uses affected GPU from Intel.

QA should test if everything works after updating. Even better if you are able to reproduce the crash.

Also, I'd be happy if someone can write a better advisory.

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

Updated mesa packages fix crash in KWin that may affect some systems with older Intel GPUs:

References:
https://bugs.freedesktop.org/show_bug.cgi?id=75723
========================

Updated packages in {core,tainted}/updates_testing:
========================
lib(64)vdpau-driver-radeonsi-10.0.5-1.1.mga4
lib(64)mesaglesv2_2-devel-10.0.5-1.1.mga4
lib(64)mesaegl1-devel-10.0.5-1.1.mga4
lib(64)mesaopenvg1-devel-10.0.5-1.1.mga4
lib(64)gbm1-10.0.5-1.1.mga4
lib(64)gbm1-devel-10.0.5-1.1.mga4
mesa-10.0.5-1.1.mga4
mesa-common-devel-10.0.5-1.1.mga4
lib(64)osmesa8-10.0.5-1.1.mga4
lib(64)vdpau-driver-r600-10.0.5-1.1.mga4
lib(64)mesaopenvg1-10.0.5-1.1.mga4
lib(64)wayland-egl1-devel-10.0.5-1.1.mga4
lib(64)dri-drivers-10.0.5-1.1.mga4
lib(64)mesaglesv1_1-10.0.5-1.1.mga4
lib(64)mesaegl1-10.0.5-1.1.mga4
lib(64)wayland-egl1-10.0.5-1.1.mga4
lib(64)mesaglesv2_2-10.0.5-1.1.mga4
lib(64)mesaglesv1_1-devel-10.0.5-1.1.mga4
lib(64)mesagl1-10.0.5-1.1.mga4
lib(64)glapi0-devel-10.0.5-1.1.mga4
lib(64)mesagl1-devel-10.0.5-1.1.mga4
lib(64)vdpau-driver-nouveau-10.0.5-1.1.mga4
lib(64)glapi0-10.0.5-1.1.mga4
lib(64)osmesa-devel-10.0.5-1.1.mga4

lib(64)gbm1-10.0.5-1.1.mga4.tainted
lib(64)mesaglesv1_1-devel-10.0.5-1.1.mga4.tainted
lib(64)mesaopenvg1-devel-10.0.5-1.1.mga4.tainted
lib(64)osmesa8-10.0.5-1.1.mga4.tainted
mesa-10.0.5-1.1.mga4.tainted
lib(64)gbm1-devel-10.0.5-1.1.mga4.tainted
lib(64)mesaglesv1_1-10.0.5-1.1.mga4.tainted
lib(64)mesagl1-10.0.5-1.1.mga4.tainted
lib(64)vdpau-driver-r600-10.0.5-1.1.mga4.tainted
mesa-common-devel-10.0.5-1.1.mga4.tainted
lib(64)vdpau-driver-nouveau-10.0.5-1.1.mga4.tainted
lib(64)mesaglesv2_2-10.0.5-1.1.mga4.tainted
lib(64)mesaegl1-devel-10.0.5-1.1.mga4.tainted
lib(64)glapi0-10.0.5-1.1.mga4.tainted
lib(64)wayland-egl1-10.0.5-1.1.mga4.tainted
lib(64)mesagl1-devel-10.0.5-1.1.mga4.tainted
lib(64)mesaglesv2_2-devel-10.0.5-1.1.mga4.tainted
lib(64)mesaopenvg1-10.0.5-1.1.mga4.tainted
lib(64)dri-drivers-10.0.5-1.1.mga4.tainted
lib(64)vdpau-driver-radeonsi-10.0.5-1.1.mga4.tainted
lib(64)glapi0-devel-10.0.5-1.1.mga4.tainted
lib(64)osmesa-devel-10.0.5-1.1.mga4.tainted
lib(64)mesaegl1-10.0.5-1.1.mga4.tainted
lib(64)wayland-egl1-devel-10.0.5-1.1.mga4.tainted


Source RPMs:
mesa-10.0.5-1.1.mga4.src.rpm
mesa-10.0.5-1.1.mga4.tainted.src.rpm

Assignee: bugsquad => qa-bugs

Comment 6 Jean-Christophe Cardot 2014-10-08 22:35:53 CEST

Thanks +Sander Lepik!
I'm currently installing these packages. Will keep the bug up to date with whatever I find.

Comment 7 Otto Leipälä 2014-10-08 22:40:53 CEST

Bug seems to affect with older intel graphics only i have hd 3000 and no any crash kwin or kwin_gles.I tested it as qa team tester with both arch 64&32 found no bugs so it's ok for me.

CC: (none) => ozkyster

Comment 8 Levchenko Roman 2014-10-09 03:04:23 CEST

After installation 
lib(64)vdpau-driver-radeonsi-10.0.5-1.1.mga4
lib(64)mesaglesv2_2-devel-10.0.5-1.1.mga4
lib(64)mesaegl1-devel-10.0.5-1.1.mga4
lib(64)mesaopenvg1-devel-10.0.5-1.1.mga4
lib(64)gbm1-10.0.5-1.1.mga4
lib(64)gbm1-devel-10.0.5-1.1.mga4
mesa-10.0.5-1.1.mga4
mesa-common-devel-10.0.5-1.1.mga4
lib(64)osmesa8-10.0.5-1.1.mga4
lib(64)vdpau-driver-r600-10.0.5-1.1.mga4
lib(64)mesaopenvg1-10.0.5-1.1.mga4
lib(64)wayland-egl1-devel-10.0.5-1.1.mga4
lib(64)dri-drivers-10.0.5-1.1.mga4
lib(64)mesaglesv1_1-10.0.5-1.1.mga4
lib(64)mesaegl1-10.0.5-1.1.mga4
lib(64)wayland-egl1-10.0.5-1.1.mga4
lib(64)mesaglesv2_2-10.0.5-1.1.mga4
lib(64)mesaglesv1_1-devel-10.0.5-1.1.mga4
lib(64)mesagl1-10.0.5-1.1.mga4
lib(64)glapi0-devel-10.0.5-1.1.mga4
lib(64)mesagl1-devel-10.0.5-1.1.mga4
lib(64)vdpau-driver-nouveau-10.0.5-1.1.mga4
lib(64)glapi0-10.0.5-1.1.mga4
lib(64)osmesa-devel-10.0.5-1.1.mga4, 

some of packeges were from tainted, KWin, all effects work good. Thank you.
Laptop Acer Trawelmate 5744Z, Intel Pentium P6200 with Intel HD Graphics.
Kernel 3.14.18-3

Comment 9 Otto Leipälä 2014-10-10 07:57:53 CEST

Testing complete for mga4 64&32 no issues found.
Update validated can sysadmin push this to updates.

CC: (none) => sysadmin-bugs
Keywords: (none) => validated_update
Whiteboard: (none) => MGA4-64-OK MGA4-32-OK

Comment 10 Jean-Christophe Cardot 2014-10-13 19:21:35 CEST

I'm late, but it is also ok here.

Comment 11 user7 2014-10-14 20:30:09 CEST

Tested on Mga4, i586.

I am affected by this bug, the update candidate fixes it. No regressions so far, I'd report back if anything comes up.

Thank you very much Jean-Christophe Cardot and Sander Lepik for fixing this bug!

CC: (none) => wassi

Comment 12 Otto Leipälä 2014-10-15 09:45:39 CEST

This update wait to pushed to updates so your testing is late it's already tested.

Comment 13 Samuel Verschelde 2014-10-15 10:05:33 CEST

(In reply to Otto LeipÃ¤lÃ¤ from comment #12)
> This update wait to pushed to updates so your testing is late it's already
> tested.

That's wrong. Nothing prevents to test an already validated update that hasn't been pushed already, it's even good to do it. If he would have spotted a regression that other testers didn't see, then we would have un-validated the update.

Could you please refrain from comments that discourage testers? That's not your first one.

CC: (none) => stormi

Comment 14 Jean-Christophe Cardot 2014-10-15 19:38:23 CEST

user7: thank you, but I did not fix, just reported the bug :)

Comment 15 user7 2014-10-15 23:33:32 CEST

Jean-Christophe Cardot: Only known problems can be fixed and you found the cause of this bug and a workaround, so you definitely contributed. I was affected and quite annoyed by this bug, but I did not succeed in finding the root cause. So I am very grateful you did. :)

Comment 16 Rémi Verschelde 2014-10-18 10:15:27 CEST

Advisory uploaded.

Whiteboard: MGA4-64-OK MGA4-32-OK => MGA4-64-OK MGA4-32-OK advisory

Comment 17 Otto Leipälä 2014-10-18 10:36:17 CEST

Thanks Rémi.

Comment 18 Sander Lepik 2014-10-18 10:43:34 CEST

Sysadmins, please push to mga4 core/updates :)

Assignee: qa-bugs => sysadmin-bugs

Comment 19 Rémi Verschelde 2014-10-18 14:13:16 CEST

Please lease it assigned to QA so that it gets listed in madb.

CC: (none) => remi
Assignee: sysadmin-bugs => qa-bugs

Comment 20 Mageia Robot 2014-10-23 15:28:26 CEST

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

http://advisories.mageia.org/MGAA-2014-0181.html

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