Bug 278386 - [kwin] Desktop Effects on Intel Graphics
Summary: [kwin] Desktop Effects on Intel Graphics
Status: RESOLVED UPSTREAM
Alias: None
Product: kwin
Classification: Plasma
Component: compositing (other bugs)
Version First Reported In: unspecified
Platform: Chakra Linux
: NOR crash
Target Milestone: ---
Assignee: KWin default assignee
URL:
Keywords:
: 280263 303534 (view as bug list)
Depends on:
Blocks:
 
Reported: 2011-07-24 11:38 UTC by Georg Grabler
Modified: 2012-07-15 11:01 UTC (History)
3 users (show)

See Also:
Latest Commit:
Version Fixed In:
Sentry Crash Report:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Georg Grabler 2011-07-24 11:38:48 UTC
Application: systemsettings (1.0)
KDE Platform Version: 4.7.00 (4.7.0)
Qt Version: 4.7.3
Operating System: Linux 2.6.39-CHAKRA x86_64
Distribution (Platform): Chakra

-- Information about the crash:
First of all, KWin started with desktop effects the first time i logged in KDE 4.7.0.
It immediately crashed, for the reason that my "graphics drivers probably have a bug". Every time i try to re-enable the desktop effects, this happens.

Well, we all know the intel drivers, that's a very likely reason, allthough, with your help, and the help of the intel guys, we probably can get it fixed (ickle said that he wants the bugs occuring for KDE with the intel drivers fixed some time soon, so a little bit backtracing maybe).

I installed the debug modules for KDE, probably it helps a little tracking down which effect crashed, where and why.

System configuration (quite recent packages there):
chakra-linux (testing)
KDE 4.7.0
kernel 2.6.39.3
intel-dri 7.10.3
xf86-video-intel 2.15.0
mesa 7.10.3

Maybe somebody of you can give a hint what the Intel guys need to look into. I'm fine doing a bug report in the freedesktop bug tracker for that and/or talk to ickle myself doing the required backtraces for them.

Best regards,
Georg

The crash can be reproduced every time.

-- Backtrace:
Application: System Settings (systemsettings), signal: Segmentation fault
[KCrash Handler]
#6  0x00007fcd98c8e8c0 in xcb_glx_query_server_string_string_length () from /usr/lib/libxcb-glx.so.0
#7  0x00007fcd994eeb44 in ?? () from /usr/lib/libGL.so.1
#8  0x00007fcd994cface in ?? () from /usr/lib/libGL.so.1
#9  0x00007fcd994cbf69 in glXQueryVersion () from /usr/lib/libGL.so.1
#10 0x00007fcd99762bfd in KWin::CompositingPrefs::detect (this=0x1953d30) at /chakra/desktop-testing/kde-workspace/src/kde-workspace-4.7.0/kwin/compositingprefs.cpp:196
#11 0x00007fcd99755cf6 in KWin::KWinCompositingConfig::updateStatusUI (this=0x1953aa0, compositingIsPossible=<value optimized out>)
    at /chakra/desktop-testing/kde-workspace/src/kde-workspace-4.7.0/kwin/kcmkwin/kwincompositing/main.cpp:415
#12 0x00007fcd99758f5b in KWin::KWinCompositingConfig::load (this=0x1953aa0) at /chakra/desktop-testing/kde-workspace/src/kde-workspace-4.7.0/kwin/kcmkwin/kwincompositing/main.cpp:433
#13 0x00007fcd9975524e in KWin::KWinCompositingConfig::rearmGlSupport (this=0x1953aa0) at /chakra/desktop-testing/kde-workspace/src/kde-workspace-4.7.0/kwin/kcmkwin/kwincompositing/main.cpp:347
#14 0x00007fcd99759e68 in KWin::KWinCompositingConfig::qt_metacall (this=0x1953aa0, _c=QMetaObject::InvokeMetaMethod, _id=15, _a=0x7fff061efe30)
    at /chakra/desktop-testing/kde-workspace/src/kde-workspace-4.7.0/kwin/kcmkwin/kwincompositing/main.moc:165
#15 0x00007fcdab8fde6f in QMetaObject::activate(QObject*, QMetaObject const*, int, void**) () from /usr/lib/libQtCore.so.4
#16 0x00007fcdacb872b2 in QAbstractButton::clicked(bool) () from /usr/lib/libQtGui.so.4
#17 0x00007fcdac8cd43b in ?? () from /usr/lib/libQtGui.so.4
#18 0x00007fcdac8cea0b in ?? () from /usr/lib/libQtGui.so.4
#19 0x00007fcdac8cec7c in QAbstractButton::mouseReleaseEvent(QMouseEvent*) () from /usr/lib/libQtGui.so.4
#20 0x00007fcdac58354c in QWidget::event(QEvent*) () from /usr/lib/libQtGui.so.4
#21 0x00007fcdac531b04 in QApplicationPrivate::notify_helper(QObject*, QEvent*) () from /usr/lib/libQtGui.so.4
#22 0x00007fcdac536f4a in QApplication::notify(QObject*, QEvent*) () from /usr/lib/libQtGui.so.4
#23 0x00007fcdad24aca6 in KApplication::notify(QObject*, QEvent*) () from /usr/lib/libkdeui.so.5
#24 0x00007fcdab8e896c in QCoreApplication::notifyInternal(QObject*, QEvent*) () from /usr/lib/libQtCore.so.4
#25 0x00007fcdac532b15 in QApplicationPrivate::sendMouseEvent(QWidget*, QMouseEvent*, QWidget*, QWidget*, QWidget**, QPointer<QWidget>&, bool) () from /usr/lib/libQtGui.so.4
#26 0x00007fcdac5b1808 in ?? () from /usr/lib/libQtGui.so.4
#27 0x00007fcdac5afec9 in QApplication::x11ProcessEvent(_XEvent*) () from /usr/lib/libQtGui.so.4
#28 0x00007fcdac5d74d2 in ?? () from /usr/lib/libQtGui.so.4
#29 0x00007fcda8dd3a43 in g_main_context_dispatch () from /usr/lib/libglib-2.0.so.0
#30 0x00007fcda8dd4220 in ?? () from /usr/lib/libglib-2.0.so.0
#31 0x00007fcda8dd44bd in g_main_context_iteration () from /usr/lib/libglib-2.0.so.0
#32 0x00007fcdab913a6f in QEventDispatcherGlib::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib/libQtCore.so.4
#33 0x00007fcdac5d717e in ?? () from /usr/lib/libQtGui.so.4
#34 0x00007fcdab8e7d02 in QEventLoop::processEvents(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib/libQtCore.so.4
#35 0x00007fcdab8e7f4c in QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) () from /usr/lib/libQtCore.so.4
#36 0x00007fcdab8ec44b in QCoreApplication::exec() () from /usr/lib/libQtCore.so.4
#37 0x0000000000411a29 in main (argc=5, argv=0x7fff061f1758) at /chakra/desktop-testing/kde-workspace/src/kde-workspace-4.7.0/systemsettings/app/main.cpp:49

Reported using DrKonqi
Comment 1 Martin Flöser 2011-07-24 17:55:46 UTC
please install also the debug symbols of Mesa. From what we have right now we only see that it crashes in the driver
Comment 2 Thomas Lübking 2011-07-24 19:45:24 UTC
It's by 99.9% chance a dupe of bug #229796

According to a apparently related mozilla bug report[1] it might be because "Mesa versions up to 7.10 cause a BadLength error during the first GLX call when the server GLX version < 1.3."

And a post suggests that it has been fixed upstream [2] but i don't know whether that is true and part of Mesa 10.3 (it's in the 7.10 branch and comitted before the 10.3 release at least - theoretically it should be in 10.1)

@ Georg: does "glxinfo" segfault for you? (part of "mesa-demos")


[1] https://bugzilla.mozilla.org/show_bug.cgi?id=632867
[2] http://cgit.freedesktop.org/mesa/mesa/commit/?id=cbe9fc12a64c3ae89fd1b20e9e165aa4b76293a5
Comment 3 Georg Grabler 2011-07-26 09:25:21 UTC
Hi, and sorry for the late answer.

Yes, glxinfo segfaults for me. I'll have to rebuild mesa here for debugging symbols, what can take some time (chakra build system needs some time to be set up).

Will report back with a new backtrace as soon as I've done that.
Comment 4 Thomas Lübking 2011-07-26 11:37:35 UTC
*sigh*

Please try whether mesa from git master works for you.
If not, exporting (best globally) LIBGL_ALWAYS_INDIRECT=1 will likely "workaround" this...
Comment 5 Georg Grabler 2011-07-26 17:39:28 UTC
Hi Thomas,

I just recompiled mesa from the git master, and the effects work, the crash is not reproducable any longer.
Thanks for the hint. Can be set on "solved" i guess, even though, most people won't have it solved :-).
Comment 6 Thomas Lübking 2011-07-26 18:17:22 UTC
At least a bug which is _really_ "resolved upstream" then ;-)
Comment 7 Christoph Feck 2011-08-18 23:59:51 UTC
*** Bug 280263 has been marked as a duplicate of this bug. ***
Comment 8 Christoph Feck 2012-07-15 11:01:19 UTC
*** Bug 303534 has been marked as a duplicate of this bug. ***