Summary: | Kwin crashs after login | ||
---|---|---|---|
Product: | [Plasma] kwin | Reporter: | Dimitrios Glentadakis <dglent> |
Component: | general | Assignee: | KWin default assignee <kwin-bugs-null> |
Status: | RESOLVED WORKSFORME | ||
Severity: | crash | CC: | dglent |
Priority: | NOR | ||
Version: | unspecified | ||
Target Milestone: | --- | ||
Platform: | Mageia RPMs | ||
OS: | Linux | ||
Latest Commit: | Version Fixed In: | ||
Attachments: |
Output of glxinfo
Output of 'kwin --replace&' New crash information added by DrKonqi Output of /var/log/syslog |
Description
Dimitrios Glentadakis
2012-12-18 04:05:55 UTC
> QPrinter::paperSize =) #9 0x00007f058ef51d6b in KWin::Scene::paintSimpleScreen (this=0x0, orig_mask=8, region=) at /usr/src/debug/kde-workspace-4.8.5/kwin/scene.cpp:329 #10 0x00007f058ef50504 in KWin::Scene::finalPaintScreen (this=0x1cf60d0, mask=8, region=<optimized out>, data=<optimized out>) at /usr/src/debug/kde-workspace-4.8.5/kwin/scene.cpp:198 Seems like stack corruption. > after login the plasma crashs Only the windowmanager (kwin, "titlebars") or also the desktop (wallpaper & panels)? Does this only happen after a cold login (first time ater re/boot of the system)? > and after a few seconds the open windows freeze really freeze, ie. the don't even react on mouse events? (and did kwin + compositor restart, ie do you have shadows under the windows?) can you restart kwin afterwards (resp. "does it happen automatically") #12 0x00007f0570328565 in KWin::BeClock::paintScreen (this=0x2352570, mask=8, region=<optimized out>, data=...) at /usr/src/debug/beclock-kwin-fx-0.18/beclock.cpp:530 While it's normal to have the effect active at that time, please try to deactivate it to rule out eg. some ABI incompatibility (ie. 3rd party, may require a recompile - effects are not ABI stable) please attach the output of glxinfo and "kwin --replace&" from konsole. I did nt have a new crash yet (In reply to comment #1) > > after login the plasma crashs > Only the windowmanager (kwin, "titlebars") or also the desktop (wallpaper & > panels)? Also the desktop. Everything is black except the windows. Also with alt+tab i can switch through the windows always with a black desktop background > Does this only happen after a cold login (first time ater re/boot of the > system)? Yes > > and after a few seconds the open windows freeze > really freeze, ie. the don't even react on mouse events? (and did kwin + > compositor restart, ie do you have shadows under the windows?) Yes there is a really freeze, no mouse events, no keyboard events (ctrl+alt+f1 or f2 or f3 ...) > can you restart kwin afterwards (resp. "does it happen automatically") No, i have to restart manyally the computer from the power button Created attachment 75915 [details]
Output of glxinfo
Created attachment 75916 [details]
Output of 'kwin --replace&'
Sounds like a kernel freeze - can you still ssh into the machine? plasma & kwin likely crash because of libgl which then might induce the issue in the nvidia kernel module. I assume you performed some system update including kernel and/or nvidia blob? -> check the resolution of /usr/lib/libGL.so - should point the nvidia lib (suffixed by driver version) In doubt up or downgrade the nvidia blob (buggy version, broken update process) This is very unlikely a problem with kwin or even KDE (gl abstraction and user/kernelspace cross) Created attachment 75968 [details]
New crash information added by DrKonqi
kwin (4.8.5 (4.8.5)) on KDE Platform 4.8.5 (4.8.5) using Qt 4.8.2
This time kwin crashed just after fresh boot, but i don't have any freeze, the desktop became operational immediately
-- Backtrace (Reduced):
#6 0x00007f50cf60b045 in QPrinter::paperSize (this=<optimized out>) at painting/qprinter.cpp:1083
#7 0x00007f50cf60b365 in QPrinter::paperSize (this=<optimized out>, unit=3752702064) at painting/qprinter.cpp:1173
[...]
#9 0x00007f50d4fdcd6b in KWin::Scene::paintSimpleScreen (this=0xffffffffffffffff, orig_mask=8, region=) at /usr/src/debug/kde-workspace-4.8.5/kwin/scene.cpp:329
#10 0x00007f50d4fdb504 in KWin::Scene::finalPaintScreen (this=0xc91e80, mask=8, region=<optimized out>, data=<optimized out>) at /usr/src/debug/kde-workspace-4.8.5/kwin/scene.cpp:198
#11 0x00007f50d4ffee87 in KWin::EffectsHandlerImpl::paintScreen (this=0xece920, mask=8, region=..., data=...) at /usr/src/debug/kde-workspace-4.8.5/kwin/effects.cpp:222
(In reply to comment #6) > #9 0x00007f50d4fdcd6b in KWin::Scene::paintSimpleScreen > (this=0xffffffffffffffff, orig_mask=8, region=) at stack corruption for sure (though the constant QPrinter calls are impressive) - since this is after some actual painting, you'll unfortunately have to valgrind kwin to figure where the stack corruption occurs :-( You could try to "bisect" the issue before - try using the XRender backend. If the issue does not remain, switch back to GL and disable *all* effect plugins (notably beclock, in case it's ABI broken, ie. "incompatible build". also set the scale method to "smooth" in the 3rd tab) and see whether the problem reoccurs. If yes, alter the nvidia driver - there'd be a good chance it's the cuplrit) If not, re-enable the effect plugins (and start with be-clock) to check for the culprit. Personal gut feeling is that it smells like a false binary offset (but in case it only happens at cold starts, could just as well be in the driver instead of the plugin) Until now i don't have the crash. I have the beclock effect disabled. But i have many freezes, after a cold boot and may be is related. I have these lines in syslog just before the hard reset that i did due the freeze: Dec 27 17:58:32 localhost kernel: [ 327.404364] virtuoso-t[4093]: segfault at 7fd79cf438ff ip 00007fd7d9371481 sp 00007fd7d37780c8 error 6 in libc-2.14.1.so[7fd7d92f2000+182000] Dec 27 17:58:32 localhost kernel: [ 327.508585] nepomukservices[4062]: segfault at ffff80517df94fcf ip 00007faed0e03481 sp 00007fae96ffcb08 error 6 in libc-2.14.1.so[7faed0d84000+182000] Dec 27 17:58:32 localhost kernel: [ 327.555582] nepomukserver[3576]: segfault at fffffffffffee3cf ip 00007fb286424481 sp 00007fff8a347a68 error 6 in libc-2.14.1.so[7fb2863a5000+182000] Dec 27 17:58:57 localhost kernel: [ 352.424022] nepomukindexer[4693]: segfault at 7fdb92dc9ffd ip 00007fdb92dc8851 sp 00007ffff3ac9800 error 6 in libexiv2.so.11.0.0[7fdb92cbd000+213000] Please can you suggest me if there is anything that i have to search for ? Created attachment 76041 [details]
Output of /var/log/syslog
The last entry is at 17:58:57
after, i did a hard reset.
(In reply to comment #8) > Until now i don't have the crash. > I have the beclock effect disabled. Ok, this *could* point an incompatible build. Where did you obtain the plugin - does it come "legally" with your distro? Is the build newer than the kwin (kdebase-workspace) one? > But i have many freezes, after a cold boot and may be is related. > > I have these lines in syslog just before the hard reset that i did due the > freeze: Looks more like virtuoso/nepomukindexer is runnig wild - i don't use nepomuk, can't say anything about that. Sorry. It's however not related to the WM at all. Only a wild guess would be that there are more mixed up libs/binaries (eg /usr/local shadowing /usr or an incomplete update or similar) (In reply to comment #10) > (In reply to comment #8) > > Until now i don't have the crash. > > I have the beclock effect disabled. > Ok, this *could* point an incompatible build. > Where did you obtain the plugin - does it come "legally" with your distro? > Is the build newer than the kwin (kdebase-workspace) one? > I have these rpm installed (beclock installed from official repos): kwin-effect-beclock-0.18-1.mga2 kdebase4-workspace-4.8.5-1.2.mga2 > > But i have many freezes, after a cold boot and may be is related. > > > > I have these lines in syslog just before the hard reset that i did due the > > freeze: > Looks more like virtuoso/nepomukindexer is runnig wild - i don't use > nepomuk, can't say anything about that. Sorry. > It's however not related to the WM at all. > I disabled nepomuk. > Only a wild guess would be that there are more mixed up libs/binaries (eg > /usr/local shadowing /usr or an incomplete update or similar) (In reply to comment #11) > I have these rpm installed (beclock installed from official repos): > kwin-effect-beclock-0.18-1.mga2 https://bugs.mageia.org/show_bug.cgi?id=1802 Did you compile it yourself? Is there any equivalent to "packages.mageia.org" (like debian, arch or suse which allow you to inspect packages online) > I disabled nepomuk. Before you got those segfaults from comment 8?? -> Nepomuk or just strigi? (In reply to comment #12) > (In reply to comment #11) > > > I have these rpm installed (beclock installed from official repos): > > kwin-effect-beclock-0.18-1.mga2 > https://bugs.mageia.org/show_bug.cgi?id=1802 > Did you compile it yourself? > Is there any equivalent to "packages.mageia.org" (like debian, arch or suse > which allow you to inspect packages online) > I use tis site: http://pkgs.org/ I know that exist this one too but it does nt work for me: http://mageia.madb.org/ > > I disabled nepomuk. > Before you got those segfaults from comment 8?? > -> Nepomuk or just strigi? I disabled Nepomuk and strigi (i uncheck the three checkboxes from systemsettings→Desktop search) after the segfaults Until now i don't have the crash any more. May be the reason was the beclock, i will turn it on again to test and if there is a problem i will let you know. The nepomuk is always disabled Also, i installed these packages as i was be suggested from here: https://bugs.mageia.org/show_bug.cgi?id=6285#c25 and since that i don't have the freezes anymore. Thank you very much for the support , and for your advices. Happy new year :) I close the report... |