Bug 257794

Created attachment 53696 [details]
New crash information added by DrKonqi

kwin (4.5.3 (KDE 4.5.3)) on KDE Platform 4.5.3 (KDE 4.5.3) using Qt 4.7.1

- What I was doing when the application crashed:
When you are changing from one task to other in the taskbar sometimes get that crash error

-- Backtrace (Reduced):
#12 0x00007f45cadcd473 in QEventLoop::exec (this=0x7fffaa8f8ec0, flags=...) at kernel/qeventloop.cpp:214
#13 0x00007f45cadd180b in QCoreApplication::exec () at kernel/qcoreapplication.cpp:1009
#14 0x00007f45ce8c1c14 in kdemain () from /usr/lib64/libkdeinit4_kwin.so
[...]
#16 0x0000000000400699 in _start ()

The backtrace is rather useless as debug symbols are missing. Please install at least kdebase-workspace-dbg and paste a new backtrace.

I have -dbg packages installed. This is not a crash from within KDE libraries, this is an exception thrown from KDE, catched by QT in eventloop, rethrown and catched by stdc++ library.
So, when program being killed in __GI_raise, original exception is lost and stack unwinded.

this is correct and actually that particular (default*) behaviour of Qt's eventloop got bashed before by smarter ppl. than us before ;-)

- what titlebar decoration do you use? (ain't much stuff in KDE throwing 'round things ;-)
- entirely other direction (wonder about the cpu load): is de/activating some client combinations more likely to crash than others?

*define QT_NO_EXCEPTIONS as cflag to get rid of this

I use "sculpture" window decorations and standard oxygen as desktop theme.
My working set is rather stable, so can't say about de/activation. It is include opera/chrome/firefox (yes, all 3), skype, thunderbird, psi, netbeans, openoffice, copule of other java programs and other office stuff like okular, konsole...

(In reply to comment #5)

> I use "sculpture" window decorations
"skulpture" - ok, that's unlikely it...

> My working set is rather stable, so can't say about de/activation.
is maybe a virtual desktop switch invoked all the time?

I suppose the problem is deeper in the kde/qt/x11. Unlocking session on morning:
No any additional actions were performed. Session unlocked, kcrash handler welcomes ;)

Regarding last question -- no, desktop switching wasn't involved. My taskbar shows only windows from current desktop and crash was when I clicked on taskbar.

Application: Рабочее пространство Plasma (plasma-desktop), signal: Aborted
[KCrash Handler]
#6  0x00007ff4b032b175 in *__GI_raise (sig=<value optimized out>) at ../nptl/sysdeps/unix/sysv/linux/raise.c:64
#7  0x00007ff4b032df80 in *__GI_abort () at abort.c:92
#8  0x00007ff4acb30dc5 in __gnu_cxx::__verbose_terminate_handler() () from /usr/lib/libstdc++.so.6
#9  0x00007ff4acb2f166 in ?? () from /usr/lib/libstdc++.so.6
#10 0x00007ff4acb2f193 in std::terminate() () from /usr/lib/libstdc++.so.6
#11 0x00007ff4acb2f216 in __cxa_rethrow () from /usr/lib/libstdc++.so.6
#12 0x00007ff4adc14c6d in QEventLoop::exec (this=<value optimized out>, flags=<value optimized out>) at kernel/qeventloop.cpp:214
#13 0x00007ff4adc18bbb in QCoreApplication::exec () at kernel/qcoreapplication.cpp:1009
#14 0x00007ff4b0696816 in kdemain (argc=<value optimized out>, argv=<value optimized out>) at ../../../../plasma/desktop/shell/main.cpp:118
#15 0x00007ff4b0317c4d in __libc_start_main (main=<value optimized out>, argc=<value optimized out>, ubp_av=<value optimized out>, init=<value optimized out>, fini=<value optimized out>, 
    rtld_fini=<value optimized out>, stack_end=0x7fffb505b458) at libc-start.c:228
#16 0x00000000004006b9 in _start ()

actually fits... (somehow)

the thing in KDE throwing exceptions is (next to the obvious kjs/khtml) plasma contexts - probably for unrelated reasons but throwing uncatched exceptions causes problems when done in slots ...

this is why i was after the decoration -aurorae- and the desktop switching -indicator osd- ... but oc the problem could be some mem corruption (and some new operator throwing the exception on top of some event...)

*** This bug has been marked as a duplicate of bug 241163 ***

*** Bug 268384 has been marked as a duplicate of this bug. ***