Version: (using KDE KDE 3.5.2) Installed from: Slackware Packages OS: Linux Hello, I think I have found a nasty bug in kwin but I have unfortunately no way to reliably reproduce it :( Basically, kwin goes stuck in a seemingly infinite loop, flooding the main console with "QGArray:at: Absolute index XXX out of range" messages. This effectively lock the keyboard and mouse input in X11. I googled for it but found nothing relevant (apparently bugs with the same symptoms were found in the KMail and Konqueror codebase by the past, though => http://bugs.kde.org/show_bug.cgi?id=71549 for example). I triggered this this afternoon by pressing alt tab while I was playing Counter Strike in wine (windowed). I successively killed wineserver and X using ssh, and I found my main console printing the aforementioned message. I then killed kwin which has survived to the death of X, and all went back to normal, so it should be the culprit. But I also encoutered this bug before while simply clicking on the icon of a mount point on my desktop while the CPU load was high. This led me to the impression that this bug is somewhat related to the CPU usage. Here is my setup: Slackware-current using a custom built 2.6.16.16 kernel (SMP) running on a pentium 4 with hyper threading enabled KDE 3.5.2 with QT 3.3.6 Hope this helps :) Have a nice day.
Can you switch to text mode, attach gdb to kwin and provide a backtrace?
I am sorry, I did not think to use gdb at the time, I will make sure to do that if I manage to reproduce it. I however successfully reproduced the bug which froze my desktop by clicking on a mount point icon while CPU load was high, but it is way more benign than I remembered :) It's simply kdesktop which get stuck with this one. I made a backtrace of the frozen kdesktop and will file a new entry.
OK, I don't know if this is a variant of the bug I encountered or a different one, but I pressed Alt Tab while Counter Strike was enumerating servers and this led kwin to segfault. Here is the backtrace: [KCrash handler] #5 0xa672fc25 in KWinInternal::Workspace::nextFocusChainClient () from /opt/kde/lib/libkdeinit_kwin.so #6 0xa6730b0e in KWinInternal::TabBox::createClientList () from /opt/kde/lib/libkdeinit_kwin.so #7 0xa6732729 in KWinInternal::TabBox::reset () from /opt/kde/lib/libkdeinit_kwin.so #8 0xa673320c in KWinInternal::Workspace::startKDEWalkThroughWindows () from /opt/kde/lib/libkdeinit_kwin.so #9 0xa6733328 in KWinInternal::Workspace::slotWalkThroughWindows () from /opt/kde/lib/libkdeinit_kwin.so #10 0xa6713c82 in KWinInternal::Workspace::qt_invoke () from /opt/kde/lib/libkdeinit_kwin.so #11 0xa778617a in KGlobalAccelPrivate::activate () from /opt/kde/lib/libkdecore.so.4 #12 0xa7787079 in KGlobalAccelPrivate::x11KeyPress () from /opt/kde/lib/libkdecore.so.4 #13 0xa778720c in KGlobalAccelPrivate::x11Event () from /opt/kde/lib/libkdecore.so.4 #14 0xa76f1e77 in KApplication::x11EventFilter () from /opt/kde/lib/libkdecore.so.4 #15 0xa672d532 in KWinInternal::Application::x11EventFilter () from /opt/kde/lib/libkdeinit_kwin.so #16 0xa7008c75 in qt_set_x11_event_filter () from /usr/lib/qt/lib/libqt-mt.so.3 #17 0xa70163f8 in QApplication::x11ProcessEvent () from /usr/lib/qt/lib/libqt-mt.so.3 #18 0xa7029f25 in QEventLoop::processEvents () from /usr/lib/qt/lib/libqt-mt.so.3 #19 0xa7094951 in QEventLoop::enterLoop () from /usr/lib/qt/lib/libqt-mt.so.3 #20 0xa70948a6 in QEventLoop::exec () from /usr/lib/qt/lib/libqt-mt.so.3 #21 0xa707d38f in QApplication::exec () from /usr/lib/qt/lib/libqt-mt.so.3 #22 0xa672dba2 in kdemain () from /opt/kde/lib/libkdeinit_kwin.so #23 0xa75ec774 in kdeinitmain () from /opt/kde/lib/kde3/kwin.so #24 0x0804e474 in ?? () #25 0x00000001 in ?? () #26 0x080ed790 in ?? () #27 0x00000001 in ?? () #28 0x00000000 in ?? () Hope this helps :)
The backtrace looks exactly like from bug #124511, so unless you can provide a backtrace that's different from this I'll assume a memory corruption caused by that bug. *** This bug has been marked as a duplicate of 124511 ***