Application: kwin (4.6.3 (4.6.3)) KDE Platform Version: 4.6.3 (4.6.3) Qt Version: 4.7.2 Operating System: Linux 2.6.35.13-91.fc14.x86_64 x86_64 Distribution (Platform): Fedora RPMs -- Information about the crash: - What I was doing when the application crashed: 1. enter system settings / desktop effects 2. select "Advanced" 3. try different "scale method" settings 4. select "OpenGL mode" as "fallback" 5. the crash happened -- Backtrace: Application: KWin (kwin), signal: Segmentation fault 82 T_PSEUDO (SYSCALL_SYMBOL, SYSCALL_NAME, SYSCALL_NARGS) [Current thread is 1 (Thread 0x7f8e36244840 (LWP 5187))] Thread 2 (Thread 0x7f8e29003700 (LWP 5191)): #0 pthread_cond_wait@@GLIBC_2.3.2 () at ../nptl/sysdeps/unix/sysv/linux/x86_64/pthread_cond_wait.S:162 #1 0x00007f8e3f263bb4 in QTWTF::TCMalloc_PageHeap::scavengerThread (this=0x7f8e3f57bd80) at ../3rdparty/javascriptcore/JavaScriptCore/wtf/FastMalloc.cpp:2359 #2 0x00007f8e3f263be9 in QTWTF::TCMalloc_PageHeap::runScavengerThread (context=0x7f8e3f589e54) at ../3rdparty/javascriptcore/JavaScriptCore/wtf/FastMalloc.cpp:1464 #3 0x0000003952e06ccb in start_thread (arg=0x7f8e29003700) at pthread_create.c:301 #4 0x00000039522e0c2d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:115 Thread 1 (Thread 0x7f8e36244840 (LWP 5187)): [KCrash Handler] #6 0x00007f8e29c572f6 in intelDRI2Flush (drawable=0x254d8a0) at intel_screen.c:108 #7 0x0000003217c43b9a in dri2SwapBuffers (pdraw=0x254d7e0, target_msc=0, divisor=0, remainder=0) at dri2_glx.c:529 #8 0x00007f8e3f8a2b87 in KWin::SceneOpenGL::flushBuffer (this=<value optimized out>, mask=<value optimized out>, damage=...) at /usr/src/debug/kdebase-workspace-4.6.3/kwin/scene_opengl.cpp:860 #9 0x00007f8e3f8a3395 in KWin::SceneOpenGL::paint (this=0x2543f40, damage=..., toplevels=<value optimized out>) at /usr/src/debug/kdebase-workspace-4.6.3/kwin/scene_opengl.cpp:789 #10 0x00007f8e3f88a3d7 in KWin::Workspace::performCompositing (this=0x2519880) at /usr/src/debug/kdebase-workspace-4.6.3/kwin/composite.cpp:448 #11 0x00007f8e3d24b2f9 in QObject::event (this=0x2519880, e=<value optimized out>) at kernel/qobject.cpp:1181 #12 0x00007f8e3c5ffcd4 in QApplicationPrivate::notify_helper (this=0x245b3a0, receiver=0x2519880, e=0x7fff75b2dd10) at kernel/qapplication.cpp:4462 #13 0x00007f8e3c60488a in QApplication::notify (this=<value optimized out>, receiver=0x2519880, e=0x7fff75b2dd10) at kernel/qapplication.cpp:4341 #14 0x00007f8e3def69f6 in KApplication::notify (this=0x7fff75b2e170, receiver=0x2519880, event=0x7fff75b2dd10) at /usr/src/debug/kdelibs-4.6.3/kdeui/kernel/kapplication.cpp:311 #15 0x00007f8e3d23719c in QCoreApplication::notifyInternal (this=0x7fff75b2e170, receiver=0x2519880, event=0x7fff75b2dd10) at kernel/qcoreapplication.cpp:731 #16 0x00007f8e3d264eae in sendEvent (this=0x245ec60) at kernel/qcoreapplication.h:215 #17 QTimerInfoList::activateTimers (this=0x245ec60) at kernel/qeventdispatcher_unix.cpp:604 #18 0x00007f8e3d264ff0 in QEventDispatcherUNIX::processEvents (this=0x2431bb0, flags=...) at kernel/qeventdispatcher_unix.cpp:923 #19 0x00007f8e3c6a615c in QEventDispatcherX11::processEvents (this=0x2431bb0, flags=...) at kernel/qeventdispatcher_x11.cpp:152 #20 0x00007f8e3d236532 in QEventLoop::processEvents (this=<value optimized out>, flags=...) at kernel/qeventloop.cpp:149 #21 0x00007f8e3d23677c in QEventLoop::exec (this=0x7fff75b2e060, flags=...) at kernel/qeventloop.cpp:201 #22 0x00007f8e3d23ac3b in QCoreApplication::exec () at kernel/qcoreapplication.cpp:1008 #23 0x00007f8e3f8284da in kdemain (argc=<value optimized out>, argv=0x7fff75b2e1d0) at /usr/src/debug/kdebase-workspace-4.6.3/kwin/main.cpp:542 #24 0x000000395221ee5d in __libc_start_main (main=0x400910 <main(int, char**)>, argc=3, ubp_av=0x7fff75b2e808, init=<value optimized out>, fini=<value optimized out>, rtld_fini=<value optimized out>, stack_end=0x7fff75b2e7f8) at libc-start.c:226 #25 0x0000000000400839 in _start () Possible duplicates by query: bug 266362, bug 265296, bug 253703. Reported using DrKonqi
Created attachment 60336 [details] New crash information added by DrKonqi kwin (4.6.3 (4.6.3)) on KDE Platform 4.6.3 (4.6.3) using Qt 4.7.2 - What I was doing when the application crashed: i was trying different advanced kwin settings -- Backtrace (Reduced): #6 0x00007f7df53fe2f6 in intelDRI2Flush (drawable=0x21c77b0) at intel_screen.c:108 #7 0x0000003217c43b9a in dri2SwapBuffers (pdraw=0x2ec2790, target_msc=0, divisor=0, remainder=0) at dri2_glx.c:529 #8 0x00007f7e0c0a2b87 in KWin::SceneOpenGL::flushBuffer (this=<value optimized out>, mask=<value optimized out>, damage=...) at /usr/src/debug/kdebase-workspace-4.6.3/kwin/scene_opengl.cpp:860 #9 0x00007f7e0c0a3395 in KWin::SceneOpenGL::paint (this=0x31202d0, damage=..., toplevels=<value optimized out>) at /usr/src/debug/kdebase-workspace-4.6.3/kwin/scene_opengl.cpp:789 #10 0x00007f7e0c08a3d7 in KWin::Workspace::performCompositing (this=0x1e23610) at /usr/src/debug/kdebase-workspace-4.6.3/kwin/composite.cpp:448
driver bug. *** This bug has been marked as a duplicate of bug 253703 ***
Sorry, but I cannot resist: why do you think the tab is labeled "Advanced"? What do you think that means? Probably not that users are supposed to "play around". It means that only users knowing what they do should change the settings! Sigh, because of such reports we had to remove the option in 4.7.
Hello Martin, This bug report is about a SEGFAULT in kwin. I played with advanced settings :) is this a crime ? I think KDE (and linux in general) should be o robust piece of software. In view of this, the worst case scenario "advanced playing" should be the display stop working for, let's say, 10 secs (the timeout for no confirmation of video changes). The user shouldn't be able to trigger a SEGFAULT, in any robust software.
It's not a crime to play with a loaded gun, but don't cry if you shoot yourself ;-) Aside this and quoting comment #2 "driver bug". So just get fixed drivers/mesa for your GPU then and or export LIBGL_ALWAYS_INDIRECT=1 (what will likely workaround this issue which is likely rather *not* restricted to the Texture Generation mode but to the recreation of the GL context) > The user shouldn't be able to trigger a SEGFAULT, in any robust software. No, that's right. Now show me robust software (Saturn V control code doesn't count - it's outdated ;-) What you expect can only be achieved by systems like Apple which are - developers - distributors - HW vendors (on a very limited HW set) ie., ship you the entire system. We cannot prevent sublying libs and/or drivers from having implementation issues (OpenGL calls /are/ -supposed to be- crash safe, ie. even /iff/ kwin did bogus shit, GL should throw an GL_ERROR and maybe paint ugly stuff, but /never/ cause a segfault)
fallback mode is removed in master -> fixed