In Compiz, re-redirecting an unredirected OpenGL window sometimes kills the X server. Test case: 1. Enable "Unredirect Fullscreen Windows" in CCSM > Composite. 2. Start an OpenGL game, or video with: mplayer -vo gl ... 3. Trigger scale (Shift+Alt+Up) or the window menu (Alt+Space) If it doesn't crash, then repeat. Expected: Either windows scale or the window menu appears Observed: Sometimes, just sometimes, X crashes and I get sent back to the login screen. Program received signal SIGSEGV, Segmentation fault. 0x00007f12ea7f1e24 in drm_intel_bo_is_reusable () from /usr/lib/x86_64-linux-gnu/libdrm_intel.so.1 (gdb) bt #0 0x00007f12ea7f1e24 in drm_intel_bo_is_reusable () from /usr/lib/x86_64-linux-gnu/libdrm_intel.so.1 #1 0x00007f12eaa1ecfb in intel_set_pixmap_bo (pixmap=0x7f12efcf2370, bo=0x0) at ../../src/intel_uxa.c:643 #2 0x00007f12eaa1eec7 in intel_uxa_destroy_pixmap (pixmap=0x7f12efcf2370) at ../../src/intel_uxa.c:1108 #3 0x00007f12ee763f60 in damageDestroyPixmap (pPixmap=0x7f12efcf2370) at ../../../miext/damage/damage.c:1652 #4 0x00007f12eb692402 in XvDestroyPixmap (pPix=0x7f12efcf2370) at ../../Xext/xvmain.c:387 #5 0x00007f12ee73980a in ShmDestroyPixmap (pPixmap=0x7f12efcf2370) at ../../Xext/shm.c:276 #6 0x00007f12eaa3070b in I830DRI2DestroyBuffer (drawable=<optimised out>, buffer=<optimised out>) at ../../src/intel_dri.c:434 #7 0x00007f12eac5e97b in DRI2DrawableGone (p=0x7f12ef829d20, id=<optimised out>) at ../../../../hw/xfree86/dri2/dri2.c:309 #8 0x00007f12ee6be5c2 in doFreeResource (res=0x7f12ef829db0, skip=0) at ../../dix/resource.c:571 #9 0x00007f12ee6bf2fa in FreeClientResources (client=0x7f12efa19d80) at ../../dix/resource.c:853 #10 0x00007f12ee69cd2a in CloseDownClient (client=0x7f12efa19d80) at ../../dix/dispatch.c:3477 #11 0x00007f12ee69d8b6 in Dispatch () at ../../dix/dispatch.c:454 #12 0x00007f12ee68c7ba in main (argc=8, argv=<optimised out>, envp=<optimised out>) at ../../dix/main.c:287 (gdb) With: libdrm-intel1 2.4.32-1ubuntu1 xserver-xorg-video-intel 2:2.17.0-1ubuntu4.1 xserver-xorg 1:7.6+12ubuntu1
Please test against upstream components before filing a distribution bug upstream.
That's excellent news. Is there an existing bug or even commit you can point me to? Obviously I would not have logged this if I could find the same bug logged already.
Just the Ubuntu gfx stack is notoriously old, and since you have only tested against that stack you have not yet confirmed whether this is an upstream bug. In particular if you look at this, this is memory corruption that looks very similar to one of the many faces of bug 50019.
I keep getting different Xorg crashes from Ubuntu 12.04 so can't get log files for this yet. On 12.10 with much newer intel drivers and similar hardware, I can't crash it any more.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.