Periodically kwin will crash. The following is in .xsession-errors: intel_do_flush_locked failed: No space left on device This appears to be a problem with the xorg Intel driver. See: https://bugs.freedesktop.org/show_bug.cgi?id=51751 https://bugzilla.redhat.com/show_bug.cgi?id=699705 https://bugzilla.redhat.com/show_bug.cgi?id=830772 However, I'm not sure if in this case it is entirely the driver's fault. We're running with the laptop display off: Screen 0: minimum 320 x 200, current 1920 x 1080, maximum 2048 x 2048 LVDS1 connected (normal left inverted right x axis y axis) 1024x768 60.0 + 800x600 60.3 56.2 640x480 59.9 VGA1 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 509mm x 286mm 1920x1080 60.0*+ 1680x1050 60.0 1680x945 60.0 1400x1050 74.9 60.0 1600x900 60.0 1280x1024 75.0 60.0 1440x900 75.0 59.9 1280x960 60.0 1366x768 60.0 1360x768 60.0 1280x800 74.9 59.8 1152x864 75.0 1280x768 74.9 59.9 1024x768 75.1 70.1 60.0 1024x576 60.0 800x600 72.2 75.0 60.3 56.2 848x480 60.0 640x480 72.8 75.0 60.0 720x400 70.1 So in normal operation everything is fine because the current screen is less than 2048 pixels wide. But perhaps kwin is asking the driver to render to a buffer greater than 2048 pixels wide at some point? Reproducible: Sometimes
Do you get a backtrace (DrKonqui dialog showing up, click on "developer information") It btw. would be entirely the drivers fault. Even if we somewhere exceeded the pbuffer or vbo limits, we should simply get a GL error and you a blck window - but no way a crash. According to google, this error btw. happens to everybody (remains the question whether it's actually related to the crash -> backtrace ;-)
We do not get a backtrace. I've disabled compositing for now.
could you gdb attach kwin, cause / wait for the crash and obtain the trace from gdb?
I'm also experiencing this crash and will try to. (KDE 4.9.3) This crash has also been reported to the Debian BTS in August: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=686157
Apparently, kwin does not really "crash" but catches an invalid condition / state and exits manually with a failure code, which also explains that Dr. Konqui does not kick in: (gdb) continue Continuing. [Thread 0x7f02dd95e700 (LWP 17132) exited] [Thread 0x7f02cd535700 (LWP 17143) exited] [Thread 0x7f02f8e9e780 (LWP 17128) exited] [Inferior 1 (process 17128) exited with code 01] (gdb) The program is not being run. (gdb) bt full No stack. (gdb) Any hints on how to diagnose that and how to get the desired backtrace? Is there any method I should / could place a breakpoint at? (I'm pretty inexperienced with gdb and kwin's / KDE's internals, so please, if possible, provide sufficient details for me to be able to follow.) BTW, it looks as if the crash could be triggered, or at least the likelyhood of the crash can be increased, by opening quite a few windows and then task-switching between them using cover flow. 00:02.0 VGA compatible controller: Intel Corporation Mobile 945GM/GMS, 943/940GML Express Integrated Graphics Controller (rev 03) 00:02.1 Display controller: Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller (rev 03)
That's a SIGCHLD - have you tried switching between SNA & UXA? Tagging upstream - that thing crashes everything and kwin cannot prevent SIGCHLDs (but pot. catch and restart on them) Also the problem is reported for compiz ("unity") and mutter (gnome3)
*** Bug 317267 has been marked as a duplicate of this bug. ***