[Intel-gfx] 945GM GPU hangs

[Nico R.]

Hello!

I can relatively easy trigger a bug which causes a hung GPU on a 945GM
using the intel driver (i915).

I just have to start an application using 3D acceleration, and then use
it for a while, then switch to another window and back. It also seems
that the bug is immediately triggered when another window uses 2D
acceleration; it always happens when a libnotify message window pops up
while another window is using Mesa.

dmesg contains the following lines:
  [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
  [drm] capturing error event; look for more information in
/debug/dri/0/i915_error_state
  [drm:i915_reset] *ERROR* Failed to reset chip.

The Xorg.0.log file contains:
  [ 14648.795] [mi] EQ overflowing. The server is probably stuck in an
infinite loop.
  [ 14648.868]
  Backtrace:
  [ 14648.868] 0: /usr/bin/X (xorg_backtrace+0x3b) [0x81ba1ab]
  [ 14648.868] 1: /usr/bin/X (mieqEnqueue+0x15c) [0x8199fcc]
  [ 14648.868] 2: /usr/bin/X (0x8048000+0x441e2) [0x808c1e2]
  [ 14648.868] 3: /usr/bin/X (xf86PostMotionEventM+0xf1) [0x80c9621]
  [ 14648.868] 4: /usr/lib/xorg/modules/input/evdev_drv.so
(0xb71af000+0x33bf) [0xb71b23bf]
  [ 14648.868] 5: /usr/lib/xorg/modules/input/evdev_drv.so
(0xb71af000+0x4609) [0xb71b3609]
  [ 14648.868] 6: /usr/bin/X (0x8048000+0x6bd9f) [0x80b3d9f]
  [ 14648.868] 7: /usr/bin/X (0x8048000+0x928fc) [0x80da8fc]
  [ 14648.868] 8: (vdso) (__kernel_sigreturn+0x0) [0xb775b400]
  [ 14648.868] 9: /usr/lib/libdrm.so.2 (drmCommandNone+0x32) [0xb72edca2]
  [ 14648.868] 10: /usr/lib/xorg/modules/drivers/intel_drv.so
(0xb71dc000+0x3cc72) [0xb7218c72]
  [ 14648.868] 11: /usr/lib/xorg/modules/drivers/intel_drv.so
(0xb71dc000+0x47c29) [0xb7223c29]
  [ 14648.868] 12: /usr/lib/xorg/modules/drivers/intel_drv.so
(0xb71dc000+0x51080) [0xb722d080]
  [ 14648.868] 13: /usr/bin/X (BlockHandler+0x58) [0x807bba8]
  [ 14648.869] 14: /usr/bin/X (WaitForSomething+0x11f) [0x81b783f]
  [ 14648.869] 15: /usr/bin/X (0x8048000+0x2f5be) [0x80775be]
  [ 14648.869] 16: /usr/bin/X (0x8048000+0x1d9fa) [0x80659fa]
  [ 14648.869] 17: /lib/libc.so.6 (__libc_start_main+0xfe) [0xb741ce5e]
  [ 14652.830] (EE) intel(0): Detected a hung GPU, disabling acceleration.
  [ 14652.830] (EE) intel(0): When reporting this, please include
i915_error_state from debugfs and the full dmesg.

From the entries in the logs, this bug seems to be similar to the
following bugs in the freedesktop.org bug tracker:
  https://bugs.freedesktop.org/show_bug.cgi?id=40075
  https://bugs.freedesktop.org/show_bug.cgi?id=32582
  https://bugs.freedesktop.org/show_bug.cgi?id=40503
  https://bugs.freedesktop.org/show_bug.cgi?id=38952
  https://bugs.freedesktop.org/show_bug.cgi?id=37807
  https://bugs.freedesktop.org/show_bug.cgi?id=29046
  https://bugs.freedesktop.org/show_bug.cgi?id=41184

Maybe a few of them are DUPLICATEs?

I am using the following package versions:
  linux             3.0.4 + tuxonice
  drm               c82ef03e4c92017bf5644f294ea04e30500f8d4c
  mesa              9c697a9d004da4aa7a26d3bda17cc473f50345ea
  xf86-video-intel  c5414ec992d935e10156a2b513d5ec2dded2f689
  xserver           fae7ed62ad476114bd37b566202cf6b6546c0a1f

This is on a Gentoo system; the following use flags are enabled for mesa:
  classic egl g3dvl gallium gbm gles1 gles2 llvm nptl openvg
  osmesa shared-dricore shared-glapi vdpau wayland xvmc
  video_cards_i915 video_cards_intel

For xf86-video-intel, the enabled use flags are:
  dri sna

I have captured dmesg, the Xorg log file, and the i915_error_state.
Should I open a /separate/ bug report? Should I send the saved logs to
this list?

I would like to help track this bug down and fix it. What can I do to be
of any assistance?

Thanks
-- 
Nico

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20111006/6ed8dac2/attachment.sig>