Comment 6 for bug 954293

Revision history for this message
Bryce Harrington (bryce) wrote :

Interesting, you're definitely getting a GPU lockup:

Xorg.0.log
[ 339.944] [mi] EQ overflowing. Additional events will be discarded until existing events are processed.
[ 339.944]
Backtrace:
[ 339.944] 0: /usr/bin/X (xorg_backtrace+0x37) [0xb7795007]
[ 339.944] 1: /usr/bin/X (mieqEnqueue+0x223) [0xb7773353]

dmesg:
[ 107.897744] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
[ 107.897804] [drm] capturing error event; look for more information in /debug/dri/0/i915_error_state
[ 107.901347] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 17410 at 17406, next 17411)
[ 243.583933] cfg80211: All devices are disconnected, going to restore regulatory settings

That means X.org couldn't write to the GPU's memory, which is a symptom of a hung GPU.

However:

Time: 1331732836 s 509176 us
PCI ID: 0x0116
EIR: 0x00000000
PGTBL_ER: 0x00000000
ERROR: 0x00000000

I assume you gathered these by ssh'ing in when the system was hung?

In both your i915_error_state files, there are no errors registered, just 0's which indicate the GPU isn't hung on a software error. Any chance it could be a hardware error? Try booting an older ubuntu livecd from before you started noticing the error, like alpha2, and see if it reproduces there. If so, a hardware problem might be a good guess, so keep an eye on the cpu temperature and see if you have a bad fan or some such.

If it doesn't reproduce on the older ubuntu, then the kernel would be the next thing to look at.