Graphic corruption with i915 (2.6.39.1)

[Chris Wilson]

On Thu, 16 Jun 2011 11:58:42 -0300, Leandro Lucarella <luca at llucax.com.ar> wrote:
> Hi, I've been experiencing sporadic graphic corruption since a few
> stable kernel versions and I always thought "maybe is fixed in the next
> release", so I upgrade, but I usually experience a different problem
> then. This problems are exposed very rarely, maybe I have no problems
> like for a month or so.
> 
> Now I'm using kernel 2.6.39.1 and I experience one of this problem so
> I though it was time to report it. This is what I got in dmesg:
> 
> [368693.756008] [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed... GPU hung
> [368693.756014] render error detected, EIR: 0x00000010
> [368693.756017]   IPEIR: 0x00000000
> [368693.756019]   IPEHR: 0x02000004
> [368693.756020]   INSTDONE: 0xcc8fffff
> [368693.756022]   INSTPS: 0x8001e025
> [368693.756024]   INSTDONE1: 0xbfffffbf
> [368693.756026]   ACTHD: 0x00715128
> [368693.756027] page table error
> [368693.756029]   PGTBL_ER: 0x00000002
> [368693.756038] [drm:i915_wait_request] *ERROR* i915_wait_request returns -11 (awaiting 17700193 at 17700192, next 17700196)
> [368694.258010] [drm:i915_reset] *ERROR* Failed to reset chip.

If you look in /sys/kernel/debug/dri/0/i915_error_state we can see what
garbage the DDX [xf86-video-intel] fed the gpu.
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre