[Intel-gfx] frequent x crashes with gen6/sna

[Daniel Vetter]

The attached log shouldn't result in a dead X server - it's the rather
harmless forcewake warning when reading certain debugfs files. Please
grab the next hang.

Thanks, Daniel

On Thu, Oct 13, 2011 at 16:55, Dieter Mummenschanz <mummenschanz at gmx.net> wrote:
> hi daniel,
>
> roger that. the last x death however was not preceeded by a gpu hang (see message log, dmesg.log also attached). Maybe I ran into something different than experienced before.No i915_error_state was logged. I remember I ran into something similar a couple of days ago and someone provided a patch that seemed to work. however I have no idea if the patch has been included in the latest Dave's latest drm-next git.
>
> Oct 13 16:40:22 gonzo kernel: WARNING: at drivers/gpu/drm/i915/i915_drv.c:354 gen6_gt_force_wake_get+0x1f/0x3c [i915]()
> Oct 13 16:40:22 gonzo kernel: Hardware name: 4236NGG
> Oct 13 16:40:22 gonzo kernel: Modules linked in: i915 fbcon font bitblit softcursor drm_kms_helper drm fb fbdev snd_hda_codec_conexant thinkpad_acpi nvram e1000e iwlagn intel_agp cfbcopyarea intel_gtt video snd_hda_intel backlight agpgart snd_hda_codec cfbimgblt cfbfillrect
> Oct 13 16:40:22 gonzo kernel: Pid: 5164, comm: cat Not tainted 3.1.0-rc4+ #1
> Oct 13 16:40:22 gonzo kernel: Call Trace:
> Oct 13 16:40:22 gonzo kernel: [<ffffffff8102fcf2>] ? warn_slowpath_common+0x78/0x8c
> Oct 13 16:40:22 gonzo kernel: [<ffffffffa00ff484>] ? gen6_gt_force_wake_get+0x1f/0x3c [i915]
> Oct 13 16:40:22 gonzo kernel: [<ffffffffa0108d0e>] ? i915_rstdby_delays+0x2e/0x77 [i915]
> Oct 13 16:40:22 gonzo kernel: [<ffffffff810ba9d1>] ? seq_read+0x16e/0x355
> Oct 13 16:40:22 gonzo kernel: [<ffffffff810a1eaa>] ? vfs_read+0xac/0x126
> Oct 13 16:40:22 gonzo kernel: [<ffffffff810a1f69>] ? sys_read+0x45/0x6e
> Oct 13 16:40:22 gonzo kernel: [<ffffffff813b3c7b>] ? system_call_fastpath+0x16/0x1b
> Oct 13 16:40:22 gonzo kernel: ---[ end trace 056b0c70c095c365 ]---
>
> Regards,
> Dieter
>
> -------- Original-Nachricht --------
>> Datum: Thu, 13 Oct 2011 16:00:23 +0200
>> Von: Daniel Vetter <daniel at ffwll.ch>
>> An: Dieter Mummenschanz <mummenschanz at gmx.net>
>> CC: intel-gfx at lists.freedesktop.org
>> Betreff: Re: [Intel-gfx]  frequent x crashes with gen6/sna
>
>> On Thu, Oct 13, 2011 at 03:45:22PM +0200, Dieter Mummenschanz wrote:
>> > hello,
>> >
>> > using latest drm-core-next git from Dave Airlie as well as the latest
>> xf86-video-intel git my system (Sandybridge Mobile (GT2+)) still hangs
>> frequently right before X dies. This happens every time the GPU is under
>> considerable stress, like decoding a va-api stream and running OpenGL apps at the
>> same time.
>> > syslog records the hang:
>> >
>> > kernel: [drm:i915_hangcheck_elapsed] *ERROR* Hangcheck timer elapsed...
>> GPU hung
>> > kernel: [drm] capturing error event; look for more information in
>> /debug/dri/0/i915_error_state
>> > kernel: [drm:i915_wait_request] *ERROR* i915_wait_request returns -11
>> (awaiting 3263296 at 3263293, next 3263297)
>> >
>> > after 3-4 seconds the system resumes normal following the X crash a
>> couple of minutes later:
>> > kernel: [drm:ironlake_update_pch_refclk] *ERROR* enabling SSC on PCH
>> > kernel: [drm:ironlake_update_pch_refclk] *ERROR* enabling SSC on PCH
>> >
>> > gpu related module opts are:
>> > i915.i915_enable_rc6=1 pcie_aspm=force i915.i915_enable_fbc=1
>> i915.lvds_downclock=1
>> >
>> > at least rc6 enabled is mandatory for me since I ran a mobile gpu.
>> >
>> > any ideas?
>>
>> We have (unfortunately) still a few outstanding issues. To check whether
>> yours is new, please attach the full dmesg and the i915_error_state of the
>> gpu hang.
>>
>> Thanks, Daniel
>> --
>> Daniel Vetter
>> Mail: daniel at ffwll.ch
>> Mobile: +41 (0)79 365 57 48
>
> --
> Empfehlen Sie GMX DSL Ihren Freunden und Bekannten und wir
> belohnen Sie mit bis zu 50,- Euro! https://freundschaftswerbung.gmx.de
>
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx
>
>



-- 
Daniel Vetter
daniel.vetter at ffwll.ch - +41 (0) 79 364 57 48 - http://blog.ffwll.ch