CI igt-all on drm-intel-next-fixes gem_exec_* and other gem_*

CI igt-all on drm-intel-next-fixes gem_exec_* and other gem_*

[Rodrigo Vivi]

Hi guys,

As we run more rounds of CI on drm-intel-next-fixes
a standard gets visible there.

This round of dinf started with CI_DINF_90.

Specially gem_exec_* but also other gem_* tests fails randomly
on gen9 platforms:

https://intel-gfx-ci.01.org/tree/drm-intel-next-fixes/shards.html
(last 5 columns of gen9 platforms)

while it is pure clean green on drm-tip:

https://intel-gfx-ci.01.org/tree/drm-tip/shards-all.html

From the logs I looked it seems that it is always during hibernation
and most of the times some strange page faults and also not
only i915:

<6>[   36.647941] PM: Image saving progress:  70%
<3>[   36.707234] e1000e 0000:00:1f.6 eno1: Detected Hardware Unit Hang:

Do you know what could justify this difference?

Is there anything we could we be missing on our side?

Thanks,
Rodrigo.
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: CI igt-all on drm-intel-next-fixes gem_exec_* and other gem_*

[Rodrigo Vivi]

Ok, things look much cleaner(greener) on last run after I rebased on
top of a more recent drm-next.

https://intel-gfx-ci.01.org/tree/drm-intel-next-fixes/shards.html

My only concern right now is this one:

https://intel-gfx-ci.01.org/tree/drm-intel-next-fixes/CI_DINF_97/shard-kbl5/igt@gem_softpin@noreloc-s3.html

<7>[  256.521580] [IGT] gem_softpin: starting subtest noreloc-S3
<7>[  256.561583] [drm:gmbus_xfer [i915]] GMBUS [i915 gmbus dpc] timed
out, falling back to bit banging on pin 4
<7>[  256.564321] [drm:drm_dp_dual_mode_detect] DP dual mode HDMI ID:  (err -6)
<7>[  256.564326] [drm:drm_helper_hpd_irq_event]
[CONNECTOR:70:HDMI-A-1] status updated from disconnected to
disconnected
�������������������������������������������

Could this be caused by?:

commit 90024a5951029685acc5396258f1b0de9b23cf4a
Author: Stefan Brüns <stefan.bruens@rwth-aachen.de>
Date:   Sun Dec 31 23:34:54 2017 +0100

    drm/i915: Try EDID bitbanging on HDMI after failed read

On Wed, Jan 31, 2018 at 10:10 PM, Rodrigo Vivi <rodrigo.vivi@intel.com> wrote:
> Hi guys,
>
> As we run more rounds of CI on drm-intel-next-fixes
> a standard gets visible there.
>
> This round of dinf started with CI_DINF_90.
>
> Specially gem_exec_* but also other gem_* tests fails randomly
> on gen9 platforms:
>
> https://intel-gfx-ci.01.org/tree/drm-intel-next-fixes/shards.html
> (last 5 columns of gen9 platforms)
>
> while it is pure clean green on drm-tip:
>
> https://intel-gfx-ci.01.org/tree/drm-tip/shards-all.html
>
> From the logs I looked it seems that it is always during hibernation
> and most of the times some strange page faults and also not
> only i915:
>
> <6>[   36.647941] PM: Image saving progress:  70%
> <3>[   36.707234] e1000e 0000:00:1f.6 eno1: Detected Hardware Unit Hang:
>
> Do you know what could justify this difference?
>
> Is there anything we could we be missing on our side?
>
> Thanks,
> Rodrigo.
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/intel-gfx



-- 
Rodrigo Vivi
Blog: http://blog.vivi.eng.br
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

Re: CI igt-all on drm-intel-next-fixes gem_exec_* and other gem_*

[Chris Wilson]

Quoting Rodrigo Vivi (2018-02-01 21:00:45)
> Ok, things look much cleaner(greener) on last run after I rebased on
> top of a more recent drm-next.
> 
> https://intel-gfx-ci.01.org/tree/drm-intel-next-fixes/shards.html
> 
> My only concern right now is this one:
> 
> https://intel-gfx-ci.01.org/tree/drm-intel-next-fixes/CI_DINF_97/shard-kbl5/igt@gem_softpin@noreloc-s3.html
> 
> <7>[  256.521580] [IGT] gem_softpin: starting subtest noreloc-S3
> <7>[  256.561583] [drm:gmbus_xfer [i915]] GMBUS [i915 gmbus dpc] timed
> out, falling back to bit banging on pin 4
> <7>[  256.564321] [drm:drm_dp_dual_mode_detect] DP dual mode HDMI ID:  (err -6)
> <7>[  256.564326] [drm:drm_helper_hpd_irq_event]
> [CONNECTOR:70:HDMI-A-1] status updated from disconnected to
> disconnected
> �������������������������������������������
> 
> Could this be caused by?:
> 
> commit 90024a5951029685acc5396258f1b0de9b23cf4a
> Author: Stefan Brüns <stefan.bruens@rwth-aachen.de>
> Date:   Sun Dec 31 23:34:54 2017 +0100
> 
>     drm/i915: Try EDID bitbanging on HDMI after failed read

Indeterminate. The loss of the log does suggest that something went
catastrophically wrong on suspend, and we have the circumspect evidence
that hpd was active at that time. But we do flush those workqueues,
right? So it may just be the grue that sometimes likes to eat machines
for no reason.
-Chris
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx