[Intel-gfx] Bad flicker on skylake HQD due to code in the 4.7 merge window

[James Hogan]

On 7 July 2016 at 20:19, James Bottomley
<James.Bottomley at hansenpartnership.com> wrote:
> On Thu, 2016-07-07 at 09:55 -0700, James Bottomley wrote:
>> On Thu, 2016-07-07 at 19:14 +0300, Ville Syrjälä wrote:
>> > On Tue, Jun 21, 2016 at 06:44:34PM +0300, Ville Syrjälä wrote:
>> > > On Tue, Jun 21, 2016 at 09:53:15AM -0400, James Bottomley wrote:
>> > > > On Mon, 2016-06-20 at 11:03 +0300, Jani Nikula wrote:
>> > > > > Cc: Ville
>> > > > >
>> > > > > On Mon, 20 Jun 2016, James Bottomley <
>> > > > > James.Bottomley at HansenPartnership.com> wrote:
>> > > > > > OK, my candidate bad commit is this one:
>> > > > > >
>> > > > > > commit a05628195a0d9f3173dd9aa76f482aef692e46ee
>> > > > > > Author: Ville Syrjälä <ville.syrjala at linux.intel.com>
>> > > > > > Date:   Mon Apr 11 10:23:51 2016 +0300
>> > > > > >
>> > > > > >     drm/i915: Get panel_type from OpRegion panel details
>> > > > > >
>> > > > > > After being more careful about waiting to identify flicker,
>> > > > > > this one seems to be the one the bisect finds.  I'm now
>> > > > > > running v4.7-rc3 with this one reverted and am currently
>> > > > > > seeing no flicker problems.   It is, however, early days
>> > > > > > because the flicker can hide for long periods, so I 'll
>> > > > > > wait
>> > > > > > until Monday evening and a few reboots before declaring
>> > > > > > victory.
>> > > > >
>> > > > > If that turns out to be the bad commit, it doesn't really
>> > > > > surprise me, and that in itself is depressing.
>> > > >
>> > > > As far as I can tell, after running for a day with this
>> > > > reverted,
>> > > > this is the problem.  The flicker hasn't appeared with it
>> > > > reverted.  It's pretty noticeable with this commit included.
>> > >
>> > > Hmm. The only difference I can see is low vs. normal vswing.
>> > > Panel
>> > > 0 has low, panel 2 has normal. So either the VBT or opregion is
>> > > telling utter lies, or there's some other bug in our low vswing
>> > > support.
>> >
>> > I did a quick once over of out DDI vswing stuff and didn't find
>> > anything too serious. There were some buglets in the iboost
>> > handling,
>> > but I'm not very hopeful that fixing those would help with your
>> > machine.
>> >
>> > Here's a branch anyway in case you want to give it a go:
>> > git://github.com/vsyrjala/linux.git ddi_iboost_fixes
>> >
>> > Actually, I think the only patch in there that might make a
>> > difference is 15d887855180 ("drm/i915: Fix iboost setting for DDI
>> > with 4 lanes on SKL")
>>
>> Running with it now (the entire branch).  So far it looks OK, but
>> I'll give it a couple of days to see if anything manifests before
>> declaring victory.
>
> Bad news, I'm afraid: after a couple of hours of run time, there is now
> noticeable flicker on the display, so although the iboost fixes may
> have lessened it, it's still present.
>
> James
>
>

I've just bisected a similar (same?) problem (slow increase and
decrease of screen brightness with a period of a few seconds) to the
same commit (this is on a Dell XPS 13 laptop)

commit a05628195a0d9f3173dd9aa76f482aef692e46ee
Author: Ville Syrjälä <ville.syrjala at linux.intel.com>
Date:   Mon Apr 11 10:23:51 2016 +0300

    drm/i915: Get panel_type from OpRegion panel details

the difference for me is that good commits used the acpi backlight,
and bad ones used the intel backlight (/sys/class/backlight/). Its
like intel_backlight is fighting with the firmware to control the
backlight or something. Reverting the commit on v4.7.3 switches it
back to the acpi backlight and all works nicely again.

let me know if I can provide anything else to help debug.

Cheers
James