[Intel-gfx] commit 15512021eb3975a8c2366e3883337e252bb0eee5 causes white spots in console screens

[Lisovskiy, Stanislav]

On Thu, Apr 14, 2022 at 08:33:35AM +0200, François Valenduc wrote:
> Le 14/04/22 à 08:31, Lisovskiy, Stanislav a écrit :
> > On Wed, Apr 13, 2022 at 08:12:20PM +0300, Jani Nikula wrote:
> > > On Wed, 13 Apr 2022, François Valenduc <francoisvalenduc at gmail.com> wrote:
> > > > Commit 15512021eb3975a8c2366e3883337e252bb0eee5
> > > > (15512021eb3975a8c2366e3883337e252bb0eee5) causes a lof of white spots
> > > > to appears on the right upper corner of all console screens (see
> > > > https://drive.google.com/file/d/13GabEvOIKSAj5yox6ybAZEDu3Ixncw5Q/view).
> > > > git-bisect shows that this is the offending commit and if I revert it,
> > > > the problem goes away. The problem still occurs with kernel 5.18-rc2 and
> > > > to all stable trees where it was applied.
> > > > Can somebody explains what happens ?
> > > > 
> > > > The video card is the following: VGA compatible controller: Intel
> > > > Corporation WhiskeyLake-U GT2 [UHD Graphics 620] (rev 02) (prog-if 00
> > > > [VGA controller])
> > > > 
> > > > Please tell me if you need more info.
> > > That's commit 15512021eb39 ("drm/i915: Workaround broken BIOS DBUF
> > > configuration on TGL/RKL"), adding Cc's.
> > > 
> > > Please file a report at fdo gitlab [1] and attach dmesg etc. there.
> > That commit looks like it is just disabling all the planes, if wrong
> > dbuf/wm configuration is detected.
> > However it should do that only once during boot as I understand.
> > 
> > Are you sure that is exactly this commit which is causing this?
> > Does the issue appear always?
> > 
> > Ville Syrjälä, thoughts?
> > 
> > Stan
> > 
> > > Thanks,
> > > Jani.
> > > 
> > > 
> > > [1] https://gitlab.freedesktop.org/drm/intel/wikis/How-to-file-i915-bugs
> > > 
> > > 
> > > -- 
> > > Jani Nikula, Intel Open Source Graphics Center
> 
> As I said, git-bisect shows it's the offending commit and if I revert it,
> the problem doesn't happens. Otherwise, it always occurs.
> 
> François
>

Does it just happen all the time or some steps/certain circumstances needed
for it to happen?

Only suspicion after looking briefly is that once suspend/resume is done
it might be messing something up. Just a quick guess..

Stan