[RFC 0/3] Enable darkscreen detection

[Kandpal, Suraj]

> -----Original Message-----
> From: Jani Nikula <jani.nikula at linux.intel.com>
> Sent: Wednesday, April 24, 2024 3:19 PM
> To: Kandpal, Suraj <suraj.kandpal at intel.com>; intel-gfx at lists.freedesktop.org
> Cc: Murthy, Arun R <arun.r.murthy at intel.com>; Kandpal, Suraj
> <suraj.kandpal at intel.com>
> Subject: Re: [RFC 0/3] Enable darkscreen detection
> 
> On Wed, 24 Apr 2024, Suraj Kandpal <suraj.kandpal at intel.com> wrote:
> > Fifo underrun can lead to flicker, screen corruption and blank screen
> > in case of pipe hang.
> > This patch series aims to enable dark screen detection whenever FIFO
> > underruns are hit and in case we face a darkscreen at this point
> > instead of disabling irq and fbc we want to try reset the pipe on
> > which the underrun has occured and keep a track on the number of times
> > underrun occurs post reset and if this persists more than 5 times we
> > go ahead and reset the whole display. If this still persists we then
> > fallback to disabling irq and the fb.
> > To summarize we are using the dark screen detction feature to detect
> > the pipe hung scenario of FIFO underrun. After detection as corrective
> > measurements following is done.
> > 1)pipe reset
> > 2)display reset
> > Comments on this design and ideas on how else this can be implemented
> > or diffrent scenarios dark screen detection can be used are also
> > welcome.
> 
> I don't know what the above has to do with the patches posted. Please don't
> send this raw patches even as RFC. If you only want to discuss, let's discuss,
> but in this case the patches were a distraction, sorry.
> 
> One of the big opens is that AFAICT this still detects desired black screen as
> dark screen, and flags it as a problem. Now, the patches do only log something
> to dmesg, so it's neither here or there, but if you want to do something more
> clever, this is a consideration.
> 

Hi Jani,
The patches posted were a little clean up of the work we already have and want to move forward from here.
This was posted to initiate a discussion if using this dark screen detection can we reset a pipe or display based on what was written on the cover letter
.So yes we do want to replace the simple print and do a pipe reset whenever darkscreen is detected during FIFO underrun.
Wanted to have a consent if that would be okay before going ahead and putting efforts on that.

Regards,
Suraj Kandpal

> BR,
> Jani.
> 
> 
> >
> > Signed-off-by: Suraj Kandpal <suraj.kandpal at intel.com>
> >
> > Suraj Kandpal (3):
> >   drm/i915: Add Dark screen detection registers
> >   drm/i915/darkscreen: Enable darkscreen detection
> >   drm/i915: Schedule darkscreen detection work
> >
> >  drivers/gpu/drm/i915/Makefile                 |   1 +
> >  .../gpu/drm/i915/display/intel_darkscreen.c   | 139 ++++++++++++++++++
> >  .../gpu/drm/i915/display/intel_darkscreen.h   |  25 ++++
> >  .../drm/i915/display/intel_display_types.h    |   3 +
> >  .../drm/i915/display/intel_fifo_underrun.c    |   2 +
> >  drivers/gpu/drm/i915/i915_reg.h               |   8 +
> >  drivers/gpu/drm/xe/Makefile                   |   1 +
> >  7 files changed, 179 insertions(+)
> >  create mode 100644 drivers/gpu/drm/i915/display/intel_darkscreen.c
> >  create mode 100644 drivers/gpu/drm/i915/display/intel_darkscreen.h
> 
> --
> Jani Nikula, Intel