2 cases were raised internally here.
1 - When fixing PSR2 on SKL/KBL Vathsala noticed that we were receiving
long pulses from the panel what was trigerring link training. If we
try to train the link while in deep sleep state that PSR2 provides
we will lock the display hard and not get any further screen update.
2 - When testing PSR on BXT Tarun noticed some flickering related to short
pulse that was also triggering some aux communications while PSR was
active with link off. This was causing flickerings.
So, we definitely need to block PSR in cases we know that we need aux
channel or any other link communication or change.
I believe many other flickering and current PSR issues might be related
to this.
On this approach I decided to use the runtime pm domains instead of
reinventing the wheel on PSR.
In the past we had psr_exit spread all over the code and that was
really bad.
Please let me know what are the results and any feedback on the idea.
Thanks,
Rodrigo.
Rodrigo Vivi (4):
drm/i915/psr: rename intel_dig_port to dig_port.
drm/i915/psr: Replace dev usage per dev_priv.
drm/i915/psr: Provide block functionality for RPM.
drm/i915: Introduce PSR Block Power Domains.
drivers/gpu/drm/i915/i915_drv.c | 1 +
drivers/gpu/drm/i915/i915_drv.h | 1 +
drivers/gpu/drm/i915/i915_reg.h | 1 +
drivers/gpu/drm/i915/intel_display.c | 2 -
drivers/gpu/drm/i915/intel_drv.h | 10 +-
drivers/gpu/drm/i915/intel_frontbuffer.c | 6 +-
drivers/gpu/drm/i915/intel_psr.c | 161 ++++++++++++++++++-------------
drivers/gpu/drm/i915/intel_runtime_pm.c | 56 +++++++++++
8 files changed, 161 insertions(+), 77 deletions(-)
--
2.4.3
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
