[Intel-gfx] [PATCH] drm/i915: Don't recheck link status for eDP display.

[Puthikorn Voravootivat]

+ Vathsala for PSR related code.

On Tue, Oct 17, 2017 at 4:29 PM, Manasi Navare
<manasi.d.navare at intel.com> wrote:
> On Tue, Oct 17, 2017 at 03:46:00PM -0700, Puthikorn Voravootivat wrote:
>> I tried  https://patchwork.freedesktop.org/series/30670/ but it doesn't work.
>>
>> > When you see the display blinking, do you know for sure that drm_dp_channel_eq_ok() returns false?
>> Yes. Repro is running "modetest" while PSR is on (easier to repro with
>> PSR2 but PSR1 is fine too)
>>
>> > lane_align = dp_link_status(link_status,
>> >    DP_LANE_ALIGN_STATUS_UPDATED);
>> > if ((lane_align & DP_INTERLANE_ALIGN_DONE) == 0)
>> >    return false;
>>
>> I always got lane_align = 128 (DP_LINK_STATUS_UPDATED)
>> So drm_dp_channel_eq_ok() always return false.
>
> Hmm, it looks like there is a bug somewhere else that is causing the
> link status to get updated and bit LINK_STATUS_UPDATED from LANE_ALIGN_STATUS_UPDATED register
> is not being read so not being cleared causing the link to be retrained again.
>
> So I still feel that this patch is a workaround for another bug in the PSR code.
> But I will let others comment on this.
>
> Regards
> Manasi
>
>>
>> If I disable PSR, drm_dp_channel_eq_ok() will return true and no
>> blinking occurs.
>>
>> On Tue, Oct 17, 2017 at 2:21 PM, Manasi Navare
>> <manasi.d.navare at intel.com> wrote:
>> > On Tue, Oct 17, 2017 at 02:01:56PM -0700, Puthikorn Voravootivat wrote:
>> >> intel_dp_long_pulse() is always checking link status because
>> >> there has been known issues of link loss triggerring long pulse.
>> >>
>> >> However this is not needed for eDP display since we won't have
>> >> link loss for internal display. Also there are reports that
>> >> screens are flickering during link status check. (repro by
>> >> running modetest command repeatedly)
>> >>
>> >> Signed-off-by: Puthikorn Voravootivat <puthik at chromium.org>
>> >> ---
>> >>  drivers/gpu/drm/i915/intel_dp.c | 7 ++++++-
>> >>  1 file changed, 6 insertions(+), 1 deletion(-)
>> >>
>> >> diff --git a/drivers/gpu/drm/i915/intel_dp.c b/drivers/gpu/drm/i915/intel_dp.c
>> >> index 4b65cf137f79..75a77ef257e2 100644
>> >> --- a/drivers/gpu/drm/i915/intel_dp.c
>> >> +++ b/drivers/gpu/drm/i915/intel_dp.c
>> >> @@ -4763,7 +4763,8 @@ intel_dp_long_pulse(struct intel_connector *intel_connector)
>> >>                */
>> >>               status = connector_status_disconnected;
>> >>               goto out;
>> >> -     } else {
>> >> +     } else if (status != connector_status_connected ||
>> >> +                intel_encoder->type != INTEL_OUTPUT_EDP) {
>> >>               /*
>> >>                * If display is now connected check links status,
>> >>                * there has been known issues of link loss triggerring
>> >> @@ -4775,6 +4776,10 @@ intel_dp_long_pulse(struct intel_connector *intel_connector)
>> >>                * going back up soon after. And once that happens we must
>> >>                * retrain the link to get a picture. That's in case no
>> >>                * userspace component reacted to intermittent HPD dip.
>> >> +              *
>> >> +              * Skip checking links status for connected eDP display.
>> >> +              * There are known issues of display blinking during checking
>> >> +              * link status and we don't have link loss for internal display.
>> >>                */
>> >
>> > Inside intel_dp_check_link_status(), it actually checks if link status is good by
>> > checking both clock recovery and Channel EQ bits in DPCD, only then it will retrain.
>> > So in case of eDP panel, if there is no link loss then it should always return link
>> > status as good and not retrain.
>> > So IMHO I dont think we need to explicitly avoid this for eDP.
>> >
>> > When you see the display blinking, do you know for sure that drm_dp_channel_eq_ok() returns false?
>> > Also look at this patch: https://patchwork.freedesktop.org/series/30670/
>> > This makes sure none of the eDP training parameters change unless link is bad. So with this patch, if the
>> > link is good then it should never retrain even in intel_dp_check_link_status() for eDP.
>> >
>> > Manasi
>> >>               intel_dp_check_link_status(intel_dp);
>> >>       }
>> >> --
>> >> 2.15.0.rc0.271.g36b669edcc-goog
>> >>
>> >> _______________________________________________
>> >> Intel-gfx mailing list
>> >> Intel-gfx at lists.freedesktop.org
>> >> https://lists.freedesktop.org/mailman/listinfo/intel-gfx