Re: [PATCH 1/2] drm/i915/gt: Don't do display work on platforms without display

["Dixit, Ashutosh" <ashutosh.dixit@xxxxxxxxx>]

On Tue, 11 Oct 2022 00:22:34 -0700, Jani Nikula wrote:
>

Hi Jani,

> On Mon, 10 Oct 2022, Ashutosh Dixit <ashutosh.dixit@xxxxxxxxx> wrote:
> > Do display work only on platforms with display. This avoids holding the
> > runtime PM wakeref for an additional 100+ ms after GT has been parked.
> >
> > Bug: https://gitlab.freedesktop.org/drm/intel/-/issues/7025
> > Signed-off-by: Ashutosh Dixit <ashutosh.dixit@xxxxxxxxx>
> > ---
> >  drivers/gpu/drm/i915/gt/intel_gt_pm.c | 36 +++++++++++++++------------
> >  1 file changed, 20 insertions(+), 16 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/gt/intel_gt_pm.c b/drivers/gpu/drm/i915/gt/intel_gt_pm.c
> > index f553e2173bdad..26aa2e979a148 100644
> > --- a/drivers/gpu/drm/i915/gt/intel_gt_pm.c
> > +++ b/drivers/gpu/drm/i915/gt/intel_gt_pm.c
> > @@ -70,19 +70,21 @@ static int __gt_unpark(struct intel_wakeref *wf)
> >
> >	GT_TRACE(gt, "\n");
> >
> > -	/*
> > -	 * It seems that the DMC likes to transition between the DC states a lot
> > -	 * when there are no connected displays (no active power domains) during
> > -	 * command submission.
> > -	 *
> > -	 * This activity has negative impact on the performance of the chip with
> > -	 * huge latencies observed in the interrupt handler and elsewhere.
> > -	 *
> > -	 * Work around it by grabbing a GT IRQ power domain whilst there is any
> > -	 * GT activity, preventing any DC state transitions.
> > -	 */
> > -	gt->awake = intel_display_power_get(i915, POWER_DOMAIN_GT_IRQ);
> > -	GEM_BUG_ON(!gt->awake);
> > +	if (HAS_DISPLAY(i915) && INTEL_DISPLAY_ENABLED(i915)) {
>
> Feels like something's wrong if you need both of those.

Don't think so:

	/* Only valid when HAS_DISPLAY() is true */
	#define INTEL_DISPLAY_ENABLED(dev_priv) \
		(drm_WARN_ON(&(dev_priv)->drm, !HAS_DISPLAY(dev_priv)),         \
	         !(dev_priv)->params.disable_display &&                         \
	         !intel_opregion_headless_sku(dev_priv))

Maybe inside display code INTEL_DISPLAY_ENABLED is sufficient since code
paths have previously invoked HAS_DISPLAY, but not in non-display code.

Thanks.
--
Ashutosh

> > +		/*
> > +		 * It seems that the DMC likes to transition between the DC states a lot
> > +		 * when there are no connected displays (no active power domains) during
> > +		 * command submission.
> > +		 *
> > +		 * This activity has negative impact on the performance of the chip with
> > +		 * huge latencies observed in the interrupt handler and elsewhere.
> > +		 *
> > +		 * Work around it by grabbing a GT IRQ power domain whilst there is any
> > +		 * GT activity, preventing any DC state transitions.
> > +		 */
> > +		gt->awake = intel_display_power_get(i915, POWER_DOMAIN_GT_IRQ);
> > +		GEM_BUG_ON(!gt->awake);
> > +	}
> >
> >	intel_rc6_unpark(&gt->rc6);
> >	intel_rps_unpark(&gt->rps);
> > @@ -115,9 +117,11 @@ static int __gt_park(struct intel_wakeref *wf)
> >	/* Everything switched off, flush any residual interrupt just in case */
> >	intel_synchronize_irq(i915);
> >
> > -	/* Defer dropping the display power well for 100ms, it's slow! */
> > -	GEM_BUG_ON(!wakeref);
> > -	intel_display_power_put_async(i915, POWER_DOMAIN_GT_IRQ, wakeref);
> > +	if (HAS_DISPLAY(i915) && INTEL_DISPLAY_ENABLED(i915)) {
> > +		/* Defer dropping the display power well for 100ms, it's slow! */
> > +		GEM_BUG_ON(!wakeref);
> > +		intel_display_power_put_async(i915, POWER_DOMAIN_GT_IRQ, wakeref);
> > +	}
> >
> >	return 0;
> >  }
>
> --
> Jani Nikula, Intel Open Source Graphics Center