[PATCH] drm/i915: Boost GPU frequency if we detect outstanding pageflips

[PATCH] drm/i915: Boost GPU frequency if we detect outstanding pageflips

[Chris Wilson]

If we hit a vblank and see that have a pageflip queue but not yet
processed, ensure that the GPU is running at maximum in order to clear
the backlog. Pageflips are only queued for the following vblank, if we
miss it, there will be a visible stutter. Boosting the GPU frequency
doesn't prevent us from missing the target vblank, but it should help
the subsequent frames hitting theirs.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
---
 drivers/gpu/drm/i915/intel_display.c |    5 +++++
 1 file changed, 5 insertions(+)

diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index e15b3ab..215526f 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -8960,6 +8960,7 @@ void intel_check_page_flip(struct drm_device *dev, int pipe)
 	struct drm_crtc *crtc = dev_priv->pipe_to_crtc_mapping[pipe];
 	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
 	unsigned long flags;
+	bool outstanding;
 
 	if (crtc == NULL)
 		return;
@@ -8972,7 +8973,11 @@ void intel_check_page_flip(struct drm_device *dev, int pipe)
 		page_flip_completed(dev_priv, intel_crtc, intel_crtc->unpin_work);
 		intel_crtc->unpin_work = NULL;
 	}
+	outstanding = intel_crtc->unpin_work != NULL;
 	spin_unlock_irqrestore(&dev->event_lock, flags);
+
+	if (outstanding && INTEL_INFO(dev)->gen >= 6)
+		gen6_rps_boost(dev_priv);
 }
 
 static int intel_crtc_page_flip(struct drm_crtc *crtc,
-- 
1.7.9.5

Re: [PATCH] drm/i915: Boost GPU frequency if we detect outstanding pageflips

[Daniel Vetter]

On Thu, Mar 27, 2014 at 12:01:11PM +0000, Chris Wilson wrote:
> If we hit a vblank and see that have a pageflip queue but not yet
> processed, ensure that the GPU is running at maximum in order to clear
> the backlog. Pageflips are only queued for the following vblank, if we
> miss it, there will be a visible stutter. Boosting the GPU frequency
> doesn't prevent us from missing the target vblank, but it should help
> the subsequent frames hitting theirs.
> 
> Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Does kms_flip already have a synthetic subtest which reliably hits this?
One of the dummy load things should be good enough.

I just want to make sure we exercise this code, actual functional testing
is a bit too tricky I think.
-Daniel

> ---
>  drivers/gpu/drm/i915/intel_display.c |    5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
> index e15b3ab..215526f 100644
> --- a/drivers/gpu/drm/i915/intel_display.c
> +++ b/drivers/gpu/drm/i915/intel_display.c
> @@ -8960,6 +8960,7 @@ void intel_check_page_flip(struct drm_device *dev, int pipe)
>  	struct drm_crtc *crtc = dev_priv->pipe_to_crtc_mapping[pipe];
>  	struct intel_crtc *intel_crtc = to_intel_crtc(crtc);
>  	unsigned long flags;
> +	bool outstanding;
>  
>  	if (crtc == NULL)
>  		return;
> @@ -8972,7 +8973,11 @@ void intel_check_page_flip(struct drm_device *dev, int pipe)
>  		page_flip_completed(dev_priv, intel_crtc, intel_crtc->unpin_work);
>  		intel_crtc->unpin_work = NULL;
>  	}
> +	outstanding = intel_crtc->unpin_work != NULL;
>  	spin_unlock_irqrestore(&dev->event_lock, flags);
> +
> +	if (outstanding && INTEL_INFO(dev)->gen >= 6)
> +		gen6_rps_boost(dev_priv);
>  }
>  
>  static int intel_crtc_page_flip(struct drm_crtc *crtc,
> -- 
> 1.7.9.5
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch