[Intel-gfx] [PATCH 3/3] drm/i915: optimise intel_runtime_pm_{get, put}

[Chris Wilson]

On Fri, Nov 18, 2016 at 12:06:03PM +0200, David Weinehall wrote:
> Benchmarking shows that on resume we spend quite a bit of time
> just taking and dropping these references, leaving us two options;
> either rewriting the code not to take these references more than
> once, which would be a rather invasive change since the involved
> functions are used from other places, or to optimise
> intel_runtime_pm_{get,put}().  This patch does the latter.
> Initial benchmarking indicate improvements of a couple
> of milliseconds on resume.
> 
> Original patch by Chris, with slight fixes by me.
> 
> Signed-off-by: David Weinehall <david.weinehall at linux.intel.com>
> CC: Chris Wilson <chris at chris-wilson.co.uk>
> ---
>  drivers/gpu/drm/i915/intel_runtime_pm.c | 12 ++++++++++++
>  1 file changed, 12 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/intel_runtime_pm.c b/drivers/gpu/drm/i915/intel_runtime_pm.c
> index 356c662ad453..4bf279023b39 100644
> --- a/drivers/gpu/drm/i915/intel_runtime_pm.c
> +++ b/drivers/gpu/drm/i915/intel_runtime_pm.c
> @@ -2632,6 +2632,9 @@ void intel_runtime_pm_get(struct drm_i915_private *dev_priv)
>  	struct pci_dev *pdev = dev_priv->drm.pdev;
>  	struct device *kdev = &pdev->dev;
>  
> +	if (atomic_inc_not_zero(&dev_priv->pm.wakeref_count))
> +		return;
> +
>  	pm_runtime_get_sync(kdev);
>  
>  	atomic_inc(&dev_priv->pm.wakeref_count);
> @@ -2653,6 +2656,9 @@ bool intel_runtime_pm_get_if_in_use(struct drm_i915_private *dev_priv)
>  	struct pci_dev *pdev = dev_priv->drm.pdev;
>  	struct device *kdev = &pdev->dev;
>  
> +	if (atomic_inc_not_zero(&dev_priv->pm.wakeref_count))
> +		return;
> +
>  	if (IS_ENABLED(CONFIG_PM)) {
>  		int ret = pm_runtime_get_if_in_use(kdev);
>  
> @@ -2695,6 +2701,9 @@ void intel_runtime_pm_get_noresume(struct drm_i915_private *dev_priv)
>  	struct pci_dev *pdev = dev_priv->drm.pdev;
>  	struct device *kdev = &pdev->dev;
>  
> +	if (atomic_inc_not_zero(&dev_priv->pm.wakeref_count))
> +		return;
> +
>  	assert_rpm_wakelock_held(dev_priv);
>  	pm_runtime_get_noresume(kdev);
>  
> @@ -2714,6 +2723,9 @@ void intel_runtime_pm_put(struct drm_i915_private *dev_priv)
>  	struct pci_dev *pdev = dev_priv->drm.pdev;
>  	struct device *kdev = &pdev->dev;
>  
> +	if (!atomic_dec_and_test(&dev_priv->pm.wakeref_count))
> +		return;
> +
>  	assert_rpm_wakelock_held(dev_priv);
>  	atomic_dec(&dev_priv->pm.wakeref_count);

Have to remove this dec. Time to retest ;-)
-Chris

-- 
Chris Wilson, Intel Open Source Technology Centre