[Intel-gfx] [PATCH v2 4/5] drm/i915/bxt: Block D3 during suspend.

[David Weinehall]

On Thu, Apr 07, 2016 at 08:22:11PM +0530, Animesh Manna wrote:
> For BXT, display engine can not generate interrupt when in D3.
> On the othen hand S0ix can be achieved without display in D3. So,

other

> Display should not put into D3 for HPD to work and will not
> have any power impact.
> 
> Signed-off-by: Animesh Manna <animesh.manna at intel.com>
> ---
>  drivers/gpu/drm/i915/i915_drv.c | 6 ++++++
>  1 file changed, 6 insertions(+)
> 
> diff --git a/drivers/gpu/drm/i915/i915_drv.c b/drivers/gpu/drm/i915/i915_drv.c
> index 020a31c..6b8c906 100644
> --- a/drivers/gpu/drm/i915/i915_drv.c
> +++ b/drivers/gpu/drm/i915/i915_drv.c
> @@ -1547,6 +1547,9 @@ static int intel_runtime_suspend(struct device *device)
>  
>  	assert_forcewakes_inactive(dev_priv);
>  
> +	if (dev_priv->vbt.hpd_wakeup_enabled)
> +		pci_save_state(pdev);
> +
>  	DRM_DEBUG_KMS("Device suspended\n");
>  	return 0;
>  }
> @@ -1563,6 +1566,9 @@ static int intel_runtime_resume(struct device *device)
>  
>  	DRM_DEBUG_KMS("Resuming device\n");
>  
> +	if (dev_priv->vbt.hpd_wakeup_enabled)
> +		pci_restore_state(pdev);

If the state hasn't been saved pci_restore_state() will return
without doing anything, so the conditional shouldn't be necessary.

> +
>  	WARN_ON_ONCE(atomic_read(&dev_priv->pm.wakeref_count));
>  	disable_rpm_wakeref_asserts(dev_priv);