[Intel-gfx] [PATCH 05/12] drm/i915: Clear VLV_IER around irq processing

[Chris Wilson]

On Wed, Apr 13, 2016 at 09:19:51PM +0300, ville.syrjala at linux.intel.com wrote:
> From: Ville Syrjälä <ville.syrjala at linux.intel.com>
> 
> On VLV/CHV the master interrupt enable bit only affects GT/PM
> interrupts. Display interrupts are not affected by the master
> irq control.
> 
> Also it seems that the CPU interrupt will only be generated when
> the combined result of all GT/PM/display interrupts has a 0->1
> edge. We already use the master interrupt enable bit to make sure
> GT/PM interrupt can generate such an edge if we don't end up clearing
> all IIR bits. We must do the same for display interrupts, and for
> that we can simply clear out VLV_IER, and restore after we've acked
> all the interrupts we are about to process.
> 
> So with both master interrupt enable and VLV_IER cleared out, we will
> guarantee that there will be a 0->1 edge if any IIR bits are still set
> at the end, and thus another CPU interrupt will be generated.
> 
> Cc: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
> Fixes: 579de73b048a ("drm/i915: Exit cherryview_irq_handler() after one pass")
> Signed-off-by: Ville Syrjälä <ville.syrjala at linux.intel.com>
> ---
>  drivers/gpu/drm/i915/i915_irq.c | 36 +++++++++++++++++++++++++++++++++++-
>  1 file changed, 35 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_irq.c b/drivers/gpu/drm/i915/i915_irq.c
> index 626775039919..46be03c616f4 100644
> --- a/drivers/gpu/drm/i915/i915_irq.c
> +++ b/drivers/gpu/drm/i915/i915_irq.c
> @@ -1778,7 +1778,7 @@ static irqreturn_t valleyview_irq_handler(int irq, void *arg)
>  	disable_rpm_wakeref_asserts(dev_priv);
>  
>  	while (true) {
> -		/* Find, clear, then process each source of interrupt */
> +		u32 ier = 0;
>  
>  		gt_iir = I915_READ(GTIIR);
>  		pm_iir = I915_READ(GEN6_PMIIR);
> @@ -1789,7 +1789,22 @@ static irqreturn_t valleyview_irq_handler(int irq, void *arg)
>  
>  		ret = IRQ_HANDLED;
>  
> +		/*
> +		 * Theory on interrupt generation, based on empirical evidence:
> +		 *
> +		 * x = ((VLV_IIR & VLV_IER) ||
> +		 *      (((GT_IIR & GT_IER) || (GEN6_PMIIR & GEN6_PMIER)) &&
> +		 *       (VLV_MASTER_IER & MASTER_INTERRUPT_ENABLE)));
> +		 *
> +		 * A CPU interrupt will only be raised when 'x' has a 0->1 edge.
> +		 * Hence we clear MASTER_INTERRUPT_ENABLE and VLV_IER to
> +		 * guarantee the CPU interrupt will be raised again even if we
> +		 * don't end up clearing all the VLV_IIR, GT_IIR, GEN6_PMIIR
> +		 * bits this time around.

Following this logic, we want to enable MASTER_IER before VLV_IER such
that we get an immediate irq if there is a residual VLV_IIR.

> +		 */
>  		I915_WRITE(VLV_MASTER_IER, 0);
> +		ier = I915_READ(VLV_IER);

+wishlist: dev_priv->irq_enabled to save adding another mmio read.

> +		I915_WRITE(VLV_IER, 0);
>  
>  		if (gt_iir)
>  			I915_WRITE(GTIIR, gt_iir);
> @@ -1815,6 +1830,7 @@ static irqreturn_t valleyview_irq_handler(int irq, void *arg)
>  		if (iir)
>  			I915_WRITE(VLV_IIR, iir);
>  
> +		I915_WRITE(VLV_IER, ier);
>  		I915_WRITE(VLV_MASTER_IER, MASTER_INTERRUPT_ENABLE);
>  		POSTING_READ(VLV_MASTER_IER);
>  	}
> 

-- 
Chris Wilson, Intel Open Source Technology Centre