Re: [PATCH] drm/i915: dont call irq_put when irq test is on

["Ville Syrjälä" <ville.syrjala@linux.intel.com>]

On Mon, Dec 09, 2013 at 08:47:22PM +0200, Mika Kuoppala wrote:
> If test is running, irq_get was not called so we should
> gain balance by not doing irq_put
> 
> v2: use local variable so it can't change during test (Chris)
> 
> Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
> ---
>  drivers/gpu/drm/i915/i915_gem.c |    7 ++++---
>  1 file changed, 4 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index 92149bc..ab8fd3d 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -1015,6 +1015,7 @@ static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno,
>  			struct drm_i915_file_private *file_priv)
>  {
>  	drm_i915_private_t *dev_priv = ring->dev->dev_private;
> +	const bool irq_test_in_progress = dev_priv->gpu_error.test_irq_rings & intel_ring_flag(ring);

Should this use ACCESS_ONCE() to make sure the value doesn't get
reloaded from the source later?

>  	struct timespec before, now;
>  	DEFINE_WAIT(wait);
>  	long timeout_jiffies;
> @@ -1035,8 +1036,7 @@ static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno,
>  					 msecs_to_jiffies(100));
>  	}
>  
> -	if (!(dev_priv->gpu_error.test_irq_rings & intel_ring_flag(ring)) &&
> -	    WARN_ON(!ring->irq_get(ring)))
> +	if (!irq_test_in_progress && WARN_ON(!ring->irq_get(ring)))
>  		return -ENODEV;
>  
>  	/* Record current time in case interrupted by signal, or wedged */
> @@ -1095,7 +1095,8 @@ static int __wait_seqno(struct intel_ring_buffer *ring, u32 seqno,
>  	getrawmonotonic(&now);
>  	trace_i915_gem_request_wait_end(ring, seqno);
>  
> -	ring->irq_put(ring);
> +	if (!irq_test_in_progress)
> +		ring->irq_put(ring);
>  
>  	finish_wait(&ring->irq_queue, &wait);
>  
> -- 
> 1.7.9.5
> 
> _______________________________________________
> Intel-gfx mailing list
> Intel-gfx@lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/intel-gfx

-- 
Ville Syrjälä
Intel OTC