[Intel-gfx] [PATCH 1/6] drm/i915: Finish the wait-for-wedge by retiring all the inflight requests

[Mika Kuoppala]

Chris Wilson <chris at chris-wilson.co.uk> writes:

> Before we reset the GPU after marking the device as wedged, we wait for
> all the remaining requests to be completed (and marked as EIO).
> Afterwards, we should flush the request lists so the next batch start
> with the driver in an idle start.

s/start/state?

>
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Mika Kuoppala <mika.kuoppala at linux.intel.com>

Reviewed-by: Mika Kuoppala <mika.kuoppala at linux.intel.com>

> ---
>  drivers/gpu/drm/i915/i915_gem.c | 11 ++++++++---
>  1 file changed, 8 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index a5bd07338b46..30cd07ebcb8e 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -3273,7 +3273,8 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
>  	if (!test_bit(I915_WEDGED, &i915->gpu_error.flags))
>  		return true;
>  
> -	/* Before unwedging, make sure that all pending operations
> +	/*
> +	 * Before unwedging, make sure that all pending operations
>  	 * are flushed and errored out - we may have requests waiting upon
>  	 * third party fences. We marked all inflight requests as EIO, and
>  	 * every execbuf since returned EIO, for consistency we want all
> @@ -3291,7 +3292,8 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
>  			if (!rq)
>  				continue;
>  
> -			/* We can't use our normal waiter as we want to
> +			/*
> +			 * We can't use our normal waiter as we want to
>  			 * avoid recursively trying to handle the current
>  			 * reset. The basic dma_fence_default_wait() installs
>  			 * a callback for dma_fence_signal(), which is
> @@ -3306,8 +3308,11 @@ bool i915_gem_unset_wedged(struct drm_i915_private *i915)
>  				return false;
>  		}
>  	}
> +	i915_retire_requests(i915);
> +	GEM_BUG_ON(i915->gt.active_requests);
>  
> -	/* Undo nop_submit_request. We prevent all new i915 requests from
> +	/*
> +	 * Undo nop_submit_request. We prevent all new i915 requests from
>  	 * being queued (by disallowing execbuf whilst wedged) so having
>  	 * waited for all active requests above, we know the system is idle
>  	 * and do not have to worry about a thread being inside
> -- 
> 2.16.2