[Intel-gfx] [PATCH] drm/i915: use a separate context for gpu relocs

[Chris Wilson]

Quoting Daniele Ceraolo Spurio (2019-08-24 01:20:22)
> @@ -943,6 +945,7 @@ static void reloc_gpu_flush(struct reloc_cache *cache)
>  static void reloc_cache_reset(struct reloc_cache *cache)
>  {
>         void *vaddr;
> +       struct intel_context *ce;
>  
>         if (cache->rq)
>                 reloc_gpu_flush(cache);
> @@ -973,6 +976,10 @@ static void reloc_cache_reset(struct reloc_cache *cache)
>                 }
>         }
>  
> +       ce = fetch_and_zero(&cache->ce);
> +       if (ce)
> +               intel_context_put(ce);

For peace of mind, this is too late. For pure gpu relocs, cache->vaddr is 0
and so we took the short-circuit at the beginning of the function.
-Chris