[Intel-gfx] [PATCH v2] drm/i915: Allow compaction upto SWIOTLB max segment size

[Konrad Rzeszutek Wilk]

On Mon, Oct 10, 2016 at 11:27:00PM +0100, Chris Wilson wrote:
> commit 1625e7e549c5 ("drm/i915: make compact dma scatter lists creation
> work with SWIOTLB backend") took a heavy handed approach to undo the
> scatterlist compaction in the face of SWIOTLB. (The compaction hit a bug
> whereby we tried to pass a segment larger than SWIOTLB could handle.) We
> can be a little more intelligent and try compacting the scatterlist up
> to the maximum SWIOTLB segment size (when using SWIOTLB).
> 

Won't this cause a bigger usage of the SWIOTLB bounce buffer ?

> v2: Tidy sg_mark_end() and cpp
> 
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> CC: Imre Deak <imre.deak at intel.com>
> CC: Daniel Vetter <daniel.vetter at ffwll.ch>
> Cc: Konrad Rzeszutek Wilk <konrad.wilk at oracle.com>
> Cc: Tvrtko Ursulin <tvrtko.ursulin at linux.intel.com>
> ---
>  drivers/gpu/drm/i915/i915_gem.c | 30 ++++++++++++++++++------------
>  1 file changed, 18 insertions(+), 12 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
> index dff8d05d80ee..50fd611926cb 100644
> --- a/drivers/gpu/drm/i915/i915_gem.c
> +++ b/drivers/gpu/drm/i915/i915_gem.c
> @@ -2201,6 +2201,15 @@ unlock:
>  	mutex_unlock(&obj->mm.lock);
>  }
>  
> +static unsigned long swiotlb_max_size(void)
> +{
> +#if IS_ENABLED(CONFIG_SWIOTLB)
> +	return swiotlb_nr_tbl() << IO_TLB_SHIFT;
> +#else
> +	return 0;
> +#endif
> +}
> +
>  static struct sg_table *
>  i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
>  {
> @@ -2212,6 +2221,7 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
>  	struct sgt_iter sgt_iter;
>  	struct page *page;
>  	unsigned long last_pfn = 0;	/* suppress gcc warning */
> +	unsigned long max_segment;
>  	int ret;
>  	gfp_t gfp;
>  
> @@ -2222,6 +2232,10 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
>  	GEM_BUG_ON(obj->base.read_domains & I915_GEM_GPU_DOMAINS);
>  	GEM_BUG_ON(obj->base.write_domain & I915_GEM_GPU_DOMAINS);
>  
> +	max_segment = swiotlb_max_size();
> +	if (!max_segment)
> +		max_segment = obj->base.size;
> +
>  	st = kmalloc(sizeof(*st), GFP_KERNEL);
>  	if (st == NULL)
>  		return ERR_PTR(-ENOMEM);
> @@ -2263,15 +2277,9 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
>  				goto err_pages;
>  			}
>  		}
> -#ifdef CONFIG_SWIOTLB
> -		if (swiotlb_nr_tbl()) {
> -			st->nents++;
> -			sg_set_page(sg, page, PAGE_SIZE, 0);
> -			sg = sg_next(sg);
> -			continue;
> -		}
> -#endif
> -		if (!i || page_to_pfn(page) != last_pfn + 1) {
> +		if (!i ||
> +		    sg->length >= max_segment ||
> +		    page_to_pfn(page) != last_pfn + 1) {
>  			if (i)
>  				sg = sg_next(sg);
>  			st->nents++;
> @@ -2284,9 +2292,7 @@ i915_gem_object_get_pages_gtt(struct drm_i915_gem_object *obj)
>  		/* Check that the i965g/gm workaround works. */
>  		WARN_ON((gfp & __GFP_DMA32) && (last_pfn >= 0x00100000UL));
>  	}
> -#ifdef CONFIG_SWIOTLB
> -	if (!swiotlb_nr_tbl())
> -#endif
> +	if (sg) /* loop terminated early; short sg table */
>  		sg_mark_end(sg);
>  
>  	ret = i915_gem_gtt_prepare_pages(obj, st);
> -- 
> 2.9.3
>