[1/2] drm/i915: Skip clflushes for all non-page backed objects

Commit Message

Chris Wilson Feb. 17, 2017, 2:07 p.m. UTC

Generalise the skip for physical and stolen objects by skipping anything
we do not have a valid address inside the sg.

Signed-off-by: Chris Wilson <[e-mail]>
---
 drivers/gpu/drm/i915/i915_gem.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

Comments

Matthew Auld Feb. 20, 2017, 6:28 p.m. UTC | #1

On 17 February 2017 at 14:07, Chris Wilson <[e-mail]> wrote:
> Generalise the skip for physical and stolen objects by skipping anything
> we do not have a valid address inside the sg.
>
> Signed-off-by: Chris Wilson <[e-mail]>
Reviewed-by: Matthew Auld <[e-mail]>

Patch

diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index 6745dcbf3799..96098a7e0bc3 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3162,14 +3162,19 @@  void i915_gem_clflush_object(struct drm_i915_gem_object *obj,
 	 * to GPU, and we can ignore the cache flush because it'll happen
 	 * again at bind time.
 	 */
-	if (!obj->mm.pages)
+	if (!obj->mm.pages) {
+		GEM_BUG_ON(obj->base.write_domain != I915_GEM_DOMAIN_CPU);
 		return;
+	}
 
 	/*
 	 * Stolen memory is always coherent with the GPU as it is explicitly
 	 * marked as wc by the system, or the system is cache-coherent.
+	 * Similarly, we only access struct pages through the CPU cache, so
+	 * anything not backed by physical memory we consider to be always
+	 * coherent and not need clflushing.
 	 */
-	if (obj->stolen || obj->phys_handle)
+	if (!i915_gem_object_has_struct_page(obj))
 		return;
 
 	/* If the GPU is snooping the contents of the CPU cache,