On Fri 16 Jun 2017, Jason Ekstrand wrote:
> Any form of CCS on gen9+ only works on Y-tiled images. The only caller
> of create_for_bo which uses Y-tiled BOs is create_for_dri_image.
If I understand ARC++ correctly, then intel_update_image_buffer() also
calls intel_miptree_create_for_bo() for Android Y-tiled winsys buffers.
(I've confirmed it with code inspection, but not with actual debug
logging). That should be noted in the commit message.
This patch shouldn't degrade ARC++ performance, though, because ARC++ is
still using an old Mesa that never allocated CCS for Android winsys
buffers.
At the end of the patch series, will Android's Y-tiled winsys buffers
get the benefit of a private CCS?
> + /* Since CCS_E can compress more than just clear color, we create the
> + * CCS for it up-front. For CCS_D which only compresses clears, we
> + * create the CCS on-demand when a clear occurs that wants one.
> + */
> + if (mt->aux_usage == ISL_AUX_USAGE_CCS_E) {
> + if (!intel_miptree_alloc_ccs(brw, mt)) {
> + intel_miptree_release(&mt);
> + return NULL;
> + }
> + }
> +
The above hunk is a duplicate. The same 'if' tree appears immediately
above it.
With the hunk de-duplicated, this patch is
Reviewed-by: Chad Versace <chadvers...@chromium.org>
> return mt;
> }
>
> --
> 2.5.0.400.gff86faf
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
