[Mesa-dev] [PATCH 2/2] anv/blorp: consider multiview and view masks for tracking pending clear aspects

[Iago Toral]

Ignore this patch, there is more to this than I thought. I'll see if I
can find a proper solution and send a v2.

Iago

On Fri, 2018-01-05 at 10:28 +0100, Iago Toral Quiroga wrote:
> When multiview is active a subpass clear may only clear a subset of
> the
> attachment layers. Other subpasses in the same render pass may also
> clear too and we want to honor those clears. This means that when
> multiview
> is active we cannot mark an attachment as already cleared after a
> subpass
> clear operation, unless we know that all subpasses share the same
> view mask,
> in which case only one clear is really required.
> 
> Fixes work-in-progress CTS tests:
> dEQP-VK.multiview.readback_implicit_clear.*
> ---
>  src/intel/vulkan/anv_blorp.c | 15 ++++++++++++++-
>  1 file changed, 14 insertions(+), 1 deletion(-)
> 
> diff --git a/src/intel/vulkan/anv_blorp.c
> b/src/intel/vulkan/anv_blorp.c
> index 18fa4a4ae5..4575476240 100644
> --- a/src/intel/vulkan/anv_blorp.c
> +++ b/src/intel/vulkan/anv_blorp.c
> @@ -1249,7 +1249,20 @@ anv_cmd_buffer_clear_subpass(struct
> anv_cmd_buffer *cmd_buffer)
>           }
>        }
>  
> -      att_state->pending_clear_aspects = 0;
> +      if (!cmd_state->subpass->view_mask) {
> +         att_state->pending_clear_aspects = 0;
> +      } else {
> +         bool all_subpasses_use_same_view_mask = true;
> +         for (uint32_t i = 0; i < cmd_state->pass->subpass_count;
> i++) {
> +            if (cmd_state->pass->subpasses[i].view_mask !=
> +                cmd_state->subpass->view_mask) {
> +               all_subpasses_use_same_view_mask = false;
> +               break;
> +            }
> +         }
> +         if (all_subpasses_use_same_view_mask)
> +            att_state->pending_clear_aspects = 0;
> +      }
>     }
>  
>     const uint32_t ds = cmd_state->subpass-
> >depth_stencil_attachment.attachment;