[Mesa-dev] [PATCH 2/2] radv/winsys: skip padded empty buffers in chain.

[Bas Nieuwenhuizen]

On Thu, Jan 11, 2018 at 5:56 AM, Dave Airlie <airlied at gmail.com> wrote:
> From: Dave Airlie <airlied at redhat.com>
>
> When chaining buffers if we see an empty one (just padded)
> skip over it.
> ---
>  src/amd/vulkan/winsys/amdgpu/radv_amdgpu_cs.c | 19 +++++++++++++++----
>  1 file changed, 15 insertions(+), 4 deletions(-)
>
> diff --git a/src/amd/vulkan/winsys/amdgpu/radv_amdgpu_cs.c b/src/amd/vulkan/winsys/amdgpu/radv_amdgpu_cs.c
> index 9a39d237ae8..9a7c4a4d779 100644
> --- a/src/amd/vulkan/winsys/amdgpu/radv_amdgpu_cs.c
> +++ b/src/amd/vulkan/winsys/amdgpu/radv_amdgpu_cs.c
> @@ -58,6 +58,7 @@ struct radv_amdgpu_cs {
>         unsigned                    *ib_size_ptr;
>         bool                        failed;
>         bool                        is_chained;
> +       bool                        empty_padded;
>
>         int                         buffer_hash_table[1024];
>         unsigned                    hw_ip;
> @@ -326,6 +327,8 @@ static bool radv_amdgpu_cs_finalize(struct radeon_winsys_cs *_cs)
>         struct radv_amdgpu_cs *cs = radv_amdgpu_cs(_cs);
>
>         if (cs->ws->use_ib_bos) {
> +
> +               cs->empty_padded = cs->base.cdw == 0;

Note that cdw can also be 0 here directly after a grow. You probably
need ot set empty_padded true on create and reset, and then back to
false on a grow or if cdw != 0 here.


>                 while (!cs->base.cdw || (cs->base.cdw & 7) != 0)
>                         cs->base.buf[cs->base.cdw++] = 0xffff1000;
>
> @@ -681,17 +684,25 @@ static int radv_amdgpu_winsys_cs_submit_chained(struct radeon_winsys_ctx *_ctx,
>         amdgpu_bo_list_handle bo_list;
>         struct amdgpu_cs_request request = {0};
>         struct amdgpu_cs_ib_info ibs[2];
> -
> +       int skip_count = 0;
>         for (unsigned i = cs_count; i--;) {
>                 struct radv_amdgpu_cs *cs = radv_amdgpu_cs(cs_array[i]);
> -
> +               int next_idx = 1;
>                 if (cs->is_chained) {
>                         *cs->ib_size_ptr -= 4;
>                         cs->is_chained = false;
>                 }
>
> -               if (i + 1 < cs_count) {
> -                       struct radv_amdgpu_cs *next = radv_amdgpu_cs(cs_array[i + 1]);
> +               if (cs->empty_padded)
> +                       skip_count++;
> +               else {
> +                       next_idx += skip_count;
> +                       skip_count = 0;
> +               }
> +
> +               if (i + next_idx < cs_count) {
> +                       struct radv_amdgpu_cs *next = radv_amdgpu_cs(cs_array[i + next_idx]);
> +
>                         assert(cs->base.cdw + 4 <= cs->base.max_dw);
>
>                         cs->is_chained = true;
> --
> 2.14.3
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev