[Mesa-dev] [PATCH v2 4/4] anv/pipeline: Bounds-check resource indices when robuts_buffer_access is enabled

[Michael Schellenberger Costa]

Hi Jason,

Am 19.05.2016 um 09:22 schrieb Jason Ekstrand:
> ---
>  src/intel/vulkan/anv_nir_apply_pipeline_layout.c | 52 ++++++++++++++++--------
>  1 file changed, 35 insertions(+), 17 deletions(-)
> 
> diff --git a/src/intel/vulkan/anv_nir_apply_pipeline_layout.c b/src/intel/vulkan/anv_nir_apply_pipeline_layout.c
> index 91f4322..7e66149 100644
> --- a/src/intel/vulkan/anv_nir_apply_pipeline_layout.c
> +++ b/src/intel/vulkan/anv_nir_apply_pipeline_layout.c
> @@ -29,6 +29,9 @@ struct apply_pipeline_layout_state {
>     nir_shader *shader;
>     nir_builder builder;
>  
> +   struct anv_pipeline_layout *layout;
> +   bool add_bounds_checks;
> +
>     struct {
>        BITSET_WORD *used;
>        uint8_t *surface_offsets;
> @@ -110,17 +113,15 @@ lower_res_index_intrinsic(nir_intrinsic_instr *intrin,
>     uint32_t binding = nir_intrinsic_binding(intrin);
>  
>     uint32_t surface_index = state->set[set].surface_offsets[binding];
> +   uint32_t array_size =
> +      state->layout->set[set].layout->binding[binding].array_size;
>  
> -   nir_const_value *const_block_idx =
> -      nir_src_as_const_value(intrin->src[0]);
> +   nir_ssa_def *block_index = nir_ssa_for_src(b, intrin->src[0], 1);
>  
> -   nir_ssa_def *block_index;
> -   if (const_block_idx) {
> -      block_index = nir_imm_int(b, surface_index + const_block_idx->u32[0]);
> -   } else {
> -      block_index = nir_iadd(b, nir_imm_int(b, surface_index),
> -                             nir_ssa_for_src(b, intrin->src[0], 1));
> -   }
> +   if (state->add_bounds_checks)
> +      block_index = nir_umax(b, block_index, nir_imm_int(b, array_size - 1));
> +
> +   block_index = nir_iadd(b, nir_imm_int(b, surface_index), block_index);
Here you do
|	if (state->add_bounds_checks)
|      	    block_index = nir_umax(...);
|	block_index = nir_iadd(...);

Below you do
|	nir_ssa_def *index = iadd(...);
|       if (state->add_bounds_checks)
|           index = nir_umax(...);

Are both functionally equivalent? Also why do you one time do

| nir_ssa_def *block_index = nir_ssa_for_src()

and the other time put that directly into nir_iadd()?

Could you unify the code so that both cases are equivalent?
--Michael


>  
>     assert(intrin->dest.is_ssa);
>     nir_ssa_def_rewrite_uses(&intrin->dest.ssa, nir_src_for_ssa(block_index));
> @@ -129,16 +130,24 @@ lower_res_index_intrinsic(nir_intrinsic_instr *intrin,
>  
>  static void
>  lower_tex_deref(nir_tex_instr *tex, nir_deref_var *deref,
> -                unsigned *const_index, nir_tex_src_type src_type,
> +                unsigned *const_index, unsigned array_size,
> +                nir_tex_src_type src_type,
>                  struct apply_pipeline_layout_state *state)
>  {
> +   nir_builder *b = &state->builder;
> +
>     if (deref->deref.child) {
>        assert(deref->deref.child->deref_type == nir_deref_type_array);
>        nir_deref_array *deref_array = nir_deref_as_array(deref->deref.child);
>  
> -      *const_index += deref_array->base_offset;
> -
>        if (deref_array->deref_array_type == nir_deref_array_type_indirect) {
> +         nir_ssa_def *index =
> +            nir_iadd(b, nir_imm_int(b, deref_array->base_offset),
> +                        nir_ssa_for_src(b, deref_array->indirect, 1));
> +
> +         if (state->add_bounds_checks)
> +            index = nir_umax(b, index, nir_imm_int(b, array_size - 1));
> +
>           nir_tex_src *new_srcs = rzalloc_array(tex, nir_tex_src,
>                                                 tex->num_srcs + 1);
>  
> @@ -154,10 +163,11 @@ lower_tex_deref(nir_tex_instr *tex, nir_deref_var *deref,
>            * first-class texture source.
>            */
>           tex->src[tex->num_srcs].src_type = src_type;
> +         nir_instr_rewrite_src(&tex->instr, &tex->src[tex->num_srcs],
> +                               nir_src_for_ssa(index));
>           tex->num_srcs++;
> -         assert(deref_array->indirect.is_ssa);
> -         nir_instr_rewrite_src(&tex->instr, &tex->src[tex->num_srcs - 1].src,
> -                               deref_array->indirect);
> +      } else {
> +         *const_index += MIN2(deref_array->base_offset, array_size - 1);
>        }
>     }
>  }
> @@ -182,17 +192,23 @@ lower_tex(nir_tex_instr *tex, struct apply_pipeline_layout_state *state)
>     /* No one should have come by and lowered it already */
>     assert(tex->texture);
>  
> +   state->builder.cursor = nir_before_instr(&tex->instr);
> +
>     unsigned set = tex->texture->var->data.descriptor_set;
>     unsigned binding = tex->texture->var->data.binding;
> +   unsigned array_size =
> +      state->layout->set[set].layout->binding[binding].array_size;
>     tex->texture_index = state->set[set].surface_offsets[binding];
> -   lower_tex_deref(tex, tex->texture, &tex->texture_index,
> +   lower_tex_deref(tex, tex->texture, &tex->texture_index, array_size,
>                     nir_tex_src_texture_offset, state);
>  
>     if (tex->sampler) {
>        unsigned set = tex->sampler->var->data.descriptor_set;
>        unsigned binding = tex->sampler->var->data.binding;
> +      unsigned array_size =
> +         state->layout->set[set].layout->binding[binding].array_size;
>        tex->sampler_index = state->set[set].sampler_offsets[binding];
> -      lower_tex_deref(tex, tex->sampler, &tex->sampler_index,
> +      lower_tex_deref(tex, tex->sampler, &tex->sampler_index, array_size,
>                        nir_tex_src_sampler_offset, state);
>     }
>  
> @@ -254,6 +270,8 @@ anv_nir_apply_pipeline_layout(struct anv_pipeline *pipeline,
>  
>     struct apply_pipeline_layout_state state = {
>        .shader = shader,
> +      .layout = layout,
> +      .add_bounds_checks = pipeline->device->robust_buffer_access,
>     };
>  
>     void *mem_ctx = ralloc_context(NULL);
>