[Mesa-dev] [PATCH 3/3] i965: Don't use nir_assign_var_locations for VS/TES/GS outputs.

Timothy Arceri timothy.arceri at collabora.com
Mon Oct 24 10:31:36 UTC 2016


On Sun, 2016-10-23 at 23:44 -0700, Kenneth Graunke wrote:
> Signed-off-by: Kenneth Graunke <kenneth at whitecape.org>
> ---
>  src/mesa/drivers/dri/i965/brw_fs.cpp     | 13 ---------
>  src/mesa/drivers/dri/i965/brw_fs_nir.cpp | 46 ++++++--------------
> ------------
>  src/mesa/drivers/dri/i965/brw_nir.c      | 13 +++------
>  src/mesa/drivers/dri/i965/brw_nir.h      |  1 -
>  4 files changed, 12 insertions(+), 61 deletions(-)
> 
> diff --git a/src/mesa/drivers/dri/i965/brw_fs.cpp
> b/src/mesa/drivers/dri/i965/brw_fs.cpp
> index 31d21ec..f4e7122 100644
> --- a/src/mesa/drivers/dri/i965/brw_fs.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_fs.cpp
> @@ -492,19 +492,6 @@ type_size_scalar(const struct glsl_type *type)
>     return 0;
>  }
>  
> -/**
> - * Returns the number of scalar components needed to store type,
> assuming
> - * that vectors are padded out to vec4.
> - *
> - * This has the packing rules of type_size_vec4(), but counts
> components
> - * similar to type_size_scalar().
> - */
> -extern "C" int
> -type_size_vec4_times_4(const struct glsl_type *type)
> -{
> -   return 4 * type_size_vec4(type);
> -}
> -
>  /* Attribute arrays are loaded as one vec4 per element (or matrix
> column),
>   * except for double-precision types, which are loaded as one dvec4.
>   */
> diff --git a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> index 9cad1a6..4baadc9 100644
> --- a/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> +++ b/src/mesa/drivers/dri/i965/brw_fs_nir.cpp
> @@ -49,47 +49,18 @@ fs_visitor::emit_nir_code()
>  }
>  
>  void
> -fs_visitor::nir_setup_single_output_varying(fs_reg *reg,
> -                                            const glsl_type *type,
> -                                            unsigned *location)
> -{
> -   if (type->is_array() || type->is_matrix()) {
> -      const struct glsl_type *elem_type =
> glsl_get_array_element(type);
> -      const unsigned length = glsl_get_length(type);
> -
> -      for (unsigned i = 0; i < length; i++) {
> -         nir_setup_single_output_varying(reg, elem_type, location);
> -      }
> -   } else if (type->is_record()) {
> -      for (unsigned i = 0; i < type->length; i++) {
> -         const struct glsl_type *field_type = type-
> >fields.structure[i].type;
> -         nir_setup_single_output_varying(reg, field_type, location);
> -      }
> -   } else {
> -      assert(type->is_scalar() || type->is_vector());
> -      unsigned num_iter = 1;
> -      if (type->is_dual_slot())
> -         num_iter = 2;
> -      for (unsigned count = 0; count < num_iter; count++) {
> -         this->outputs[*location] = *reg;
> -         *reg = offset(*reg, bld, 4);
> -         (*location)++;
> -      }
> -   }
> -}
> -
> -void
>  fs_visitor::nir_setup_outputs()
>  {
>     if (stage == MESA_SHADER_TESS_CTRL || stage ==
> MESA_SHADER_FRAGMENT)
>        return;
>  
> -   nir_outputs = bld.vgrf(BRW_REGISTER_TYPE_F, nir->num_outputs);

We can remove nir_outputs from brw_fs.h after these changes. It looks
like we can even remove num_outputs from nir_shader.

While you are at it a follow-up patch to remove nir_inputs would have
my r-b also.

Right I think I follow the series ... since we changed things a little
while back to just pad out outputs this just cleans up the left over
bits an pieces.

With the mentioned fields removed this series is:

Reviewed-by: Timothy Arceri <timothy.arceri at collabora.com> 

You probably want to Cc stable for 13.0 as this fixes some locations
issues for ARB_enhanced_layouts 

> -
>     nir_foreach_variable(var, &nir->outputs) {
> -      fs_reg reg = offset(nir_outputs, bld, var-
> >data.driver_location);
> -      unsigned location = var->data.location;
> -      nir_setup_single_output_varying(&reg, var->type, &location);
> +      const unsigned vec4s = type_size_vec4(var->type);
> +      fs_reg reg = bld.vgrf(BRW_REGISTER_TYPE_F, 4 * vec4s);
> +      for (unsigned i = 0; i < vec4s; i++) {
> +         if (outputs[var->data.driver_location + i].file ==
> BAD_FILE)
> +            outputs[var->data.driver_location + i] = offset(reg,
> bld, 4 * i);
> +      }
>     }
>  }
>  
> @@ -4242,12 +4213,11 @@ fs_visitor::nir_emit_intrinsic(const
> fs_builder &bld, nir_intrinsic_instr *instr
>  
>     case nir_intrinsic_store_output: {
>        fs_reg src = get_nir_src(instr->src[0]);
> -      fs_reg new_dest = offset(retype(nir_outputs, src.type), bld,
> -                               instr->const_index[0]);
>  
>        nir_const_value *const_offset = nir_src_as_const_value(instr-
> >src[1]);
>        assert(const_offset && "Indirect output stores not allowed");
> -      new_dest = offset(new_dest, bld, const_offset->u32[0]);
> +      fs_reg new_dest = retype(offset(outputs[instr-
> >const_index[0]], bld,
> +                                      4 * const_offset->u32[0]),
> src.type);
>  
>        unsigned num_components = instr->num_components;
>        unsigned first_component = nir_intrinsic_component(instr);
> diff --git a/src/mesa/drivers/dri/i965/brw_nir.c
> b/src/mesa/drivers/dri/i965/brw_nir.c
> index a935f42..de0e235 100644
> --- a/src/mesa/drivers/dri/i965/brw_nir.c
> +++ b/src/mesa/drivers/dri/i965/brw_nir.c
> @@ -327,16 +327,11 @@ void
>  brw_nir_lower_vue_outputs(nir_shader *nir,
>                            bool is_scalar)
>  {
> -   if (is_scalar) {
> -      nir_assign_var_locations(&nir->outputs, &nir->num_outputs,
> -                               VARYING_SLOT_VAR0,
> -                               type_size_vec4_times_4);
> -      nir_lower_io(nir, nir_var_shader_out, type_size_vec4_times_4,
> 0);
> -   } else {
> -      nir_foreach_variable(var, &nir->outputs)
> -         var->data.driver_location = var->data.location;
> -      nir_lower_io(nir, nir_var_shader_out, type_size_vec4, 0);
> +   nir_foreach_variable(var, &nir->outputs) {
> +      var->data.driver_location = var->data.location;
>     }
> +
> +   nir_lower_io(nir, nir_var_shader_out, type_size_vec4, 0);
>  }
>  
>  void
> diff --git a/src/mesa/drivers/dri/i965/brw_nir.h
> b/src/mesa/drivers/dri/i965/brw_nir.h
> index aef5c53..645a818 100644
> --- a/src/mesa/drivers/dri/i965/brw_nir.h
> +++ b/src/mesa/drivers/dri/i965/brw_nir.h
> @@ -34,7 +34,6 @@ extern "C" {
>  int type_size_scalar(const struct glsl_type *type);
>  int type_size_vec4(const struct glsl_type *type);
>  int type_size_dvec4(const struct glsl_type *type);
> -int type_size_vec4_times_4(const struct glsl_type *type);
>  int type_size_vs_input(const struct glsl_type *type);
>  
>  static inline int


More information about the mesa-dev mailing list