[Mesa-dev] [PATCH 25/46] glsl: lower gl_TessLevel* from float[n] to vecn.
Kenneth Graunke
kenneth at whitecape.org
Fri Jun 19 15:08:19 PDT 2015
On Wednesday, June 17, 2015 01:01:21 AM Marek Olšák wrote:
> From: Fabian Bieler <fabianbieler at fastmail.fm>
>
> Similar to gl_ClipDistance -> gl_ClipDistanceMESA
> ---
> src/glsl/Makefile.sources | 1 +
> src/glsl/ir_optimization.h | 1 +
> src/glsl/link_varyings.cpp | 51 +++-
> src/glsl/link_varyings.h | 13 +-
> src/glsl/linker.cpp | 4 +
> src/glsl/lower_tess_level.cpp | 462 ++++++++++++++++++++++++++++++++
> src/mesa/drivers/dri/i965/brw_context.c | 1 +
> src/mesa/main/mtypes.h | 1 +
> 8 files changed, 521 insertions(+), 13 deletions(-)
> create mode 100644 src/glsl/lower_tess_level.cpp
>
> diff --git a/src/glsl/Makefile.sources b/src/glsl/Makefile.sources
> index d784a81..b3b84d6 100644
> --- a/src/glsl/Makefile.sources
> +++ b/src/glsl/Makefile.sources
> @@ -154,6 +154,7 @@ LIBGLSL_FILES = \
> lower_packed_varyings.cpp \
> lower_named_interface_blocks.cpp \
> lower_packing_builtins.cpp \
> + lower_tess_level.cpp \
> lower_texture_projection.cpp \
> lower_variable_index_to_cond_assign.cpp \
> lower_vec_index_to_cond_assign.cpp \
> diff --git a/src/glsl/ir_optimization.h b/src/glsl/ir_optimization.h
> index e6939f3..688a5e1 100644
> --- a/src/glsl/ir_optimization.h
> +++ b/src/glsl/ir_optimization.h
> @@ -132,6 +132,7 @@ bool optimize_split_arrays(exec_list *instructions, bool linked);
> bool lower_offset_arrays(exec_list *instructions);
> void optimize_dead_builtin_variables(exec_list *instructions,
> enum ir_variable_mode other);
> +bool lower_tess_level(gl_shader *shader);
>
> bool lower_vertex_id(gl_shader *shader);
>
> diff --git a/src/glsl/link_varyings.cpp b/src/glsl/link_varyings.cpp
> index 373d337..d439b62 100644
> --- a/src/glsl/link_varyings.cpp
> +++ b/src/glsl/link_varyings.cpp
> @@ -318,7 +318,7 @@ tfeedback_decl::init(struct gl_context *ctx, const void *mem_ctx,
>
> this->location = -1;
> this->orig_name = input;
> - this->is_clip_distance_mesa = false;
> + this->is_mesa_var = none;
This seems like a bad name..."is_foo" suggests a boolean (rather than an
enum), and "mesa_var" is pretty generic. Perhaps calling it
"lowered_builtin_array_variable" would be better?
> this->skip_components = 0;
> this->next_buffer_separator = false;
> this->matched_candidate = NULL;
> @@ -367,8 +367,15 @@ tfeedback_decl::init(struct gl_context *ctx, const void *mem_ctx,
> */
> if (ctx->Const.ShaderCompilerOptions[MESA_SHADER_VERTEX].LowerClipDistance &&
> strcmp(this->var_name, "gl_ClipDistance") == 0) {
> - this->is_clip_distance_mesa = true;
> + this->is_mesa_var = clip_distance;
> }
> +
> + if (ctx->Const.ShaderCompilerOptions[MESA_SHADER_VERTEX].LowerTessLevel &&
> + (strcmp(this->var_name, "gl_TessLevelOuter") == 0))
> + this->is_mesa_var = tess_level_outer;
> + if (ctx->Const.ShaderCompilerOptions[MESA_SHADER_VERTEX].LowerTessLevel &&
> + (strcmp(this->var_name, "gl_TessLevelInner") == 0))
> + this->is_mesa_var = tess_level_inner;
Using ShaderCompilerOptions[MESA_SHADER_VERTEX] for LowerTessLevel seems
odd - perhaps LowerClipDistance and LowerTessLevel really shouldn't be
per-stage options, and just live in ctx->Const? *shrug*
> }
>
>
> @@ -415,9 +422,22 @@ tfeedback_decl::assign_location(struct gl_context *ctx,
> this->matched_candidate->type->fields.array->matrix_columns;
> const unsigned vector_elements =
> this->matched_candidate->type->fields.array->vector_elements;
> - unsigned actual_array_size = this->is_clip_distance_mesa ?
> - prog->LastClipDistanceArraySize :
> - this->matched_candidate->type->array_size();
> + unsigned actual_array_size;
> + switch (this->is_mesa_var) {
> + case clip_distance:
> + actual_array_size = prog->LastClipDistanceArraySize;
> + break;
> + case tess_level_outer:
> + actual_array_size = 4;
> + break;
> + case tess_level_inner:
> + actual_array_size = 2;
> + break;
> + case none:
> + default:
> + actual_array_size = this->matched_candidate->type->array_size();
> + break;
> + }
>
> if (this->is_subscripted) {
> /* Check array bounds. */
> @@ -428,7 +448,7 @@ tfeedback_decl::assign_location(struct gl_context *ctx,
> actual_array_size);
> return false;
> }
> - unsigned array_elem_size = this->is_clip_distance_mesa ?
> + unsigned array_elem_size = this->is_mesa_var ?
> 1 : vector_elements * matrix_cols;
> fine_location += array_elem_size * this->array_subscript;
> this->size = 1;
> @@ -437,7 +457,7 @@ tfeedback_decl::assign_location(struct gl_context *ctx,
> }
> this->vector_elements = vector_elements;
> this->matrix_columns = matrix_cols;
> - if (this->is_clip_distance_mesa)
> + if (this->is_mesa_var)
> this->type = GL_FLOAT;
> else
> this->type = this->matched_candidate->type->fields.array->gl_type;
> @@ -560,8 +580,21 @@ const tfeedback_candidate *
> tfeedback_decl::find_candidate(gl_shader_program *prog,
> hash_table *tfeedback_candidates)
> {
> - const char *name = this->is_clip_distance_mesa
> - ? "gl_ClipDistanceMESA" : this->var_name;
> + const char *name = this->var_name;
> + switch (this->is_mesa_var) {
> + case none:
> + name = this->var_name;
> + break;
> + case clip_distance:
> + name = "gl_ClipDistanceMESA";
> + break;
> + case tess_level_outer:
> + name = "gl_TessLevelOuterMESA";
> + break;
> + case tess_level_inner:
> + name = "gl_TessLevelInnerMESA";
> + break;
> + }
> this->matched_candidate = (const tfeedback_candidate *)
> hash_table_find(tfeedback_candidates, name);
> if (!this->matched_candidate) {
> diff --git a/src/glsl/link_varyings.h b/src/glsl/link_varyings.h
> index afc16a8..e0d57ce 100644
> --- a/src/glsl/link_varyings.h
> +++ b/src/glsl/link_varyings.h
> @@ -128,7 +128,7 @@ public:
> */
> unsigned num_components() const
> {
> - if (this->is_clip_distance_mesa)
> + if (this->is_mesa_var)
> return this->size;
> else
> return this->vector_elements * this->matrix_columns * this->size;
> @@ -161,10 +161,15 @@ private:
> unsigned array_subscript;
>
> /**
> - * True if the variable is gl_ClipDistance and the driver lowers
> - * gl_ClipDistance to gl_ClipDistanceMESA.
> + * Non-zero if the variable is gl_ClipDistance, glTessLevelOuter or
> + * gl_TessLevelInner and the driver lowers it to gl_*MESA.
> */
> - bool is_clip_distance_mesa;
> + enum {
> + none,
> + clip_distance,
> + tess_level_outer,
> + tess_level_inner,
> + } is_mesa_var;
>
> /**
> * The vertex shader output location that the linker assigned for this
> diff --git a/src/glsl/linker.cpp b/src/glsl/linker.cpp
> index 49a88b9..acf6f81 100644
> --- a/src/glsl/linker.cpp
> +++ b/src/glsl/linker.cpp
> @@ -3224,6 +3224,10 @@ link_shaders(struct gl_context *ctx, struct gl_shader_program *prog)
> lower_clip_distance(prog->_LinkedShaders[i]);
> }
>
> + if (ctx->Const.ShaderCompilerOptions[i].LowerTessLevel) {
> + lower_tess_level(prog->_LinkedShaders[i]);
> + }
> +
> while (do_common_optimization(prog->_LinkedShaders[i]->ir, true, false,
> &ctx->Const.ShaderCompilerOptions[i],
> ctx->Const.NativeIntegers))
> diff --git a/src/glsl/lower_tess_level.cpp b/src/glsl/lower_tess_level.cpp
> new file mode 100644
> index 0000000..47ff93e
> --- /dev/null
> +++ b/src/glsl/lower_tess_level.cpp
> @@ -0,0 +1,462 @@
> +/*
> + * Copyright © 2014 Intel Corporation
> + *
> + * Permission is hereby granted, free of charge, to any person obtaining a
> + * copy of this software and associated documentation files (the "Software"),
> + * to deal in the Software without restriction, including without limitation
> + * the rights to use, copy, modify, merge, publish, distribute, sublicense,
> + * and/or sell copies of the Software, and to permit persons to whom the
> +
> + *
> + * The above copyright notice and this permission notice (including the next
> + * paragraph) shall be included in all copies or substantial portions of the
> + * Software.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
> + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
> + * DEALINGS IN THE SOFTWARE.
> + */
> +
> +/**
> + * \file lower_tess_level.cpp
> + *
> + * This pass accounts for the difference between the way gl_TessLevelOuter
> + * and gl_TessLevelInner is declared in standard GLSL (as an array of
> + * floats), and the way it is frequently implemented in hardware (as a vec4
> + * and vec2).
> + *
> + * The declaration of gl_TessLevel* is replaced with a declaration
> + * of gl_TessLevel*MESA, and any references to gl_TessLevel* are
> + * translated to refer to gl_TessLevel*MESA with the appropriate
> + * swizzling of array indices. For instance:
> + *
> + * gl_TessLevelOuter[i]
> + *
> + * is translated into:
> + *
> + * gl_TessLevelOuterMESA[i]
> + *
> + * Since some hardware may not internally represent gl_TessLevel* as a pair
> + * of vec4's, this lowering pass is optional. To enable it, set the
> + * LowerTessLevel flag in gl_shader_compiler_options to true.
> + */
> +
> +#include "glsl_symbol_table.h"
> +#include "ir_rvalue_visitor.h"
> +#include "ir.h"
> +#include "program/prog_instruction.h" /* For WRITEMASK_* */
> +
> +namespace {
> +
> +class lower_tess_level_visitor : public ir_rvalue_visitor {
> +public:
> + explicit lower_tess_level_visitor(gl_shader_stage shader_stage)
> + : progress(false), old_tess_level_outer_var(NULL),
> + old_tess_level_inner_var(NULL), new_tess_level_outer_var(NULL),
> + new_tess_level_inner_var(NULL), shader_stage(shader_stage)
> + {
> + }
> +
> + virtual ir_visitor_status visit(ir_variable *);
> + bool is_tess_level_array(ir_rvalue *ir);
> + ir_rvalue *lower_tess_level_array(ir_rvalue *ir);
> + virtual ir_visitor_status visit_leave(ir_assignment *);
> + void visit_new_assignment(ir_assignment *ir);
> + virtual ir_visitor_status visit_leave(ir_call *);
> +
> + virtual void handle_rvalue(ir_rvalue **rvalue);
> +
> + void fix_lhs(ir_assignment *);
> +
> + bool progress;
> +
> + /**
> + * Pointer to the declaration of gl_TessLevel*, if found.
> + */
> + ir_variable *old_tess_level_outer_var;
> + ir_variable *old_tess_level_inner_var;
> +
> + /**
> + * Pointer to the newly-created gl_TessLevel*MESA variables.
> + */
> + ir_variable *new_tess_level_outer_var;
> + ir_variable *new_tess_level_inner_var;
> +
> + /**
> + * Type of shader we are compiling (e.g. MESA_SHADER_TESS_CTRL)
> + */
> + const gl_shader_stage shader_stage;
> +};
> +
> +} /* anonymous namespace */
> +
> +/**
> + * Replace any declaration of gl_TessLevel* as an array of floats with a
> + * declaration of gl_TessLevel*MESA as a vec4.
> + */
> +ir_visitor_status
> +lower_tess_level_visitor::visit(ir_variable *ir)
> +{
> + if ((!ir->name) ||
> + ((strcmp(ir->name, "gl_TessLevelInner") != 0) &&
> + (strcmp(ir->name, "gl_TessLevelOuter") != 0)))
> + return visit_continue;
> +
> + assert (ir->type->is_array());
> +
> + if (strcmp(ir->name, "gl_TessLevelOuter") == 0) {
> + if (this->old_tess_level_outer_var)
> + return visit_continue;
> +
> + old_tess_level_outer_var = ir;
> + assert(ir->type->fields.array == glsl_type::float_type);
> +
> + /* Clone the old var so that we inherit all of its properties */
> + new_tess_level_outer_var = ir->clone(ralloc_parent(ir), NULL);
> +
> + /* And change the properties that we need to change */
> + new_tess_level_outer_var->name = ralloc_strdup(new_tess_level_outer_var,
> + "gl_TessLevelOuterMESA");
> + new_tess_level_outer_var->type = glsl_type::vec4_type;
> + new_tess_level_outer_var->data.max_array_access = 0;
> +
> + ir->replace_with(new_tess_level_outer_var);
> + } else if (strcmp(ir->name, "gl_TessLevelInner") == 0) {
> + if (this->old_tess_level_inner_var)
> + return visit_continue;
> +
> + old_tess_level_inner_var = ir;
> + assert(ir->type->fields.array == glsl_type::float_type);
> +
> + /* Clone the old var so that we inherit all of its properties */
> + new_tess_level_inner_var = ir->clone(ralloc_parent(ir), NULL);
> +
> + /* And change the properties that we need to change */
> + new_tess_level_inner_var->name = ralloc_strdup(new_tess_level_inner_var,
> + "gl_TessLevelInnerMESA");
> + new_tess_level_inner_var->type = glsl_type::vec2_type;
> + new_tess_level_inner_var->data.max_array_access = 0;
> +
> + ir->replace_with(new_tess_level_inner_var);
> + } else {
> + assert(0);
> + }
> +
> + this->progress = true;
> +
> + return visit_continue;
> +}
> +
> +
> +/**
> + * Determine whether the given rvalue describes an array of floats that
> + * needs to be lowered to a vec4; that is, determine whether it
> + * matches one of the following patterns:
> + *
> + * - gl_TessLevelOuter
> + * - gl_TessLevelInner
> + */
> +bool
> +lower_tess_level_visitor::is_tess_level_array(ir_rvalue *ir)
> +{
> + if (!ir->type->is_array())
> + return false;
> + if (ir->type->fields.array != glsl_type::float_type)
> + return false;
> +
> + if (this->old_tess_level_outer_var) {
> + if (ir->variable_referenced() == this->old_tess_level_outer_var)
> + return true;
> + }
> + if (this->old_tess_level_inner_var) {
> + if (ir->variable_referenced() == this->old_tess_level_inner_var)
> + return true;
> + }
> + return false;
> +}
> +
> +
> +/**
> + * If the given ir satisfies is_tess_level_array(), return new ir
> + * representing its lowered equivalent. That is, map:
> + *
> + * - gl_TessLevelOuter => gl_TessLevelOuterMESA
> + * - gl_TessLevelInner => gl_TessLevelInnerMESA
> + *
> + * Otherwise return NULL.
> + */
> +ir_rvalue *
> +lower_tess_level_visitor::lower_tess_level_array(ir_rvalue *ir)
> +{
> + if (!ir->type->is_array())
> + return NULL;
> + if (ir->type->fields.array != glsl_type::float_type)
> + return NULL;
> +
> + ir_variable **new_var = NULL;
> +
> + if (this->old_tess_level_outer_var) {
> + if (ir->variable_referenced() == this->old_tess_level_outer_var)
> + new_var = &this->new_tess_level_outer_var;
> + }
> + if (this->old_tess_level_inner_var) {
> + if (ir->variable_referenced() == this->old_tess_level_inner_var)
> + new_var = &this->new_tess_level_inner_var;
> + }
> +
> + if (new_var == NULL)
> + return NULL;
> +
> + assert(ir->as_dereference_variable());
> + return new(ralloc_parent(ir)) ir_dereference_variable(*new_var);
> +}
> +
> +
> +void
> +lower_tess_level_visitor::handle_rvalue(ir_rvalue **rv)
> +{
> + if (*rv == NULL)
> + return;
> +
> + ir_dereference_array *const array_deref = (*rv)->as_dereference_array();
> + if (array_deref == NULL)
> + return;
> +
> + /* Replace any expression that indexes one of the floats in gl_TessLevel*
> + * with an expression that indexes into one of the vec4's
> + * gl_TessLevel*MESA and accesses the appropriate component.
> + */
> + ir_rvalue *lowered_vec4 =
> + this->lower_tess_level_array(array_deref->array);
> + if (lowered_vec4 != NULL) {
> + this->progress = true;
> + void *mem_ctx = ralloc_parent(array_deref);
> +
> + ir_expression *const expr =
> + new(mem_ctx) ir_expression(ir_binop_vector_extract,
> + lowered_vec4,
> + array_deref->array_index);
> +
> + *rv = expr;
> + }
> +}
> +
> +void
> +lower_tess_level_visitor::fix_lhs(ir_assignment *ir)
> +{
> + if (ir->lhs->ir_type != ir_type_expression)
> + return;
> + void *mem_ctx = ralloc_parent(ir);
> + ir_expression *const expr = (ir_expression *) ir->lhs;
> +
> + /* The expression must be of the form:
> + *
> + * (vector_extract gl_TessLevel*MESA, j).
> + */
> + assert(expr->operation == ir_binop_vector_extract);
> + assert(expr->operands[0]->ir_type == ir_type_dereference_variable);
> + assert((expr->operands[0]->type == glsl_type::vec4_type) ||
> + (expr->operands[0]->type == glsl_type::vec2_type));
> +
> + ir_dereference *const new_lhs = (ir_dereference *) expr->operands[0];
> +
> + ir_constant *old_index_constant = expr->operands[1]->constant_expression_value();
> + if (!old_index_constant) {
> + ir->rhs = new(mem_ctx) ir_expression(ir_triop_vector_insert,
> + glsl_type::vec4_type,
This seems wrong, shouldn't it be vec2 sometimes? i.e. I think we want
expr->operands[0]->type here...
> + new_lhs->clone(mem_ctx, NULL),
> + ir->rhs,
> + expr->operands[1]);
> + }
> + ir->set_lhs(new_lhs);
> +
> + if (old_index_constant) {
> + /* gl_TessLevel* is being accessed via a constant index. Don't bother
> + * creating a vector insert op. Just use a write mask.
> + */
> + ir->write_mask = 1 << old_index_constant->get_int_component(0);
> + } else {
> + if(expr->operands[0]->type == glsl_type::vec4_type)
> + ir->write_mask = WRITEMASK_XYZW;
> + else
> + ir->write_mask = WRITEMASK_XY;
This could be shortened to:
ir->writemask = (1 << expr->operands[0]->type->vector_elements) - 1;
> + }
> +}
> +
> +/**
> + * Replace any assignment having a gl_TessLevel* (undereferenced) as
> + * its LHS or RHS with a sequence of assignments, one for each component of
> + * the array. Each of these assignments is lowered to refer to
> + * gl_TessLevel*MESA as appropriate.
> + */
> +ir_visitor_status
> +lower_tess_level_visitor::visit_leave(ir_assignment *ir)
> +{
> + /* First invoke the base class visitor. This causes handle_rvalue() to be
> + * called on ir->rhs and ir->condition.
> + */
> + ir_rvalue_visitor::visit_leave(ir);
> +
> + if (this->is_tess_level_array(ir->lhs) ||
> + this->is_tess_level_array(ir->rhs)) {
> + /* LHS or RHS of the assignment is the entire gl_TessLevel* array.
> + * Since we are
> + * reshaping gl_TessLevel* from an array of floats to a
> + * vec4, this isn't going to work as a bulk assignment anymore, so
> + * unroll it to element-by-element assignments and lower each of them.
> + *
> + * Note: to unroll into element-by-element assignments, we need to make
> + * clones of the LHS and RHS. This is safe because expressions and
> + * l-values are side-effect free.
> + */
> + void *ctx = ralloc_parent(ir);
> + int array_size = ir->lhs->type->array_size();
> + for (int i = 0; i < array_size; ++i) {
> + ir_dereference_array *new_lhs = new(ctx) ir_dereference_array(
> + ir->lhs->clone(ctx, NULL), new(ctx) ir_constant(i));
> + ir_dereference_array *new_rhs = new(ctx) ir_dereference_array(
> + ir->rhs->clone(ctx, NULL), new(ctx) ir_constant(i));
> + this->handle_rvalue((ir_rvalue **) &new_rhs);
> +
> + /* Handle the LHS after creating the new assignment. This must
> + * happen in this order because handle_rvalue may replace the old LHS
> + * with an ir_expression of ir_binop_vector_extract. Since this is
> + * not a valide l-value, this will cause an assertion in the
> + * ir_assignment constructor to fail.
> + *
> + * If this occurs, replace the mangled LHS with a dereference of the
> + * vector, and replace the RHS with an ir_triop_vector_insert.
> + */
> + ir_assignment *const assign = new(ctx) ir_assignment(new_lhs, new_rhs);
> + this->handle_rvalue((ir_rvalue **) &assign->lhs);
> + this->fix_lhs(assign);
> +
> + this->base_ir->insert_before(assign);
> + }
> + ir->remove();
> +
> + return visit_continue;
> + }
> +
> + /* Handle the LHS as if it were an r-value. Normally
> + * rvalue_visit(ir_assignment *) only visits the RHS, but we need to lower
> + * expressions in the LHS as well.
> + *
> + * This may cause the LHS to get replaced with an ir_expression of
> + * ir_binop_vector_extract. If this occurs, replace it with a dereference
> + * of the vector, and replace the RHS with an ir_triop_vector_insert.
> + */
> + handle_rvalue((ir_rvalue **)&ir->lhs);
> + this->fix_lhs(ir);
> +
> + return rvalue_visit(ir);
> +}
> +
> +
> +/**
> + * Set up base_ir properly and call visit_leave() on a newly created
> + * ir_assignment node. This is used in cases where we have to insert an
> + * ir_assignment in a place where we know the hierarchical visitor won't see
> + * it.
> + */
> +void
> +lower_tess_level_visitor::visit_new_assignment(ir_assignment *ir)
> +{
> + ir_instruction *old_base_ir = this->base_ir;
> + this->base_ir = ir;
> + ir->accept(this);
> + this->base_ir = old_base_ir;
> +}
> +
> +
> +/**
> + * If a gl_TessLevel* variable appears as an argument in an ir_call
> + * expression, replace it with a temporary variable, and make sure the ir_call
> + * is preceded and/or followed by assignments that copy the contents of the
> + * temporary variable to and/or from gl_TessLevel*. Each of these
> + * assignments is then lowered to refer to gl_TessLevel*MESA.
> + */
> +ir_visitor_status
> +lower_tess_level_visitor::visit_leave(ir_call *ir)
> +{
It seems odd that we're handling calls - though the old clip distance
code seems to do that too. This is post-linking, so calls should have
been inlined. We use ir_call for a couple of "intrinsics" (for atomic
counters and image load/store) that never get inlined, but those should
never match this signature.
i.e. I wonder if we can just delete all of this code.
> + void *ctx = ralloc_parent(ir);
> +
> + const exec_node *formal_param_node = ir->callee->parameters.head;
> + const exec_node *actual_param_node = ir->actual_parameters.head;
> + while (!actual_param_node->is_tail_sentinel()) {
> + ir_variable *formal_param = (ir_variable *) formal_param_node;
> + ir_rvalue *actual_param = (ir_rvalue *) actual_param_node;
> +
> + /* Advance formal_param_node and actual_param_node now so that we can
> + * safely replace actual_param with another node, if necessary, below.
> + */
> + formal_param_node = formal_param_node->next;
> + actual_param_node = actual_param_node->next;
> +
> + if (!this->is_tess_level_array(actual_param))
> + continue;
> +
> + /* User is trying to pass a whole gl_TessLevel* array to a function
> + * call. Since we are reshaping gl_TessLevel* from an array of floats
> + * to a vec4, this isn't going to work anymore, so use a temporary
> + * array instead.
> + */
> + ir_variable *temp_clip_distance = new(ctx) ir_variable(
> + actual_param->type, "temp_clip_distance", ir_var_temporary);
Maybe call this temp or temp_tess_level...
With those things fixed, this is
Reviewed-by: Kenneth Graunke <kenneth at whitecape.org>
> + this->base_ir->insert_before(temp_clip_distance);
> + actual_param->replace_with(
> + new(ctx) ir_dereference_variable(temp_clip_distance));
> + if (formal_param->data.mode == ir_var_function_in
> + || formal_param->data.mode == ir_var_function_inout) {
> + /* Copy from gl_TessLevel* to the temporary before the call.
> + * Since we are going to insert this copy before the current
> + * instruction, we need to visit it afterwards to make sure it
> + * gets lowered.
> + */
> + ir_assignment *new_assignment = new(ctx) ir_assignment(
> + new(ctx) ir_dereference_variable(temp_clip_distance),
> + actual_param->clone(ctx, NULL));
> + this->base_ir->insert_before(new_assignment);
> + this->visit_new_assignment(new_assignment);
> + }
> + if (formal_param->data.mode == ir_var_function_out
> + || formal_param->data.mode == ir_var_function_inout) {
> + /* Copy from the temporary to gl_TessLevel* after the call.
> + * Since visit_list_elements() has already decided which
> + * instruction it's going to visit next, we need to visit
> + * afterwards to make sure it gets lowered.
> + */
> + ir_assignment *new_assignment = new(ctx) ir_assignment(
> + actual_param->clone(ctx, NULL),
> + new(ctx) ir_dereference_variable(temp_clip_distance));
> + this->base_ir->insert_after(new_assignment);
> + this->visit_new_assignment(new_assignment);
> + }
> + }
> +
> + return rvalue_visit(ir);
> +}
> +
> +
> +bool
> +lower_tess_level(gl_shader *shader)
> +{
> + if ((shader->Stage != MESA_SHADER_TESS_CTRL) &&
> + (shader->Stage != MESA_SHADER_TESS_EVAL))
> + return false;
> +
> + lower_tess_level_visitor v(shader->Stage);
> +
> + visit_list_elements(&v, shader->ir);
> +
> + if (v.new_tess_level_outer_var)
> + shader->symbols->add_variable(v.new_tess_level_outer_var);
> + if (v.new_tess_level_inner_var)
> + shader->symbols->add_variable(v.new_tess_level_inner_var);
> +
> + return v.progress;
> +}
> diff --git a/src/mesa/drivers/dri/i965/brw_context.c b/src/mesa/drivers/dri/i965/brw_context.c
> index f39b350..016303b 100644
> --- a/src/mesa/drivers/dri/i965/brw_context.c
> +++ b/src/mesa/drivers/dri/i965/brw_context.c
> @@ -581,6 +581,7 @@ brw_initialize_context_constants(struct brw_context *brw)
> (i == MESA_SHADER_FRAGMENT);
> ctx->Const.ShaderCompilerOptions[i].EmitNoIndirectUniform = false;
> ctx->Const.ShaderCompilerOptions[i].LowerClipDistance = true;
> + ctx->Const.ShaderCompilerOptions[i].LowerTessLevel = true;
> }
>
> ctx->Const.ShaderCompilerOptions[MESA_SHADER_VERTEX].OptimizeForAOS = true;
> diff --git a/src/mesa/main/mtypes.h b/src/mesa/main/mtypes.h
> index a727b8b..25afd61 100644
> --- a/src/mesa/main/mtypes.h
> +++ b/src/mesa/main/mtypes.h
> @@ -2980,6 +2980,7 @@ struct gl_shader_compiler_options
> GLboolean EmitNoPow; /**< Emit POW opcodes? */
> GLboolean EmitNoSat; /**< Emit SAT opcodes? */
> GLboolean LowerClipDistance; /**< Lower gl_ClipDistance from float[8] to vec4[2]? */
> + GLboolean LowerTessLevel; /**< Lower gl_TessLevel* from float[n] to vecn? */
>
> /**
> * \name Forms of indirect addressing the driver cannot do.
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20150619/fca6669e/attachment-0001.sig>
More information about the mesa-dev
mailing list