src/glsl/ir_vec_index_to_cond_assign.cpp

   1 /*
   2  * Copyright © 2010 Intel Corporation
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining a
   5  * copy of this software and associated documentation files (the "Software"),
   6  * to deal in the Software without restriction, including without limitation
   7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
   8  * and/or sell copies of the Software, and to permit persons to whom the
   9  * Software is furnished to do so, subject to the following conditions:
  10  *
  11  * The above copyright notice and this permission notice (including the next
  12  * paragraph) shall be included in all copies or substantial portions of the
  13  * Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
  21  * DEALINGS IN THE SOFTWARE.
  22  */
  23
  24 /**
  25  * \file ir_vec_index_to_cond_assign.cpp
  26  *
  27  * Turns indexing into vector types to a series of conditional moves
  28  * of each channel's swizzle into a temporary.
  29  *
  30  * Most GPUs don't have a native way to do this operation, and this
  31  * works around that.  For drivers using both this pass and
  32  * ir_vec_index_to_swizzle, there's a risk that this pass will happen
  33  * before sufficient constant folding to find that the array index is
  34  * constant.  However, we hope that other optimization passes,
  35  * particularly constant folding of assignment conditions and copy
  36  * propagation, will result in the same code in the end.
  37  */
  38
  39 #include "ir.h"
  40 #include "ir_visitor.h"
  41 #include "ir_optimization.h"
  42 #include "glsl_types.h"
  43
  44 /**
  45  * Visitor class for replacing expressions with ir_constant values.
  46  */
  47
  48 class ir_vec_index_to_cond_assign_visitor : public ir_hierarchical_visitor {
  49 public:
  50    ir_vec_index_to_cond_assign_visitor()
  51    {
  52       progress = false;
  53    }
  54
  55    ir_rvalue *convert_vec_index_to_cond_assign(ir_rvalue *val);
  56
  57    virtual ir_visitor_status visit_enter(ir_expression *);
  58    virtual ir_visitor_status visit_enter(ir_swizzle *);
  59    virtual ir_visitor_status visit_enter(ir_assignment *);
  60    virtual ir_visitor_status visit_enter(ir_return *);
  61    virtual ir_visitor_status visit_enter(ir_call *);
  62    virtual ir_visitor_status visit_enter(ir_if *);
  63
  64    bool progress;
  65 };
  66
  67 ir_rvalue *
  68 ir_vec_index_to_cond_assign_visitor::convert_vec_index_to_cond_assign(ir_rvalue *ir)
  69 {
  70    ir_dereference_array *orig_deref = ir->as_dereference_array();
  71    ir_assignment *assign;
  72    ir_variable *index, *var;
  73    ir_dereference *deref;
  74    ir_expression *condition;
  75    ir_swizzle *swizzle;
  76    int i;
  77
  78    if (!orig_deref)
  79       return ir;
  80
  81    if (orig_deref->array->type->is_matrix() ||
  82        orig_deref->array->type->is_array())
  83       return ir;
  84
  85    assert(orig_deref->array_index->type->base_type == GLSL_TYPE_INT);
  86
  87    /* Store the index to a temporary to avoid reusing its tree. */
  88    index = new(base_ir) ir_variable(glsl_type::int_type,
  89                                     "vec_index_tmp_i");
  90    base_ir->insert_before(index);
  91    deref = new(base_ir) ir_dereference_variable(index);
  92    assign = new(base_ir) ir_assignment(deref, orig_deref->array_index, NULL);
  93    base_ir->insert_before(assign);
  94
  95    /* Temporary where we store whichever value we swizzle out. */
  96    var = new(base_ir) ir_variable(ir->type, "vec_index_tmp_v");
  97    base_ir->insert_before(var);
  98
  99    /* Generate a conditional move of each vector element to the temp. */
 100    for (i = 0; i < orig_deref->array->type->vector_elements; i++) {
 101       deref = new(base_ir) ir_dereference_variable(index);
 102       condition = new(base_ir) ir_expression(ir_binop_equal,
 103                                              glsl_type::bool_type,
 104                                              deref,
 105                                              new(base_ir) ir_constant(i));
 106
 107       /* Just clone the rest of the deref chain when trying to get at the
 108        * underlying variable.
 109        */
 110       deref = (ir_dereference *)orig_deref->array->clone(NULL);
 111       swizzle = new(base_ir) ir_swizzle(deref, i, 0, 0, 0, 1);
 112
 113       deref = new(base_ir) ir_dereference_variable(var);
 114       assign = new(base_ir) ir_assignment(deref, swizzle, condition);
 115       base_ir->insert_before(assign);
 116    }
 117
 118    this->progress = true;
 119    return new(base_ir) ir_dereference_variable(var);
 120 }
 121
 122 ir_visitor_status
 123 ir_vec_index_to_cond_assign_visitor::visit_enter(ir_expression *ir)
 124 {
 125    unsigned int i;
 126
 127    for (i = 0; i < ir->get_num_operands(); i++) {
 128       ir->operands[i] = convert_vec_index_to_cond_assign(ir->operands[i]);
 129    }
 130
 131    return visit_continue;
 132 }
 133
 134 ir_visitor_status
 135 ir_vec_index_to_cond_assign_visitor::visit_enter(ir_swizzle *ir)
 136 {
 137    /* Can't be hit from normal GLSL, since you can't swizzle a scalar (which
 138     * the result of indexing a vector is.  But maybe at some point we'll end up
 139     * using swizzling of scalars for vector construction.
 140     */
 141    ir->val = convert_vec_index_to_cond_assign(ir->val);
 142
 143    return visit_continue;
 144 }
 145
 146 ir_visitor_status
 147 ir_vec_index_to_cond_assign_visitor::visit_enter(ir_assignment *ir)
 148 {
 149    /* FINISHME: Handle it on the LHS. */
 150    ir->rhs = convert_vec_index_to_cond_assign(ir->rhs);
 151
 152    return visit_continue;
 153 }
 154
 155 ir_visitor_status
 156 ir_vec_index_to_cond_assign_visitor::visit_enter(ir_call *ir)
 157 {
 158    foreach_iter(exec_list_iterator, iter, *ir) {
 159       ir_rvalue *param = (ir_rvalue *)iter.get();
 160       ir_rvalue *new_param = convert_vec_index_to_cond_assign(param);
 161
 162       if (new_param != param) {
 163          param->insert_before(new_param);
 164          param->remove();
 165       }
 166    }
 167
 168    return visit_continue;
 169 }
 170
 171 ir_visitor_status
 172 ir_vec_index_to_cond_assign_visitor::visit_enter(ir_return *ir)
 173 {
 174    if (ir->value) {
 175       ir->value = convert_vec_index_to_cond_assign(ir->value);
 176    }
 177
 178    return visit_continue;
 179 }
 180
 181 ir_visitor_status
 182 ir_vec_index_to_cond_assign_visitor::visit_enter(ir_if *ir)
 183 {
 184    ir->condition = convert_vec_index_to_cond_assign(ir->condition);
 185
 186    return visit_continue;
 187 }
 188
 189 bool
 190 do_vec_index_to_cond_assign(exec_list *instructions)
 191 {
 192    ir_vec_index_to_cond_assign_visitor v;
 193
 194    visit_list_elements(&v, instructions);
 195
 196    return v.progress;
 197 }