1/*
2 * Copyright © 2013 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#include "ir.h"
24#include "ir_builder.h"
25#include "ir_rvalue_visitor.h"
26#include "ir_optimization.h"
27
28using namespace ir_builder;
29
30namespace {
31
32class vector_insert_visitor : public ir_rvalue_visitor {
33public:
34   vector_insert_visitor(bool lower_nonconstant_index)
35      : progress(false), lower_nonconstant_index(lower_nonconstant_index)
36   {
37      factory.instructions = &factory_instructions;
38   }
39
40   virtual ~vector_insert_visitor()
41   {
42      assert(factory_instructions.is_empty());
43   }
44
45   virtual void handle_rvalue(ir_rvalue **rv);
46
47   ir_factory factory;
48   exec_list factory_instructions;
49   bool progress;
50   bool lower_nonconstant_index;
51};
52
53} /* anonymous namespace */
54
55void
56vector_insert_visitor::handle_rvalue(ir_rvalue **rv)
57{
58   if (*rv == NULL || (*rv)->ir_type != ir_type_expression)
59      return;
60
61   ir_expression *const expr = (ir_expression *) *rv;
62
63   if (likely(expr->operation != ir_triop_vector_insert))
64      return;
65
66   factory.mem_ctx = ralloc_parent(expr);
67
68   ir_constant *const idx =
69      expr->operands[2]->constant_expression_value(factory.mem_ctx);
70   if (idx != NULL) {
71      /* Replace (vector_insert (vec) (scalar) (index)) with a dereference of
72       * a new temporary.  The new temporary gets assigned as
73       *
74       *     t = vec
75       *     t.mask = scalar
76       *
77       * where mask is the component selected by index.
78       */
79      ir_variable *const temp =
80         factory.make_temp(expr->operands[0]->type, "vec_tmp");
81
82      const int mask = 1 << idx->value.i[0];
83
84      factory.emit(assign(temp, expr->operands[0]));
85      factory.emit(assign(temp, expr->operands[1], mask));
86
87      this->progress = true;
88      *rv = new(factory.mem_ctx) ir_dereference_variable(temp);
89   } else if (this->lower_nonconstant_index) {
90      /* Replace (vector_insert (vec) (scalar) (index)) with a dereference of
91       * a new temporary.  The new temporary gets assigned as
92       *
93       *     t = vec
94       *     if (index == 0)
95       *         t.x = scalar
96       *     if (index == 1)
97       *         t.y = scalar
98       *     if (index == 2)
99       *         t.z = scalar
100       *     if (index == 3)
101       *         t.w = scalar
102       */
103      ir_variable *const temp =
104         factory.make_temp(expr->operands[0]->type, "vec_tmp");
105
106      ir_variable *const src_temp =
107         factory.make_temp(expr->operands[1]->type, "src_temp");
108
109      factory.emit(assign(temp, expr->operands[0]));
110      factory.emit(assign(src_temp, expr->operands[1]));
111
112      assert(expr->operands[2]->type == glsl_type::int_type ||
113             expr->operands[2]->type == glsl_type::uint_type);
114
115      for (unsigned i = 0; i < expr->type->vector_elements; i++) {
116         ir_constant *const cmp_index =
117            ir_constant::zero(factory.mem_ctx, expr->operands[2]->type);
118         cmp_index->value.u[0] = i;
119
120         ir_variable *const cmp_result =
121            factory.make_temp(glsl_type::bool_type, "index_condition");
122
123         factory.emit(assign(cmp_result,
124                             equal(expr->operands[2]->clone(factory.mem_ctx,
125                                                            NULL),
126                                   cmp_index)));
127
128         factory.emit(if_tree(cmp_result,
129                              assign(temp, src_temp, WRITEMASK_X << i)));
130      }
131
132      this->progress = true;
133      *rv = new(factory.mem_ctx) ir_dereference_variable(temp);
134   }
135
136   base_ir->insert_before(factory.instructions);
137}
138
139bool
140lower_vector_insert(exec_list *instructions, bool lower_nonconstant_index)
141{
142   vector_insert_visitor v(lower_nonconstant_index);
143
144   visit_list_elements(&v, instructions);
145
146   return v.progress;
147}
148