2 * Copyright © 2012 Intel Corporation
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:
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
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 DEALINGS
27 /** @file brw_fs_cse.cpp
29 * Support for local common subexpression elimination.
31 * See Muchnik's Advanced Compiler Design and Implementation, section
36 struct aeb_entry : public exec_node {
37 /** The instruction that generates the expression value. */
40 /** The temporary where the value is stored. */
46 is_expression(const fs_inst *const inst)
48 switch (inst->opcode) {
70 case FS_OPCODE_UNIFORM_PULL_CONSTANT_LOAD:
71 case FS_OPCODE_CINTERP:
72 case FS_OPCODE_LINTERP:
80 operands_match(fs_reg *xs, fs_reg *ys)
82 return xs[0].equals(ys[0]) && xs[1].equals(ys[1]) && xs[2].equals(ys[2]);
86 fs_visitor::opt_cse_local(bblock_t *block, exec_list *aeb)
88 bool progress = false;
90 void *mem_ctx = ralloc_context(this->mem_ctx);
92 for (fs_inst *inst = (fs_inst *)block->start;
93 inst != block->end->next;
94 inst = (fs_inst *) inst->next) {
96 /* Skip some cases. */
97 if (is_expression(inst) && !inst->predicate && inst->mlen == 0 &&
98 !inst->force_uncompressed && !inst->force_sechalf &&
99 !inst->conditional_mod)
104 foreach_list(entry_node, aeb) {
105 entry = (aeb_entry *) entry_node;
107 /* Match current instruction's expression against those in AEB. */
108 if (inst->opcode == entry->generator->opcode &&
109 inst->saturate == entry->generator->saturate &&
110 inst->dst.type == entry->generator->dst.type &&
111 operands_match(entry->generator->src, inst->src)) {
120 /* Our first sighting of this expression. Create an entry. */
121 aeb_entry *entry = ralloc(mem_ctx, aeb_entry);
122 entry->tmp = reg_undef;
123 entry->generator = inst;
124 aeb->push_tail(entry);
126 /* This is at least our second sighting of this expression.
127 * If we don't have a temporary already, make one.
129 bool no_existing_temp = entry->tmp.file == BAD_FILE;
130 if (no_existing_temp) {
131 entry->tmp = fs_reg(this, glsl_type::float_type);
132 entry->tmp.type = inst->dst.type;
134 fs_inst *copy = new(ralloc_parent(inst))
135 fs_inst(BRW_OPCODE_MOV, entry->generator->dst, entry->tmp);
136 entry->generator->insert_after(copy);
137 entry->generator->dst = entry->tmp;
141 assert(inst->dst.type == entry->tmp.type);
142 fs_inst *copy = new(ralloc_parent(inst))
143 fs_inst(BRW_OPCODE_MOV, inst->dst, entry->tmp);
144 copy->force_writemask_all = inst->force_writemask_all;
145 inst->replace_with(copy);
147 /* Appending an instruction may have changed our bblock end. */
148 if (inst == block->end) {
152 /* Continue iteration with copy->next */
157 /* Kill all AEB entries that use the destination. */
158 foreach_list_safe(entry_node, aeb) {
159 aeb_entry *entry = (aeb_entry *)entry_node;
161 for (int i = 0; i < 3; i++) {
162 if (inst->overwrites_reg(entry->generator->src[i])) {
171 ralloc_free(mem_ctx);
174 this->live_intervals_valid = false;
180 fs_visitor::opt_cse()
182 bool progress = false;
186 for (int b = 0; b < cfg.num_blocks; b++) {
187 bblock_t *block = cfg.blocks[b];
190 progress = opt_cse_local(block, &aeb) || progress;