src/gallium/drivers/vc4/vc4_opt_coalesce_ff_writes.c

   1 /*
   2  * Copyright © 2014 Broadcom
   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 DEALINGS
  21  * IN THE SOFTWARE.
  22  */
  23
  24 /**
  25  * @file vc4_opt_coalesce_ff_writes.c
  26  *
  27  * This modifies instructions that generate the value consumed by a VPM or TMU
  28  * coordinate write to write directly into the VPM or TMU.
  29  */
  30
  31 #include "vc4_qir.h"
  32
  33 bool
  34 qir_opt_coalesce_ff_writes(struct vc4_compile *c)
  35 {
  36         /* For now, only do this pass when we don't have control flow. */
  37         struct qblock *block = qir_entry_block(c);
  38         if (block != qir_exit_block(c))
  39                 return false;
  40
  41         bool progress = false;
  42         uint32_t use_count[c->num_temps];
  43         memset(&use_count, 0, sizeof(use_count));
  44
  45         qir_for_each_inst_inorder(inst, c) {
  46                 for (int i = 0; i < qir_get_nsrc(inst); i++) {
  47                         if (inst->src[i].file == QFILE_TEMP) {
  48                                 uint32_t temp = inst->src[i].index;
  49                                 use_count[temp]++;
  50                         }
  51                 }
  52         }
  53
  54         qir_for_each_inst_inorder(mov_inst, c) {
  55                 if (!qir_is_raw_mov(mov_inst) || mov_inst->sf)
  56                         continue;
  57                 if (mov_inst->src[0].file != QFILE_TEMP)
  58                         continue;
  59
  60                 if (!(mov_inst->dst.file == QFILE_VPM ||
  61                       mov_inst->dst.file == QFILE_TLB_COLOR_WRITE ||
  62                       mov_inst->dst.file == QFILE_TLB_COLOR_WRITE_MS ||
  63                       qir_is_tex(mov_inst)))
  64                         continue;
  65
  66                 uint32_t temp = mov_inst->src[0].index;
  67                 if (use_count[temp] != 1)
  68                         continue;
  69
  70                 struct qinst *inst = c->defs[temp];
  71                 if (!inst)
  72                         continue;
  73
  74                 /* Don't bother trying to fold in an ALU op using a uniform to
  75                  * a texture op, as we'll just have to lower the uniform back
  76                  * out.
  77                  */
  78                 if (qir_is_tex(mov_inst) && qir_has_uniform_read(inst))
  79                         continue;
  80
  81                 if (qir_depends_on_flags(inst) || inst->sf)
  82                         continue;
  83
  84                 if (qir_has_side_effects(c, inst) ||
  85                     qir_has_side_effect_reads(c, inst) ||
  86                     inst->op == QOP_TLB_COLOR_READ ||
  87                     inst->op == QOP_VARY_ADD_C) {
  88                         continue;
  89                 }
  90
  91                 /* Move the generating instruction into the position of the FF
  92                  * write.
  93                  */
  94                 c->defs[inst->dst.index] = NULL;
  95                 inst->dst.file = mov_inst->dst.file;
  96                 inst->dst.index = mov_inst->dst.index;
  97                 if (qir_has_implicit_tex_uniform(mov_inst)) {
  98                         inst->src[qir_get_tex_uniform_src(inst)] =
  99                                 mov_inst->src[qir_get_tex_uniform_src(mov_inst)];
 100                 }
 101
 102                 list_del(&inst->link);
 103                 list_addtail(&inst->link, &mov_inst->link);
 104
 105                 qir_remove_instruction(c, mov_inst);
 106
 107                 progress = true;
 108         }
 109
 110         return progress;
 111 }