src/gallium/drivers/vc4/vc4_opt_vpm_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_vpm_writes.c
  26  *
  27  * This modifies instructions that generate the value consumed by a VPM write
  28  * to write directly into the VPM.
  29  */
  30
  31 #include "vc4_qir.h"
  32
  33 bool
  34 qir_opt_vpm_writes(struct vc4_compile *c)
  35 {
  36         if (c->stage == QSTAGE_FRAG)
  37                 return false;
  38
  39         bool progress = false;
  40         struct qinst *vpm_writes[64] = { 0 };
  41         uint32_t use_count[c->num_temps];
  42         uint32_t vpm_write_count = 0;
  43         memset(&use_count, 0, sizeof(use_count));
  44
  45         list_for_each_entry(struct qinst, inst, &c->instructions, link) {
  46                 switch (inst->dst.file) {
  47                 case QFILE_VPM:
  48                         vpm_writes[vpm_write_count++] = inst;
  49                         break;
  50                 default:
  51                         break;
  52                 }
  53
  54                 for (int i = 0; i < qir_get_op_nsrc(inst->op); i++) {
  55                         if (inst->src[i].file == QFILE_TEMP)
  56                                 use_count[inst->src[i].index]++;
  57                 }
  58         }
  59
  60         for (int i = 0; i < vpm_write_count; i++) {
  61                 if (vpm_writes[i]->op != QOP_MOV ||
  62                     vpm_writes[i]->src[0].file != QFILE_TEMP) {
  63                         continue;
  64                 }
  65
  66                 uint32_t temp = vpm_writes[i]->src[0].index;
  67                 if (use_count[temp] != 1)
  68                         continue;
  69
  70                 struct qinst *inst = c->defs[temp];
  71                 if (qir_is_multi_instruction(inst))
  72                         continue;
  73
  74                 if (qir_depends_on_flags(inst) || inst->sf)
  75                         continue;
  76
  77                 if (qir_has_side_effects(c, inst) ||
  78                     qir_has_side_effect_reads(c, inst)) {
  79                         continue;
  80                 }
  81
  82                 /* A QOP_TEX_RESULT destination is r4, so we can't move
  83                  * accesses to it past another QOP_TEX_RESULT which would
  84                  * update it.
  85                  */
  86                 int src;
  87                 for (src = 0; src < qir_get_op_nsrc(inst->op); src++) {
  88                         if (inst->src[src].file == QFILE_TEMP) {
  89                                 if (c->defs[inst->src[src].index]->op ==
  90                                     QOP_TEX_RESULT) {
  91                                         break;
  92                                 }
  93                         }
  94                 }
  95                 if (src != qir_get_op_nsrc(inst->op))
  96                         continue;
  97
  98                 /* Move the generating instruction to the end of the program
  99                  * to maintain the order of the VPM writes.
 100                  */
 101                 assert(!vpm_writes[i]->sf);
 102                 list_del(&inst->link);
 103                 list_addtail(&inst->link, &vpm_writes[i]->link);
 104                 qir_remove_instruction(c, vpm_writes[i]);
 105
 106                 c->defs[inst->dst.index] = NULL;
 107                 inst->dst.file = QFILE_VPM;
 108                 inst->dst.index = 0;
 109
 110                 progress = true;
 111         }
 112
 113         return progress;
 114 }