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 simple_node *node;
  41         struct qinst *defs[c->num_temps];
  42         struct qinst *vpm_writes[64] = { 0 };
  43         uint32_t use_count[c->num_temps];
  44         uint32_t vpm_write_count = 0;
  45         memset(&defs, 0, sizeof(defs));
  46         memset(&use_count, 0, sizeof(use_count));
  47
  48         foreach(node, &c->instructions) {
  49                 struct qinst *inst = (struct qinst *)node;
  50
  51                 switch (inst->dst.file) {
  52                 case QFILE_TEMP:
  53                         defs[inst->dst.index] = inst;
  54                         break;
  55                 case QFILE_VPM:
  56                         vpm_writes[vpm_write_count++] = inst;
  57                         break;
  58                 default:
  59                         break;
  60                 }
  61
  62                 for (int i = 0; i < qir_get_op_nsrc(inst->op); i++) {
  63                         if (inst->src[i].file == QFILE_TEMP)
  64                                 use_count[inst->src[i].index]++;
  65                 }
  66         }
  67
  68         for (int i = 0; i < vpm_write_count; i++) {
  69                 if (vpm_writes[i]->op != QOP_MOV ||
  70                     vpm_writes[i]->src[0].file != QFILE_TEMP) {
  71                         continue;
  72                 }
  73
  74                 uint32_t temp = vpm_writes[i]->src[0].index;
  75                 if (use_count[temp] != 1)
  76                         continue;
  77
  78                 struct qinst *inst = defs[temp];
  79                 if (qir_is_multi_instruction(inst))
  80                         continue;
  81
  82                 if (qir_depends_on_flags(inst))
  83                         continue;
  84
  85                 if (qir_has_side_effects(c, inst))
  86                         continue;
  87
  88                 /* A QOP_TEX_RESULT destination is r4, so we can't move
  89                  * accesses to it past another QOP_TEX_RESULT which would
  90                  * update it.
  91                  */
  92                 int src;
  93                 for (src = 0; src < qir_get_op_nsrc(inst->op); src++) {
  94                         if (inst->src[src].file == QFILE_TEMP) {
  95                                 if (defs[inst->src[src].index]->op ==
  96                                     QOP_TEX_RESULT) {
  97                                         break;
  98                                 }
  99                         }
 100                 }
 101                 if (src != qir_get_op_nsrc(inst->op))
 102                         continue;
 103
 104                 /* Move the generating instruction to the end of the program
 105                  * to maintain the order of the VPM writes.
 106                  */
 107                 move_to_tail(&vpm_writes[i]->link, &inst->link);
 108                 qir_remove_instruction(vpm_writes[i]);
 109
 110                 inst->dst.file = QFILE_VPM;
 111                 inst->dst.index = 0;
 112
 113                 progress = true;
 114         }
 115
 116         return progress;
 117 }