src/mesa/drivers/dri/i965/brw_fs_saturate_propagation.cpp

   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 DEALINGS
  21  * IN THE SOFTWARE.
  22  */
  23
  24 #include "brw_fs.h"
  25 #include "brw_fs_live_variables.h"
  26 #include "brw_cfg.h"
  27
  28 /** @file brw_fs_saturate_propagation.cpp
  29  *
  30  * Implements a pass that propagates the SAT modifier from a MOV.SAT into the
  31  * instruction that produced the source of the MOV.SAT, thereby allowing the
  32  * MOV's src and dst to be coalesced and the MOV removed.
  33  *
  34  * For instance,
  35  *
  36  *    ADD     tmp, src0, src1
  37  *    MOV.SAT dst, tmp
  38  *
  39  * would be transformed into
  40  *
  41  *    ADD.SAT tmp, src0, src1
  42  *    MOV     dst, tmp
  43  */
  44
  45 static bool
  46 opt_saturate_propagation_local(fs_visitor *v, bblock_t *block)
  47 {
  48    bool progress = false;
  49    int ip = block->end_ip + 1;
  50
  51    foreach_inst_in_block_reverse(fs_inst, inst, block) {
  52       ip--;
  53
  54       if (inst->opcode != BRW_OPCODE_MOV ||
  55           inst->dst.file != GRF ||
  56           inst->src[0].file != GRF ||
  57           inst->src[0].abs ||
  58           inst->src[0].negate ||
  59           !inst->saturate)
  60          continue;
  61
  62       int src_var = v->live_intervals->var_from_reg(inst->src[0]);
  63       int src_end_ip = v->live_intervals->end[src_var];
  64
  65       bool interfered = false;
  66       foreach_inst_in_block_reverse_starting_from(fs_inst, scan_inst, inst, block) {
  67          if (scan_inst->dst.file == GRF &&
  68              scan_inst->dst.reg == inst->src[0].reg &&
  69              scan_inst->dst.reg_offset == inst->src[0].reg_offset &&
  70              !scan_inst->is_partial_write()) {
  71             if (scan_inst->saturate) {
  72                inst->saturate = false;
  73                progress = true;
  74             } else if (src_end_ip <= ip || inst->dst.equals(inst->src[0])) {
  75                if (scan_inst->can_do_saturate()) {
  76                   scan_inst->saturate = true;
  77                   inst->saturate = false;
  78                   progress = true;
  79                }
  80             }
  81             break;
  82          }
  83          for (int i = 0; i < scan_inst->sources; i++) {
  84             if ((scan_inst->opcode != BRW_OPCODE_MOV || !scan_inst->saturate) &&
  85                 scan_inst->src[i].file == GRF &&
  86                 scan_inst->src[i].reg == inst->src[0].reg &&
  87                 scan_inst->src[i].reg_offset == inst->src[0].reg_offset) {
  88                interfered = true;
  89                break;
  90             }
  91          }
  92
  93          if (interfered)
  94             break;
  95       }
  96    }
  97
  98    return progress;
  99 }
 100
 101 bool
 102 fs_visitor::opt_saturate_propagation()
 103 {
 104    bool progress = false;
 105
 106    calculate_live_intervals();
 107
 108    foreach_block (block, cfg) {
 109       progress = opt_saturate_propagation_local(this, block) || progress;
 110    }
 111
 112    /* Live intervals are still valid. */
 113
 114    return progress;
 115 }