Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 5564 | serge | 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 *vpm_writes[64] = { 0 }; |
||
| 42 | uint32_t use_count[c->num_temps]; |
||
| 43 | uint32_t vpm_write_count = 0; |
||
| 44 | memset(&use_count, 0, sizeof(use_count)); |
||
| 45 | |||
| 46 | foreach(node, &c->instructions) { |
||
| 47 | struct qinst *inst = (struct qinst *)node; |
||
| 48 | |||
| 49 | switch (inst->dst.file) { |
||
| 50 | case QFILE_VPM: |
||
| 51 | vpm_writes[vpm_write_count++] = inst; |
||
| 52 | break; |
||
| 53 | default: |
||
| 54 | break; |
||
| 55 | } |
||
| 56 | |||
| 57 | for (int i = 0; i < qir_get_op_nsrc(inst->op); i++) { |
||
| 58 | if (inst->src[i].file == QFILE_TEMP) |
||
| 59 | use_count[inst->src[i].index]++; |
||
| 60 | } |
||
| 61 | } |
||
| 62 | |||
| 63 | for (int i = 0; i < vpm_write_count; i++) { |
||
| 64 | if (vpm_writes[i]->op != QOP_MOV || |
||
| 65 | vpm_writes[i]->src[0].file != QFILE_TEMP) { |
||
| 66 | continue; |
||
| 67 | } |
||
| 68 | |||
| 69 | uint32_t temp = vpm_writes[i]->src[0].index; |
||
| 70 | if (use_count[temp] != 1) |
||
| 71 | continue; |
||
| 72 | |||
| 73 | struct qinst *inst = c->defs[temp]; |
||
| 74 | if (qir_is_multi_instruction(inst)) |
||
| 75 | continue; |
||
| 76 | |||
| 77 | if (qir_depends_on_flags(inst) || inst->sf) |
||
| 78 | continue; |
||
| 79 | |||
| 80 | if (qir_has_side_effects(c, inst) || |
||
| 81 | qir_has_side_effect_reads(c, inst)) { |
||
| 82 | continue; |
||
| 83 | } |
||
| 84 | |||
| 85 | /* A QOP_TEX_RESULT destination is r4, so we can't move |
||
| 86 | * accesses to it past another QOP_TEX_RESULT which would |
||
| 87 | * update it. |
||
| 88 | */ |
||
| 89 | int src; |
||
| 90 | for (src = 0; src < qir_get_op_nsrc(inst->op); src++) { |
||
| 91 | if (inst->src[src].file == QFILE_TEMP) { |
||
| 92 | if (c->defs[inst->src[src].index]->op == |
||
| 93 | QOP_TEX_RESULT) { |
||
| 94 | break; |
||
| 95 | } |
||
| 96 | } |
||
| 97 | } |
||
| 98 | if (src != qir_get_op_nsrc(inst->op)) |
||
| 99 | continue; |
||
| 100 | |||
| 101 | /* Move the generating instruction to the end of the program |
||
| 102 | * to maintain the order of the VPM writes. |
||
| 103 | */ |
||
| 104 | assert(!vpm_writes[i]->sf); |
||
| 105 | move_to_tail(&vpm_writes[i]->link, &inst->link); |
||
| 106 | qir_remove_instruction(c, vpm_writes[i]); |
||
| 107 | |||
| 108 | c->defs[inst->dst.index] = NULL; |
||
| 109 | inst->dst.file = QFILE_VPM; |
||
| 110 | inst->dst.index = 0; |
||
| 111 | |||
| 112 | progress = true; |
||
| 113 | } |
||
| 114 | |||
| 115 | return progress; |
||
| 116 | } |