1 /*
2 * Copyright (C) 2020 Collabora Ltd.
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 FROM,
20 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 * SOFTWARE.
22 *
23 * Authors (Collabora):
24 * Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
25 */
26
27 #include "pan_ir.h"
28 #include "util/macros.h"
29
30 /* Converts a per-component mask to a byte mask */
31
32 uint16_t
pan_to_bytemask(unsigned bytes,unsigned mask)33 pan_to_bytemask(unsigned bytes, unsigned mask)
34 {
35 switch (bytes) {
36 case 0:
37 assert(mask == 0);
38 return 0;
39
40 case 8:
41 return mask;
42
43 case 16: {
44 unsigned space =
45 (mask & 0x1) |
46 ((mask & 0x2) << (2 - 1)) |
47 ((mask & 0x4) << (4 - 2)) |
48 ((mask & 0x8) << (6 - 3)) |
49 ((mask & 0x10) << (8 - 4)) |
50 ((mask & 0x20) << (10 - 5)) |
51 ((mask & 0x40) << (12 - 6)) |
52 ((mask & 0x80) << (14 - 7));
53
54 return space | (space << 1);
55 }
56
57 case 32: {
58 unsigned space =
59 (mask & 0x1) |
60 ((mask & 0x2) << (4 - 1)) |
61 ((mask & 0x4) << (8 - 2)) |
62 ((mask & 0x8) << (12 - 3));
63
64 return space | (space << 1) | (space << 2) | (space << 3);
65 }
66
67 case 64: {
68 unsigned A = (mask & 0x1) ? 0xFF : 0x00;
69 unsigned B = (mask & 0x2) ? 0xFF : 0x00;
70 return A | (B << 8);
71 }
72
73 default:
74 unreachable("Invalid register mode");
75 }
76 }
77
78 void
pan_block_add_successor(pan_block * block,pan_block * successor)79 pan_block_add_successor(pan_block *block, pan_block *successor)
80 {
81 assert(block);
82 assert(successor);
83
84 /* Cull impossible edges */
85 if (block->unconditional_jumps)
86 return;
87
88 for (unsigned i = 0; i < ARRAY_SIZE(block->successors); ++i) {
89 if (block->successors[i]) {
90 if (block->successors[i] == successor)
91 return;
92 else
93 continue;
94 }
95
96 block->successors[i] = successor;
97 _mesa_set_add(successor->predecessors, block);
98 return;
99 }
100
101 unreachable("Too many successors");
102 }
103
104 /* Prints a NIR ALU type in Bifrost-style ".f32" ".i8" etc */
105
106 void
pan_print_alu_type(nir_alu_type t,FILE * fp)107 pan_print_alu_type(nir_alu_type t, FILE *fp)
108 {
109 unsigned size = nir_alu_type_get_type_size(t);
110 nir_alu_type base = nir_alu_type_get_base_type(t);
111
112 switch (base) {
113 case nir_type_int:
114 fprintf(fp, ".i");
115 break;
116 case nir_type_uint:
117 fprintf(fp, ".u");
118 break;
119 case nir_type_bool:
120 fprintf(fp, ".b");
121 break;
122 case nir_type_float:
123 fprintf(fp, ".f");
124 break;
125 default:
126 fprintf(fp, ".unknown");
127 break;
128 }
129
130 fprintf(fp, "%u", size);
131 }
132
133 /* Could optimize with a better data structure if anyone cares, TODO: profile */
134
135 unsigned
pan_lookup_pushed_ubo(struct panfrost_ubo_push * push,unsigned ubo,unsigned offs)136 pan_lookup_pushed_ubo(struct panfrost_ubo_push *push, unsigned ubo, unsigned offs)
137 {
138 struct panfrost_ubo_word word = {
139 .ubo = ubo,
140 .offset = offs
141 };
142
143 for (unsigned i = 0; i < push->count; ++i) {
144 if (memcmp(push->words + i, &word, sizeof(word)) == 0)
145 return i;
146 }
147
148 unreachable("UBO not pushed");
149
150 }
151