panfrost: Rewrite varying assembly
[mesa.git] / src / gallium / drivers / panfrost / pan_assemble.c
1 /*
2 * © Copyright 2018 Alyssa Rosenzweig
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 */
24
25 #include <stdio.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include "pan_context.h"
29
30 #include "compiler/nir/nir.h"
31 #include "nir/tgsi_to_nir.h"
32 #include "midgard/midgard_compile.h"
33 #include "util/u_dynarray.h"
34
35 #include "tgsi/tgsi_dump.h"
36
37 void
38 panfrost_shader_compile(struct panfrost_context *ctx, struct mali_shader_meta *meta, const char *src, int type, struct panfrost_shader_state *state)
39 {
40 uint8_t *dst;
41
42 nir_shader *s;
43
44 struct pipe_shader_state *cso = state->base;
45
46 if (cso->type == PIPE_SHADER_IR_NIR) {
47 s = nir_shader_clone(NULL, cso->ir.nir);
48 } else {
49 assert (cso->type == PIPE_SHADER_IR_TGSI);
50 //tgsi_dump(cso->tokens, 0);
51 s = tgsi_to_nir(cso->tokens, ctx->base.screen);
52 }
53
54 s->info.stage = type == JOB_TYPE_VERTEX ? MESA_SHADER_VERTEX : MESA_SHADER_FRAGMENT;
55
56 if (s->info.stage == MESA_SHADER_FRAGMENT) {
57 /* Inject the alpha test now if we need to */
58
59 if (state->alpha_state.enabled) {
60 NIR_PASS_V(s, nir_lower_alpha_test, state->alpha_state.func, false);
61 }
62 }
63
64 /* Call out to Midgard compiler given the above NIR */
65
66 midgard_program program = {
67 .alpha_ref = state->alpha_state.ref_value
68 };
69
70 midgard_compile_shader_nir(s, &program, false);
71
72 /* Prepare the compiled binary for upload */
73 int size = program.compiled.size;
74 dst = program.compiled.data;
75
76 /* Upload the shader. The lookahead tag is ORed on as a tagged pointer.
77 * I bet someone just thought that would be a cute pun. At least,
78 * that's how I'd do it. */
79
80 meta->shader = panfrost_upload(&ctx->shaders, dst, size, true) | program.first_tag;
81
82 util_dynarray_fini(&program.compiled);
83
84 meta->midgard1.uniform_count = MIN2(program.uniform_count, program.uniform_cutoff);
85 meta->attribute_count = program.attribute_count;
86 meta->varying_count = program.varying_count;
87 meta->midgard1.work_count = program.work_register_count;
88
89 state->can_discard = program.can_discard;
90 state->writes_point_size = program.writes_point_size;
91 state->reads_point_coord = false;
92
93 /* Separate as primary uniform count is truncated */
94 state->uniform_count = program.uniform_count;
95
96 meta->midgard1.unknown2 = 8; /* XXX */
97
98 unsigned default_vec1_swizzle = panfrost_get_default_swizzle(1);
99 unsigned default_vec2_swizzle = panfrost_get_default_swizzle(2);
100 unsigned default_vec4_swizzle = panfrost_get_default_swizzle(4);
101
102 /* Iterate the varyings and emit the corresponding descriptor */
103 unsigned general_purpose_count = 0;
104
105 for (unsigned i = 0; i < program.varying_count; ++i) {
106 unsigned location = program.varyings[i];
107
108 /* Default to a vec4 varying */
109 struct mali_attr_meta v = {
110 .format = MALI_RGBA16F,
111 .swizzle = default_vec4_swizzle,
112 .unknown1 = 0x2,
113 };
114
115 /* Check for special cases, otherwise assume general varying */
116
117 if (location == VARYING_SLOT_POS) {
118 v.index = 1;
119 v.format = MALI_VARYING_POS;
120 } else if (location == VARYING_SLOT_PSIZ) {
121 v.index = 2;
122 v.format = MALI_R16F;
123 v.swizzle = default_vec1_swizzle;
124
125 state->writes_point_size = true;
126 } else if (location == VARYING_SLOT_PNTC) {
127 v.index = 3;
128 v.format = MALI_RG16F;
129 v.swizzle = default_vec2_swizzle;
130
131 state->reads_point_coord = true;
132 } else {
133 v.index = 0;
134 v.src_offset = 8 * (general_purpose_count++);
135 }
136
137 state->varyings[i] = v;
138 }
139
140 /* Set the stride for the general purpose fp16 vec4 varyings */
141 state->general_varying_stride = (2 * 4) * general_purpose_count;
142 }