1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
|
/*
* Copyright © 2013 Intel Corporation
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
* DEALINGS IN THE SOFTWARE.
*/
/**
* \file brw_vec4_gs.c
*
* State atom for client-programmable geometry shaders, and support code.
*/
#include "brw_gs.h"
#include "brw_context.h"
#include "brw_vec4_gs_visitor.h"
#include "brw_state.h"
#include "brw_ff_gs.h"
#include "brw_nir.h"
#include "brw_program.h"
#include "compiler/glsl/ir_uniform.h"
static void
brw_gs_debug_recompile(struct brw_context *brw,
struct gl_shader_program *shader_prog,
const struct brw_gs_prog_key *key)
{
struct brw_cache_item *c = NULL;
const struct brw_gs_prog_key *old_key = NULL;
bool found = false;
perf_debug("Recompiling geometry shader for program %d\n",
shader_prog->Name);
for (unsigned int i = 0; i < brw->cache.size; i++) {
for (c = brw->cache.items[i]; c; c = c->next) {
if (c->cache_id == BRW_CACHE_GS_PROG) {
old_key = c->key;
if (old_key->program_string_id == key->program_string_id)
break;
}
}
if (c)
break;
}
if (!c) {
perf_debug(" Didn't find previous compile in the shader cache for "
"debug\n");
return;
}
found |= brw_debug_recompile_sampler_key(brw, &old_key->tex, &key->tex);
if (!found) {
perf_debug(" Something else\n");
}
}
static void
assign_gs_binding_table_offsets(const struct gen_device_info *devinfo,
const struct gl_shader_program *shader_prog,
const struct gl_program *prog,
struct brw_gs_prog_data *prog_data)
{
/* In gen6 we reserve the first BRW_MAX_SOL_BINDINGS entries for transform
* feedback surfaces.
*/
uint32_t reserved = devinfo->gen == 6 ? BRW_MAX_SOL_BINDINGS : 0;
brw_assign_common_binding_table_offsets(MESA_SHADER_GEOMETRY, devinfo,
shader_prog, prog,
&prog_data->base.base,
reserved);
}
bool
brw_codegen_gs_prog(struct brw_context *brw,
struct gl_shader_program *prog,
struct brw_geometry_program *gp,
struct brw_gs_prog_key *key)
{
struct brw_compiler *compiler = brw->screen->compiler;
const struct gen_device_info *devinfo = &brw->screen->devinfo;
struct brw_stage_state *stage_state = &brw->gs.base;
struct brw_gs_prog_data prog_data;
bool start_busy = false;
double start_time = 0;
memset(&prog_data, 0, sizeof(prog_data));
assign_gs_binding_table_offsets(devinfo, prog,
&gp->program.Base, &prog_data);
/* Allocate the references to the uniforms that will end up in the
* prog_data associated with the compiled program, and which will be freed
* by the state cache.
*
* Note: param_count needs to be num_uniform_components * 4, since we add
* padding around uniform values below vec4 size, so the worst case is that
* every uniform is a float which gets padded to the size of a vec4.
*/
struct gl_linked_shader *gs = prog->_LinkedShaders[MESA_SHADER_GEOMETRY];
struct brw_shader *bgs = (struct brw_shader *) gs;
int param_count = gp->program.Base.nir->num_uniforms / 4;
prog_data.base.base.param =
rzalloc_array(NULL, const gl_constant_value *, param_count);
prog_data.base.base.pull_param =
rzalloc_array(NULL, const gl_constant_value *, param_count);
prog_data.base.base.image_param =
rzalloc_array(NULL, struct brw_image_param, gs->NumImages);
prog_data.base.base.nr_params = param_count;
prog_data.base.base.nr_image_params = gs->NumImages;
brw_nir_setup_glsl_uniforms(gp->program.Base.nir, prog, &gp->program.Base,
&prog_data.base.base,
compiler->scalar_stage[MESA_SHADER_GEOMETRY]);
GLbitfield64 outputs_written = gp->program.Base.OutputsWritten;
prog_data.base.cull_distance_mask =
((1 << gp->program.Base.CullDistanceArraySize) - 1) <<
gp->program.Base.ClipDistanceArraySize;
brw_compute_vue_map(devinfo,
&prog_data.base.vue_map, outputs_written,
prog->SeparateShader);
if (unlikely(INTEL_DEBUG & DEBUG_GS))
brw_dump_ir("geometry", prog, gs, NULL);
int st_index = -1;
if (INTEL_DEBUG & DEBUG_SHADER_TIME)
st_index = brw_get_shader_time_index(brw, prog, NULL, ST_GS);
if (unlikely(brw->perf_debug)) {
start_busy = brw->batch.last_bo && drm_intel_bo_busy(brw->batch.last_bo);
start_time = get_time();
}
void *mem_ctx = ralloc_context(NULL);
unsigned program_size;
char *error_str;
const unsigned *program =
brw_compile_gs(brw->screen->compiler, brw, mem_ctx, key,
&prog_data, gs->Program->nir, prog,
st_index, &program_size, &error_str);
if (program == NULL) {
ralloc_strcat(&prog->InfoLog, error_str);
_mesa_problem(NULL, "Failed to compile geometry shader: %s\n", error_str);
ralloc_free(mem_ctx);
return false;
}
if (unlikely(brw->perf_debug)) {
if (bgs->compiled_once) {
brw_gs_debug_recompile(brw, prog, key);
}
if (start_busy && !drm_intel_bo_busy(brw->batch.last_bo)) {
perf_debug("GS compile took %.03f ms and stalled the GPU\n",
(get_time() - start_time) * 1000);
}
bgs->compiled_once = true;
}
/* Scratch space is used for register spilling */
brw_alloc_stage_scratch(brw, stage_state,
prog_data.base.base.total_scratch,
devinfo->max_gs_threads);
brw_upload_cache(&brw->cache, BRW_CACHE_GS_PROG,
key, sizeof(*key),
program, program_size,
&prog_data, sizeof(prog_data),
&stage_state->prog_offset, &brw->gs.prog_data);
ralloc_free(mem_ctx);
return true;
}
static bool
brw_gs_state_dirty(const struct brw_context *brw)
{
return brw_state_dirty(brw,
_NEW_TEXTURE,
BRW_NEW_GEOMETRY_PROGRAM |
BRW_NEW_TRANSFORM_FEEDBACK);
}
void
brw_gs_populate_key(struct brw_context *brw,
struct brw_gs_prog_key *key)
{
struct gl_context *ctx = &brw->ctx;
struct brw_geometry_program *gp =
(struct brw_geometry_program *) brw->geometry_program;
struct gl_program *prog = &gp->program.Base;
memset(key, 0, sizeof(*key));
key->program_string_id = gp->id;
/* _NEW_TEXTURE */
brw_populate_sampler_prog_key_data(ctx, prog, &key->tex);
}
void
brw_upload_gs_prog(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
struct gl_shader_program **current = ctx->_Shader->CurrentProgram;
struct brw_stage_state *stage_state = &brw->gs.base;
struct brw_gs_prog_key key;
/* BRW_NEW_GEOMETRY_PROGRAM */
struct brw_geometry_program *gp =
(struct brw_geometry_program *) brw->geometry_program;
if (!brw_gs_state_dirty(brw))
return;
if (gp == NULL) {
/* No geometry shader. Vertex data just passes straight through. */
if (brw->gen == 6 &&
(brw->ctx.NewDriverState & BRW_NEW_TRANSFORM_FEEDBACK)) {
gen6_brw_upload_ff_gs_prog(brw);
return;
}
/* Other state atoms had better not try to access prog_data, since
* there's no GS program.
*/
brw->gs.prog_data = NULL;
brw->gs.base.prog_data = NULL;
return;
}
brw_gs_populate_key(brw, &key);
if (!brw_search_cache(&brw->cache, BRW_CACHE_GS_PROG,
&key, sizeof(key),
&stage_state->prog_offset, &brw->gs.prog_data)) {
bool success = brw_codegen_gs_prog(brw, current[MESA_SHADER_GEOMETRY],
gp, &key);
assert(success);
(void)success;
}
brw->gs.base.prog_data = &brw->gs.prog_data->base.base;
}
bool
brw_gs_precompile(struct gl_context *ctx,
struct gl_shader_program *shader_prog,
struct gl_program *prog)
{
struct brw_context *brw = brw_context(ctx);
struct brw_gs_prog_key key;
uint32_t old_prog_offset = brw->gs.base.prog_offset;
struct brw_gs_prog_data *old_prog_data = brw->gs.prog_data;
bool success;
struct gl_geometry_program *gp = (struct gl_geometry_program *) prog;
struct brw_geometry_program *bgp = brw_geometry_program(gp);
memset(&key, 0, sizeof(key));
brw_setup_tex_for_precompile(brw, &key.tex, prog);
key.program_string_id = bgp->id;
success = brw_codegen_gs_prog(brw, shader_prog, bgp, &key);
brw->gs.base.prog_offset = old_prog_offset;
brw->gs.prog_data = old_prog_data;
return success;
}
|