summaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/etnaviv/etnaviv_shader.c
blob: b5256e4b510bb937c6ad739c76734e0ee2d7f358 (plain)
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
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
/*
 * Copyright (c) 2012-2015 Etnaviv Project
 *
 * 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, sub license,
 * 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 NON-INFRINGEMENT. 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.
 *
 * Authors:
 *    Wladimir J. van der Laan <laanwj@gmail.com>
 */

#include "etnaviv_shader.h"

#include "etnaviv_compiler.h"
#include "etnaviv_context.h"
#include "etnaviv_debug.h"
#include "etnaviv_util.h"

#include "tgsi/tgsi_parse.h"
#include "util/u_math.h"
#include "util/u_memory.h"

/* Link vs and fs together: fill in shader_state from vs and fs
 * as this function is called every time a new fs or vs is bound, the goal is to
 * do little processing as possible here, and to precompute as much as possible in
 * the vs/fs shader_object.
 *
 * XXX we could cache the link result for a certain set of VS/PS; usually a pair
 * of VS and PS will be used together anyway.
 */
static bool
etna_link_shaders(struct etna_context *ctx, struct compiled_shader_state *cs,
                  const struct etna_shader_variant *vs, const struct etna_shader_variant *fs)
{
   struct etna_shader_link_info link = { };

   assert(vs->processor == PIPE_SHADER_VERTEX);
   assert(fs->processor == PIPE_SHADER_FRAGMENT);

#ifdef DEBUG
   if (DBG_ENABLED(ETNA_DBG_DUMP_SHADERS)) {
      etna_dump_shader(vs);
      etna_dump_shader(fs);
   }
#endif

   if (etna_link_shader(&link, vs, fs)) {
      /* linking failed: some fs inputs do not have corresponding
       * vs outputs */
      assert(0);

      return false;
   }

   if (DBG_ENABLED(ETNA_DBG_LINKER_MSGS)) {
      debug_printf("link result:\n");
      debug_printf("  vs  -> fs  comps use     pa_attr\n");

      for (int idx = 0; idx < link.num_varyings; ++idx)
         debug_printf("  t%-2u -> t%-2u %-5.*s %u,%u,%u,%u 0x%08x\n",
                      link.varyings[idx].reg, idx + 1,
                      link.varyings[idx].num_components, "xyzw",
                      link.varyings[idx].use[0], link.varyings[idx].use[1],
                      link.varyings[idx].use[2], link.varyings[idx].use[3],
                      link.varyings[idx].pa_attributes);
   }

   /* set last_varying_2x flag if the last varying has 1 or 2 components */
   bool last_varying_2x = false;
   if (link.num_varyings > 0 && link.varyings[link.num_varyings - 1].num_components <= 2)
      last_varying_2x = true;

   cs->RA_CONTROL = VIVS_RA_CONTROL_UNK0 |
                    COND(last_varying_2x, VIVS_RA_CONTROL_LAST_VARYING_2X);

   cs->PA_ATTRIBUTE_ELEMENT_COUNT = VIVS_PA_ATTRIBUTE_ELEMENT_COUNT_COUNT(link.num_varyings);
   for (int idx = 0; idx < link.num_varyings; ++idx)
      cs->PA_SHADER_ATTRIBUTES[idx] = link.varyings[idx].pa_attributes;

   cs->VS_END_PC = vs->code_size / 4;
   cs->VS_OUTPUT_COUNT = 1 + link.num_varyings; /* position + varyings */

   /* vs outputs (varyings) */
   DEFINE_ETNA_BITARRAY(vs_output, 16, 8) = {0};
   int varid = 0;
   etna_bitarray_set(vs_output, 8, varid++, vs->vs_pos_out_reg);
   for (int idx = 0; idx < link.num_varyings; ++idx)
      etna_bitarray_set(vs_output, 8, varid++, link.varyings[idx].reg);
   if (vs->vs_pointsize_out_reg >= 0)
      etna_bitarray_set(vs_output, 8, varid++, vs->vs_pointsize_out_reg); /* pointsize is last */

   for (int idx = 0; idx < ARRAY_SIZE(cs->VS_OUTPUT); ++idx)
      cs->VS_OUTPUT[idx] = vs_output[idx];

   if (vs->vs_pointsize_out_reg != -1) {
      /* vertex shader outputs point coordinate, provide extra output and make
       * sure PA config is
       * not masked */
      cs->PA_CONFIG = ~0;
      cs->VS_OUTPUT_COUNT_PSIZE = cs->VS_OUTPUT_COUNT + 1;
   } else {
      /* vertex shader does not output point coordinate, make sure thate
       * POINT_SIZE_ENABLE is masked
       * and no extra output is given */
      cs->PA_CONFIG = ~VIVS_PA_CONFIG_POINT_SIZE_ENABLE;
      cs->VS_OUTPUT_COUNT_PSIZE = cs->VS_OUTPUT_COUNT;
   }

   cs->VS_LOAD_BALANCING = vs->vs_load_balancing;
   cs->VS_START_PC = 0;

   cs->PS_END_PC = fs->code_size / 4;
   cs->PS_OUTPUT_REG = fs->ps_color_out_reg;
   cs->PS_INPUT_COUNT =
      VIVS_PS_INPUT_COUNT_COUNT(link.num_varyings + 1) | /* Number of inputs plus position */
      VIVS_PS_INPUT_COUNT_UNK8(fs->input_count_unk8);
   cs->PS_TEMP_REGISTER_CONTROL =
      VIVS_PS_TEMP_REGISTER_CONTROL_NUM_TEMPS(MAX2(fs->num_temps, link.num_varyings + 1));
   cs->PS_CONTROL = VIVS_PS_CONTROL_UNK1; /* XXX when can we set BYPASS? */
   cs->PS_START_PC = 0;

   /* Precompute PS_INPUT_COUNT and TEMP_REGISTER_CONTROL in the case of MSAA
    * mode, avoids some fumbling in sync_context. */
   cs->PS_INPUT_COUNT_MSAA =
      VIVS_PS_INPUT_COUNT_COUNT(link.num_varyings + 2) | /* MSAA adds another input */
      VIVS_PS_INPUT_COUNT_UNK8(fs->input_count_unk8);
   cs->PS_TEMP_REGISTER_CONTROL_MSAA =
      VIVS_PS_TEMP_REGISTER_CONTROL_NUM_TEMPS(MAX2(fs->num_temps, link.num_varyings + 2));

   uint32_t total_components = 0;
   DEFINE_ETNA_BITARRAY(num_components, ETNA_NUM_VARYINGS, 4) = {0};
   DEFINE_ETNA_BITARRAY(component_use, 4 * ETNA_NUM_VARYINGS, 2) = {0};
   for (int idx = 0; idx < link.num_varyings; ++idx) {
      const struct etna_varying *varying = &link.varyings[idx];

      etna_bitarray_set(num_components, 4, idx, varying->num_components);
      for (int comp = 0; comp < varying->num_components; ++comp) {
         etna_bitarray_set(component_use, 2, total_components, varying->use[comp]);
         total_components += 1;
      }
   }

   cs->GL_VARYING_TOTAL_COMPONENTS =
      VIVS_GL_VARYING_TOTAL_COMPONENTS_NUM(align(total_components, 2));
   cs->GL_VARYING_NUM_COMPONENTS = num_components[0];
   cs->GL_VARYING_COMPONENT_USE[0] = component_use[0];
   cs->GL_VARYING_COMPONENT_USE[1] = component_use[1];

   /* reference instruction memory */
   cs->vs_inst_mem_size = vs->code_size;
   cs->VS_INST_MEM = vs->code;
   cs->ps_inst_mem_size = fs->code_size;
   cs->PS_INST_MEM = fs->code;

   return true;
}

bool
etna_shader_link(struct etna_context *ctx)
{
   if (!ctx->shader.vs || !ctx->shader.fs)
      return false;

   /* re-link vs and fs if needed */
   return etna_link_shaders(ctx, &ctx->shader_state, ctx->shader.vs, ctx->shader.fs);
}

static bool
etna_shader_update_vs_inputs(struct etna_context *ctx,
                             struct compiled_shader_state *cs,
                             const struct etna_shader_variant *vs,
                             const struct compiled_vertex_elements_state *ves)
{
   unsigned num_temps, cur_temp, num_vs_inputs;

   if (!vs)
      return false;

   /* Number of vertex elements determines number of VS inputs. Otherwise,
    * the GPU crashes. Allocate any unused vertex elements to VS temporary
    * registers. */
   num_vs_inputs = MAX2(ves->num_elements, vs->infile.num_reg);
   if (num_vs_inputs != ves->num_elements) {
      BUG("Number of elements %u does not match the number of VS inputs %zu",
          ctx->vertex_elements->num_elements, ctx->shader.vs->infile.num_reg);
      return false;
   }

   cur_temp = vs->num_temps;
   num_temps = num_vs_inputs - vs->infile.num_reg + cur_temp;

   cs->VS_INPUT_COUNT = VIVS_VS_INPUT_COUNT_COUNT(num_vs_inputs) |
                        VIVS_VS_INPUT_COUNT_UNK8(vs->input_count_unk8);
   cs->VS_TEMP_REGISTER_CONTROL =
      VIVS_VS_TEMP_REGISTER_CONTROL_NUM_TEMPS(num_temps);

   /* vs inputs (attributes) */
   DEFINE_ETNA_BITARRAY(vs_input, 16, 8) = {0};
   for (int idx = 0; idx < num_vs_inputs; ++idx) {
      if (idx < vs->infile.num_reg)
         etna_bitarray_set(vs_input, 8, idx, vs->infile.reg[idx].reg);
      else
         etna_bitarray_set(vs_input, 8, idx, cur_temp++);
   }

   for (int idx = 0; idx < ARRAY_SIZE(cs->VS_INPUT); ++idx)
      cs->VS_INPUT[idx] = vs_input[idx];

   return true;
}

static inline const char *
etna_shader_stage(struct etna_shader_variant *shader)
{
   switch (shader->processor) {
   case PIPE_SHADER_VERTEX:     return "VERT";
   case PIPE_SHADER_FRAGMENT:   return "FRAG";
   case PIPE_SHADER_COMPUTE:    return "CL";
   default:
      unreachable("invalid type");
      return NULL;
   }
}

static void
dump_shader_info(struct etna_shader_variant *v, struct pipe_debug_callback *debug)
{
   if (!unlikely(etna_mesa_debug & ETNA_DBG_SHADERDB))
      return;

   pipe_debug_message(debug, SHADER_INFO, "\n"
         "SHADER-DB: %s prog %d/%d: %u instructions %u temps\n"
         "SHADER-DB: %s prog %d/%d: %u immediates %u consts\n"
         "SHADER-DB: %s prog %d/%d: %u loops\n",
         etna_shader_stage(v),
         v->shader->id, v->id,
         v->code_size,
         v->num_temps,
         etna_shader_stage(v),
         v->shader->id, v->id,
         v->uniforms.imm_count,
         v->uniforms.const_count,
         etna_shader_stage(v),
         v->shader->id, v->id,
         v->num_loops);
}

bool
etna_shader_update_vertex(struct etna_context *ctx)
{
   return etna_shader_update_vs_inputs(ctx, &ctx->shader_state, ctx->shader.vs,
                                       ctx->vertex_elements);
}

static struct etna_shader_variant *
create_variant(struct etna_shader *shader, struct etna_shader_key key)
{
   struct etna_shader_variant *v = CALLOC_STRUCT(etna_shader_variant);
   int ret;

   if (!v)
      return NULL;

   v->shader = shader;
   v->key = key;

   ret = etna_compile_shader(v);
   if (!ret) {
      debug_error("compile failed!");
      goto fail;
   }

   v->id = ++shader->variant_count;

   return v;

fail:
   FREE(v);
   return NULL;
}

struct etna_shader_variant *
etna_shader_variant(struct etna_shader *shader, struct etna_shader_key key,
                   struct pipe_debug_callback *debug)
{
   struct etna_shader_variant *v;

   for (v = shader->variants; v; v = v->next)
      if (etna_shader_key_equal(&key, &v->key))
         return v;

   /* compile new variant if it doesn't exist already */
   v = create_variant(shader, key);
   if (v) {
      v->next = shader->variants;
      shader->variants = v;
      dump_shader_info(v, debug);
   }

   return v;
}

static void *
etna_create_shader_state(struct pipe_context *pctx,
                         const struct pipe_shader_state *pss)
{
   struct etna_context *ctx = etna_context(pctx);
   struct etna_shader *shader = CALLOC_STRUCT(etna_shader);

   if (!shader)
      return NULL;

   static uint32_t id;
   shader->id = id++;
   shader->specs = &ctx->specs;
   shader->tokens = tgsi_dup_tokens(pss->tokens);

   if (etna_mesa_debug & ETNA_DBG_SHADERDB) {
      /* if shader-db run, create a standard variant immediately
       * (as otherwise nothing will trigger the shader to be
       * actually compiled).
       */
      struct etna_shader_key key = {};
      etna_shader_variant(shader, key, &ctx->debug);
   }

   return shader;
}

static void
etna_delete_shader_state(struct pipe_context *pctx, void *ss)
{
   struct etna_shader *shader = ss;
   struct etna_shader_variant *v, *t;

   v = shader->variants;
   while (v) {
      t = v;
      v = v->next;
      etna_destroy_shader(t);
   }

   FREE(shader->tokens);
   FREE(shader);
}

static void
etna_bind_fs_state(struct pipe_context *pctx, void *hwcso)
{
   struct etna_context *ctx = etna_context(pctx);

   ctx->shader.bind_fs = hwcso;
   ctx->dirty |= ETNA_DIRTY_SHADER;
}

static void
etna_bind_vs_state(struct pipe_context *pctx, void *hwcso)
{
   struct etna_context *ctx = etna_context(pctx);

   ctx->shader.bind_vs = hwcso;
   ctx->dirty |= ETNA_DIRTY_SHADER;
}

void
etna_shader_init(struct pipe_context *pctx)
{
   pctx->create_fs_state = etna_create_shader_state;
   pctx->bind_fs_state = etna_bind_fs_state;
   pctx->delete_fs_state = etna_delete_shader_state;
   pctx->create_vs_state = etna_create_shader_state;
   pctx->bind_vs_state = etna_bind_vs_state;
   pctx->delete_vs_state = etna_delete_shader_state;
}