summaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/svga/svga_tgsi.c
blob: 2e2ff5e467381c6be96074ce3bc5bdb83228d9b2 (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
/**********************************************************
 * Copyright 2008-2009 VMware, Inc.  All rights reserved.
 *
 * 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 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.
 *
 **********************************************************/


#include "pipe/p_compiler.h"
#include "pipe/p_shader_tokens.h"
#include "pipe/p_defines.h"
#include "tgsi/tgsi_parse.h"
#include "tgsi/tgsi_dump.h"
#include "tgsi/tgsi_scan.h"
#include "util/u_math.h"
#include "util/u_memory.h"
#include "util/u_bitmask.h"

#include "svgadump/svga_shader_dump.h"

#include "svga_context.h"
#include "svga_tgsi.h"
#include "svga_tgsi_emit.h"
#include "svga_debug.h"

#include "svga_hw_reg.h"
#include "svga3d_shaderdefs.h"


/* Sinkhole used only in error conditions.
 */
static char err_buf[128];

#if 0
static void
svga_destroy_shader_emitter(struct svga_shader_emitter *emit)
{
   if (emit->buf != err_buf)
      FREE(emit->buf);
}
#endif


static boolean
svga_shader_expand(struct svga_shader_emitter *emit)
{
   char *new_buf;
   unsigned newsize = emit->size * 2;

   if (emit->buf != err_buf)
      new_buf = REALLOC(emit->buf, emit->size, newsize);
   else
      new_buf = NULL;

   if (new_buf == NULL) {
      emit->ptr = err_buf;
      emit->buf = err_buf;
      emit->size = sizeof(err_buf);
      return FALSE;
   }

   emit->size = newsize;
   emit->ptr = new_buf + (emit->ptr - emit->buf);
   emit->buf = new_buf;
   return TRUE;
}


static inline boolean
reserve(struct svga_shader_emitter *emit, unsigned nr_dwords)
{
   if (emit->ptr - emit->buf + nr_dwords * sizeof(unsigned) >= emit->size) {
      if (!svga_shader_expand(emit)) {
         return FALSE;
      }
   }

   return TRUE;
}


boolean
svga_shader_emit_dword(struct svga_shader_emitter * emit, unsigned dword)
{
   if (!reserve(emit, 1))
      return FALSE;

   *(unsigned *) emit->ptr = dword;
   emit->ptr += sizeof dword;
   return TRUE;
}


boolean
svga_shader_emit_dwords(struct svga_shader_emitter * emit,
                        const unsigned *dwords, unsigned nr)
{
   if (!reserve(emit, nr))
      return FALSE;

   memcpy(emit->ptr, dwords, nr * sizeof *dwords);
   emit->ptr += nr * sizeof *dwords;
   return TRUE;
}


boolean
svga_shader_emit_opcode(struct svga_shader_emitter * emit, unsigned opcode)
{
   SVGA3dShaderInstToken *here;

   if (!reserve(emit, 1))
      return FALSE;

   here = (SVGA3dShaderInstToken *) emit->ptr;
   here->value = opcode;

   if (emit->insn_offset) {
      SVGA3dShaderInstToken *prev =
         (SVGA3dShaderInstToken *) (emit->buf + emit->insn_offset);
      prev->size = (here - prev) - 1;
   }

   emit->insn_offset = emit->ptr - emit->buf;
   emit->ptr += sizeof(unsigned);
   return TRUE;
}


static boolean
svga_shader_emit_header(struct svga_shader_emitter *emit)
{
   SVGA3dShaderVersion header;

   memset(&header, 0, sizeof header);

   switch (emit->unit) {
   case PIPE_SHADER_FRAGMENT:
      header.value = SVGA3D_PS_30;
      break;
   case PIPE_SHADER_VERTEX:
      header.value = SVGA3D_VS_30;
      break;
   }

   return svga_shader_emit_dword(emit, header.value);
}


/**
 * Use the shader info to generate a bitmask indicating which generic
 * inputs are used by the shader.  A set bit indicates that GENERIC[i]
 * is used.
 */
unsigned
svga_get_generic_inputs_mask(const struct tgsi_shader_info *info)
{
   unsigned i, mask = 0x0;

   for (i = 0; i < info->num_inputs; i++) {
      if (info->input_semantic_name[i] == TGSI_SEMANTIC_GENERIC) {
         unsigned j = info->input_semantic_index[i];
         assert(j < sizeof(mask) * 8);
         mask |= 1 << j;
      }
   }

   return mask;
}


/**
 * Given a mask of used generic variables (as returned by the above functions)
 * fill in a table which maps those indexes to small integers.
 * This table is used by the remap_generic_index() function in
 * svga_tgsi_decl_sm30.c
 * Example: if generics_mask = binary(1010) it means that GENERIC[1] and
 * GENERIC[3] are used.  The remap_table will contain:
 *   table[1] = 0;
 *   table[3] = 1;
 * The remaining table entries will be filled in with the next unused
 * generic index (in this example, 2).
 */
void
svga_remap_generics(unsigned generics_mask,
                    int8_t remap_table[MAX_GENERIC_VARYING])
{
   /* Note texcoord[0] is reserved so start at 1 */
   unsigned count = 1, i;

   for (i = 0; i < MAX_GENERIC_VARYING; i++) {
      remap_table[i] = -1;
   }

   /* for each bit set in generic_mask */
   while (generics_mask) {
      unsigned index = ffs(generics_mask) - 1;
      remap_table[index] = count++;
      generics_mask &= ~(1 << index);
   }
}


/**
 * Use the generic remap table to map a TGSI generic varying variable
 * index to a small integer.  If the remapping table doesn't have a
 * valid value for the given index (the table entry is -1) it means
 * the fragment shader doesn't use that VS output.  Just allocate
 * the next free value in that case.  Alternately, we could cull
 * VS instructions that write to register, or replace the register
 * with a dummy temp register.
 * XXX TODO: we should do one of the later as it would save precious
 * texcoord registers.
 */
int
svga_remap_generic_index(int8_t remap_table[MAX_GENERIC_VARYING],
                         int generic_index)
{
   assert(generic_index < MAX_GENERIC_VARYING);

   if (generic_index >= MAX_GENERIC_VARYING) {
      /* just don't return a random/garbage value */
      generic_index = MAX_GENERIC_VARYING - 1;
   }

   if (remap_table[generic_index] == -1) {
      /* This is a VS output that has no matching PS input.  Find a
       * free index.
       */
      int i, max = 0;
      for (i = 0; i < MAX_GENERIC_VARYING; i++) {
         max = MAX2(max, remap_table[i]);
      }
      remap_table[generic_index] = max + 1;
   }

   return remap_table[generic_index];
}


/**
 * Parse TGSI shader and translate to SVGA/DX9 serialized
 * representation.
 *
 * In this function SVGA shader is emitted to an in-memory buffer that
 * can be dynamically grown.  Once we've finished and know how large
 * it is, it will be copied to a hardware buffer for upload.
 */
static struct svga_shader_variant *
svga_tgsi_translate(const struct svga_shader *shader,
                    const struct svga_compile_key *key, unsigned unit)
{
   struct svga_shader_variant *variant = NULL;
   struct svga_shader_emitter emit;

   memset(&emit, 0, sizeof(emit));

   emit.size = 1024;
   emit.buf = MALLOC(emit.size);
   if (emit.buf == NULL) {
      goto fail;
   }

   emit.ptr = emit.buf;
   emit.unit = unit;
   emit.key = *key;

   tgsi_scan_shader(shader->tokens, &emit.info);

   emit.imm_start = emit.info.file_max[TGSI_FILE_CONSTANT] + 1;

   if (unit == PIPE_SHADER_FRAGMENT)
      emit.imm_start += key->fkey.num_unnormalized_coords;

   if (unit == PIPE_SHADER_VERTEX) {
      emit.imm_start += key->vkey.need_prescale ? 2 : 0;
   }

   emit.nr_hw_float_const =
      (emit.imm_start + emit.info.file_max[TGSI_FILE_IMMEDIATE] + 1);

   emit.nr_hw_temp = emit.info.file_max[TGSI_FILE_TEMPORARY] + 1;

   if (emit.nr_hw_temp >= SVGA3D_TEMPREG_MAX) {
      debug_printf("svga: too many temporary registers (%u)\n",
                   emit.nr_hw_temp);
      goto fail;
   }

   emit.in_main_func = TRUE;

   if (!svga_shader_emit_header(&emit)) {
      debug_printf("svga: emit header failed\n");
      goto fail;
   }

   if (!svga_shader_emit_instructions(&emit, shader->tokens)) {
      debug_printf("svga: emit instructions failed\n");
      goto fail;
   }

   variant = CALLOC_STRUCT(svga_shader_variant);
   if (variant == NULL)
      goto fail;

   variant->shader = shader;
   variant->tokens = (const unsigned *) emit.buf;
   variant->nr_tokens = (emit.ptr - emit.buf) / sizeof(unsigned);
   memcpy(&variant->key, key, sizeof(*key));
   variant->id = UTIL_BITMASK_INVALID_INDEX;

   if (SVGA_DEBUG & DEBUG_TGSI) {
      debug_printf("#####################################\n");
      debug_printf("Shader %u below\n", shader->id);
      tgsi_dump(shader->tokens, 0);
      if (SVGA_DEBUG & DEBUG_TGSI) {
         debug_printf("Shader %u compiled below\n", shader->id);
         svga_shader_dump(variant->tokens, variant->nr_tokens, FALSE);
      }
      debug_printf("#####################################\n");
   }

   return variant;

 fail:
   FREE(variant);
   FREE(emit.buf);
   return NULL;
}


struct svga_shader_variant *
svga_translate_fragment_program(const struct svga_fragment_shader *fs,
                                const struct svga_fs_compile_key *fkey)
{
   struct svga_compile_key key;

   memset(&key, 0, sizeof(key));

   memcpy(&key.fkey, fkey, sizeof *fkey);

   memcpy(key.generic_remap_table, fs->generic_remap_table,
          sizeof(fs->generic_remap_table));

   return svga_tgsi_translate(&fs->base, &key, PIPE_SHADER_FRAGMENT);
}


struct svga_shader_variant *
svga_translate_vertex_program(const struct svga_vertex_shader *vs,
                              const struct svga_vs_compile_key *vkey)
{
   struct svga_compile_key key;

   memset(&key, 0, sizeof(key));

   memcpy(&key.vkey, vkey, sizeof *vkey);

   /* Note: we could alternately store the remap table in the vkey but
    * that would make it larger.  We just regenerate it here instead.
    */
   svga_remap_generics(vkey->fs_generic_inputs, key.generic_remap_table);

   return svga_tgsi_translate(&vs->base, &key, PIPE_SHADER_VERTEX);
}