summaryrefslogtreecommitdiff
path: root/src/mesa/drivers/dri/i965/brw_fs_fp.cpp
blob: 834f290a31547789191c9901807e982018c5e948 (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
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
/*
 * Copyright © 2012 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_fs_fp.cpp
 *
 * Implementation of the compiler for GL_ARB_fragment_program shaders on top
 * of the GLSL compiler backend.
 */

#include "brw_context.h"
#include "brw_fs.h"

static fs_reg
regoffset(fs_reg reg, int i)
{
   reg.reg_offset += i;
   return reg;
}

void
fs_visitor::emit_fp_alu1(enum opcode opcode,
                         const struct prog_instruction *fpi,
                         fs_reg dst, fs_reg src)
{
   for (int i = 0; i < 4; i++) {
      if (fpi->DstReg.WriteMask & (1 << i))
         emit(opcode, regoffset(dst, i), regoffset(src, i));
   }
}

void
fs_visitor::emit_fp_alu2(enum opcode opcode,
                         const struct prog_instruction *fpi,
                         fs_reg dst, fs_reg src0, fs_reg src1)
{
   for (int i = 0; i < 4; i++) {
      if (fpi->DstReg.WriteMask & (1 << i))
         emit(opcode, regoffset(dst, i),
              regoffset(src0, i), regoffset(src1, i));
   }
}

void
fs_visitor::emit_fp_minmax(const prog_instruction *fpi,
                           fs_reg dst, fs_reg src0, fs_reg src1)
{
   uint32_t conditionalmod;
   if (fpi->Opcode == OPCODE_MIN)
      conditionalmod = BRW_CONDITIONAL_L;
   else
      conditionalmod = BRW_CONDITIONAL_GE;

   for (int i = 0; i < 4; i++) {
      if (fpi->DstReg.WriteMask & (1 << i)) {
         emit_minmax(conditionalmod, regoffset(dst, i),
                     regoffset(src0, i), regoffset(src1, i));
      }
   }
}

void
fs_visitor::emit_fp_sop(uint32_t conditional_mod,
                        const struct prog_instruction *fpi,
                        fs_reg dst, fs_reg src0, fs_reg src1,
                        fs_reg one)
{
   for (int i = 0; i < 4; i++) {
      if (fpi->DstReg.WriteMask & (1 << i)) {
         fs_inst *inst;

         emit(CMP(reg_null_d, regoffset(src0, i), regoffset(src1, i),
                  conditional_mod));

         inst = emit(BRW_OPCODE_SEL, regoffset(dst, i), one, fs_reg(0.0f));
         inst->predicate = BRW_PREDICATE_NORMAL;
      }
   }
}

void
fs_visitor::emit_fp_scalar_write(const struct prog_instruction *fpi,
                                 fs_reg dst, fs_reg src)
{
   for (int i = 0; i < 4; i++) {
      if (fpi->DstReg.WriteMask & (1 << i))
         emit(MOV(regoffset(dst, i), src));
   }
}

void
fs_visitor::emit_fp_scalar_math(enum opcode opcode,
                                const struct prog_instruction *fpi,
                                fs_reg dst, fs_reg src)
{
   fs_reg temp = fs_reg(this, glsl_type::float_type);
   emit_math(opcode, temp, src);
   emit_fp_scalar_write(fpi, dst, temp);
}

void
fs_visitor::emit_fragment_program_code()
{
   setup_fp_regs();

   fs_reg null = fs_reg(brw_null_reg());

   /* Keep a reg with 1.0 around, for reuse by emit_fp_sop so that it can just
    * be:
    *
    * sel.f0 dst 1.0 0.0
    *
    * instead of
    *
    * mov    dst 0.0
    * mov.f0 dst 1.0
    */
   fs_reg one = fs_reg(this, glsl_type::float_type);
   emit(MOV(one, fs_reg(1.0f)));

   for (unsigned int insn = 0; insn < fp->Base.NumInstructions; insn++) {
      const struct prog_instruction *fpi = &fp->Base.Instructions[insn];
      base_ir = fpi;

      //_mesa_print_instruction(fpi);

      fs_reg dst;
      fs_reg src[3];

      /* We always emit into a temporary destination register to avoid
       * aliasing issues.
       */
      dst = fs_reg(this, glsl_type::vec4_type);

      for (int i = 0; i < 3; i++)
         src[i] = get_fp_src_reg(&fpi->SrcReg[i]);

      switch (fpi->Opcode) {
      case OPCODE_ABS:
         src[0].abs = true;
         src[0].negate = false;
         emit_fp_alu1(BRW_OPCODE_MOV, fpi, dst, src[0]);
         break;

      case OPCODE_ADD:
         emit_fp_alu2(BRW_OPCODE_ADD, fpi, dst, src[0], src[1]);
         break;

      case OPCODE_CMP:
         for (int i = 0; i < 4; i++) {
            if (fpi->DstReg.WriteMask & (1 << i)) {
               fs_inst *inst;

               emit(CMP(null, regoffset(src[0], i), fs_reg(0.0f),
                        BRW_CONDITIONAL_L));

               inst = emit(BRW_OPCODE_SEL, regoffset(dst, i),
                           regoffset(src[1], i), regoffset(src[2], i));
               inst->predicate = BRW_PREDICATE_NORMAL;
            }
         }
         break;

      case OPCODE_COS:
         emit_fp_scalar_math(SHADER_OPCODE_COS, fpi, dst, src[0]);
         break;

      case OPCODE_DP2:
      case OPCODE_DP3:
      case OPCODE_DP4:
      case OPCODE_DPH: {
         fs_reg mul = fs_reg(this, glsl_type::float_type);
         fs_reg acc = fs_reg(this, glsl_type::float_type);
         int count;

         switch (fpi->Opcode) {
         case OPCODE_DP2: count = 2; break;
         case OPCODE_DP3: count = 3; break;
         case OPCODE_DP4: count = 4; break;
         case OPCODE_DPH: count = 3; break;
         default: assert(!"not reached"); count = 0; break;
         }

         emit(MUL(acc, regoffset(src[0], 0), regoffset(src[1], 0)));
         for (int i = 1; i < count; i++) {
            emit(MUL(mul, regoffset(src[0], i), regoffset(src[1], i)));
            emit(ADD(acc, acc, mul));
         }

         if (fpi->Opcode == OPCODE_DPH)
            emit(ADD(acc, acc, regoffset(src[1], 3)));

         emit_fp_scalar_write(fpi, dst, acc);
         break;
      }

      case OPCODE_DST:
         if (fpi->DstReg.WriteMask & WRITEMASK_X)
            emit(MOV(dst, fs_reg(1.0f)));
         if (fpi->DstReg.WriteMask & WRITEMASK_Y) {
            emit(MUL(regoffset(dst, 1),
                     regoffset(src[0], 1), regoffset(src[1], 1)));
         }
         if (fpi->DstReg.WriteMask & WRITEMASK_Z)
            emit(MOV(regoffset(dst, 2), regoffset(src[0], 2)));
         if (fpi->DstReg.WriteMask & WRITEMASK_W)
            emit(MOV(regoffset(dst, 3), regoffset(src[1], 3)));
         break;

      case OPCODE_EX2:
         emit_fp_scalar_math(SHADER_OPCODE_EXP2, fpi, dst, src[0]);
         break;

      case OPCODE_FLR:
         emit_fp_alu1(BRW_OPCODE_RNDD, fpi, dst, src[0]);
         break;

      case OPCODE_FRC:
         emit_fp_alu1(BRW_OPCODE_FRC, fpi, dst, src[0]);
         break;

      case OPCODE_KIL: {
         for (int i = 0; i < 4; i++) {
            /* In most cases the argument to a KIL will be something like
             * TEMP[0].wwww, so there's no point in checking whether .w is < 0
             * 4 times in a row.
             */
            if (i > 0 &&
                GET_SWZ(fpi->SrcReg[0].Swizzle, i) ==
                GET_SWZ(fpi->SrcReg[0].Swizzle, i - 1) &&
                ((fpi->SrcReg[0].Negate >> i) & 1) ==
                ((fpi->SrcReg[0].Negate >> (i - 1)) & 1)) {
               continue;
            }

            emit(CMP(null, regoffset(src[0], i), fs_reg(0.0f),
                     BRW_CONDITIONAL_L));

            if (intel->gen < 6 && c->dispatch_width == 16)
               fail("Can't support (non-uniform) control flow on 16-wide");
            emit(IF(BRW_PREDICATE_NORMAL));
            emit(FS_OPCODE_DISCARD);
            emit(BRW_OPCODE_ENDIF);
         }
         break;
      }

      case OPCODE_LG2:
         emit_fp_scalar_math(SHADER_OPCODE_LOG2, fpi, dst, src[0]);
         break;

      case OPCODE_LIT:
         /* From the ARB_fragment_program spec:
          *
          *      tmp = VectorLoad(op0);
          *      if (tmp.x < 0) tmp.x = 0;
          *      if (tmp.y < 0) tmp.y = 0;
          *      if (tmp.w < -(128.0-epsilon)) tmp.w = -(128.0-epsilon);
          *      else if (tmp.w > 128-epsilon) tmp.w = 128-epsilon;
          *      result.x = 1.0;
          *      result.y = tmp.x;
          *      result.z = (tmp.x > 0) ? RoughApproxPower(tmp.y, tmp.w) : 0.0;
          *      result.w = 1.0;
          *
          * Note that we don't do the clamping to +/- 128.  We didn't in
          * brw_wm_emit.c either.
          */
         if (fpi->DstReg.WriteMask & WRITEMASK_X)
            emit(MOV(regoffset(dst, 0), fs_reg(1.0f)));

         if (fpi->DstReg.WriteMask & WRITEMASK_YZ) {
            fs_inst *inst;
            emit(CMP(null, regoffset(src[0], 0), fs_reg(0.0f),
                     BRW_CONDITIONAL_LE));

            if (fpi->DstReg.WriteMask & WRITEMASK_Y) {
               emit(MOV(regoffset(dst, 1), regoffset(src[0], 0)));
               inst = emit(MOV(regoffset(dst, 1), fs_reg(0.0f)));
               inst->predicate = BRW_PREDICATE_NORMAL;
            }

            if (fpi->DstReg.WriteMask & WRITEMASK_Z) {
               emit_math(SHADER_OPCODE_POW, regoffset(dst, 2),
                         regoffset(src[0], 1), regoffset(src[0], 3));

               inst = emit(MOV(regoffset(dst, 2), fs_reg(0.0f)));
               inst->predicate = BRW_PREDICATE_NORMAL;
            }
         }

         if (fpi->DstReg.WriteMask & WRITEMASK_W)
            emit(MOV(regoffset(dst, 3), fs_reg(1.0f)));

         break;

      case OPCODE_LRP:
         for (int i = 0; i < 4; i++) {
            if (fpi->DstReg.WriteMask & (1 << i)) {
               fs_reg neg_src0 = regoffset(src[0], i);
               neg_src0.negate = !neg_src0.negate;
               fs_reg temp = fs_reg(this, glsl_type::float_type);
               fs_reg temp2 = fs_reg(this, glsl_type::float_type);
               emit(ADD(temp, neg_src0, fs_reg(1.0f)));
               emit(MUL(temp, temp, regoffset(src[2], i)));
               emit(MUL(temp2, regoffset(src[0], i), regoffset(src[1], i)));
               emit(ADD(regoffset(dst, i), temp, temp2));
            }
         }
         break;

      case OPCODE_MAD:
         for (int i = 0; i < 4; i++) {
            if (fpi->DstReg.WriteMask & (1 << i)) {
               fs_reg temp = fs_reg(this, glsl_type::float_type);
               emit(MUL(temp, regoffset(src[0], i), regoffset(src[1], i)));
               emit(ADD(regoffset(dst, i), temp, regoffset(src[2], i)));
            }
         }
         break;

      case OPCODE_MAX:
         emit_fp_minmax(fpi, dst, src[0], src[1]);
         break;

      case OPCODE_MOV:
         emit_fp_alu1(BRW_OPCODE_MOV, fpi, dst, src[0]);
         break;

      case OPCODE_MIN:
         emit_fp_minmax(fpi, dst, src[0], src[1]);
         break;

      case OPCODE_MUL:
         emit_fp_alu2(BRW_OPCODE_MUL, fpi, dst, src[0], src[1]);
         break;

      case OPCODE_POW: {
         fs_reg temp = fs_reg(this, glsl_type::float_type);
         emit_math(SHADER_OPCODE_POW, temp, src[0], src[1]);
         emit_fp_scalar_write(fpi, dst, temp);
         break;
      }

      case OPCODE_RCP:
         emit_fp_scalar_math(SHADER_OPCODE_RCP, fpi, dst, src[0]);
         break;

      case OPCODE_RSQ:
         emit_fp_scalar_math(SHADER_OPCODE_RSQ, fpi, dst, src[0]);
         break;

      case OPCODE_SCS:
         if (fpi->DstReg.WriteMask & WRITEMASK_X) {
            emit_math(SHADER_OPCODE_COS, regoffset(dst, 0),
                      regoffset(src[0], 0));
         }

         if (fpi->DstReg.WriteMask & WRITEMASK_Y) {
            emit_math(SHADER_OPCODE_SIN, regoffset(dst, 1),
                      regoffset(src[0], 1));
         }
         break;

      case OPCODE_SGE:
         emit_fp_sop(BRW_CONDITIONAL_GE, fpi, dst, src[0], src[1], one);
         break;

      case OPCODE_SIN:
         emit_fp_scalar_math(SHADER_OPCODE_SIN, fpi, dst, src[0]);
         break;

      case OPCODE_SLT:
         emit_fp_sop(BRW_CONDITIONAL_L, fpi, dst, src[0], src[1], one);
         break;

      case OPCODE_SUB: {
         fs_reg neg_src1 = src[1];
         neg_src1.negate = !src[1].negate;

         emit_fp_alu2(BRW_OPCODE_ADD, fpi, dst, src[0], neg_src1);
         break;
      }

      case OPCODE_TEX:
      case OPCODE_TXB:
      case OPCODE_TXP: {
         /* We piggy-back on the GLSL IR support for texture setup.  To do so,
          * we have to cook up an ir_texture that has the coordinate field
          * with appropriate type, and shadow_comparitor set or not.  All the
          * other properties of ir_texture are passed in as arguments to the
          * emit_texture_gen* function.
          */
         ir_texture *ir = NULL;

         fs_reg lod;
         fs_reg dpdy;
         fs_reg coordinate = src[0];
         fs_reg shadow_c;

         switch (fpi->Opcode) {
         case OPCODE_TEX:
            ir = new(mem_ctx) ir_texture(ir_tex);
            break;
         case OPCODE_TXP: {
            ir = new(mem_ctx) ir_texture(ir_tex);

            coordinate = fs_reg(this, glsl_type::vec3_type);
            fs_reg invproj = fs_reg(this, glsl_type::float_type);
            emit_math(SHADER_OPCODE_RCP, invproj, regoffset(src[0], 3));
            for (int i = 0; i < 3; i++) {
               emit(MUL(regoffset(coordinate, i),
                        regoffset(src[0], i), invproj));
            }
            break;
         }
         case OPCODE_TXB:
            ir = new(mem_ctx) ir_texture(ir_txb);
            lod = regoffset(src[0], 3);
            break;
         default:
            assert(!"not reached");
            break;
         }

         const glsl_type *coordinate_type;
         switch (fpi->TexSrcTarget) {
         case TEXTURE_1D_INDEX:
            coordinate_type = glsl_type::float_type;
            break;

         case TEXTURE_2D_INDEX:
         case TEXTURE_1D_ARRAY_INDEX:
         case TEXTURE_RECT_INDEX:
         case TEXTURE_EXTERNAL_INDEX:
            coordinate_type = glsl_type::vec2_type;
            break;

         case TEXTURE_3D_INDEX:
         case TEXTURE_2D_ARRAY_INDEX:
            coordinate_type = glsl_type::vec3_type;
            break;

         case TEXTURE_CUBE_INDEX: {
            coordinate_type = glsl_type::vec3_type;

            fs_reg temp = fs_reg(this, glsl_type::float_type);
            fs_reg cubecoord = fs_reg(this, glsl_type::vec3_type);
            fs_reg abscoord = coordinate;
            abscoord.negate = false;
            abscoord.abs = true;
            emit_minmax(BRW_CONDITIONAL_GE, temp,
                        regoffset(abscoord, 0), regoffset(abscoord, 1));
            emit_minmax(BRW_CONDITIONAL_GE, temp,
                        temp, regoffset(abscoord, 2));
            emit_math(SHADER_OPCODE_RCP, temp, temp);
            for (int i = 0; i < 3; i++) {
               emit(MUL(regoffset(cubecoord, i),
                        regoffset(coordinate, i), temp));
            }

            coordinate = cubecoord;
            break;
         }

         default:
            assert(!"not reached");
            coordinate_type = glsl_type::vec2_type;
            break;
         }

         ir_constant_data junk_data;
         ir->coordinate = new(mem_ctx) ir_constant(coordinate_type, &junk_data);

         coordinate = rescale_texcoord(ir, coordinate,
                                       fpi->TexSrcTarget == TEXTURE_RECT_INDEX,
                                       fpi->TexSrcUnit, fpi->TexSrcUnit);

         if (fpi->TexShadow) {
            shadow_c = regoffset(coordinate, 2);
            ir->shadow_comparitor = new(mem_ctx) ir_constant(0.0f);
         }

         fs_inst *inst;
         if (intel->gen >= 7) {
            inst = emit_texture_gen7(ir, dst, coordinate, shadow_c, lod, dpdy);
         } else if (intel->gen >= 5) {
            inst = emit_texture_gen5(ir, dst, coordinate, shadow_c, lod, dpdy);
         } else {
            inst = emit_texture_gen4(ir, dst, coordinate, shadow_c, lod, dpdy);
         }

         inst->sampler = fpi->TexSrcUnit;
         inst->shadow_compare = fpi->TexShadow;

         /* Reuse the GLSL swizzle_result() handler. */
         swizzle_result(ir, dst, fpi->TexSrcUnit);
         dst = this->result;

         break;
      }

      case OPCODE_SWZ:
         /* Note that SWZ's extended swizzles are handled in the general
          * get_src_reg() code.
          */
         emit_fp_alu1(BRW_OPCODE_MOV, fpi, dst, src[0]);
         break;

      case OPCODE_XPD:
         for (int i = 0; i < 3; i++) {
            if (fpi->DstReg.WriteMask & (1 << i)) {
               int i1 = (i + 1) % 3;
               int i2 = (i + 2) % 3;

               fs_reg temp = fs_reg(this, glsl_type::float_type);
               fs_reg neg_src1_1 = regoffset(src[1], i1);
               neg_src1_1.negate = !neg_src1_1.negate;
               emit(MUL(temp, regoffset(src[0], i2), neg_src1_1));
               emit(MUL(regoffset(dst, i),
                        regoffset(src[0], i1), regoffset(src[1], i2)));
               emit(ADD(regoffset(dst, i), regoffset(dst, i), temp));
            }
         }
         break;

      case OPCODE_END:
         break;

      default:
         _mesa_problem(ctx, "Unsupported opcode %s in fragment program\n",
                       _mesa_opcode_string(fpi->Opcode));
      }

      /* To handle saturates, we emit a MOV with a saturate bit, which
       * optimization should fold into the preceding instructions when safe.
       */
      if (fpi->Opcode != OPCODE_END) {
         fs_reg real_dst = get_fp_dst_reg(&fpi->DstReg);

         for (int i = 0; i < 4; i++) {
            if (fpi->DstReg.WriteMask & (1 << i)) {
               fs_inst *inst = emit(MOV(regoffset(real_dst, i),
                                        regoffset(dst, i)));
               inst->saturate = fpi->SaturateMode;
            }
         }
      }
   }

   /* Epilogue:
    *
    * Fragment depth has this strange convention of being the .z component of
    * a vec4.  emit_fb_write() wants to see a float value, instead.
    */
   this->current_annotation = "result.depth write";
   if (frag_depth.file != BAD_FILE) {
      fs_reg temp = fs_reg(this, glsl_type::float_type);
      emit(MOV(temp, regoffset(frag_depth, 2)));
      frag_depth = temp;
   }
}

void
fs_visitor::setup_fp_regs()
{
   /* PROGRAM_TEMPORARY */
   int num_temp = fp->Base.NumTemporaries;
   fp_temp_regs = rzalloc_array(mem_ctx, fs_reg, num_temp);
   for (int i = 0; i < num_temp; i++)
      fp_temp_regs[i] = fs_reg(this, glsl_type::vec4_type);

   /* PROGRAM_STATE_VAR etc. */
   if (c->dispatch_width == 8) {
      for (unsigned p = 0;
           p < c->fp->program.Base.Parameters->NumParameters; p++) {
         for (unsigned int i = 0; i < 4; i++) {
            this->param_index[c->prog_data.nr_params] = p;
            this->param_offset[c->prog_data.nr_params] = i;
            c->prog_data.nr_params++;
         }
      }
   }

   fp_input_regs = rzalloc_array(mem_ctx, fs_reg, FRAG_ATTRIB_MAX);
   for (int i = 0; i < FRAG_ATTRIB_MAX; i++) {
      if (fp->Base.InputsRead & BITFIELD64_BIT(i)) {
         /* Make up a dummy instruction to reuse code for emitting
          * interpolation.
          */
         ir_variable *ir = new(mem_ctx) ir_variable(glsl_type::vec4_type,
                                                    "fp_input",
                                                    ir_var_in);
         ir->location = i;

         this->current_annotation = ralloc_asprintf(ctx, "interpolate input %d",
                                                    i);

         switch (i) {
         case FRAG_ATTRIB_WPOS:
            ir->pixel_center_integer = fp->PixelCenterInteger;
            ir->origin_upper_left = fp->OriginUpperLeft;
            fp_input_regs[i] = *emit_fragcoord_interpolation(ir);
            break;
         case FRAG_ATTRIB_FACE:
            fp_input_regs[i] = *emit_frontfacing_interpolation(ir);
            break;
         default:
            fp_input_regs[i] = *emit_general_interpolation(ir);

            if (i == FRAG_ATTRIB_FOGC) {
               emit(MOV(regoffset(fp_input_regs[i], 1), fs_reg(0.0f)));
               emit(MOV(regoffset(fp_input_regs[i], 2), fs_reg(0.0f)));
               emit(MOV(regoffset(fp_input_regs[i], 3), fs_reg(1.0f)));
            }

            break;
         }

         this->current_annotation = NULL;
      }
   }
}

fs_reg
fs_visitor::get_fp_dst_reg(const prog_dst_register *dst)
{
   switch (dst->File) {
   case PROGRAM_TEMPORARY:
      return fp_temp_regs[dst->Index];

   case PROGRAM_OUTPUT:
      if (dst->Index == FRAG_RESULT_DEPTH) {
         if (frag_depth.file == BAD_FILE)
            frag_depth = fs_reg(this, glsl_type::vec4_type);
         return frag_depth;
      } else if (dst->Index == FRAG_RESULT_COLOR) {
         if (outputs[0].file == BAD_FILE) {
            outputs[0] = fs_reg(this, glsl_type::vec4_type);
            output_components[0] = 4;

            /* Tell emit_fb_writes() to smear fragment.color across all the
             * color attachments.
             */
            for (int i = 1; i < c->key.nr_color_regions; i++) {
               outputs[i] = outputs[0];
               output_components[i] = output_components[0];
            }
         }
         return outputs[0];
      } else {
         int output_index = dst->Index - FRAG_RESULT_DATA0;
         if (outputs[output_index].file == BAD_FILE) {
            outputs[output_index] = fs_reg(this, glsl_type::vec4_type);
         }
         output_components[output_index] = 4;
         return outputs[output_index];
      }

   case PROGRAM_UNDEFINED:
      return fs_reg();

   default:
      _mesa_problem(ctx, "bad dst register file: %s\n",
                    _mesa_register_file_name((gl_register_file)dst->File));
      return fs_reg(this, glsl_type::vec4_type);
   }
}

fs_reg
fs_visitor::get_fp_src_reg(const prog_src_register *src)
{
   struct gl_program_parameter_list *plist = c->fp->program.Base.Parameters;

   fs_reg result;

   assert(!src->Abs);

   switch (src->File) {
   case PROGRAM_UNDEFINED:
      return fs_reg();
   case PROGRAM_TEMPORARY:
      result = fp_temp_regs[src->Index];
      break;

   case PROGRAM_INPUT:
      result = fp_input_regs[src->Index];
      break;

   case PROGRAM_STATE_VAR:
   case PROGRAM_UNIFORM:
   case PROGRAM_CONSTANT:
      /* We actually want to look at the type in the Parameters list for this,
       * because this lets us upload constant builtin uniforms, as actual
       * constants.
       */
      switch (plist->Parameters[src->Index].Type) {
      case PROGRAM_CONSTANT: {
         result = fs_reg(this, glsl_type::vec4_type);

         for (int i = 0; i < 4; i++) {
            emit(MOV(regoffset(result, i),
                     fs_reg(plist->ParameterValues[src->Index][i].f)));
         }
         break;
      }

      case PROGRAM_STATE_VAR:
      case PROGRAM_UNIFORM:
         result = fs_reg(UNIFORM, src->Index * 4);
         break;

      default:
         _mesa_problem(ctx, "bad uniform src register file: %s\n",
                       _mesa_register_file_name((gl_register_file)src->File));
         return fs_reg(this, glsl_type::vec4_type);
      }
      break;

   default:
      _mesa_problem(ctx, "bad src register file: %s\n",
                    _mesa_register_file_name((gl_register_file)src->File));
      return fs_reg(this, glsl_type::vec4_type);
   }

   if (src->Swizzle != SWIZZLE_NOOP || src->Negate) {
      fs_reg unswizzled = result;
      result = fs_reg(this, glsl_type::vec4_type);
      for (int i = 0; i < 4; i++) {
         bool negate = src->Negate & (1 << i);
         /* The ZERO, ONE, and Negate options are only used for OPCODE_SWZ,
          * but it costs us nothing to support it.
          */
         int src_swiz = GET_SWZ(src->Swizzle, i);
         if (src_swiz == SWIZZLE_ZERO) {
            emit(MOV(regoffset(result, i), fs_reg(0.0f)));
         } else if (src_swiz == SWIZZLE_ONE) {
            emit(MOV(regoffset(result, i),
                     negate ? fs_reg(-1.0f) : fs_reg(1.0f)));
         } else {
            fs_reg src = regoffset(unswizzled, src_swiz);
            if (negate)
               src.negate = !src.negate;
            emit(MOV(regoffset(result, i), src));
         }
      }
   }

   return result;
}