summaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/radeon/MCTargetDesc/R600MCCodeEmitter.cpp
blob: dcf833876cee7feda1204cc40c5d6a9ae1d06cbf (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
//===- R600MCCodeEmitter.cpp - Code Emitter for R600->Cayman GPU families -===//
//
//                     The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This code emitters outputs bytecode that is understood by the r600g driver
// in the Mesa [1] project.  The bytecode is very similar to the hardware's ISA,
// except that the size of the instruction fields are rounded up to the
// nearest byte.
//
// [1] http://www.mesa3d.org/
//
//===----------------------------------------------------------------------===//

#include "R600Defines.h"
#include "MCTargetDesc/AMDGPUMCTargetDesc.h"
#include "MCTargetDesc/AMDGPUMCCodeEmitter.h"
#include "llvm/MC/MCCodeEmitter.h"
#include "llvm/MC/MCContext.h"
#include "llvm/MC/MCInst.h"
#include "llvm/MC/MCInstrInfo.h"
#include "llvm/MC/MCRegisterInfo.h"
#include "llvm/MC/MCSubtargetInfo.h"
#include "llvm/Support/raw_ostream.h"

#include <stdio.h>

#define SRC_BYTE_COUNT 11
#define DST_BYTE_COUNT 5

using namespace llvm;

namespace {

class R600MCCodeEmitter : public AMDGPUMCCodeEmitter {
  R600MCCodeEmitter(const R600MCCodeEmitter &); // DO NOT IMPLEMENT
  void operator=(const R600MCCodeEmitter &); // DO NOT IMPLEMENT
  const MCInstrInfo &MCII;
  const MCSubtargetInfo &STI;
  MCContext &Ctx;

public:

  R600MCCodeEmitter(const MCInstrInfo &mcii, const MCSubtargetInfo &sti,
                    MCContext &ctx)
    : MCII(mcii), STI(sti), Ctx(ctx) { }

  /// EncodeInstruction - Encode the instruction and write it to the OS.
  virtual void EncodeInstruction(const MCInst &MI, raw_ostream &OS,
                         SmallVectorImpl<MCFixup> &Fixups) const;

  /// getMachineOpValue - Reutrn the encoding for an MCOperand.
  virtual uint64_t getMachineOpValue(const MCInst &MI, const MCOperand &MO,
                                     SmallVectorImpl<MCFixup> &Fixups) const;
private:

  void EmitALUInstr(const MCInst &MI, SmallVectorImpl<MCFixup> &Fixups,
                    raw_ostream &OS) const;
  void EmitSrc(const MCInst &MI, unsigned OpIdx, raw_ostream &OS) const;
  void EmitSrcISA(const MCInst &MI, unsigned OpIdx, uint64_t &Value,
                  raw_ostream &OS) const;
  void EmitDst(const MCInst &MI, raw_ostream &OS) const;
  void EmitTexInstr(const MCInst &MI, SmallVectorImpl<MCFixup> &Fixups,
                    raw_ostream &OS) const;
  void EmitFCInstr(const MCInst &MI, raw_ostream &OS) const;

  void EmitNullBytes(unsigned int byteCount, raw_ostream &OS) const;

  void EmitByte(unsigned int byte, raw_ostream &OS) const;

  void EmitTwoBytes(uint32_t bytes, raw_ostream &OS) const;

  void Emit(uint32_t value, raw_ostream &OS) const;
  void Emit(uint64_t value, raw_ostream &OS) const;

  unsigned getHWRegIndex(unsigned reg) const;
  unsigned getHWRegChan(unsigned reg) const;
  unsigned getHWReg(unsigned regNo) const;

  bool isFCOp(unsigned opcode) const;
  bool isTexOp(unsigned opcode) const;
  bool isFlagSet(const MCInst &MI, unsigned Operand, unsigned Flag) const;

  /// getHWRegIndexGen - Get the register's hardware index.  Implemented in
  /// R600HwRegInfo.include.
  unsigned getHWRegIndexGen(unsigned int Reg) const;

  /// getHWRegChanGen - Get the register's channel.  Implemented in
  /// R600HwRegInfo.include.
  unsigned getHWRegChanGen(unsigned int Reg) const;
};

} // End anonymous namespace

enum RegElement {
  ELEMENT_X = 0,
  ELEMENT_Y,
  ELEMENT_Z,
  ELEMENT_W
};

enum InstrTypes {
  INSTR_ALU = 0,
  INSTR_TEX,
  INSTR_FC,
  INSTR_NATIVE,
  INSTR_VTX
};

enum FCInstr {
  FC_IF = 0,
  FC_IF_INT,
  FC_ELSE,
  FC_ENDIF,
  FC_BGNLOOP,
  FC_ENDLOOP,
  FC_BREAK,
  FC_BREAK_NZ_INT,
  FC_CONTINUE,
  FC_BREAK_Z_INT,
  FC_BREAK_NZ
};

enum TextureTypes {
  TEXTURE_1D = 1,
  TEXTURE_2D,
  TEXTURE_3D,
  TEXTURE_CUBE,
  TEXTURE_RECT,
  TEXTURE_SHADOW1D,
  TEXTURE_SHADOW2D,
  TEXTURE_SHADOWRECT,
  TEXTURE_1D_ARRAY,
  TEXTURE_2D_ARRAY,
  TEXTURE_SHADOW1D_ARRAY,
  TEXTURE_SHADOW2D_ARRAY
};

MCCodeEmitter *llvm::createR600MCCodeEmitter(const MCInstrInfo &MCII,
                                           const MCSubtargetInfo &STI,
                                           MCContext &Ctx) {
  return new R600MCCodeEmitter(MCII, STI, Ctx);
}

void R600MCCodeEmitter::EncodeInstruction(const MCInst &MI, raw_ostream &OS,
                                       SmallVectorImpl<MCFixup> &Fixups) const {
  if (isTexOp(MI.getOpcode())) {
    EmitTexInstr(MI, Fixups, OS);
  } else if (isFCOp(MI.getOpcode())){
    EmitFCInstr(MI, OS);
  } else if (MI.getOpcode() == AMDGPU::RETURN ||
    MI.getOpcode() == AMDGPU::BUNDLE ||
    MI.getOpcode() == AMDGPU::KILL) {
    return;
  } else {
    switch(MI.getOpcode()) {
    case AMDGPU::RAT_WRITE_CACHELESS_eg:
      {
        uint64_t inst = getBinaryCodeForInstr(MI, Fixups);
        EmitByte(INSTR_NATIVE, OS);
        Emit(inst, OS);
        break;
      }
    case AMDGPU::VTX_READ_PARAM_i32_eg:
    case AMDGPU::VTX_READ_PARAM_f32_eg:
    case AMDGPU::VTX_READ_GLOBAL_i32_eg:
    case AMDGPU::VTX_READ_GLOBAL_f32_eg:
    case AMDGPU::VTX_READ_GLOBAL_v4i32_eg:
    case AMDGPU::VTX_READ_GLOBAL_v4f32_eg:
      {
        uint64_t InstWord01 = getBinaryCodeForInstr(MI, Fixups);
        uint32_t InstWord2 = MI.getOperand(2).getImm(); // Offset

        EmitByte(INSTR_VTX, OS);
        Emit(InstWord01, OS);
        Emit(InstWord2, OS);
        break;
      }

    default:
      EmitALUInstr(MI, Fixups, OS);
      break;
    }
  }
}

void R600MCCodeEmitter::EmitALUInstr(const MCInst &MI,
                                     SmallVectorImpl<MCFixup> &Fixups,
                                     raw_ostream &OS) const {
  const MCInstrDesc &MCDesc = MCII.get(MI.getOpcode());
  unsigned NumOperands = MI.getNumOperands();

  if(MCDesc.findFirstPredOperandIdx() > -1)
    NumOperands--;

  if (GET_FLAG_OPERAND_IDX(MCDesc.TSFlags) != 0)
    NumOperands--;

  if(MI.getOpcode() == AMDGPU::PRED_X)
    NumOperands = 2;

  // XXX Check if instruction writes a result
  if (NumOperands < 1) {
    return;
  }

  // Emit instruction type
  EmitByte(INSTR_ALU, OS);

  uint64_t InstWord01 = getBinaryCodeForInstr(MI, Fixups);

  //older alu have different encoding for instructions with one or two src
  //parameters.
  if (STI.getFeatureBits() & AMDGPU::FeatureR600ALUInst &&
      MI.getNumOperands() < 4) {
    uint64_t ISAOpCode = InstWord01 & (0x3FFULL << 39);
    InstWord01 &= ~(0x3FFULL << 39);
    InstWord01 |= ISAOpCode << 1;
  }

  unsigned int OpIndex;
  for (OpIndex = 1; OpIndex < NumOperands; OpIndex++) {
    // Literal constants are always stored as the last operand.
    if (MI.getOperand(OpIndex).isImm() || MI.getOperand(OpIndex).isFPImm()) {
      break;
    }
    EmitSrcISA(MI, OpIndex, InstWord01, OS);
  }

  // Emit zeros for unused sources
  for ( ; OpIndex < 4; OpIndex++) {
    EmitNullBytes(SRC_BYTE_COUNT - 6, OS);
  }

  // Emit destination register
  const MCOperand &dstOp = MI.getOperand(0);
  if (dstOp.isReg() && dstOp.getReg() != AMDGPU::PREDICATE_BIT) {
    //element of destination register
    InstWord01 |= uint64_t(getHWRegChan(dstOp.getReg())) << 61;

    // isClamped
    if (isFlagSet(MI, 0, MO_FLAG_CLAMP)) {
      InstWord01 |= 1ULL << 63;
    }

    // write mask
    if (!isFlagSet(MI, 0, MO_FLAG_MASK) && NumOperands < 4) {
      InstWord01 |= 1ULL << 36;
    }

    // XXX: Emit relative addressing mode
  }

  // Emit ALU

  // Emit IsLast (for this instruction group) (1 byte)
  if (!isFlagSet(MI, 0, MO_FLAG_NOT_LAST)) {
    InstWord01 |= 1ULL << 31;
  }

  // XXX: Emit push modifier
  if(isFlagSet(MI, 1,  MO_FLAG_PUSH)) {
    InstWord01 |= 1ULL << 34;
  }

    // XXX: Emit predicate (1 byte)
  int PredIdx = MCDesc.findFirstPredOperandIdx();
  if (PredIdx != -1) {
    switch(MI.getOperand(PredIdx).getReg()) {
    case AMDGPU::PRED_SEL_ZERO:
      InstWord01 |= 2ULL << 29;
      break;
    case AMDGPU::PRED_SEL_ONE:
      InstWord01 |= 3ULL << 29;
      break;
    }
  }

  //XXX: predicate
  //XXX: bank swizzle
  //XXX: OMOD
  //XXX: index mode

  Emit(InstWord01, OS);
}

void R600MCCodeEmitter::EmitSrc(const MCInst &MI, unsigned OpIdx,
                                raw_ostream &OS) const {
  const MCOperand &MO = MI.getOperand(OpIdx);
  union {
    float f;
    uint32_t i;
  } Value;
  Value.i = 0;
  // Emit the source select (2 bytes).  For GPRs, this is the register index.
  // For other potential instruction operands, (e.g. constant registers) the
  // value of the source select is defined in the r600isa docs.
  if (MO.isReg()) {
    unsigned reg = MO.getReg();
    EmitTwoBytes(getHWReg(reg), OS);
    if (reg == AMDGPU::ALU_LITERAL_X) {
      unsigned ImmOpIndex = MI.getNumOperands() - 1;
      MCOperand ImmOp = MI.getOperand(ImmOpIndex);
      if (ImmOp.isFPImm()) {
        Value.f = ImmOp.getFPImm();
      } else {
        assert(ImmOp.isImm());
        Value.i = ImmOp.getImm();
      }
    }
  } else {
    // XXX: Handle other operand types.
    EmitTwoBytes(0, OS);
  }

  // Emit the source channel (1 byte)
  if (MO.isReg()) {
    EmitByte(getHWRegChan(MO.getReg()), OS);
  } else {
    EmitByte(0, OS);
  }

  // XXX: Emit isNegated (1 byte)
  if ((!(isFlagSet(MI, OpIdx, MO_FLAG_ABS)))
      && (isFlagSet(MI, OpIdx, MO_FLAG_NEG) ||
     (MO.isReg() &&
      (MO.getReg() == AMDGPU::NEG_ONE || MO.getReg() == AMDGPU::NEG_HALF)))){
    EmitByte(1, OS);
  } else {
    EmitByte(0, OS);
  }

  // Emit isAbsolute (1 byte)
  if (isFlagSet(MI, OpIdx, MO_FLAG_ABS)) {
    EmitByte(1, OS);
  } else {
    EmitByte(0, OS);
  }

  // XXX: Emit relative addressing mode (1 byte)
  EmitByte(0, OS);

  // Emit kc_bank, This will be adjusted later by r600_asm
  EmitByte(0, OS);

  // Emit the literal value, if applicable (4 bytes).
  Emit(Value.i, OS);

}

void R600MCCodeEmitter::EmitSrcISA(const MCInst &MI, unsigned OpIdx,
                                   uint64_t &Value, raw_ostream &OS) const {
  const MCOperand &MO = MI.getOperand(OpIdx);
  union {
    float f;
    uint32_t i;
  } InlineConstant;
  InlineConstant.i = 0;
  // Emit the source select (2 bytes).  For GPRs, this is the register index.
  // For other potential instruction operands, (e.g. constant registers) the
  // value of the source select is defined in the r600isa docs.
  if (MO.isReg()) {
    unsigned Reg = MO.getReg();
    if (AMDGPUMCRegisterClasses[AMDGPU::R600_CReg32RegClassID].contains(Reg)) {
      EmitByte(1, OS);
    } else {
      EmitByte(0, OS);
    }

    if (Reg == AMDGPU::ALU_LITERAL_X) {
      unsigned ImmOpIndex = MI.getNumOperands() - 1;
      MCOperand ImmOp = MI.getOperand(ImmOpIndex);
      if (ImmOp.isFPImm()) {
        InlineConstant.f = ImmOp.getFPImm();
      } else {
        assert(ImmOp.isImm());
        InlineConstant.i = ImmOp.getImm();
      }
    }
  } else {
    // XXX: Handle other operand types.
    EmitTwoBytes(0, OS);
  }

  // source channel
  uint64_t sourceChannelValue = getHWRegChan(MO.getReg());
  if (OpIdx == 1)
    Value |= sourceChannelValue << 10;
  if (OpIdx == 2)
    Value |= sourceChannelValue << 23;
  if (OpIdx == 3)
    Value |= sourceChannelValue << 42;

  // isNegated
  if ((!(isFlagSet(MI, OpIdx, MO_FLAG_ABS)))
      && (isFlagSet(MI, OpIdx, MO_FLAG_NEG) ||
     (MO.isReg() &&
      (MO.getReg() == AMDGPU::NEG_ONE || MO.getReg() == AMDGPU::NEG_HALF)))){
    if (OpIdx == 1)
      Value |= 1ULL << 12;
    else if (OpIdx == 2)
      Value |= 1ULL << 25;
    else if (OpIdx == 3)
      Value |= 1ULL << 44;
  }

  // isAbsolute
  if (isFlagSet(MI, OpIdx, MO_FLAG_ABS)) {
    assert(OpIdx < 3);
    Value |= 1ULL << (32+OpIdx-1);
  }

  // XXX: relative addressing mode
  // XXX: kc_bank

  // Emit the literal value, if applicable (4 bytes).
  Emit(InlineConstant.i, OS);

}

void R600MCCodeEmitter::EmitTexInstr(const MCInst &MI,
                                     SmallVectorImpl<MCFixup> &Fixups,
                                     raw_ostream &OS) const {

  unsigned opcode = MI.getOpcode();
  bool hasOffsets = (opcode == AMDGPU::TEX_LD);
  unsigned op_offset = hasOffsets ? 3 : 0;
  int64_t sampler = MI.getOperand(op_offset+2).getImm();
  int64_t textureType = MI.getOperand(op_offset+3).getImm();
  unsigned srcSelect[4] = {0, 1, 2, 3};

  // Emit instruction type
  EmitByte(1, OS);

  // Emit instruction
  EmitByte(getBinaryCodeForInstr(MI, Fixups), OS);

  // XXX: Emit resource id r600_shader.c uses sampler + 1.  Why?
  EmitByte(sampler + 1 + 1, OS);

  // Emit source register
  EmitByte(getHWReg(MI.getOperand(1).getReg()), OS);

  // XXX: Emit src isRelativeAddress
  EmitByte(0, OS);

  // Emit destination register
  EmitByte(getHWReg(MI.getOperand(0).getReg()), OS);

  // XXX: Emit dst isRealtiveAddress
  EmitByte(0, OS);

  // XXX: Emit dst select
  EmitByte(0, OS); // X
  EmitByte(1, OS); // Y
  EmitByte(2, OS); // Z
  EmitByte(3, OS); // W

  // XXX: Emit lod bias
  EmitByte(0, OS);

  // XXX: Emit coord types
  unsigned coordType[4] = {1, 1, 1, 1};

  if (textureType == TEXTURE_RECT
      || textureType == TEXTURE_SHADOWRECT) {
    coordType[ELEMENT_X] = 0;
    coordType[ELEMENT_Y] = 0;
  }

  if (textureType == TEXTURE_1D_ARRAY
      || textureType == TEXTURE_SHADOW1D_ARRAY) {
    if (opcode == AMDGPU::TEX_SAMPLE_C_L || opcode == AMDGPU::TEX_SAMPLE_C_LB) {
      coordType[ELEMENT_Y] = 0;
    } else {
      coordType[ELEMENT_Z] = 0;
      srcSelect[ELEMENT_Z] = ELEMENT_Y;
    }
  } else if (textureType == TEXTURE_2D_ARRAY
             || textureType == TEXTURE_SHADOW2D_ARRAY) {
    coordType[ELEMENT_Z] = 0;
  }

  for (unsigned i = 0; i < 4; i++) {
    EmitByte(coordType[i], OS);
  }

  // XXX: Emit offsets
  if (hasOffsets)
	  for (unsigned i = 2; i < 5; i++)
		  EmitByte(MI.getOperand(i).getImm()<<1, OS);
  else
	  EmitNullBytes(3, OS);

  // Emit sampler id
  EmitByte(sampler, OS);

  // XXX:Emit source select
  if ((textureType == TEXTURE_SHADOW1D
      || textureType == TEXTURE_SHADOW2D
      || textureType == TEXTURE_SHADOWRECT
      || textureType == TEXTURE_SHADOW1D_ARRAY)
      && opcode != AMDGPU::TEX_SAMPLE_C_L
      && opcode != AMDGPU::TEX_SAMPLE_C_LB) {
    srcSelect[ELEMENT_W] = ELEMENT_Z;
  }

  for (unsigned i = 0; i < 4; i++) {
    EmitByte(srcSelect[i], OS);
  }
}

void R600MCCodeEmitter::EmitFCInstr(const MCInst &MI, raw_ostream &OS) const {

  // Emit instruction type
  EmitByte(INSTR_FC, OS);

  // Emit SRC
  unsigned NumOperands = MI.getNumOperands();
  if (NumOperands > 0) {
    assert(NumOperands == 1);
    EmitSrc(MI, 0, OS);
  } else {
    EmitNullBytes(SRC_BYTE_COUNT, OS);
  }

  // Emit FC Instruction
  enum FCInstr instr;
  switch (MI.getOpcode()) {
  case AMDGPU::BREAK_LOGICALZ_f32:
    instr = FC_BREAK;
    break;
  case AMDGPU::BREAK_LOGICALNZ_f32:
    instr = FC_BREAK_NZ;
    break;
  case AMDGPU::BREAK_LOGICALNZ_i32:
    instr = FC_BREAK_NZ_INT;
    break;
  case AMDGPU::BREAK_LOGICALZ_i32:
    instr = FC_BREAK_Z_INT;
    break;
  case AMDGPU::CONTINUE_LOGICALNZ_f32:
  case AMDGPU::CONTINUE_LOGICALNZ_i32:
    instr = FC_CONTINUE;
    break;
  case AMDGPU::IF_LOGICALNZ_f32:
    instr = FC_IF;
  case AMDGPU::IF_LOGICALNZ_i32:
    instr = FC_IF_INT;
    break;
  case AMDGPU::IF_LOGICALZ_f32:
    abort();
    break;
  case AMDGPU::ELSE:
    instr = FC_ELSE;
    break;
  case AMDGPU::ENDIF:
    instr = FC_ENDIF;
    break;
  case AMDGPU::ENDLOOP:
    instr = FC_ENDLOOP;
    break;
  case AMDGPU::WHILELOOP:
    instr = FC_BGNLOOP;
    break;
  default:
    abort();
    break;
  }
  EmitByte(instr, OS);
}

void R600MCCodeEmitter::EmitNullBytes(unsigned int ByteCount,
                                      raw_ostream &OS) const {

  for (unsigned int i = 0; i < ByteCount; i++) {
    EmitByte(0, OS);
  }
}

void R600MCCodeEmitter::EmitByte(unsigned int Byte, raw_ostream &OS) const {
  OS.write((uint8_t) Byte & 0xff);
}

void R600MCCodeEmitter::EmitTwoBytes(unsigned int Bytes,
                                     raw_ostream &OS) const {
  OS.write((uint8_t) (Bytes & 0xff));
  OS.write((uint8_t) ((Bytes >> 8) & 0xff));
}

void R600MCCodeEmitter::Emit(uint32_t Value, raw_ostream &OS) const {
  for (unsigned i = 0; i < 4; i++) {
    OS.write((uint8_t) ((Value >> (8 * i)) & 0xff));
  }
}

void R600MCCodeEmitter::Emit(uint64_t Value, raw_ostream &OS) const {
  for (unsigned i = 0; i < 8; i++) {
    EmitByte((Value >> (8 * i)) & 0xff, OS);
  }
}

unsigned R600MCCodeEmitter::getHWRegIndex(unsigned reg) const {
  switch(reg) {
  case AMDGPU::ZERO: return 248;
  case AMDGPU::ONE:
  case AMDGPU::NEG_ONE: return 249;
  case AMDGPU::ONE_INT: return 250;
  case AMDGPU::HALF:
  case AMDGPU::NEG_HALF: return 252;
  case AMDGPU::ALU_LITERAL_X: return 253;
  case AMDGPU::PREDICATE_BIT:
  case AMDGPU::PRED_SEL_OFF:
  case AMDGPU::PRED_SEL_ZERO:
  case AMDGPU::PRED_SEL_ONE:
    return 0;
  default: return getHWRegIndexGen(reg);
  }
}

unsigned R600MCCodeEmitter::getHWRegChan(unsigned reg) const {
  switch(reg) {
  case AMDGPU::ZERO:
  case AMDGPU::ONE:
  case AMDGPU::ONE_INT:
  case AMDGPU::NEG_ONE:
  case AMDGPU::HALF:
  case AMDGPU::NEG_HALF:
  case AMDGPU::ALU_LITERAL_X:
  case AMDGPU::PREDICATE_BIT:
  case AMDGPU::PRED_SEL_OFF:
  case AMDGPU::PRED_SEL_ZERO:
  case AMDGPU::PRED_SEL_ONE:
    return 0;
  default: return getHWRegChanGen(reg);
  }
}
unsigned R600MCCodeEmitter::getHWReg(unsigned RegNo) const {
  unsigned HWReg;

  HWReg = getHWRegIndex(RegNo);
  if (AMDGPUMCRegisterClasses[AMDGPU::R600_CReg32RegClassID].contains(RegNo)) {
    HWReg += 512;
  }
  return HWReg;
}

uint64_t R600MCCodeEmitter::getMachineOpValue(const MCInst &MI,
                                              const MCOperand &MO,
                                        SmallVectorImpl<MCFixup> &Fixup) const {
  if (MO.isReg()) {
    return getHWRegIndex(MO.getReg());
  } else if (MO.isImm()) {
    return MO.getImm();
  } else {
    assert(0);
    return 0;
  }
}

//===----------------------------------------------------------------------===//
// Encoding helper functions
//===----------------------------------------------------------------------===//

bool R600MCCodeEmitter::isFCOp(unsigned opcode) const {
  switch(opcode) {
  default: return false;
  case AMDGPU::BREAK_LOGICALZ_f32:
  case AMDGPU::BREAK_LOGICALNZ_i32:
  case AMDGPU::BREAK_LOGICALZ_i32:
  case AMDGPU::BREAK_LOGICALNZ_f32:
  case AMDGPU::CONTINUE_LOGICALNZ_f32:
  case AMDGPU::IF_LOGICALNZ_i32:
  case AMDGPU::IF_LOGICALZ_f32:
  case AMDGPU::ELSE:
  case AMDGPU::ENDIF:
  case AMDGPU::ENDLOOP:
  case AMDGPU::IF_LOGICALNZ_f32:
  case AMDGPU::WHILELOOP:
    return true;
  }
}

bool R600MCCodeEmitter::isTexOp(unsigned opcode) const {
  switch(opcode) {
  default: return false;
  case AMDGPU::TEX_LD:
  case AMDGPU::TEX_GET_TEXTURE_RESINFO:
  case AMDGPU::TEX_SAMPLE:
  case AMDGPU::TEX_SAMPLE_C:
  case AMDGPU::TEX_SAMPLE_L:
  case AMDGPU::TEX_SAMPLE_C_L:
  case AMDGPU::TEX_SAMPLE_LB:
  case AMDGPU::TEX_SAMPLE_C_LB:
  case AMDGPU::TEX_SAMPLE_G:
  case AMDGPU::TEX_SAMPLE_C_G:
  case AMDGPU::TEX_GET_GRADIENTS_H:
  case AMDGPU::TEX_GET_GRADIENTS_V:
  case AMDGPU::TEX_SET_GRADIENTS_H:
  case AMDGPU::TEX_SET_GRADIENTS_V:
    return true;
  }
}

bool R600MCCodeEmitter::isFlagSet(const MCInst &MI, unsigned Operand,
                                  unsigned Flag) const {
  const MCInstrDesc &MCDesc = MCII.get(MI.getOpcode());
  unsigned FlagIndex = GET_FLAG_OPERAND_IDX(MCDesc.TSFlags);
  if (FlagIndex == 0) {
    return false;
  }
  assert(MI.getOperand(FlagIndex).isImm());
  return !!((MI.getOperand(FlagIndex).getImm() >>
            (NUM_MO_FLAGS * Operand)) & Flag);
}
#define R600RegisterInfo R600MCCodeEmitter
#include "R600HwRegInfo.include"
#undef R600RegisterInfo

#include "AMDGPUGenMCCodeEmitter.inc"